Abstract
A high-resolution multi-proxy record from sediments of a small-sized lake situated in Telemark, southeastern Norway, was used to reconstruct the local landscape development of the past c. 10,500 years. Our data demonstrate that changes in vegetation composition and structure in the first two-thirds of the Holocene are principally attributable to climatic changes and high erosion rates, as deduced from geochemical and physical (loss-on-ignition) proxy analyses. The highest signals of erosional inputs to the lake (c. 8030–5760 cal. BP) can be correlated with the first part of the Holocene Thermal Maximum. Nevertheless, evidence from pollen, non-pollen palynomorphs and microscopic charcoal analyses indicates the presence of nutrient-rich and disturbed environments already during the middle Mesolithic (c. 10,050–9400 cal. BP). It also shows traces of animal husbandry (c. 5580 cal. BP) and small-scale cereal cultivation (c. 5520 cal. BP) in the early Neolithic. In subsequent periods, human impact remains at a relatively low level and does not generate significant palaeo-environmental changes. Not until the second half of the Bronze Age (c. 2840 cal. BP) is some intensification in animal husbandry recorded, whereas crop cultivation continues to play a minor role in the second millennium BP. The establishment of a full farming economy took place during the Roman Iron Age (c. 1790 cal. BP), characterised by extensive forest clearance and local fires, crop cultivation in permanent fields and the presence of open pastures. This establishment is associated with advanced soil degeneration and increased erosion rates.
Introduction
Our knowledge about Holocene vegetational development and human impact in southeastern Norway is undoubtedly considerable because the palynological research has a long tradition in Scandinavia (e.g. Bjune et al., 2013, 2009a; Danielsen, 1970; Eide et al., 2006; Hafsten, 1956, 1992; Hafsten et al., 1979; Henningsmoen, 1979, 1988; Høeg, 1978, 1980, 1982, 1989; Høeg and Stabell, 1978; Molinari et al., 2005; Prøsch-Danielsen, 1996; Soltvedt and Henningsmoen, 2016). However, despite the existence of several pollen records from lake or bog deposits, some specific questions have not yet been definitively clarified for some regions of southeastern Norway, including the character of local vegetation cover, the replacement processes within the forest, the onset of anthropogenic impact on the environment, the degree of human activity throughout time and the duration of settlement phases. Furthermore, studies of landscape development have also illustrated that the degree of impact, in particular human agency, has varied considerably in space and time (e.g. Henningsmoen, 1980, 1988; Høeg, 1989; Soltvedt and Henningsmoen, 2016; Sørensen et al., 2007). This demonstrates that there is still a great need to improve our coverage of pollen records with high temporal resolution in this part of Scandinavia.
The response of vegetation to climate change, soil erosion and human impact can be smooth, stepwise or abrupt. The low temporal resolution of pollen records and the application of a single method, however, entail a risk of missing in short-lived palaeo-environmental events and processes in particular. Well-dated sediment sequences with high temporal resolution complemented by multi-proxy records are therefore essential to identify as exactly as possible the timing and extent of vegetational signals in relation to external effects. The potential of multi-proxy records from lake sediments to improve our understanding of the complex interaction between climate, vegetation, erosion and humans over the millennia has been highlighted by several recent studies (e.g. Bajard et al., 2015; Enters et al., 2010; Koinig et al., 2003; Rasmussen and Bradshaw, 2005; Szal et al., 2014; Wacnik et al., 2016; Wick et al., 2003). In this context, small lakes provide an important source of information for a precise description of local human–environment–climate interactions (Gaillard et al., 1991) within catchment areas of only a few metres in diameter around the water basin (Sugita, 1994, 1998).
This study was performed in connection with two large-scale rescue excavation archaeological projects in southeastern Norway, both conducted by the Museum of Cultural History in Oslo, namely, the Vestfoldbaneprosjektet (Melvold and Persson, 2014; Reitan and Persson, 2014), a railway construction project in the southern Norwegian Vestfold and Telemark counties, and the E18-Rugtvedt-Dørdal project (Solheim, 2014, in press), a highway construction measure through the county of Telemark. Lake Skogstjern was chosen for the palynological investigations because of its location close to the excavated sites (the nearest at a distance of c. 200 m), which consist mainly of traces of coastal settlements dating from the early Mesolithic to the early Neolithic (c. 10,950–5650 cal. BP) (Reitan, 2014; Solheim, 2014, in press). The aim of the study was to supplement the results from the archaeological excavations with knowledge of the vegetation and erosion history. Our study focused on three topics in particular: the relationship of the development of Mesolithic woodlands to climatic developments and available plant resources, the introduction of agriculture and the establishment of a fully agrarian economy.
In this paper, we present a detailed reconstruction of landscape development in the catchment of a small-sized southeastern Norwegian lake based on a multi-proxy approach, consisting of the analyses of pollen, spores, non-pollen palynomorphs (NPPs) and microscopic charcoal, data from loss-on-ignition (LOI) and geochemistry. We compare environmental changes recorded during the past c. 10,500 years with climate history and archaeological records of anthropogenic activities in order to obtain a better understanding of the factors that might have controlled the Holocene landscape transformation.
Site description
Lake Skogstjern (also locally known as Skaugtjenna; 57.2 m a.s.l.; 59°0′25.38″N, 9°38′18.77″E) is situated in the municipality of Bamble in the county of Telemark, southeastern Norway (Figure 1). It is part of a fertile region that is characterised by alkaline-rich Cambro-Silurian bedrock. The typical soil type is podzol (Låg and Norges Landbrukshøgskole, 1983). The lake is located on the proximal side of the Ra-moraine (Early Younger Dryas: at about 11,000–10,600 yr BP; Bergstrøm, 1999; Høeg, 1982) and was connected to the sea in the Late Glacial and the early Holocene. It formerly constituted the inner part of a fjord reaching the Skagerrak (Høeg, 1982; Stabell, 1980). Today, its outflow, to the southeast, reaches the Åbyfjorden after 2.4 km. The surface area of the lake is about 2.6 ha. The hilly surroundings are dominated by forest that consists of Picea, Pinus, Betula and Populus. Nearly one-third of the shoreline is formed by farmland on the more gentle slopes. In the swampy areas, peat accumulated from a bog vegetation of Sphagnum, Myrica gale, Cyperaceae, Drosera, Calluna vulgaris, Vaccinium oxycoccus and Menyanthes (Høeg, 1989).
Location of the coring site Lake Skogstjern in southeastern Norway. The dashed line displays the catchment area of the lake (map according to Philip Lüth (www.lueth-archaeologie.de) after the metadata of Kartverket (www.data.kartverket.no)).
The Telemark region is characterised by a humid continental climate with large temperature variations between seasons. During the past 50 years, the mean July and January temperatures reached highs of 11.7°C and −7.2°C, respectively. The annual precipitation averaged 1263.7 mm (www.met.no).
Material and methods
Coring, dating and sediment analysis
An overlapping sediment core of 7 m was obtained from the deepest part (4.1 m) of Lake Skogstjern (Figure 1) in January 2013, using an 8-cm-diameter Usinger-type piston corer (Mingram et al., 2007).
AMS radiocarbon dating of three samples was carried out at the Leibniz Laboratory for Radiometric Dating and Isotope Research, Kiel University. Another six samples were sent for AMS 14C measurements to the Poznań Radiocarbon Laboratory (Table 1). Calibration to the calendar timescale was carried out using the OxCal v.4.2.3 calibration software (Bronk Ramsey, 2013) and the IntCal13 calibration dataset (Reimer et al., 2013). The calibration results used for the age–depth model are given as mean values of the calibrated ages (cal. BP) within the 2σ range and linear interpolation and extrapolation (Figure 2). One of the dates was rejected for the age model; the 50-cm date (Poz-63990) being situated below the establishment of Picea around 1000 cal. BP (Bjune et al., 2009a) but dated to 151–55 cal. BP provided an age that was approximately 900 years too young.
Radiocarbon dates from the core Lake Skogstjern (Calibration: OxCal, Bronk Ramsey, 2013; IntCal13, Reimer et al., 2013).
Lithology as well as age–depth model, accumulation rates (mm/yr) and time resolution (yr/cm) for the sediment record of Lake Skogstjern (OxCal, Bronk Ramsey, 2013; IntCal13, Reimer et al., 2013). Archaeological periods according to Gjerpe (2013).
The geochemical distribution of major elements (Al, Rb, Zr, Ti, K, Si, S and Br) in the sediment core was determined with an Avaatech x-ray fluorescence (XRF) core scanner at the Institute for Geosciences, Kiel University. The scanning resolution was 1 cm, with varying tube voltage (20 s, 30 kV, 1500 mA and 10 s, 10 kV, 750 mA) to detect different element spectra. As this technique is not capable of differentiating between the different chemical forms of the elements, the element data produced by the scanner are expressed as element intensities in counts per second (cps). For a better correlation of the different datasets, the sample intervals presented here are the same as those used for the pollen samples.
LOI was applied to estimate the organic/mineral matter content of the sediment (Bengtsson and Enell, 1986; Dean, 1974). The implementation of LOI followed the procedure described by Heiri et al. (2001).
Pollen analysis
To gain a high temporal resolution for certain archaeological periods, pollen samples were taken at 2 cm intervals throughout the sections of 18–238 and 324–418 cm, respectively. The section between 240 and 323 cm was sampled at 1 cm intervals. Thus, a record with a temporal resolution of c. 12–38 yr/cm (Figure 2) is available for Lake Skogstjern.
The samples were prepared for pollen analysis according to the standard techniques outlined by Moore et al. (1991). Pollen counts were carried out under 400-fold or, in ambiguous cases, 1000-fold magnification. For the identification of Cerealia-type pollen, phase-contrast microscopy was applied. On average, 500 arboreal pollen (AP) grains per sample were counted, excluding Corylus. In addition, the analysis included scanning for rare pollen types. The calculation of pollen percentages is based on the sum of total terrestrial pollen (trees, shrubs and dwarf-shrubs + pollen of herbaceous terrestrial plants (or non-arboreal pollen (NAP)), excluding wetland and aquatic pollen types. Tablets containing a known number of Lycopodium spores were added to enable calculation of pollen concentrations (Figure 4). The data from the microscopic charcoal analysis are expressed as concentration per cubic centimetre of sediment. In addition to pollen and spores, NPPs were studied for a more profound reconstruction of vegetation and environmental conditions. The diagrams were produced with the help of the program CountPol (Feeser, unpublished). In order to make the plots easier to read, quantities of AP, NAP and NPP are plotted at 10 times their actual values. The nomenclature of pollen types follows Beug (2004) and Faegri and Iversen (1989); that of spores, Moore et al. (1991); and that of NPP, Van Geel (2001), Van Geel and Aptroot (2006) and Van Geel et al. (1980).
Results
Sediment analysis
The sediment was composed of brownish fine detritus gyttja in the upper part of the core and light-grey silty deposits below c. 403.4 cm. This lithostratigraphical transition, representing the isolation of the lake basin from the sea, was dated to c. 9950 cal. BP. Furthermore, the gyttja deposits included several lithological units and consisted of fine and coarse laminations and non-laminated sections in the uppermost part of the sediment (Figure 2).
Based on the major changes in the ratio of both mineral matter content and the elemental intensities, 10 lake sediment zones (LSZ I–X) were distinguished in the LOI/geochemical diagram (Figure 3). The main features of the mineralogical and geochemical composition of the sediment are summarised in Table 2. Among the geochemical record, the elemental concentrations of Al, Rb, Zr, Ti, K and Si show the highest values in LSZ I (⩽234 × 103 cps) and LSZ IV (⩽210 × 103 cps). The highest concentrations of S are in LSZ I (⩽7.2 × 103 cps), followed by LSZ II (⩽4.4 × 103 cps), whereas the values of Br reach their highest levels in LSZ VII (⩽21.5 × 103 cps). Furthermore, the elements Al, Rb, Zr, Ti, K and Si, along with the profiles of Ti/Br, Ti/S, Si/Br and Si/S, show similar patterns throughout the core and correlate positively with high mineral matter content in the sediment (LSZ 1: ⩽99%; LSZ IV: ⩽ 4.5%).
Loss-on-ignition (LOI) and geochemical (XRF scan) diagram for Lake Skogstjern, showing selected elemental ratios and the intensities of selected elements. The ratios are expressed in percentage values and the geochemical measurements in counts per second (cps).
Main characteristics of the lake sediment zones I-X of Lake Skogstjern.
Pollen analysis
Because of low pollen concentrations within the marine section of the sediment core (Figure 4), the base of the pollen diagram was defined to occur at a depth of 418 cm. The lowermost part of the pollen record has been dated to 10,510 cal. BP based on extrapolation from the oldest AMS date (Figure 2). The pollen record consists of 245 pollen spectra. Based on major changes in AP proportions, 10 local pollen assemblage zones (PAZs) were distinguished in the pollen diagram (Figure 4). These zones, A–J, were further subdivided into subzones based on minor changes. Their main characteristics are summarised in Table 3.
Pollen record for Lake Skogstjern. The diagram shows pollen concentrations (grains/cm3 × 103) of selected sum curves and the simple moving average of the total terrestrial pollen concentration (in red) in the left part. The adjacent pollen percentage diagram to the right shows selected taxa. Percentages are based on the sum of terrestrial pollen (AP + NAP) excluding wetland and aquatic pollen types. Coloured curves are exaggerated by factor 10. Microscopic charcoal particles are presented as bars showing concentration in cubic centimetre of sediment.
Main characteristics of the local pollen assemblage zones A-J of Lake Skogstjern.
Discussion
Pollen source area of Lake Skogstjern
Pollen percentages do not correspond to the vegetation abundance directly because they are affected by pollen dispersal characteristics and pollen productivity (Prentice, 1985). Research has shown that indicators for agriculture or settlement activity are not reflected in pollen diagrams if these activities happened only a few kilometres away from the sampled location (Behre and Kučan, 1986). Prentice (1985) estimated the pollen source area for a point at the centre of a sedimentary basin and referred to local pollen as pollen input from within 20 m of the edge of the basin; extra-local, from 20 m to 2 km; and regional, from 2 to 200 km. Other empirical and modelling studies have demonstrated a correlation between the size of the lake and the relevant source area (Jackson, 1990; Jacobson and Bradshaw, 1981; Parsons and Prentice, 1981; Tauber, 1965). Thus, the palynological data can only be interpreted usefully if the size of the area around the sampled point for which the pollen assemblages are relevant is taken into account. Sugita (1994, 1998) has defined the relevant source area of pollen (RSAP) as the area (or distance) beyond which correlations between pollen loading and vegetation abundance for all taxa do not continue to improve. Sugita (1994) has shown that small-sized lakes (50 m in radius) reflect the vegetation within 300–400 m from the lake edge and medium-sized lakes (250 m in radius) within 600–800 m. Thus, the pollen spectra from Lake Skogstjern, which measures c. 130 m × 270 m, represent mainly the vegetation cover of an area of approximately 800 m around the lakeshore (Figure 1).
Landscape development in the catchment of Lake Skogstjern 10,510–9930 cal. BP (PAZ A): Coastal environment
The silty deposits (⩾9950 cal. BP) from this period present the maximum concentrations of Al, Rb, Zr, Ti, K and Si as well as the highest amounts of mineral matter content, representing the sedimentation of eroded terrigenous particles during the marine stage of the basin (LSZ I). At the same time, the elevated concentrations of S and Br can be explained by the fact that both elements are components of seawater (Feil et al., 2017; Kaim and Schwederski, 1991). Among the NPP, HdV-704A and HdV-704C (both representing cysts of dinoflagellates) and HdV-116 occur, in part, in extremely high amounts (Figure 4), also demonstrating the marine character of the sediments (Bakker and van Smeerdijk, 1982).
The earliest stage of the pollen record is dominated by Pinus (up to 75.3%). Nevertheless, the strong coniferous signal is admixed with pollen of deciduous taxa, especially those of Betula and Corylus (Figure 4). Because Corylus starts to grow in early spring and requires a long growing season (Dahl, 1998), its presence already indicates increasingly oceanic conditions (Antonsson and Seppä, 2007). At the same time, the elevated values of the extreme heliophilous and salt-tolerant shrub Hippophaë (up to 1.8%), which is native to sand dunes or sea cliffs and other locations where landslides are of frequent occurrence (e.g. Hafsten, 1956), reflect open vegetation of seashore habitats. In addition, the pollen spectra feature a relatively high content of dwarf-shrub pollen and NAP from Vaccinium-type, Poaceae and Chenopodiaceae (up to 8.2%) – the latter probably mainly reflecting halophilous vegetation, because many representatives of this family grow as pioneers on the seashore (Danielsen, 1970). Consequently, such more or less open vegetation points to a possible marine over-representation of Pinus pollen, which may be due, to some extent, to a higher degree of long-distance transport. In addition, the relatively high frequencies of Polypodiaceae indet., together with Cyperaceae, reflect the presence of damp areas.
Between c. 10,050 and 9970 cal. BP, visibly elevated amounts of microscopic charcoal particles (9.2 × 103 particles/cm3) appear in the sediment, which may be indicative of local burning. This is accompanied by distinct rises in Poaceae (6.7%) and in herbs (1%) such as Artemisia, Rumex acetosa-type and Liguliflorae (Subzone A2), pointing to the presence of disturbed ground. At the same time, there is a decrease in pollen from deciduous trees, in particular Betula and Corylus, whereas light-demanding shrubs (up to 2.1%), such as Juniperus, Sorbus/Rubus-type and Salix, increase. These changes in pollen taxa composition may have been induced by burning, as evidenced by the presence of micro-charcoal. However, it is virtually impossible to assess unambiguously whether the charcoal dust represents natural or anthropogenic fire events. The increased microscopic charcoal record could also be associated with domestic fires from hearths (Bennett et al., 1990; Edwards, 1990) within camps located at the shoreline.
9930–8790 cal. BP (PAZ B): Light woodland
The decreasing marine influence during the following stage is shown, on one hand, by the change from minerogenic sediments to gyttja deposits (Figure 2) and, on the other hand, by the decrease in Hippophaë, Chenopodiaceae and NPP indicative of marine conditions. At the same time, the values of fern spores of Polypodiaceae indet. drop markedly. Their decline could be because of a general reduction in coastal wetlands triggered by the withdrawal of the sea during this period. The first records of pollen from aquatic plants, such as Nymphaea and Potamogeton, accompany these changes, demonstrating the colonisation of the lake basin by freshwater plant communities. Among the geochemical record, the profiles of Br and S in the elemental ratios increase (LSZ II–III), also demonstrating changes in the depositional processes. Elemental bromine is typically enriched in organic-rich sediments (Edmunds, 1996; Rittenhouse, 1967), whereas sulphur persists in freshwater aquatic environments in a variety of forms (Bura-Nakić et al., 2015). In accordance with this, the organic component in the sediment increases visibly, whereas the allochthonous input diminishes as shown by the decreasing ratios of Ti/Br, Ti/S, Si/Br and Si/S (Figure 3).
The time span reflects increasingly warmer conditions, with the abrupt rise in the warmth-demanding Corylus (up to 48.1%). It also shows a distinct increase in the light-dependent Betula (up to 38%), reflecting the expansion of open hazel-rich birch forests. Single finds of Hedera helix pollen indicate the oceanic character of the climate and the increase in temperature (Iversen, 1944; Zagwijn, 1994), both of which are also attested to by the slightly elevated values of the thermophilous forest elements Ulmus and Quercus, showing that while these taxa were also present they played only a minor role in the forest inventory. The climatic amelioration is also reflected in a significant rise in the values of pollen concentrations (up to 378 × 103 grains/cm3), probably resulting from increased pollen production and its dispersal to the lake.
The period between c. 9930 and 9400 cal. BP (Subzone B1) is characterised by the occurrence of some heliophilous shrubs and dwarf-shrubs, such as Salix, Virbunum opulus-type, Juniperus, Calluna and Vaccinium-type (up to 3.1%), which, along with relatively high values of Poaceae (up to 4.8%), reflect gaps in the forest canopy. This is coupled with the presence of pollen of herbaceous taxa, particularly Rumex acetosa-type and Artemisia, and also other ruderals, such as Urtica, Caryophyllaceae, Chenopodiaceae, Senecio-type and Geum-type, all of which appear occasionally. This palynological record may indicate natural disturbances, such as open areas along the lakeshore, precipices or rock outcrops with thin soil cover, talus formations and landslides where sufficient light is present, enabling these plants to grow. However, the abundance of pollen of those representatives of disturbed vegetation appear together with elevated levels of microscopic charcoal particles (up to 2.6 × 103 particles/cm3) and Pteridium, a common fire-adapted fern (Bińka and Nitychoruk, 2013; Tinner et al., 2000). The increase in charcoal dust could be a result of natural burning, but the pollen stratigraphical evidence shows a deciduous tree cover with birch and hazel stands dominating. Betula and Corylus do not burn as readily as do conifers and their litter (Ryan and Blackford, 2010); therefore, natural fires in such environments appear unlikely (Edlin, 1970; Rackham, 1986) and the possibility of anthropogenically induced small-scale openings in the woodland created by the use of fire also has to be taken into consideration. Furthermore, because Artemisia, Chenopodiaceae and Urtica are primarily nitrophilous species, their presence in the pollen record can be associated with nitrogen-rich areas around dwellings (Behre, 1981), pointing to the presence of Mesolithic inhabitants. This evidence of potential human agency is in a good agreement with the archaeological record of the E18-Rugtvedt-Dørdal project, which indicates a shift from highly mobile marine hunter-gatherer groups in the early Mesolithic to a more semi-sedentary settlement in the middle Mesolithic, as evidenced by more permanent constructions being built at sites and settlement becoming more stable within the landscape (Solheim, 2013).
Another characteristic element occurring at this stage is the immigration of Alnus (Subzone B2). Because of the specific wet requirements of alder, its steep rise at c. 9400 cal. BP (up to 36.2%) demonstrates the rapid colonisation of marshy sites along the lakeshore – sites that were previously occupied by Betula and Salix (Antonsson and Seppä, 2007; Digerfeldt, 1982), whose values are now visibly reduced.
8790–7510 cal. BP (PAZ C): Increasingly dense woodland
This stage is characterised by a distinct rise in Ulmus (up to 8.9%) and, nearly simultaneously, a less pronounced increase in Fraxinus (up to 0.5%), demonstrating a gradual transition from open Corylus–Betula forest to a more dense mixed woodland, which is also reflected by a further increase in the values of AP concentrations (up to 1085 × 103 grains/cm3). Denser forest canopy is beyond that demonstrated by the decrease in the frequency of Juniperus, along with, in general, slightly decreased amounts of ruderal herbs. Both Ulmus and Fraxinus were probably restricted to damp niches, such as shaded hillside locations or the moist surroundings of the lake, because of their edaphic requirements for soil moisture (e.g. Düll and Kutzelnigg, 2005). An indication of habitat competition between these two taxa and Alnus is indicated by the coincident slight decrease in the amount of alder pollen. During this time, Tilia also arrived in the area. During the time span between 8270 and 8110 cal. BP, two sudden and distinct reductions in the frost-sensitive Corylus are recorded (Subzone C2). The Corylus declines are accompanied by a decrease in other warmth-demanding woody taxa, such as Ulmus and Fraxinus. Simultaneously, the low temperature–adapted Pinus, Betula and Juniperus increase. This transitional phase may be related to the 8.2-ka cooling event that is widely recorded in the North Atlantic and northern Europe (e.g. Alley et al., 1997; Antonsson and Seppä, 2007; Seppä et al., 2005), known as the Finse event in Norway (e.g. Dahl and Nesje, 1994, 1996; Matthews et al., 2005; Nesje and Dahl, 2001). According to Seppä et al. (2005), there was a rapid decrease in temperature during this time, and the reconstructed mean annual temperature was c. 1.5°C lower than today. Accordingly, the vegetation responded with a sudden decline in pollen values of the thermophilous tree taxa.
At the same time, the amounts of microscopic charcoal particles remain at least in part relatively high, perhaps reflecting the input of micro-charcoal fragments from domestic fires into the lake. Synchronously, slightly increased frequencies of Rumex acetosa-type, Artemisia and Humulus/Cannabis-type, together with the conspicuously high assemblages of Poaceae (up to 5.9%), may indicate the existence of permanent campsites at the lakeshore.
7510–5940 cal. BP (PAZ D): Dense and diverse forest landscape
A new type of forest characterises the period between c. 7510 and 5940 cal. BP. At the start of this stage, there is a strong decrease in pollen values of Corylus (up to 8%), along with parallel rapid increases in Tilia (up to 12.2%) and Quercus (up to 11.4%). Both these latter taxa naturally occupy a more dominant position on the base-rich soils of the Cambro-Silurian bedrock (Hafsten, 1956). At the same time, there is a considerable drop in the amount of total terrestrial pollen concentration, probably resulting to a certain extent from the decrease in the highly prolific pollen producer Corylus. The replacement of Corylus was possibly driven by the spread of Tilia. Because of their great height and dense canopy, lime trees easily shade out hazel copses (e.g. Hafsten, 1992).
The insect-pollinated Tilia does not disperse its pollen as effectively as do other forest trees, and it is therefore often underestimated as a component of the forest (Prøsch-Danielsen, 1996). Hence, from this point in time onwards, lime was probably one of the major tree taxa in the catchment of the lake, although after the reduction in Corylus, it is Betula that dominates by far in the pollen record. The Betula percentage curve even increases, presenting birch as the main constituent of the woodland in terms of pollen percentages. During this time, scattered pollen grains of Carpinus occur, and records of Picea are more frequent than before. Pollen of the insect-pollinated Acer also appears, showing that maple stands must have grown near the lake. At the end of this stage, at c. 6090 cal. BP, Fagus pollen occurs for the first time. The presence of Fagus, along with Carpinus and Picea, however, probably represents long-distance transport from the areas where these taxa have already been established.
The rise in Tilia and Quercus implies that there was a change from an oceanic climate towards more continental conditions (Dahl, 1998; Hafsten, 1956) because these trees are less sensitive to drought and generally require high mid-summer temperatures (Hintikka, 1963; Pigott, 1981; Pigott and Huntley, 1978; Prentice and Helmisaari, 1991; Seppä et al., 2005; Skre, 1979). The first presence of pollen of Viscum, for which a high summer temperature is of vital importance (Hafsten, 1956; Zagwijn, 1994), also supports the notion of the continental character of the climate. At the same time, dry conditions are attested to by low values of the representatives of wetland vegetation, such as Cyperaceae, Equisetum and Polypodiaceae indet. Thus, the occurrence of these indicator species in the pollen record of Lake Skogstjern is in good agreement with the beginning of the Holocene Thermal Maximum (HTM), a period characterised by warm and dry summer conditions (Antonsson and Seppä, 2007; Seppä et al., 2005; Wanner et al., 2011). The records of Seppä et al. (2009) and Antonsson and Seppä (2007) show that the reconstructed mean annual temperature was about 2.0–3.0°C higher than at present during the earlier part of the HTM, at 8000–6000 cal. BP, which is assumed to be related to predominantly anticyclonic summer circulation in Northern Europe during that time.
This period is moreover characterised by a significant decrease in the total terrestrial pollen concentration values (up to 50 × 103 grains/cm3). Variations in pollen production and pollen dispersal to the lake can be associated with the mean temperature in spring and the duration of spring. According to, for example, Davies and Smith (1973), it is the temperature for growth in spring that appears to be the dominant factor for pollen productivity. Under a more continental climate, the spring season is very short and moves on quickly to hot summer. As a result, the flowering period is shortened and drought-induced stress can arise. In relation to this, Ejsmond et al. (2011) have found out that desiccation stress during the flowering period has a negative effect on pollen production.
During this stage, a strong forest dynamic is furthermore reflected in the pollen record. Repetitive gaps in the forest canopy are indicated particularly by recurrent fluctuations in the values of Tilia, which alternate with rising values for pollen assemblages of the light-demanding Corylus. Equivalent phases of regeneration of the vegetation are demonstrated by strongly oscillating curves of Betula and Alnus, reflecting their nature as pioneer trees that expand in disturbed areas. At the same time, elevated amounts of Calluna and Vaccinium-type (up to 2.3%) may point to an increase in open areas with leached soils. The oscillations in the curves of the arboreal taxa may be explained by soil disturbances events, such as landslides, scree avalanches or rock falls, which would create open areas in the woodland. A series of such events of strong catchment erosion are visible in a coincidence in time in the geochemical record with extremely high concentrations of Al, Rb, Zr, Ti, K and Si (LSZ IV). Al, Rb, Zr, Ti, K and Si are commonly associated with allochthonous minerals or clastic sediments (Bajard et al., 2015; Kemp and Izumi, 2014; Unkel et al., 2011). In accordance with this, strong intensities of these elements are observed to correlate positively with high mineral matter content in the sediment, demonstrating a phase of increased erosional input from the lake catchment to the water column between 8030 and 5760 cal. BP. Among these elements, Si has the most intense signals (⩽101 × 103 cps). This element, however, can be affected by both allochthonous and autochthonous signals, representing the input of, on one hand, terrigenous material and, on the other hand, biogenic silica (e.g. diatoms). Nevertheless, the ratios of Si to other elements (Si/Br, Si/S) show a similar pattern to the ratios of Ti (Ti/Br, Ti/S), which can be regarded as an element of only terrigenous origin (Arnaud et al., 2012), mainly found in weathering resistant fine-grained silicate minerals (Brady, 1990; Kylander et al., 2013; Taboada et al., 2006). The same is true of Rb, which is related to fine clay silicates (Koinig et al., 2003). In contrast, coarse-grained sand fractions are typically enriched in Si and K because of the dominance of feldspars and quartz (Kylander et al., 2013). Accordingly, the increases in the ratios of Si/Ti, K/Ti, Si/Rb and K/Rb may be indicative of enhanced input of coarse-sized particles to the lake basin.
Erosion intensity depends on rainfall duration and intensity, soil type, slope and vegetation (Bajard et al., 2015; Cerdan et al., 2002). Heavy rainfall events during summertime, which are typical of a continental climate, coupled with the steep topography surrounding the lake may be a potential cause for the deposition of such high quantities of terrigenous material in the gyttja fraction. Furthermore, desiccation-induced low vegetation cover in the lakeshore environments, as indicated by the low representation of the elements of moist habitats (Cyperaceae, Polypodiaceae indet., Equisetum), may also have encouraged the sedimentation of eroded material to the lake deposits. This hypothesis is in accordance with the results of Støren et al. (2016), showing that intense summer rainstorms are triggers for floods, eroding and transporting material to lakes. The authors also point out that rainstorm-triggered floods are more frequent during the warm early and middle Holocene.
The pollen record of this stage also displays a visible increase in palynological diversity among herbaceous taxa. During this period, the repeated increases in pollen of several ruderals – such as Poaceae, Rumex acetosa-type, Artemisia, Caryophyllaceae, Chenopodiaceae, Urtica, Trifolium repens-type, Senecio-type, Liguliflorae, Polygonum aviculare-type, Potentilla-type, Geum-type and Thalictrum – point to the frequent presence of small-scale open locations on the lakeshore. This may indicate that small human groups repeatedly utilised the areas around the lake during this time, probably for fishing, for hunting birds or game or for collecting eggs. However, significant traces of burning, reflected by elevated amounts of microscopic charcoal particles together with high values of the fire indicator Pteridium (Bińka and Nitychoruk, 2013; Tinner et al., 2000), were recorded only between c. 7510 and 6930 cal. BP (Subzone D1). This demonstrates a clear decrease in fire activity and may be indicative of a general decrease in human impact. According to the archaeological record, the locations of the settlement sites show a clear marine orientation during the late Mesolithic (Solheim, 2013). Therefore, natural causes such as those described above also have to be taken into account in explanations for forest disturbance. Nevertheless, periods of strong erosional events seem to have been decisive for human abandonment of the catchment of the lake because the high signals of mineral input in the sediment are negatively correlated with the assemblages of herbaceous pollen and the content of micro-charcoal. It is conceivable, for instance, that rainstorm-triggered intense slope erosion occurring over lengthy periods of time could have had a negative effect on the choice of such ‘unstable’ inland locations as settlement sites. Strong erosional inputs to the lake may also have had an effect on both its water quality and biota (Wieckowska-Lüth, in preparation), changing the range of limnic food resources.
5940–4950 cal. BP (PAZ E): Small-scale forest disturbances
The beginning of this stage is marked by a distinct decrease in Tilia assemblages and a less pronounced reduction in Ulmus around 5940 cal. BP. At the same time, there is an increase in the light-demanding Betula (up to 44.7%), which may be attributed to lower competition for light after the decline in lime. Elevated amounts of Calluna (up to 1.7%) and frequent peaks of Juniperus pollen may also reflect better light conditions in the woodland. There is no concurrent evidence of increased anthropogenic impact on the vegetation visible in the pollen record, suggesting that changing climatic conditions may have affected in particular Tilia, which is the most warmth-demanding tree of the nemoral forest (Hafsten, 1992). Analogous declines in the warm temperate tree taxa Tilia and Ulmus occur contemporaneously in other pollen records from central Scandinavia and are usually associated with the start of the gradual and continuous decrease in annual mean temperatures towards present times (Antonsson and Seppä, 2007; Giesecke, 2005 Heikkilä and Seppä, 2003; Seppä and Poska, 2004). At the same time, evidence for a cold spell has been recorded for the North Atlantic and central Europe between 5800 and 5100 cal. BP (Moros et al., 2004; O’Brien et al., 1995; Oppo et al., 2003). The decrease in the abundance of pollen from Viscum and Hedera helix in the record of Lake Skogstjern supports the interpretation that the climate was becoming cooler.
At c. 5760 cal. BP, there is a marked reduction in Ulmus visible in the pollen record (Subzone E1), which may reflect the ‘classic elm decline’. At the same time, the values of Alnus increase, indicating the spread of alder on former elm sites. However, the low amounts of Ulmus persist for a couple of hundred years. The elm curve nearly reaches its original level (4%) again around 5520 cal. BP. The ‘elm decline’, a widely recognised horizon in pollen sequences from northwestern Europe, has been connected with a fungal disease, which has affected the elm trees in a woodland already stressed and disturbed by humans (e.g. Peglar and Birks, 1993). Perhaps the Ulmus stands were not so much affected by the ‘elm disease’ or, alternatively, were able to recover again because the trees were not additionally penetrated by humans, as there is no unequivocal evidence of anthropogenic activities at this time in the catchment of Lake Skogstjern. At the same time that the elm curve declines, there is a visible reduction in the terrigenous component shown by the decreases in both the mineral matter content and the ratios of Ti/Br, Ti/S, Si/Br and Si/S (LSZ V), demonstrating reduced erosional inputs to the lake and thus a largely undisturbed catchment.
A little later, just above the ‘elm decline’, the first indicators of agriculture are visible (Subzone E2). The presence of Plantago lanceolata-type, a taxon characteristic of open grazed as well as fallow land (Behre, 1981; Hjelle, 1999), occurs at c. 5580 cal. BP, together with pollen from grasses (up to 3.1%), several ruderal herbs (up to 1%) and finds of spores of decomposing fungi (Sordaria sp. (HdV-55A) and Apiosordaria verruculosa (HdV-169)), which prosper on, among other things, animal dung (Van Geel, 2001). Shortly after, at c. 5520 cal. BP, cereal pollen grains of Hordeum-type occur. At the same time, rising values of microscopic charcoal particles (up to 1.8 × 103 particles/cm3) suggest forest disturbances caused by fire. However, the amount of total AP decreases only marginally during the early Neolithic, indicating that the gaps in the forest canopy were probably limited to small-scale open areas around the lake. In addition, the evidence of agriculture indicators remains sporadic (in the form of Plantago lanceolata-type) in the further course of this stage or even disappears (in the form of pollen of cereals). This demonstrates the very limited extent of farming activity at this time, in that it had no lasting implications for the environment. Nevertheless, synchronously with the culmination of pollen indicative of agriculture, the amount of terrigenous input in the sediment increases slightly (LSZ V), possibly reflecting input of eroded lake margin material resulting from anthropogenic activities in the immediate vicinity of the lakeshore.
The number of excavated settlement sites from the early Neolithic is very low in southeastern Norway. According to the archaeological record, the sites occupied the same settings as Mesolithic sites, which are located directly on the shore and are oriented towards the sea (Glørstad, 2010). Nevertheless, some sites can be considered to have been oriented inland (Amundsen et al., 2006; Reitan, 2012, 2014). Thus, the results of the pollen analytical studies largely confirm the archaeological picture – a picture that leads to the expectation that farming played a minor role (Glørstad, 2010).
Around c. 5350 cal. BP, a marked increase in Betula, in parallel with a reduction in Poaceae, is visible, followed by small increases in Tilia and Ulmus, reflecting a regeneration phase of the forest (Subzone E3). On the other hand, the values of Quercus decrease. This may point to a selective maintenance of lime and elm trees for the purposes of harvesting fodder for livestock, but it may also be related to climatic causes. However, forest browsing by small herds of domestic animals may also be indicated by the increased abundance of Corylus and the presence of Juniperus and Calluna pollen. In any case, human influence seems to have diminished at this time, as evidenced by the general increase in AP.
A little later, around 5190 cal. BP, the amount of mineral matter increases again visibly in the sediment (LSZ VI). This, along with the high concentrations of the minerogenic elements (Al, Rb, Zr, Ti, K, Si) and the ratios indicative of inputs of both allochthonous and coarse-grained particles (Ti/Br, Ti/S, Si/Br, Si/S, Si/Ti, Si/Rb, K/Ti, K/Rb), initiates a new phase of increased contribution from terrigenous material being transported from the slopes into the lake. During this interval, there is palaeoclimatic evidence for lower winter precipitation and glacier retreat in western Norway between c. 5100 and 4700 cal. BP, which are suggested to be associated with a decreased strength of the westerly circulation over Northern Europe (Nesje et al., 2001). Over a period of reduced winter precipitation, the resulting low snow cover in general implicates a decrease in spring snowmelt floods. The consequence of this could be that the soils become sensitive to subsequent water runoff, in particular, when the precipitation is being limited to major rainstorms. In this context, Wei et al. (2007), for instance, have shown that rainfall regimes, which have features such as high intensity, short duration and high frequency, produce more runoff and soil erosion. Hence, the erosional signals recorded by the physical and geochemical proxies in the sediment of Lake Skogstjern are likely to be related to a phase of increased rainstorm events.
4950–4020 cal. BP (PAZ F): More open forest structure
During this stage, consecutive declining trends can be observed in the curves of the warmth-loving temperate taxa Tilia and Ulmus, suggesting that these trees become increasingly sparse in the forest inventory, whereas the amount of the light-demanding Quercus increases distinctly (up to 16.3%). With an exception around c. 4480 cal. BP, the values of Corylus (up to 24.1%) reach somewhat higher levels than before, indicating that the reduction in elm and lime and the associated crown thinning also favoured the spread of hazel. This stage is also characterised by a noticeable increase in the values of the total terrestrial pollen concentrations (up to 262 × 103 grains/cm3), indicating an absolute increase in pollen production and its deposition to the lake. This is probably because of the more open forest structure, enabling in particular Corylus and Quercus to flower more prolifically.
At the beginning of this period, there is a temporary increase in total AP, demonstrating closed woodland. In particular, the values of both Quercus and Corylus reach their highest levels at c. 4860 cal. BP, coincident with a distinct reduction in Betula pollen. This may therefore reflect the effect of decreased competition for light within the forest. The decrease in the pioneer tree Betula may be linked to a lower degree of human-induced forest disturbances, because the pollen record shows a nearly complete absence of herbaceous taxa and a marked reduction in Poaceae, in particular between c. 4830 and 4580 cal. BP (Subzone F1). The hypothesis that human-induced forest disturbances were reduced is contradicted, however, by the fact that the content of the microscopic charcoal particles features relatively high amounts (up to 3.6 × 103 particles/cm3), pointing to the frequent occurrence of fires. In addition, spores of Gelasinospora (HdV-1), a fungus that prefers dry conditions and layers with charred material (Van Geel and Aptroot, 2006), accompany the highest peaks of microscopic charcoal particles. The rise in the charcoal dust cannot be attributed solely to natural fires, as they would have also resulted in an increase in pioneer vegetation. In any case, the use of fire was not practised for clearing new land, as demonstrated by the rising amount of total AP and the distinct reduction in herbaceous open-land indicators. It is difficult to establish how or why these relatively high values of microscopic charcoal particles arose. One possible explanation could be that the micro-charcoal recorded in the sediment originates from domestic fires related to temporary campsites at the lakeshore. On balance, this palynological evidence is in a very good agreement with the archaeological data, which shows that people reverted to hunting and gathering during the first part of the middle Neolithic (Hinsch, 1955; Østmo, 1988).
The period between c. 4950 and 4580 cal. BP also displays an elevated accumulation of minerogenic material in the sediment, suggesting unstable soil conditions throughout this time (LSZ VI). As there is no concurrent evidence of anthropogenic activities, humans can be excluded as a potential promoter of the erosion in the catchment area. Destabilisation of slopes because of more open forest structure could be a possible driver behind this. The record of Lake Skogstjern accordingly suggests a short period of destabilisation of the local environment, probably because of changing palaeo-environmental conditions that may be related to a stage of increasing climatic instability (e.g. Antonsson, 2006; Seppä et al., 2009), initiating the end of the HTM. High summer temperature anomalies were recorded between 5000 and 4000 cal. BP, for example, by Seppä et al. (2009) for northern Europe. On the other hand, Hammarlund et al. (2003) report stable and dry conditions between c. 4950 and 4600 cal. BP in southern Sweden.
Around c. 4580 cal. BP, an increasing trend in the values of Poaceae is visible, which is accompanied by small rises in ruderal herbs and the scattered occurrence of Plantago lanceolata-type as well as single cereal pollen grain (Subzone F2). This suggests, on one hand, that only a restricted number of grazing livestock were present in the surroundings of the lake and, on the other hand, that cereal cultivation played a minor role in the economy of that time, if any at all. This seems to be confirmed by the fact that the values of the AP remain at a relatively high level, reflecting low-level human impact during the first half of the late Neolithic. From this point in time, a rather low disturbance regime in the catchment is indicated by sparse erosional inputs into the lake. This can be deduced from the low amount of terrigenous matter in the sediment (LSZ VII), shown by the declining ratios indicative of both allochthonous material (Ti/Br, Ti/S, Si/Br, Si/S) and coarse-grained sand fractions (Si/Ti, Si/Rb, K/Ti, K/Rb). Only several short-lived increases occur throughout the time, demonstrated by, in particular, peaks of the minerogenic elements (Al, Rb, Zr, Ti, K, Si). Moreover, the more abundant records of spores of Polypodiaceae indet., Sphagnum, Equisetum and Lycopodium annotinum-type from c. 4360 cal. BP onwards point to increased available moisture and the development of a peat bog around the lake.
4020–2310 cal. BP (PAZ G): Gradual increase in forest disturbances
The start of this stage is accompanied by a further decrease in Ulmus and a pronounced reduction in Tilia, which is in good agreement with the dating of the end of the HTM (Antonsson and Seppä, 2007; Hammarlund et al., 2003; Seppä et al., 2009). Nonetheless, the elm curve in particular shows more or less strong fluctuations throughout the time period, which may result from different effects, such as forest disturbances, habitat competition or changing moisture conditions. The values of Quercus remain generally high during this stage, whereas those of Betula even increase (up to 52.1%). In addition, the pollen record shows a visible rise in Fraxinus (up to 3%). The expansion of Fraxinus may be explained by its low degree of shade tolerance. As a less shade-tolerant tree (ten Hove, 1968), ash could have benefited from the better light conditions in the now more open Betula–Quercus forests. Adding to the picture, the occasional occurrence of pollen of Myrica gale suggests the existence of acidic ground. In particular, starting c. 3420 cal. BP, elevated values of the light-demanding Juniperus (up to 1.6%) are repeatedly recorded together with peaks of Corylus (Subzone G2), suggesting frequent openings in the forest canopy, as is also displayed by the fluctuating curve of Quercus. Nevertheless, there is no great change in the amount of AP, showing that the clearings of the forest were still limited to small areas.
During this time span, several steep peaks in Betula are also noted, paralleling decreases in shrub pollen of Corylus and Juniperus. These patterns point to frequent phases of secondary forest succession, during which woodland could regenerate. The first very steep Betula peak, at c. 3950 cal. BP, however, is coincident with a distinct reduction in Quercus and an increase in anthropogenic indicators, especially in the form of microscopic charcoal particles and Pteridium. This may be indicative of forest clearances by burning to create new land for settlement. At the same time, there is a temporary increase in the terrigenous inputs to the water, which may reflect the effect of anthropogenic activity near the lake.
With the exception of the transition from the late Neolithic to the Bronze Age (c. 3650–3610 cal. BP) and, again, the Middle Bronze Age (c. 3230–3180 cal. BP), when short-term but clear drops in the assemblages of anthropogenic indicators are visible, the period 4020–2310 cal. BP shows more traces of human-induced disturbances in the vegetation in the palynological record than the preceding periods. This is demonstrated by elevated percentages of ruderal herbs (up to 2.8%) and Poaceae (up to 9%), increased palynological richness (Plantago major/media-type, Rumex acetosa-type, Artemisia, Chenopodiaceae, Spergularia-type, Urtica, Caryophyllaceae, Liguliflorae, Senecio-type, Matricaria-type and Potentilla-type) and generally higher values of microscopic charcoal particles (up to 6.4 × 103 particles/cm3). Pollen of Plantago lanceolata-type occurs more frequently now, but its curve is still not closed. Further evidence of grazing is provided by a few emerging finds of potentially coprophilous fungal spores (Cercophora sp. (HdV-112), Sordaria sp. and HdV-204). This may demonstrate a higher density of grazing domestic animals in the surroundings of the lake. However, there is no question of permanent open pastures yet. In addition, consistent, scattered finds of cereal pollen grains show that crop cultivation continued to play only a minor role.
Between c. 2990 and 2310 cal. BP, the values for Pinus increase slightly (up to 19%), showing that pine became more dominant in the forest inventory (Subzone G3). The temporary expansion of Pinus might indicate the onset of somewhat drier conditions. An equivalent drop in the frequency of Alnus pollen displays a reduction in local alder stands and agrees very well with the assumption of decreased humidity. At the same time, the elm curve shows some increases, which are always synchronous with the lowest levels of Alnus assemblages, indicating less habitat competition after the reduction in alder. Changes in moisture conditions may also be indicated by the strong fluctuating values of Polypodiaceae indet. and Equisetum. Just at this time, Fagus and Carpinus gain in importance, probably representing long-distance transport because of more open forest canopy. Another characteristic feature during this time is the noticeable decrease in the values of the total terrestrial pollen concentrations (up to 70 × 103 grains/cm3; Subzones G3 and G4), pointing to reduced pollen productivity, probably because of the shift towards drier conditions. Evidence for an analogous dry phase is shown by the record of Hammarlund et al. (2003), which exhibits increased evaporation as a result of decreased net precipitation between c. 3100 and 2500 cal. BP. For the same time span, Nesje et al. (2001) report on decreased glacier activity and lower winter precipitation.
The values of Quercus remain generally high during this time. The curve, however, fluctuates considerably, indicating disturbances. The most important reduction in oak assemblages is recorded during the Late Bronze Age, between c. 2840 and 2670 cal. BP. At exactly that time, the values of Corylus show a marked decrease, whereas the elements belonging to more open vegetation, such as Juniperus, Calluna and Poaceae, increase visibly (up to 12.8%). This increase is paralleled by a rise in microscopic charcoal particles and a find of Gelasinospora (Subzone G4), both of which point to local fire activity. At c. 2840 cal. BP, the number of pollen grains from different herbs increases, as do records of potentially coprophilous fungal spores (Sordaria sp.). This increase is followed by an increase in Plantago lanceolata-type and Triticum-type at c. 2710 cal. BP. Around the same time, elevated amounts of Filipendula and Ranunculus flammula-type suggest removal of the near-shore tree growth in order to utilise the damp areas for grazing. A corresponding slight rise in the input of terrigenous material in the sediment indicates increased soil erosion during this time.
A regeneration phase of the forest is demonstrated a little later by the distinct rise in Betula pollen, succeeded by increases in Corylus and Quercus pollen. Shortly thereafter, a less pronounced rise in both anthropogenic indicators and terrigenous input in the sediment is recorded, starting c. 2630 cal. BP. However, the amounts of Plantago lanceolata-type and potentially dung-indicating fungal spores are somewhat higher than before, pointing to more animal husbandry taking place. Nevertheless, the evidence of Plantago lanceolata-type as well as cereal pollen grains is still so low that the establishment of a fully agrarian community cannot be assumed in the catchment of the lake at this stage.
On balance, the results from this portion of the lake core from Lake Skogstjern do not entirely agree with the archaeological record for the late Neolithic. The archaeological record shows great changes in settlement structure and a coincident increase in the number of finds of axes, which, together, are interpreted as indicating an increase in both population and agricultural activity (Mikkelsen, 1989). According to Mikkelsen and Høeg (1979), the great expansion of cereal growing took place in the late Neolithic and in the Bronze Age throughout eastern Norway, and perhaps throughout the entire country. The data obtained from the palynological study thus demonstrate that there are obviously considerable local differences in the extent of land use and the expansion of agriculture during this time.
After another stage of forest regeneration evidenced by increases in Betula and Quercus, which culminates around c. 2380 cal. BP, a steep rise in Poaceae is recorded, along with a rise in the values of Corylus, Juniperus, Calluna and ruderal herbs. At that same time, the sediment core contains pollen of Hordeum-type and Plantago lanceolata-type. This may be associated with the onset of increased settlement and farming activity during the pre–Roman Iron Age.
2310–1870 cal. BP (PAZ H): Incipient expansion of open land
The next zone in the core is distinguished by a further distinct reduction in Tilia, whereas the values of Alnus increase visibly (up to 33.6%), demonstrating the expansion of wetland forests. The rise in the abundance of alder may be related to a rise in the groundwater table. A synchronous rise in spores of Polypodiaceae indet. and Equisetum also indicates the development of wet biotopes, probably because of more humid conditions. Between c. 2310 and 2090 cal. BP, steep decreases in Quercus, Ulmus, Fraxinus, Pinus and Corylus are registered, while the values of Betula increase significantly, reflecting its nature as pioneer tree, colonising disturbed areas. Simultaneously, the amount of total terrestrial pollen concentrations rises visibly (up to 194 × 103 grains/cm3), possibly because of an increase in Betula and Alnus, both of which are highly prolific pollen producers.
At the same time, indicators of open land (up to. 9.9%), such as Calluna, Poaceae and ruderal herbs – particularly Rumex acetosa-type – continue to rise (Subzone H1), whereas the values of microscopic charcoal particles show no appreciable increases, although they are consistently present. In parallel, spores of Pteridium, a fire-indicating fern (Bińka and Nitychoruk, 2013; Tinner et al., 2000), rise sharply (up to 1.5%), suggesting that fire may have been an important factor in the opening up of the forest. Conversely, unequivocal indicators of farming, such as Plantago lanceolata-type and pollen of cereals, almost disappear. Instead, the elements of wet meadows (Ranunculus flammula-type, Filipendula) are present in the pollen record. The relatively high proportions of Rumex acetosa-type (up to 0.6%) probably also reflect the spread of wet meadow communities, as some representatives of this taxon grow in mineral-rich damp areas (Behre, 1981) and may be indicative for the use of near-shore locations for grazing purposes. Parallel finds of decomposing fungal spores, such as Apiosordaria verruculosa, Sordaria sp. and Cercophora sp., indicate the presence of animal dung. However, the low evidence of clearer indicators of farming suggests that any agricultural areas that did exist in the vicinity may have been situated at a greater distance from the lake. Rather, low inputs of terrigenous material affirm an undisturbed catchment, which may be a further indication of limited use of the local environment for agricultural activities.
Around 2120 cal. BP, distinct declines in the total amount of herb pollen and Calluna especially, together with a very steep rise in Betula, followed a little later by increases in Quercus, Ulmus, Fraxinus and Pinus, indicate a forest succession stage (Subzone H2). A decreasing trend in the AP together with higher proportions of open-land indicators are again recorded from c. 2030 to 1900 cal. BP. The percentages of open-land indicators are, however, lower than before (up to 5.9%), probably reflecting a somewhat lower level of anthropogenic activity. Nevertheless, Plantago lanceolata-type occurs again during this time, together with elevated evidence of potentially dung-inhabiting fungal spores (Sordaria sp.), pointing to grazing animals in the surroundings of the lake. At the same time, Alnus, which prefers wet sites, decreases temporarily, indicating that the wetland forests were also cleared, perhaps to create new grazing land.
1870–850 cal. BP (PAZ I): Significant woodland clearance and establishment of permanent fields and pastures
This stage is characterised by a further decrease in Ulmus, whereas the frequencies of Corylus (up to 29.5%) and Juniperus (up to 3.5%) increase considerably in general, demonstrating changes in vegetation structure towards a more open catchment area. The extremely high abundance of Corylus, in particular, suggests that hazel shrubs may have formed a major part of an open-cover forest because Corylus cannot regenerate under a closed canopy (Vera, 2000). During this time span, a considerable change in the landscape is also shown by distinctly rising amounts of NAP, maximal values of microscopic charcoal particles and increasing diversity of herbaceous pollen taxa. The curves of the major anthropogenic indicators, however, show repeated short regressions, demonstrating several fluctuations in settlement activity.
The woodland clearance begins at c. 1790 cal. BP, during the Roman Iron Age, with a distinct reduction in Quercus and Corylus, paralleled by a steep rise in the open-land indicators Poaceae (8.9%; Subzone I1). During this period, however, the values of microscopic charcoal particles initially remain at a relatively low level. Nevertheless, the occurrence of Gelasinospora, Pteridium and Melampyrum indicates disturbances in the local vegetation by the use of fire (Innes and Simmons, 2000). The amount of evidence for cereal pollen grains is still rather small, but a weed flora with Scleranthus and Spergularia-type may reflect local arable fields (Behre, 1981). Plantago lanceolata-type is present, together with somewhat a higher abundance of potentially coprophilous fungal spores (Cercophora sp., Podospora sp. (HdV-368), Sordaria sp. and Apiosordaria verruculosa), indicating pastoral farming. Another indication of the presence of grazing animals is the appearance of an egg of the parasitic worm Capillaria sp. (Le Bailly et al., 2007), which infects, inter alia, large herbivores (Léger et al., 1991). Then, at c. 1660 cal. BP, a short phase of forest regeneration is recorded through an increase in Betula, followed by Corylus and Quercus. But this does not, however, lead to an absolute interruption of settlement.
A second and much more substantial rise in cultural indicators can be seen between c. 1640 and 1380 cal. BP, the period encompassing the end of the Roman Ion Age and the following Migration Period. Increased anthropogenic impact is demonstrated by the maximum abundance of microscopic charcoal particles (up to 15.1 × 103 particles/cm3) and Gelasinospora (Subzone I2), reflecting significant fire activity during this phase. At this time, the curves of both Plantago lanceolata-type and cereals can be followed continuously for the first time, indicating the establishment of permanent cultivated fields and pastures. In addition, the evidence for Plantago lanceolata-type (up to 0.8%) as well as cereal pollen grains (up to 0.9%) – including Triticum-type, Hordeum-type and Avena-type– is now visibly higher than in the previous period, demonstrating general expansion and intensification of agrarian activities. In addition, high values of Poaceae (up to 9.6%) reflect large open areas, while increases in both the amount and the diversity of potentially coprophilous fungal spores (Cercophora sp., Sordaria sp., Podospora sp., Sporormiella sp. (HdV-113) and HdV-204) indicate extensive grazing. The management of these sites either as meadows or as pastures is reflected by the presence of herbs such as Rumex acetosa-type, Liguliflorae, Senecio-type, Matricaria-type, Cirsium-type, Trifolium pratense-type, Trifolium-type, Thalictrum and Centaurea jacea-type (Behre, 1981; Hjelle, 1999). The increased importance of crop cultivation may also be seen in a greater abundance than before of several potential arable indicators, such as Brassicaceae, Caryophyllaceae, Spergularia-type and Spergula (Behre, 1981). Moreover, the number of taxa characteristic of ruderal communities, such as Chenopodiaceae, Artemisia, Potentilla-type, Geum-type, Rumex scutatus-type and Epilobium, is now significantly higher than before (Behre, 1981; Düll and Kutzelnigg, 2005).
From c. 1400 cal. BP onwards, Picea becomes important in the forest inventory (Subzone I3). As already concluded by Bjune et al. (2009a), increased human impact and fire seem to have been the driving factors in the creation of openings for spruce to establish itself. The first evidence of Abies also occurs during this time, pointing to its local presence. Finally, at c. 1360 cal. BP, a strong rise in pollen of Betula and a somewhat delayed rise Quercus indicate a forest succession stage, in which woodland could regenerate. While there is no break in the continuity of settlement, its level of intensity is significantly lower.
A new phase of enhanced settlement activity is recorded around 1340–1000 cal. BP, the period covering the Merovingian Age and the first part of the Viking Age, accompanied by renewed increases in microscopic charcoal particles, Poaceae, cereal pollen grains, Plantago lanceolata-type, potentially coprophilous fungal spores and ruderal herbs (Subzone I3). The first Secale pollen is recorded around c. 1130 cal. BP. The intensity of this settlement phase, however, is somewhat diminished compared with that of the preceding phase.
The two phases of increased land use just reported are, moreover, characterised by relatively high amounts of elements of wet meadow communities (up to 1.3%), such as Ranunculus flammula-type, Filipendula and Mentha-type, suggesting that the near-shore areas were used as pastureland. This may be verified by a clear decreasing trend in the curve of Alnus, pointing to the reduction in local alder forests (Subzones I2 and I3). Extensive open habitats at the lakeshore are also demonstrated by two significant increases in fern spores, corresponding to the highest level of settlement intensity. Synchronous strong increases in the values of Calluna (up to 3.5%) may point to advanced deterioration of soils through continuous exploitation. The higher evidence of Myrica gale pollen (up to 1.3%; Subzone I3) also implies greater soil acidity associated with the development of bogs. At the same time, distinct increases in the mineral matter content, along with the rises in both the elemental concentrations of Al, Rb, Zr, Ti, K, Si and, in particular, the ratios of Ti/Br, Ti/S, Si/Br and Si/S, demonstrate increased inputs of material from terrigenous sources (LSZ VIII). This sedimentation of allochthonous material, culminating at c. 1660–1430 cal. BP and at c. 1250–1060 cal. BP, respectively, could be explained by the very extensive use of the landscape at that time, which has probably led to the destabilisation of the catchment soils.
This relatively long-lasting stage of both settlement continuity and clearly increased human activity is coincident with a period of warm and dry summer conditions, as recorded by Seppä et al. (2009) for Northern Europe between 3000 and 1000 cal. BP. According to Seppä et al. (2009), this phase can be connected with the Roman Warm Period and the Medieval Warm Period. Prevailing dry conditions may be reflected in the pollen record of Lake Skogstjern by the decrease in Alnus assemblages, as well as by the extremely high abundance of fern spores of Polypodiaceae indet., suggesting advanced filling in of the near-shore vegetation towards the centre of the lake.
A distinct drop in human pressure on the catchment of the lake is noticeable only between c. 970 and 890 cal. BP, during the second part of the Viking Age, as evidenced by the distinct reduction in both the microscopic charcoal particle content and the curves of agriculture indicators, such as Plantago lanceolata-type, potentially coprophilous fungal spores, cereals and the representatives of wet meadows (Subzone I4). The coincident small increase in Alnus suggests the colonisation of some lakeshore areas with alder. At the same time, the lake geochemical record shows an abrupt decrease in the terrigenous contribution. This decline in land-use activity may be connected with the onset of cooling at c. 1000–800 cal. BP, leading to the ‘Little Ice Age’ (Bjune et al., 2009b; Bradley et al., 2003; Seppä et al., 2009). Another plausible explanation for the diminishing human impact could, however, also be that the settlement and the agricultural areas have moved further away from the lake, probably because of destabilisation of the catchment area as a result of the preceding intensive use of the landscape.
850 cal. BP–present (PAZ J): Less diverse woodland in an open, cultivated landscape
This stage is characterised by a major change in the forest composition because of the establishment of Picea. With the expansion of Picea (up to 6.4%) from c. 850 cal. BP onwards (Subzone J1), a significant decrease especially in the abundance of Quercus is recorded, perhaps indicating the replacement of the light-demanding oak by the intermediate shade-tolerant Picea. Further declining trends can be observed during this period in the curves of the other thermophilous deciduous taxa, such as Ulmus, Fraxinus and Corylus. Tilia pollen even disappears for some centuries, whereas the values of Pinus and Betula increase temporarily (Subzone J2). The dominance of birch, together with the presence conifers in the forest inventory, from c. 850 cal. BP onwards may be attributed to climatic change during the late Holocene rather than to other factors, such as human assistance. There is a marked drop in the amounts of total terrestrial pollen concentration (up to 75 × 103 grains/cm3) recorded during this stage, probably resulting, in part, from the decrease in the highly prolific pollen producer Corylus. However, the causes for the general reduction in the concentration amount of total terrestrial pollen may also be associated with changing climatic conditions. As shown by several studies (e.g. Danielsen, 1970; Giesecke and Bennett, 2004; Hammarlund et al., 2003; Nesje et al., 2001; Seppä et al., 2005), declining temperatures and increased humidity prevailed in Scandinavia at that time, becoming noticeable particularly by cold, snow-rich winter conditions. Such circumstances must have contributed both to the spread of Picea, as this taxon favours moist soils in a humid, cool climate, and to the concomitant reduction in temperate broad-leaved trees. Nevertheless, anthropogenic pressure probably played a secondary or marginal role in the spread of spruce, especially as human impact had by now reached significant levels. In any case, confirmation for waterlogging may be found in the palynological record of the lake in the form of general increases in the assemblages of Cyperaceae and Equisetum, as well as rising values of Alnus (Subzone J2), especially between c. 670 and 290 cal. BP.
During the Medieval Period, a phase of enhanced settlement activity is indicated by the distinctly rising values of Calluna, Poaceae and representatives of ruderal communities (up to 15.1%) (Plantago lanceolata-type, Plantago major/media-type, Rumex acetosa-type, Chenopodiaceae, Artemisia, Liguliflorae) between c. 800 and 720 cal. BP. The curve for cereals is also elevated. Contemporaneously, a Juniperus maximum (13.8%) is recorded (Subzone J1). Such extraordinarily high amounts of juniper pollen may be indicative of open cultivation landscape, for example, pastures (Berglund, 1966), and may reflect the local expansion and intensification of agricultural activities. However, records of only moderate values of microscopic charcoal point to reduced fire activity when compared with the previous settlement phases. Instead, the values of Humulus/Cannabis-type increase distinctly (up to 2.2%), suggesting that the lake was probably used as a retting pond for hemp. At the same time, the amounts of Myrica gale (up to 1.7%) increase, suggesting greater soil acidity than before and a further spread of bog communities around the lake. The synchronous decrease in the input of terrigenous material in the sediment, however, indicates stable soil conditions, which may be because of the increased buffering capacity of the developing bog vegetation at the lakeshore.
Later on in the Medieval Period, settlement activity decreases visibly, showing its lowest level at c. 420 cal. BP, although even at this time, indicators for the cultivation of winter cereals, such as Centaurea cyanus and Secale, occur (Subzone J2).
The most significant increase in human agency is visible between the 17th and the 19th centuries. During Modern Times, this land-use expansion is reflected by a large-scale opening up of the landscape, manifested by unprecedentedly sharp increases in Poaceae (up to 18.6%), ruderal herbs (especially Rumex acetosa-type) (up to 9.5%), elements of wet meadows (Ranunculus flammula-type and Filipendula) (up to 2.4%) and cereal pollen grains such as Hordeum-type, Triticum-type, Avena-type and Secale (up to 1.7%; Subzone J3). These increases are accompanied, furthermore, by high values of Juniperus, along with a strong decrease in AP. This stage is also characterised by a distinct increase in the concentration values of NAP, together with a pronounced rise in terrigenous input in the sediment, indicating significant erosion processes during the last centuries covered by the core (LSZ X).
Finally, in the recent past, somewhat reduced settlement activity is recorded, and the allochthonous contribution in the sediment shows a tendency to decrease.
Summary and conclusion
The high temporal resolution record of Lake Skogstjern provides a very detailed picture of the development of both vegetation and land-use strategies during the Holocene. Moreover, it provides additional data on soil erosion processes in the catchment of the lake for more than 10,500 years.
Important vegetational changes in the first two-thirds of the Holocene seem to have been largely brought about by climatic changes. Beyond that, changing ecological factors also appear to have triggered catchment erosion during that time (Figure 5). Human beings started to play a significant role in shaping their environment only in the past two millennia (Figure 5), although the first evidence of increased human activity is visible during the middle Mesolithic (c. 10,050–9400 cal. BP), roughly synchronous with the isolation of the lake from the sea (c. 9950 cal. BP) or shortly thereafter, reflecting small-scale openings in the forest that were created by humans with the assistance of fire. Nevertheless, in the late Mesolithic (c. 6930–5940 cal. BP), a general decrease in human impact is recorded, which seems to be linked to strong erosional disturbances in the catchment of the lake (c. 8030–5760 cal. BP) during the HTM. While the first traces of farming emerge during the early Neolithic (c. 5580–5520 cal. BP), they indicate rather small-scale agriculture and only limited animal husbandry. During the following periods, traces of farming occur repeatedly. However, human impact remains more or less at a low level and does not generate any significant palaeo-environmental change. The pollen record evidences some intensification in animal husbandry only during the second half of the Bronze Age (from c. 2840 cal. BP), involving a slightly higher degree of landscape openness and the utilisation of wet meadows for grazing purposes. Crop cultivation, however, continues to play only a minor role.
Summary diagram of the record of Lake Skogstjern with a chronological overview of human impact and catchment erosion during the Holocene. The figure also shows the climate development according to Antonsson and Seppä (2007), Bjune et al. (2009b), Bradley et al. (2003), Dahl (1998), Digerfeldt (1982), Hammarlund et al. (2003), Nesje et al. (2001) and Seppä et al. (2005, 2009). Archaeological periods according to Gjerpe (2013).
The establishment of a farming community did not take place until the Roman Iron Age. There is continuity in the utilisation of the catchment area of the lake by humans from approximately the second part of that period (c. 1790 cal. BP) to the first part of the Viking Age (c. 1000 cal. BP). Furthermore, these periods are characterised by a general expansion and intensification of land use, as indicated by evidence for extensive forest clearances and fires, crop cultivation in permanent fields and the presence of open pastures, as well as associated advanced soil degeneration and increased erosion rates. The highest degree of human impact took place during modern times, in the 17th–19th centuries.
The causes behind the abrupt reorganisation in land-use practices during the course of the Roman Iron Age could be multiple. A climatic improvement could be one of the key factors for the agricultural changes because it probably made the catchment area of Lake Skogstjern more attractive for agriculture. The intensification in farming may also be explained by a cultural and economic change, in that the increasing specialisation in iron tool technology may have played an important role. Rising demographic pressure coupled with the opening up of new settlement areas may also have been decisive. However, considering the terrain of the catchment area of Lake Skogstjern, of which roughly two-thirds consist of steep, rocky hillsides, it is of only limited suitability for either settlement or farming. Furthermore, because flat, alluvial plains tend to have better farming soils, it can be concluded that the location of usable areas in the past will have complied approximately with the present-day situation. Accordingly, in areas with difficult terrain, the forest was probably never completely cleared for land-use purposes, which is confirmed in the palynological data by the relatively high abundance of AP (⩾70%) right up to the present time.
Footnotes
Acknowledgements
The authors express their sincere thanks to Suzanne Needs for her help with the English language and to the two referees for their helpful comments and suggestions. The authors are grateful to Per Persson, Almut Schülke, Steinar Solheim and Gaute Reitan for their marvellous cooperation. Sincere thanks go to the coring team of 2013 and to Samuel Müller, the ‘XRF-Scanner-HiWi’.
Funding
This work was financed by the Museum of Cultural History, University of Oslo.