A forested form of life: Southeast Norway in 2350–500 BCE

1 Introduction

Present-day Norway is largely covered by forest, making trees a central component in the landscape (Larsson and Søgnen, 2003). This does not mean that trees are always the most abundant organisms, as forested areas can have tree coverage as low as 10%−20%, but that trees are a dominant organism in the ecosystems. Where the necessary nutritional and moisture conditions are met, trees join with “tens of thousands of other plants, animals and microbes” (Thomas and Packham, 2007, p. 5) to form forests, for which as many as 624 different definitions are used worldwide (Lund, 2002).

Forests have played a central role in the development of civilizations across the globe (Ritter and Dauksta, 2012), and here they are rarely primordial and untouched ecosystems. Hunter-gatherers, for instance, often systematically exploit and actively shape the woodland through vegetation management strategies, such as creating clearings, pruning and coppicing (Bishop et al., 2015). The Amazon provides a particularly illustrative example: long regarded as “natural” and subject to little or no human impact, the forest has been redefined as constructed or “domesticated” landscapes (Heckenberger et al., 2007). Based on archaeological research, it has been demonstrated that many forest and wetland areas along the Amazon river has a biocultural diversity and ecological setting which has emerged from thousands of years of human impact within what (Heckenberger et al. 2007, p. 198) term coupled human-environmental systems (see also Fabiano et al., 2021). A comparable long-term entanglement of people and forest is also evident in our study region of southeast Norway, where living within and with the forest appears as a persistent feature of life during the Mesolithic and earlier Neolithic periods, as well as historically (e.g., Grimm and Schmölcke, 2013; Gundersen, 2013; Indrelid et al., 2015; Kaldal, 2000, 2018, 2022; Schülke, 2023).

This paper seeks to address the potential role of forests during the later prehistoric times of the Late Neolithic and Bronze Age (2350–500 BCE). Here, there has been a relatively one-sided focus on the role of deforestation and agricultural intensification in shaping civilisations. Following a series of studies that have mapped the rise of a farm-based society after 2350 BCE in southern Norway, developing through the Late Neolithic and continuing its consolidation during the Bronze Age (e.g., Bergsvik et al., 2020; Glørstad, 2012; Glørstad et al., 2020; Østmo, 1988; Prescott, 1996, 2009, 2020; Prøsch-Danielsen et al., 2020; Solheim, 2021a), there has been an emphasis on forests clearance, often portraying forests as obstacles to be overcome rather than as persistent, integral parts of human life.

In some regions, land clearing was indeed a very real and rapid phenomenon, as has been convincingly suggested for western Jutland and Thy in Denmark particularly, as well as Scania in southernmost Sweden and Jæren in southwestern Norway (e.g., Bech, 2018; Earle et al., 1998; Haak et al., 2023; Iversen, 1941; Kristiansen, 1998; Odgaard, 1985; Prøsch-Danielsen and Simonsen, 2000; Prøsch-Danielsen et al., 2020, 2018; Regnell and Sjögren, 2006; Skoglund et al., 2022). In these regions, the resulting open, pastoral landscapes had significant consequences for Bronze Age societies with regards to, for example, the development of socio-political hierarchies, infrastructure, mobility, sight, visibility, and communication.

However, we argue that the focus on specific economies from specific environments linked to agriculture and pastoralism, has obscured important regional diversities, including communities relying heavily on wild or marine resources, heath biotopes or heavily wooded landscapes. For example, substantial parts of Denmark, particularly outside of western Jutland, were still covered in forest in the Bronze Age (Nielsen and Odgaard, 2010; see also Bech and Rasmussen, 2018, p. 27), and many of these areas display high settlement densities and equally high activity levels. On Djursland in eastern Jutland, for instance, the first major opening up of the landscape occurred at the transition to the Late Bronze Age (Regnell and Sjögren, 2006, p. 148), more than a millennium later than in western Jutland, although there was also earlier, smaller scale opening of the forest here (Klassen et al., 2020).

The prevalent focus on pasture- and agrarian-led deforestation reflects a deeper tendency toward “plant blindness” in archaeology, evidenced through the general failure to recognize the role of diverse plant-based ecologies in shaping human societies (Sand-Eriksen, 2024; Taylor, 2025). It risks uncritical general application of economic and socio-political explanatory frameworks derived from radically different, and potentially irrelevant, ecological and social realities. Moreover, it leans into notions of linear evolution and progress, fantasies of capitalist modes of growth and the optimisms and temporal horizons that it entails, minimizing the important role of persistence dynamics (Løvschal, 2020, 2022).

For southeast Norway, palynological data has suggested that the landscape did not undergo comparable trajectories of extensive deforestation in the Neolithic or Early Bronze Age as in Jutland and in Scania (Høeg, 2003; Høeg et al., 2019; Wieckowska-Lüth et al., 2017). Nonetheless, forest environments have played only a minor role in shaping the developmental narrative of southeast Norway in research history, which—as outlined in brevity above and detailed below—is strongly biased toward agrarian processes. Considering that forested landscapes had radically different social and economic affordances than those of the open landscapes of western Jutland, Thy, Scania, and Jæren, it is highly plausible that the societies of southeast Norway diverge from the generalized expectations associated with the Nordic Bronze Age (1700–500 BCE). What remains an open question, however, is how people residing here would have lived in and co-created a landscape that was profoundly dominated by forests, and how this affected their livelihoods, one generation after the next.

To explore these questions, we integrate high-resolution environmental and archaeological data through a two-pronged methodology. First, we apply the REVEALS model of land-cover reconstruction to 63 pollen records to reconstruct past vegetation. Second, we integrate these environmental reconstructions with an exploratory GIS analysis of an extensive body of archaeological data. With this, we seek to uncover what living a forested “form of life” meant for the communities living in Late Neolithic and Bronze Age southeast Norway. We show how tree cover fluctuates and even increases at some points, indicating that human activity in southeast Norway produced a dynamic mosaic of managed woodlands rather than the permanently open fields predicted by standard agrarian models. By demonstrating the persistence of wooded landscapes and the management of their associated species, for example, the active maintenance of hazel (Corylus) stands and creation of woodland pastures through management and the thinning of forest canopies, these results challenge linear narratives of inevitable deforestation and propose the existence of distinct forest livelihoods during the Bronze Age. The presented research hence contributes to the broader shift and current debates in environmental archaeology, moving away from deterministic, top-down models and toward frameworks that treat nature and culture as co-emergent entities (e.g., Burke et al., 2025; Fitzhugh et al., 2019).

2 Background

The current consensus among Norwegian archaeologists is that southern Norway became a fully farming-based society after 2350 BCE (Bergsvik et al., 2020; Glørstad, 2012; Glørstad et al., 2020; Prescott, 1996, 2020; Prøsch-Danielsen et al., 2020; Sand-Eriksen and Mjærum, 2023; Solheim, 2021a), as first proposed by (Fægri 1944) based on pollen analysis. However, the process leading to this agricultural society remains contentious and is typically explained through two competing models: one proposing a swift, externally driven introduction of a complete farming package, and another arguing for a slow, internal development beginning much earlier in the Neolithic (for an overview and discussion, see Nyland et al., 2023). While the swift model has typically dominated historical narratives, there is evidence suggesting that small-scale agriculture may have existed alongside hunting, fishing, and gathering from the Middle or even Early Neolithic (Bergsvik et al., 2020; Bostwick Bjerck, 1988; Bruen Olsen, 1988, 2013; Hjelle et al., 2006; Høeg, 1989; Mjærum, 2020; Nielsen et al., 2023). This corresponds with the diachronic developments seen in other parts of southern Scandinavia, where univocal evidence of farming is present from about 4000 BCE (Fischer, 2002; Gron and Sørensen, 2018; Gron et al., 2016; Rowley-Conwy and Legge, 2015; Sjögren, 2013; Sørensen and Karg, 2014).

This debate is, however, not new. Around a century ago, scholars like (Shetelig 1922, p. 205) and (Fægri 1944) noted palaeoecological evidence that challenged the simple, late-introduction model. Fægri, for instance, was surprised by the sudden appearance of evidence for crop cultivation during the Late Neolithic and onwards in his pollen samples from southwest Norway, an area geologically and topologically not unlike the early-cleared landscapes of Jutland and Scania (see also Fægri, 1940). He speculated that this might represent a final stage of a much prolonged and multidirectional process, which later scholars suggest had a different character (Bostwick Bjerck, 1988, p. 25), one that may be less visible archaeologically (Nielsen et al., 2019, p. 88; Solheim, 2021a, p. 10). More recent research on the Late Neolithic and Early Bronze Age development supports a multi-stage process (Sand-Eriksen and Mjærum, 2023), where continuous arable farming did not commence before c. 2100 BCE (Solheim, 2021a).

In southeast Norway, there is evidence for localized landscape openings during the Late Neolithic (Høeg, 2002, p. 129), but which are often extrapolated to support claims of extensive forest clearing (Glørstad, 2004; Rønne, 2003a,b). Archaeological and pollen data from the same area are also said to convey an “impression [of an] intensive agrarian activity” in the region (Glørstad, 2012, p. 85), when the pollen report actually argues for small-scale cultivation and suggests that any agricultural activity was likely confined to a small patch within an otherwise forested environment (Høeg, 2003).

We argue that these observations—and the predominant emphasis on agriculture in research concerning the Late Neolithic and Bronze Age in southeastern Norway—overlook a critical factor: the region's enduring forest cover. We suggest that the forests were not merely passive backdrops waiting to be cleared, but rather shaped and were shaped by human and more-than-human lifeways, affording unique historical trajectories. Acknowledging humans as enormously potent niche constructors (Laland and O'Brien, 2010), we view these forests as co-created, or co-evolved, landscapes, where human activities—such as localized clearing, grazing, or selective timber harvesting, but also the establishment of farmsteads in certain areas and their farming activities—served not only to extract resources but also to actively modify the forest in specific directions, where different types of forests gave rise to different resources and social structures.

2.1 Forest as co-created niches

The persistence of denser forest cover in sedentary settlement areas in southeast Norway during the Late Neolithic and Bronze Age, compared to other regions within Scandinavia, demands a framework that recognizes that whilst the human influence on the physical environment is necessarily pervasive, the environment still plays a role in directing these actions.

Understanding this relationship benefits from Gibson's (1979) notion of affordance, defined as the possibilities for action that an environment provides or offers an organism (Gibson, 1979, p. 127; see also Chemero, 2003). Building on this idea, Rietveld and Kiverstein's (2014) concept of a “form of life”, as advanced from Gibson's idea of a “way of life”, captures how these possibilities become structured within human sociocultural practices. A form of life encompasses the shared, normative ways of acting that define a community, emerging from the actions the environment makes possible (Rietveld and Kiverstein, 2014, p. 335). In this view, relevant abilities are not confined to an individual's capacity—in which the typical critique of affordance as a concept lies (e.g., Knappett, 2004)—but are distributed across the collective practices that shape everyday life.

Affordances are hence critical for analyzing the primary mechanics of human behavior, such as where to harvest fodder or clear a small patch for cultivation, or other immediate decisions regarding resource needs and costs. While affordance helps us understand decisions, it is insufficient to explain the broader processes of human and forest coexistence (for similar arguments, see e.g., Stiner and Kuhn, 2016).

When the environmental features, such as resources or habitat, invite human activity, people often begin to reshape those surroundings through processes of niche construction (Odling-Smee et al., 2003, 2013). This perspective clarifies how a forest can evolve from being a neutral backdrop into a true ecological niche—a “forested form of life” defined by specific affordances and habitual ways of living. In this sense, the forested form of life arises as an emergent property of the human-shaped forest environment, which in turn continues to influenced community's development and evolution (Stiner and Kuhn, 2016, p. 182).

This perspective acknowledges that the activities of the Late Neolithic and Bronze Age populations—whether involving inceptive or counteractive niche constructions—result from and in turn affect long-term anthropogenic shaping of the landscape and its species distributions (e.g., Boivin et al., 2016; Hussain and Baumann, 2024; Laland and O'Brien, 2010). Crucially, the outcome is ecological inheritance, creating a specific evolutionary niche (Albuquerque et al., 2018, p. 5), which in the case of southeast Norway is persistently framed by the forest.

3 Material

To reconstruct the past vegetation in the region, we utilize data from 63 pollen records across southeast Norway, a subset of about 300 pollen cores sampled and analyzed by botanist Helge Irgens Høeg since the 1970s (for most recent update, see Høeg, 2005). These cores are selected based on strict dating criteria and qualities needed for them to be included in the REVEALS modeling (for details see Supplement information 1, henceforth SI 1). Considering the likely 50–100 km radius of the pollen source areas, the cores have been divided into eight overarching pollen groups (PG) for vegetation reconstruction (Figure 1). These comprise from three to 19 pollen cores per group (overview in SI 1).

To understand how people lived within and co-shaped this environment, we draw on a diverse set of archaeological data (Table 1), altogether 6,843 unique data entries from the Late Neolithic and the Bronze Age (Figure 2). This includes permanent settlement sites (for overview, see Sand-Eriksen, 2025), which help us address where people established their farmsteads. To explore their use of the wider surroundings, we use data on “off-sites”—these being activities occurring “between [permanent settlement] sites” (Foley, 1981; Thomas, 1975). This data consists of a wide array of archaeological contexts radiocarbon dated to the Late Neolithic and the Bronze Age (see Sand-Eriksen et al., 2025). The archaeological data further includes artifacts (Sand-Eriksen and Solheim, Forthcoming), faunal and macrofossil remains (Solheim, 2021b; Supplement information 2, henceforth SI 2), and also draws on registered hunting sites and so-called negative data.

Table 1. Summary of the archaeological datasets used in this study.

Supporting data are sourced from The Directorate for Cultural Heritage's database Askeladden (askeladden.ra.no), the museum database UniMus (unimus.no), and the University of Oslo's online repository DUO (duo.uio.no)/Norwegian Research Information Repository (https://nva.sikt.no).

4 Methods and analytical approaches

Our methodology comprises two main components: quantitative vegetation reconstruction from pollen data and GIS-based spatial analysis of archaeological data.

4.1 Vegetation reconstruction and economic potential

Translation of pollen assemblages into quantitative estimates of former vegetation cover is complicated by taxon-specific differences in pollen productivity and dispersal mechanisms, and because the pollen signals preserved in different types and sizes of sampling site reflect the surrounding landscape with differing degrees of spatial sensitivity. Taxon-specific algebraic models of the relationship between vegetation abundance and pollen assemblages are available to correct for these biases, with the most widely used to date being the Prentice-Sugita model (Prentice, 1985, 1988; Sugita, 1993, 1994). We used REVEALS (for general details, see Sugita, 2007), which implements the Prentice-Sugita model, to calculate the regional vegetation composition of selected taxa (Table 2) by correcting raw pollen counts with taxon-specific parameters including relative pollen productivity and fall speed, while also taking into account the size and type (i.e., lake or bog) of the basin that pollen records were recovered from (Mazier et al., 2012; Sugita et al., 2010). We used LRA.REVEALS v6.2.5 and relative pollen productivity (RPP) and fall speed (FSP) obtained from (Githumbi et al. 2022), using Sutton's Gaussian Plume dispersal model and assuming a wind speed of 3 m/s.

Table 2. Detailed values for the 35 taxa used in the REVEALS analysis, organized into overarching types and grouped for analysis.

RPP, SD, and FSP values are sourced from (Githumbi et al. 2022), except Brassicaceae (Wieczorek and Herzschuh, 2020). See details in SI 1 regarding Populus.

The taxa used within the REVEALS analysis are summarized in Table 2 and typically make up 95%−99% of the original pollen counts (for notes on the omitted taxa and numerical values from the vegetation reconstruction utilized in this paper, see SI1). The pollen counts were aggregated in 200-year bins covering the period 4400 to 2200 cal BP (c. 2500–300 BCE) on the basis of age-depth models created using the rbacon package (Blaauw and Christen, 2011) and visualized using ggplot2 (Wickham, 2016), developed for the R programming language (R Core Team, 2020). Radiocarbon dates were calibrated using IntCal20 (Reimer et al., 2020).

To account for potential diversity within the large REVEALS reconstructed pollen groups, we also employ a cross-point approach to explore vegetation variations at a finer spatial scale (Figure 1). Overall, we follow the methodology outlined by (Hjelle et al. 2018), utilizing a grid comprising 25 by 25 km cells across the region. For each grid cell containing two or more cores we have created a cross-point (CP); however, for spatial representativity, a CP containing data from only one pollen core has been included on the eastern side of the Oslo fjord. The total number of CPs is hence 16 (Figure 1).

4.2 Archaeological mapping

To understand how people lived within and co-shaped their environment, the vegetation reconstructions are aligned with an exploratory GIS analysis of the archaeological datasets using QGIS (Graser et al., 2025), version 3.34.14-Prizren.

To structure observations across the region and organize the extensive data, we use a grid-based model of 10 by 10 km cells. The spatial distribution of the archaeological data can then effectively be compared against the extent of the PG and CPs. The spatial point data is aggregated into grid cells to produce density heatmaps across the region (e.g., Figure 2); in addition, each data type and dataset is examined separately across multiple spatial scales—zooming in on local details and zooming out to reveal broader patterns. For temporal resolution, the data is divided into ten time-windows from c. 4200 to 2200 cal BP (c. 2350–300 BCE), following the extent of the radiocarbon dated “off site” material and enabling a direct comparison with the vegetation reconstructions diachronically.

5 Results

5.1 Vegetation reconstruction

The vegetation reconstructions are summarized below for the time windows broadly corresponding to the Late Neolithic, Early Bronze Age, and Late Bronze Age (for detailed accounts, see SI 1), with the key findings and initial interpretations synthesized at the end of this section.

5.1.1 End of the Middle Neolithic and the Late Neolithic, c. 4400–3600 cal BP (2500–1700 BCE)

During the four time windows that make up this period, tree cover fluctuates and even increases in some windows, although there there is a marked decline from >40% (often >60%) in the first window to < 40% in the final window of the period (Figures 3A, 4A, and 5A–C). In some areas, like PG3 and PG4, CP3 and CP5–6 especially, the declining tree cover is replaced by an expansion of dwarf shrub taxa (Figure 5D). Other herbaceous taxa are occasionally very abundant in the earliest Middle Neolithic window, reaching 10%−20% of land-cover, but more typically 2%−10% (Figure 5E). Disturbance weeds also show elevated levels early on (Figure 5F), and pastoral indicators (Plantago lancelota and Plantago major) are similarly pronounced (Figure 5G), coinciding with relatively large proportions of grassland, typically < 10% but reaching up to ~35% (Figure 5H). Cereals are prominent in certain locations, reaching 1% in PG7 (Figure 5I).

5.1.2 The Early Bronze Age and transition to the Late Bronze Age, c. 3600–2800 cal BP (1700–900 BCE)

Tree cover remains relatively stable during this period (Figures 3B, 4B, and 5A–C). In certain areas hazel (Corylus) becomes generally more common, although higher levels are occasionally found in earlier time windows (Figure 5J). Dwarf shrubs remain broadly stable (Figure 5D). There is an increase in grassland in some areas, reaching roughly 25%−35% vegetation cover between 1500 and 1100 BCE (Figure 5H). PG4 records a pronounced vegetation shift, with ling heather (Calluna vulgaris), which varied between ~1%−10% in the Neolithic, stabilizing at ~11%−12% in the earliest time windows of this period while Ericales dominate at ~30%. From c. 1500 BCE this relationship changes, with Calluna rising to ~30%, reflecting the establishment of Calluna-dominated heathland in some areas. Several areas experience a pronounced decline in other herbs, all falling below 10% after 1500–1300 BCE (Figure 5E). These declines coincide with an occasional increase in cereals to above 1% of the land-cover, whereas earlier intervals showed lower cereal levels more evenly distributed across sites (Figure 5I).

5.1.3 The Late Bronze Age and beginning of the Pre-Roman Iron Age, c. 2800–2200 cal BP (900–300 BCE)

In terms of tree coverage, this phase remains relatively similar to the Early Bronze Age, but with another minor decline at 700–500 BCE in the PGs (Figures 3C and 5A–C), whereas the CP-data, especially within PG3 and partly PG4, shows that this decline is more pronounced in the Pre-Roman Iron Age (Figures 4C and 5A–C). Of the trees, birch (Betula) is the dominant taxon overall, although in some areas, there is a significant increase in hazel (Corylus) during the Late Bronze Ag (Figure 5J). Although relatively stable, dwarf shrubs generally increase slightly during this period (Figure 5D). At the end of the period, however, Ericales drops significantly to 10% in CP1, with a simultaneous increase in grassland to above 40%. This increase in grassland is also reflected more generally (Figure 5H), although there is no accompanying increase in pastoral indicators (Figure 5G). There is, however, a slight increase in other herbs (Figure 5E) and a more stable, albeit low level of disturbance indicators, particularly between 900 and 500 BCE (Figure 5F). This coincides with a marked increase in cereals (Figure 5I).

5.1.4 Key findings and initial interpretations

Overall, the vegetation reconstruction of southeast Norway shows a dynamic, forest-dominated mosaic rather than a linear transition to permanent open fields. Although the tree cover generally drops from often >60% in the Middle Neolithic to below 50% and even 40% by the Late Neolithic, making these generally semi-open forests, the decline was rather localized and most often followed by regeneration, stabilization, or continued flux into the Bronze Age. In CP3–4, CP5 and CP7, for instance, deciduous species re-expanded at the Early Bronze Age transition, whereas in CP6 a pronounced regeneration occured at the Late Bronze Age transition, while a smaller on occurred in CP5 (Figure 5A).

Birch (Betula) was broadly dominant in the forests, and although woodland composition varied locally, the semi-open forests typically contained a rich understory (Figures 5C–E). Some areas were dominated by pine (Pinus), particularly at higher altitudes (Figure 5B), while sites like CP11 retained a diverse deciduous assemblage. This forest diversity is coupled with evidence of human influence. Hazel (Corylus), for instance, featured strongly around the Oslo fjord (e.g., CP3–4, CP11), with patterns that could point to active management such as coppicing or intensified foraging.

Coastal locations such as CP1 and CP16, on the other hand, exhibit rises in dwarf shrubs and ling heather (Calluna) that were converted primarily into grassland by the Late Bronze Age, with a simultaneous increase in cereals. This could point to a change in economic practice toward cereal cultivation from pastoralism, as higher grass pollen might indicate low grazing pressure, because grasses are able to flower and produce pollen. The low level of pastoral indicators in PG1 at around 1600 BCE, coinciding with a sudden increase in grassland, together with their generally inverse relationship, supports this interpretation (comparing Figures 5G and H). It is noteworthy that these southern- and easternmost coastal areas had relatively low initial tree cover (~20%), making these areas naturally more open and suited for more large-scale agro-pastoral pursuits. In other areas, the human activity created openings and managed woodlands but did not lead to wholesale, permanent deforestation.

Overall, the region remained fundamentally wooded throughout, and anthropogenic indicators, in particular cereal cultivation, typically occurred as pulses around a forest baseline that retained the ability to regenerate to full forest. Evidence of potential cereal cultivation first appears in the Middle Neolithic (c. 2500–2300 BCE) in deciduous forest settings (comparing Figures 5A–D and I). Considering that wild grasses can sometimes be misidentified as cereals due to overlap in size range both of the pollen grain itself and of the annulus, particularly between barley (Hordeum) and some wild grass species (e.g., Glyceria fluitans), the likelihood of these being true identifications is increased by the observation of various cereal pollen types (see SI 1). In some groups, especially in PG2 but also evident in PG1/CP1 and PG7, this interpretation is strengthened by a co-occurrence of disturbance weeds (comparing Figures 5F and I). Levels of cereal cultivation fluctuates through the Late Neolithic and Bronze Age, becoming more widespread from the mid to end of the Late Bronze Age (Figure 5I), especially in areas where tree cover had dropped below ~30%−40%.

5.2 Distribution and interpretation of the archaeological record

In the following sections, the archaeological material is presented under two principal categories: settlement remains and artifacts. These data are subsequently examined in relation to the vegetation reconstruction outlined above. Finally, evidence from plant macrofossils and faunal remains is introduced to assess and further refine the interpretations developed in the preceding sections.

5.2.1 Permanent and non-permanent settlement data

Permanent settlement sites are predominantly located around the Oslo fjord and the southern coast (Figure 6), largely corresponding with the extent of PG1, PG3-4 and PG6. Overall, such settlements were widely dispersed early in the study period, becoming more clustered toward the end of the Bronze Age (Figure 7), although they do not form hamlets (see also Sand-Eriksen, 2025). Moreover, buildings appear in previously unoccupied areas during the Late Bronze Age, suggesting an expansion of permanent settlements that may reflect broader changes in land use and economy (see also Sand-Eriksen, 2023).

Most of these permanent settlement sites contain post-built longhouses (n = 119), which in Scandinavian archaeology are among the main features for identifying the turn to a more farm-based society. However, some sites contain other types of buildings, which in addition to four- post buildings (n = 9), typically found connected to the houses and commonly interpreted as storage facilities (e.g., Løken, 1998), include the newly discovered categories of wall-ditch (n = 11) and trapezoidal “staveline” buildings (n = 5), which are interpreted as outbuildings, possibly connected to pastoral or other mobile occupational practices (Sand-Eriksen, 2025). In addition to their placement away from the farmstead and the houses, this interpretation is grounded in the building form as less suited to permanent residence, alongside historical analogs of buildings in the region. For example, longitudinal “staveline” constructed buildings have typically functioned as boat-houses or hay barns (Godal et al., 2018, p. 159). While some of the prehistoric examples, particularly around PG1, could with a higher shoreline be situated relatively near the contemporaneous sea level, none are shore-bound. Instead, they are located on or adjacent to alluvial plains that correspond to the extents of CP1, CP11–CP12, and CP16.

These are areas which generally have good pastoral potential, as outlined above, which considering that many show vegetational changes that correspond with the timeframe during which the buildings were in use, supports the hypothesis of a pastoral function. Local trends in the vegetation reconstruction indicate broader forest openings and an intensification of pastoral practices. For example, in CP11 a predominantly deciduous forest with substantial grassland, herbs, and shrubs—indicating a relatively open canopy—experiences a sudden decline in trees (especially hazel) and an increase in heathland around 1300–1100 BCE (Figures 5A, B, D and J). This aligns with the dating of wall-ditch houses in this area, from 1420 cal BCE (3565, 35; TUA-7581) to 830 cal BCE (2770, 35; TUA-7579), supporting the interpretation of these buildings as connected to pastoral activities. Following the decline at the end of the Early Bronze Age, there is a small regeneration of trees at the start of the Late Bronze Age, and tree cover increases again after the buildings fall out of use (Figures 5A, B and 8).

Non-permanent settlement traces, including rock shelters, have a broad distribution, occurring in all areas except PG7 (Figures 6 and 7). Within regions that also contain houses, non-permanent sites are often situated between such permanent settlements, attesting to a mobile occupational practice across the various forests. These non-permanent locations are typically distant from cultivation indicators (cultivated layers, n = 81; plow/ard marks, n = 9; cereal grains, n = 162; sickles, n = 321), which are closely associated with activities that take place near permanent farmsteads (see distribution in Figure 7). Considering the dated material, cultivation first increases from c. 1900 cal BCE before stabilizing, rising again around 850 cal BCE, and then showing a sharper increase from 400 cal BCE (Figures 8 and 9). The clustering of permanent settlements corresponds, both diachronically and spatially, with this intensified cultivation.

5.2.2 Artifacts and pitfall traps

Aspects of this land-use system are further manifest in the spatial distribution of lithic artifacts, so-called large tools primarily. Simple shaft-hole axes (n = 1,496) are the most abundant artifact type and have the widest distribution, occurring from the high mountains to the coast but with greatest density in the lowlands (Figure 6). This distribution has traditionally been interpreted as evidence for land-clearing related to agro-pastoral activities (e.g., Østmo, 1988). That interpretation is important in the light of the vegetation reconstructions, which do not show widespread deforestation—apart from around CP1. The broad, even distribution of axes is therefore more consistent with limited but ongoing forest or woodland management than with large-scale clearance, while localized clusters of axes coincide with known permanent settlements (Figures 6 and 7). A similar even spread of charcoal layers (n = 39), typically in between the farmsteads in the lowlands, could indicate that burning formed part of this management. Considering that tree cover declines slightly in areas where houses appear—such as in PG6/CP12 between 2300–2100 BCE vs. 2100–1900 BCE (Figures 3A, B), the time window when the first two-aisled post-built houses are recorded in that area—the co-occurrence of axe clusters with known houses could reflect the establishment of farmsteads, including smaller forest clearances for cultivation (Figure 8).

That hunting remained an integral part of the economy throughout the period and across the study region is indicated by the wide distribution of arrowheads (n = 892) and pitfall traps. Pitfall traps, particularly common inland and in higher mountainous areas (notably PG7 and PG8, partly PG5-6), point to a specialized, trapping-based strategy in open to semi-open, higher-altitude mixed forests where pine (Pinus) is more abundant; these habitats favor large game such as elk, with reindeer occurring still higher up.

Arrowheads are widely distributed, including in lower-lying areas, with more semi-open deciduous forest (corresponding to PG1-4), likely reflecting more active hunting of smaller game, which based on the even temporal spread must have continued throughout the Bronze Age (Figure 8). The connection between forest type and prey is illustrated by a bone from small game, suggested to be either hare, fox or goose, dated to 1420–1265 cal BCE from CP3 (Fossum, 2017). This underscores how different woodland and forest assemblages supported different hunting practices and target species. Against this, it is worth noting that blueberries are argued to be the most important species in the Norwegian forest, functioning as a keystone resource in multiple food chains and influencing animal distributions (Larsson and Dalen, 2018, pp. 43–44). Since blueberries cannot be directly identified in the REVEALS vegetation reconstruction, their likely presence must be inferred.

Empirical relationships suggest that when Ericales makes up ~40%−45% of the vegetation alongside ~50% mixed dominance of pine (Pinus) and/or birch (Betula), blueberries typically account for about 15% of the heathland vegetation. Blueberry forests are also characterized by associated taxa such as juniper (Juniperus), as well as grass-like species and certain herbs belonging to the aster family (Larsson, 2000, pp. 5, 7; Larsson and Dalen, 2018, p. 53; Larsson and Søgnen, 2003, pp. 93–94). This assemblage is most common outside the areas with permanent settlements, notably CP7-8 and CP14, and partly CP13, providing conditions ideal for elk, while CP9-10 and CP15 are opener landscapes and hence more suited for reindeer. Around permanent settlements, in PG3 and around CP4 especially, PG4 in general, and PG6 just east of CP12, the forest is generally more deciduous and support a grassland understory—conditions that, when maintained, rather favor pastoral practices (Figure 10), as outlined above, and hunting of smaller game. Taken together, different forest types would have supported distinct suites of wild resources, and therefore different hunting, foraging and herding strategies that can be further investigated using the plant macrofossil and animal bone records.

5.2.3 Substantiating interpretation through plant macrofossils and faunal remains

Plant macrofossil and faunal evidence (see SI 2) indicates a broad utilization of wild plants and animals associated with a semi-open forest environment, as described above, confirming the high foraging and harvesting potential identified from the vegetation reconstruction.

Nutshells, for example, mostly hazel (Corylus avellana), which has increased yields under open conditions (Nedkvitne and Gjerdåker, 1999), make up 23% of the plant remains. Historical sources indicate that access to productive hazel stands could provide important supplementary income: farms with good hazel woodland could collect and sell some 300–400 kg of hazelnuts per year (Nedkvitne and Gjerdåker, 1999, p. 95), in addition to what they consumed themselves. The distribution of hazel nutshell and hazel groves often align with farmsteads. While no post-built houses have been identified within CP11, there is as mentioned evidence of a wall-ditch building, a lighter, less permanent form of construction interpreted as serving economic functions, such as temporary shelter for people and animals or storage of equipment and fodder (Sand-Eriksen, 2025, p. 217). Although it is possible that people traveled from hitherto unrecognized farmsteads nearby to harvest nuts or branches for wickerwork and other wooden artifacts (see Nedkvitne and Gjerdåker, 1999, pp. 109–130), it is possible that CP11 was encompassed within a larger mobility scheme, most likely including hunting and pastoral activities.

Because of its versatility, hazel was highly valued and generally not used as a primary fodder tree; however, its nuts were occasionally used to feed pigs (Nedkvitne and Gjerdåker, 1999, pp. 92–93), and hazel woodlands could have been valuable for pannage or hunting of wild boars. Pigs (Sus scrofa) account for about 6.5% of the faunal assemblage, with their highest concentration in CP4, an area with extensive hazel stands. However, because distinguishing domestic pigs from wild boar is challenging in fragmented osteological material (Rowley-Conwy et al., 2012), these remains could represent either domesticated animals or hunted wild boar; the present data therefore do not allow firm conclusions to be drawn about husbandry or hunting practices.

Of other domesticated animals, sheep/goat (Ovis aries/Capra hircus) remains are the most common, making up about 24% of the faunal assemblage. It also shows the widest distribution, even appearing in the highlands (PG8), possibly reflecting the aforementioned Late Bronze Age expansion, as further supported by the date of the bones (2440 ± 30 BP, Beta-345280). Their remains are primarily concentrated in PG4 and partly PG3, a pattern mirrored by cattle (Bos taurus), comprising 19%, which together indicate a pastoral economy operating in the forest and maintained landscape surrounding farmsteads.

Remains of berries are found from the coast to the high mountains, with remains of raspberries (Rubus idaeus) being the most common, with lower amounts of blackberries (Rubus fruticosus agg.), sloe (Prunus spinosa), bearberries (Arctostaphylos uva-ursi), crowberries (Empetrum nigrum) and Vaccinium, which comprises amongst others blueberries and lingonberries. Other plants like sorrel (Rumex acetosa), nettle (Urtica), buttercup (Ranunculus), bog-myrtle (Myrica gale), saltbush (Atriplex) and pyramidal bugle (Ajuga pyramidalis) are concentrated in excavations corresponding with the diverse forests of PG4 and PG6, with especially CP11 having high levels of other herbs in the vegetation reconstructions. The presence of bark from unspecified trees as well as seeds from juniper (Juniperus) from the farm site Opstad Vest (Munch Havstein and Danielsen, 2024) points to a wider use of non-timber forest products. The mentioned species underlines the centrality of the forest to human subsistence. Had people relied solely on the agrarian-based diet typically reflected in the archaeological record, they would likely have been at risk of vitamin A and C deficiencies (van Amerongen, 2016, pp. 266–267).

Faunal remains also show a targeted exploitation of different forest: reindeer (Rangifer tarandus), comprising about 5% of the material, were hunted in the mountainous inland region (PG8 and northernmost part of PG5), while beaver (Castor fiber), about 6.5% of the material, could have been trapped/hunted in forested wetlands. The distribution of beaver remains is typically located near settlements. Comprising about 9% and 8% respectively, fish (Pisces) and birds (Aves) were found more widely, from mountains to coast, while marine mammals, such as seal (Phoca vitulina) and whales (Cetaceae), were exploited near the coast, comprising 4% of the material.

From an excavation in the coastal parts of PG4 on the eastern side of the Oslo fjord, there are also a few examples of flax (Linum usitatissimum) and instances of millet (Panicum miliaceum), which is among the first cultivated grasses, although not commonly found in Norway (Bårdseth et al., 2007, pp. 112, 117–118; Bårdseth and Utigård Sandvik, 2010). These species are typically underrepresented in pollen records, and it is worth noting that millet is almost impossible to distinguish from wheat pollen (Bårdseth and Utigård Sandvik, 2010, p. 3292; with further reference).

Cereals are represented by 162 directly dated grains from 51 different sites (data from Solheim, 2021b), which are here aligned with information on indirectly dated grains, that is, undated grains that can be set in relation with a dated context or site. Most of the cereal finds are located in the area corresponding to PG4, with the eastern side of the fjord showing the highest concentration, accounting for 58% of all listed cereal grains. In comparison, the western and northern sides of the Oslo Fjord each comprise 18%, with only 6% found in the southern part of the southeastern Norwegian region. This pattern does not match that of the cereal pollen-based land-cover, which is highest in the southern region.

In terms of temporal distribution, about 63% of the cereal grains belong to the Late Bronze Age, which corresponds well with the vegetation reconstruction (Figure 5I). The remaining cereal grains dates to the Early Bronze Age and the Late Neolithic, although a wheat (Triticum dicoccum) cereal grain (4351 ± 55 BP, Ua-52926) and a barley (Hordeum vulgare var. nudum) cereal grain (4551 ± 55 BP, Ua-52925) have been dated to the Middle Neolithic, of which the latter falls just outside the timeframe of this study. These grains were among 42 grains, including 13 grain fragments, found at site Kvastad A2, of which four others date to the Late Neolithic while the remaining are undated (Stokke and Bjørkli, 2016). This site is located between PG2 and PG3, and there is therefore a broad spatial overlap between these grains and the early cereal cultivation evident from the vegetation reconstruction. This may suggest that even the earliest dated cereal grains derive from local cultivation rather than import, although more detailed studies are needed to confirm this. In terms of type distribution, they largely corresponds with the proportions of the types found in the land-cover reconstructions, with barley (Hordeum) dominating. However, oat (Avena) is higher in the latter compared to the dated grains, which could reflect a presence of wild oat amongst the counted cereal pollen.

6 Discussion

The results challenge the dominant narrative of a uniform agrarian expansion and deforestation in Bronze Age Scandinavia. Instead, they point to a distinct “forested form of life” in southeast Norway, where communities pursued subsistence strategies in coexistence with, rather than displace their persistently wooded surroundings.

6.1 Forest management, not clearance

The archaeological and environmental data and results presented in this paper suggest that southeast Norwegian Bronze Age communities engaged in niche construction, actively managing the forest to maintain specific affordances rather than destroying the forest biome entirely. A central element of this forested life was the active management of woodlands, a practice that fundamentally differs from the large-scale deforestation narratives common for southern Scandinavia. Here, we reframe these interactions as a form of feedback loop between cultural and natural systems.

The direct spatial correlation of axes and settlement sites with areas showing declines in tree cover suggests clearing for farmsteads, likely for construction of buildings and establishment of cultivated plots. Their wider distribution could indicate the maintenance of woodland pastures in the surrounding forest, while a correlation between certain axe types and rivers could point to timber harvest and extraction through log driving (Sand-Eriksen, 2026). The practice of thinning woodlands, making the land suitable for grazing livestock without removing the forest itself (Figure 10), is well-known from historical sources (Dahlström, 2006, p. 20) and is a prime example of humans co-shaping their niche to enhance pastoral affordances without deforestation. Considering the vegetation record and how grass pollen may increase when grazing pressure is low, because grasses can flower and therefore produce pollen, the apparent marked rise in grassland during the Late Bronze Age and especially toward the Pre-Roman Iron Age transition may reflect a broader economic shift. Specifically, the pattern is consistent with a move away from predominantly pastoral and mobile subsistence strategies in the Late Neolithic and earliest Bronze Age toward increased cereal cultivation, possibly driven by population growth (Solheim, 2024). This interpretation is supported by both the pastoral indicators and the cereal pollen signal.

The high prevalence of hazel in the charcoal record (Mjærum et al., 2022) has previously been suggested as a form of “arboriculture” (see also Sand-Eriksen, 2023), which our results allow to be framed more explicitly as niche construction. In several places naturally abundant hazel appears to have been actively manipulated and maintained because of its multiple affordances, with communities promoting and shaping the forest structure to ensure a steady supply of nuts, wickerwork materials, and fodder. Specifically, this represents a (counteractive) feedback loop: while cooling climatic trends would have naturally created a decline in thermophilic plant species like hazel from 2200 to 200 cal BCE (e.g., ter Schure et al., 2021; Warden et al., 2017), human cultural agency, through coppicing and hence canopy thinning, altered the selective environment. This intervention created a reciprocal relationship where the modified landscape then reinforced the social and economic practices required to maintain it (Burke et al., 2025; see also Donges et al., 2017), effectively preserving the forest's ‘biocultural' diversity despite external climate pressures.

Where hazel coverage fluctuates, as seen in especially CP3–4 and CP11, this may indicate coppicing. It is important to note, however, that coppice cycles are necessarily much shorter than the analyzed 200-year time windows, and that the data presented in Figure 5J only represent some potential overarching trends. This human-hazel relationship would also involve other species. Given that light is among the primary drivers of pollen production, an understory species like hazel would depend on the management of major canopy trees, for instance, through coppicing, which suppresses their flowering and thereby allows increased light to reach the understory (Waller et al., 2012). Such deliberate tending represents a deeper, more entangled form of forest management in which people actively structured species composition and canopy openness to sustain a range of economic and ecological benefits.

6.2 A mixed and mobile forest-adapted economy

The results show that the subsistence strategy was not a transition toward intensive agriculture but a resilient adaptation to the forest's inherent affordances. Unlike the trajectory in Jutland, where early lower tree density (e.g., Roberts et al., 2018) afforded a faster and apparently linear path to open pastures, the denser forests of southeast Norway favored a trajectory of mobility and subsistence diversification, following the affordances of the forest.

Building on van Amerongen's (2016) emphasis on wild plants as an essential nutritional supplement that helped prevent dietary shortfalls, our findings broaden this to a forested livelihood marked by communities who actively shaped and co-created woodland composition and openness to secure resources rather than converting the landscape to permanent arable land. Hunting, foraging and other woodland practices that stretch back to the Mesolithic persisted alongside farming from the Neolithic onward; agrarian activities from c. 2350 BCE were therefore added to, rather than replacing wholesale, existing forest-based lifeways. This is reflected in the maintenance of woodlands rather than large-scale clearance, and in a pattern of small-scale cultivation immediately around farmsteads, which were increasingly opened and intensified especially in the later Late Bronze Age as cereal cultivation rose. The non-permanent settlements identified in the “off-site” data represent the physical manifestation of this mobility. These sites functioned as camps operating as components within logistical systems (e.g., Binford, 1980), used for herding, hunting, and harvesting, allowing the community to exploit a wide range of forest types. The macrofossil record proves that these populations were not “plant blind” but possessed an intimate knowledge of the forest's seasonal affordances, as well as different affordances pertaining to different kinds of woodland, such as the hazel groves found in CP4 and CP11.

That hunting remained an integral part of the economy is clearly reflected in the presented archaeological record and may also be mirrored in vegetation change. The Late Neolithic decline in tree cover observed in PG7–8 (Figure 3A) coincides with the first intensification of pitfalls (Post-Melby and Bergstøl, 2020, Figure 18.5), and may therefore reflect the activity of pitfall construction (Figure 11). Pitfall systems often spanned several kilometers and comprised tens or even hundreds of pits. The temporal pattern of pitfall activity, with an initial intensification c. 2500–1700 BCE, a modest stabilization thereafter, and a marked increase from c. 1200–700 BCE, corresponds with declines in forest observed in PG7 (Figures 3B, C), indicating an expansion of hunting and trapping activity (Figure 8).

Against this, the emergence and intensification of pitfall traps does not appear to reflect the assertation of territorial claims and rights to large game in response to expanding agrarian practices, as previously argued (Post-Melby and Bergstøl, 2020, p. 320). Rather, instead of merely being holes in the ground awaiting chance encounters, (Rimpi 2024, p. 6) argues that humans exploited predictable movements of herd animals. By constructing pitfalls and managing surrounding vegetation, people reshaped local landscape dynamics so that the physical layout directed prey movement and increased hunting success. The construction of the pits, their associated fences (see e.g., Jakobsen, 1989), and their upkeep, including regular maintenance of cover material and of surrounding vegetation to funnel animals from migration routes into the traps, would have created and maintained more open local conditions and contributed to reduced tree cover. Although more subtle than agrarian land modification, this illustrates that inland environments are not “just” wilderness but, like the Amazon example in the introduction, reflect co-evolution between human cultures and natural ecosystems over thousands of years. Ultimately, this illustrates a large-scale environmental feedback loop where cultural behaviors (trapping and vegetation management) influenced the spatial structure and connectivity of the environment. In turn, these environmental shifts likely dictated human population density and mobility, as higher encounter rates with prey supported the social networks necessary for the long-term persistence of this ‘forested' livelihood.

7 Conclusion

By shifting the perspective from clearance to coexistence, this study identifies a “forested form of life”, a successful, long-term adaptation in Late Neolithic and Bronze Age southeast Norway with roots in earlier periods and likely persistence into later times, although the exact chronology and form of this requires further research. These communities were not passive woodland dwellers but active niche constructors who managed forest composition and openness to sustain a diverse array of economic and ecological affordances. Their mobile, mixed economy integrated settled farmsteads with logistical exploitation of the wider landscape, showing that non-agrarian land-use practices, often archaeologically subtle and therefore requiring interdisciplinary approaches, were central to regional subsistence.

Through a framework that recognizes the deep integration of natural and cultural systems, this study contributes to current debates in environmental archaeology, viewing landscapes not as passive backdrops but as co-created environments shaped by long-term human agency. The divergent trajectory of southeast Norway compared to other parts of southern Scandinavia is not a sign of “backwardness” but a reflection of how prehistoric communities attuned their “form of life” to the specific ecological possibilities of their landscape. This underscores a resilient society in which a forest-dominated environment provided a sustainable and complex foundation for Bronze Age life rather than merely an obstacle to be overcome.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fearc.2026.1794721/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Debora Zurro, Spanish National Research Council (CSIC), Spain

Reviewed by: Mara Weinelt, University of Kiel, Germany

Svein Vatsvåg Nielsen, Stavanger Maritime Museum, Norway