Introduction

Pollen and plant macrofossils have been used both separately and together to assess past agricultural practices such as grazing and cultivation, to identify the type of crop plants cultivated and to separate between permanent and shifting cultivation (Viklund 1998; Jones 2002; Kreuz and Schäfer 2011; Halvorsen and Hjelle 2017). Based on the demands of the different arable weed taxa identified, inferences regarding agricultural regime, nutrition levels and fertilisation of fields can be drawn and contribute to archaeological discussions on prehistoric farming societies. With stable isotope analysis, further insights into past land-use practices can be made. Stable isotope values of nitrogen in crops from archaeological sites can provide information about nutrition levels and intensity of manuring and thus on prehistoric food-producing systems (e.g. Bogaard et al. 2007, 2016; Fraser et al. 2011; Kanstrup et al. 2014; Gron et al. 2017). Interpretations of the intensity of manuring are based on measurements of modern material. Nitrogen isotope values (δ15N) represent the difference in ratio between the heavier and lighter stable isotopes of nitrogen in samples compared to N2 in air (e.g. Coplen 2011). Generally, soil nitrogen is the main source of nitrogen in plants, except for plants with nitrogen-fixing capabilities such as legumes, which have microbial bacteria that can convert atmospheric N2 into ammonia (Virginia and Delwiche 1982; DeNiro 1987; Styring et al. 2014). Studies have shown how manuring can significantly increase δ15N values of crops by up to 10‰ (Bogaard et al. 2007; Fraser et al. 2011; Styring et al. 2017). This occurs due to bacterial decomposition of the manure, which leads to preferential loss of the lighter 14N, leading to enrichment of 15N within the N source pool of the soil available to plants (Fraser et al. 2011). Other processes, such as salt spray and burning of vegetation, may also increase the δ15N values of crops and vegetation (Handley et al. 1999; Szpak 2014).

Modern manuring experiments in northern and middle Europe have shown that δ15N values above 6‰ in cereals indicate high levels of manure or repeated manuring. Values below 2.5–3‰ δ15N indicate no or little manuring, or residual effects of previous manuring (Fraser et al. 2011; Bogaard et al. 2013). Stable isotope measurements of bone material from wild herbivores can give an indication of natural plant δ15N values (Styring et al. 2016). Natural vegetation has δ15N values around 0‰ (close to atmospheric N2), whereas herbivores have values of 3–5‰ due to trophic enrichment. Red deer bones (13 samples) from the Mesolithic to the Iron Age from western Norway show mean stable δ15N values of 3.8 ± 0.5‰ and 3.9 ± 1.0‰ (before and after the transition to agriculture) (Rosvold et al. 2010), indicating that the natural vegetation had δ15N values close to atmospheric N2. This is also supported by comparable δ15N values of bones of wild boar from the same periods (Rosvold et al. 2010), and bones of sheep/goats and cattle from the Middle Ages (MA) in Norway (Naumann et al. 2014; van der Sluis et al. 2016).

Soil profiles with agricultural/cultural layers and postholes of prehistoric houses are the main sources of cereal assemblages in western Norway (Prøsch-Danielsen and Soltvedt 2011; Hjelle et al. 2016; Halvorsen and Hjelle 2017). Votive offerings and cereal storages with pure cereal grain assemblages occur but are rare (Prøsch-Danielsen and Soltvedt 2011). The bedrock is generally acidic (Moen et al. 1999), which leads to poor preservation of uncharred organic material. In addition, the topography restricts areas available for habitation and agriculture to the area between sea-level and the mountains, which may be a quite narrow area of land. Thus, the same areas have been used throughout time (cf. Diinhoff 2004) and traces of prehistoric activity might be disturbed or destroyed by activities such as ploughing in more recent time-periods. Based on cereals and weeds in pollen and macrofossil assemblages from soil profiles as well as macrofossil assemblages from house remains, fertilisation of fields from the Late Neolithic/Early Bronze Age (LN/EBA; see Table 1 for time-periods) has been proposed, with increased intensity from the Pre-Roman Iron Age (PRIA) (Halvorsen and Hjelle 2017). In the present paper we attempt to substantiate these interpretations through stable isotope data of cereals from some of the same contexts.

Table 1 Time-periods and abbreviations used in the text
Full size table

The main cereal type found in prehistoric contexts in the study area is naked barley (Hordeum vulgare var. nudum, hereafter called H. nudum). Hulled barley (Hordeum vulgare var. vulgare, hereafter called H. vulgare) is found from the PRIA, and oat (Avena) from the Iron Age. Compared to naked barley, hulled barley is considered to be better suited for an agricultural regime that includes high levels of manuring as it has higher requirements for fertilisation to give a reasonable yield (Gustafsson 1998; Viklund 1998). Thus, this grain type is thought to follow the introduction of three-aisled buildings with a byre section where the animals were stalled during winter and manure was easily collected (Olausson 1998). In western Norway, three-aisled houses became common in the Early Bronze Age (Diinhoff and Slinning 2013; Olsen 2013). Household waste and not manure may have been the primary source of fertilisation in the LN-EBA, whereas intensive, full-scale agriculture with permanent fields and manuring started towards the end of the Bronze Age in southern Scandinavia (Grabowski 2011; Kanstrup et al. 2011) and probably also in our study area. In addition to a higher yield when manured compared to H. nudum, H. vulgare is more resistant to attack by pathogens (Bakkevig 1992).

Together with δ15N, carbon isotope values (δ13C) are commonly measured and included also in our study. The δ13C values have been used to identify irrigation levels in arid areas (e.g. Araus et al. 1997; Ferrio et al. 2005; Bogaard et al. 2013; Wallace et al. 2013). Although water availability is generally not a limitation, dry summers occur also in western Norway, and measurements of δ13C may potentially inform on climate conditions/watering practices in prehistory.

Since the 1950s, there has been a politically founded incentive to channel cereal production to areas in eastern and middle Norway that have the best suited soil, climate, and topographic conditions for cultivation. Western and northern Norway were considered better suited for livestock and grass production for fodder, which now represent ca. 80% and 15% respectively of the agricultural production in these areas (Rognstad et al. 2016). In recent years, a few farms have started low-scale cereal production in western Norway, mostly for fodder, but also supplying local breweries.

Through measuring δ15N and δ13C values of modern cereals with known manuring regimes, we aim to obtain reference data regarding δ15N and δ13C values for western Norway. The resulting values are discussed in relation to the stable isotope values of fossil cereals from different time-periods from archaeological contexts, with the aim of making inferences regarding manuring and land-use practices in early management systems.

Material and methods

Study area and material

The study area is situated between 59° and 63°N and 005° and 009°E in western Norway. Charred cereal grains dated from the LN to the MA form the basis of our study (Fig. 1, Table 2). The analysis is destructive, and only sites with enough cereal grains to retain grains for future research were selected. Following this, 16 archaeological sites with cereal assemblages were selected from the database at the University Museum of Bergen. From the archaeological sites, cereals were retrieved from house contexts (cultural layers from MA, otherwise postholes) or agricultural/cultural layers in soil profiles, the latter referred to in the following simply as soil profiles (Table 2). From the sampled houses, grains from two contexts were analysed if possible. From the soil profiles, two cereal grains were selected from each context/layer if possible. The number of cereals analysed from each site is given in Table 2. Both soil profiles and house contexts are nearly equally represented in the data covering the LN/EBA and VP/MA, whereas only one soil profile and several house contexts are included in the data covering the LBA and PRIA. The medieval period is poorly represented in rural areas of western Norway, and samples from the medieval town of Bergen were included to increase the amount of data from this period. We assumed these cereals were grown in manured fields and expected the δ15N values to reflect this. However, these grains have probably been imported to Bergen, and will not represent local cereal cultivation.

Fig. 1
figure 1

Map showing the geographical setting of the sites included in the paper. Numbering of the sites follows information given in Table 2. The modern cereal test-sites are marked M = Mogstad and H = Havrå

Full size image
Table 2 Overview of the study sites: site numbers correspond with those in Fig. 1
Full size table

The modern cereals of our study were grown in test-plots in two areas in western Norway (Fig. 1). The site Mogstad in Surnadal (ecological farming using only cattle manure) was maintained by NIBIO (Norwegian Institute of Bioeconomy research) in collaboration with a local farmer (Solemdal et al. 2021). The site had previously been used for conventional cereal farming for ca. 20 years and started ecological farming in 2019. The second site, Havrå in Osterøy, is in the south-western part of the study area. Havrå is a traditional cluster farm and a protected cultural environment managed by Osterøy Museum and Stiftinga Havråtunet. Only cattle manure is used at the site. Palaeobotanical and archaeological studies at Havrå have shown activity from the LN/EBA transition with increased farming during the Iron Age (Hjelle 1999; Øye et al. 2002), and written sources mention the farm in 1303 ce (Skre 1994). The farm was traditionally run until the 1940s, followed by a gradual decline in activity with cereal farming ceasing in 1950 (Austad et al. 2012). Since 1989 a small area of the farm is run after traditional practices and cereal farming (Avena) resumed in 1995 (Jensen et al. 2012). An overview of cereal varieties and the amount of manure used at these two sites is given in Table 3. The modern cereal grains from the northern site were collected in the test-plots in September 2020, whereas grains from an already harvested field were taken at Havrå, also in September 2020.

Table 3 Modern cereal varieties and the manuring regimes of the experimental plots
Full size table

In western Norway, the main climatic gradient is from west to east rather than from south to north. Most study sites have an oceanic climate (precipitation > 1,200 mm per year, mean annual temperature 6.1–7.7 °C, Tables 2, 3). A few sites in the inner fjord areas have a weakly oceanic climate (precipitation 500–1,200 mm per year, mean annual temperature 5.7–6.3 °C) (Moen and Odland 1993), thus all sites are within zone Cfb (temperate oceanic climate) of the Köppen-Geiger climate classification (Peel et al. 2007; Beck et al. 2018). The vegetation is characterised by mixed deciduous and pine forests, spruce plantations, coastal heathlands, and open grass-dominated agricultural land. The soils in the coastal areas are marine beach deposits or of morainic origin, whilst sites in the inner fjord areas have (glacio-)fluvial deposits.

Laboratory methods

Single-grain analysis of 76 charred fossil cereal grains were carried out (Table 2, ESM 1 Table 1.3). Half of the grains from each context were subjected to ABA (Acid–Base-Acid) pre-treatment following the protocol described in Fraser et al. (2013b), the remaining grains were analysed without pre-treatment (Fraser et al. 2013a), and any visible contaminants were removed before analysis. Twenty-two grains (two single grains and two assemblages of 10 grains) from each of the eight modern cereal types were selected for analysis (ESM 1 Table 1.4). The modern cereals were wrapped in aluminium foil and charred at 245 °C for 4 h in sand-filled crucibles (following Fraser et al. 2013a; Nitsch et al. 2015). Initial tests showed this to be the appropriate temperature for achieving complete charring with little distortion in shape.

The measurements for δ15N and δ13C were done at the Department of Geosciences at the University of Bergen using a Thermo Scientific Flash 1112 Elemental Analyser connected to a Delta V plus isotope ratio mass spectrometer (IRMS). Homogenised samples were weighed into tin capsules (0.4–1.4 mg of crushed fossil cereal grain, modern grains 1.2–3.8 mg), combusted at 1,020 °C, catalysed by chromium oxide and silvered cobaltous oxide (Elemental Microanalysis, UK) to form CO2 and N-gasses, before N-oxides were reduced to N at 650 °C using metallic copper (Elemental Microanalysis, UK). Water was removed and gasses were separated by a GC column before entering the IRMS. A cut-off value of 1,000 mV in the amplitude of mass 28 (14N2) was implemented as lower values were considered to give unreliable readings. Multiple measurements were performed for samples with sufficient material. A full list of number of measurements is given in ESM 1.

IAEA-N1 (ammonium sulphate, δ15N =  + 0.43‰), IAEA-N2 (ammonium sulphate, δ15N =  + 20.41‰) and IAEA-600 (caffeine, δ15N =  + 1.0‰) were used to calibrate the N-measurements, and USGS-24 (graphite, δ13C = − 16.05‰), IAEA-CH6 (sucrose, δ13C = − 10.45‰) and IAEA-600 (caffeine, δ13C = − 27.77‰) to calibrate the C-measurements. A House Glycine standard (mean and SD measured over eight runs, δ15N = 0.98 ± 0.21‰ and δ13C = − 29.87 ± 0.37‰) was used as a check standard. See ESM 1 Tables 1.1–1.5 for calibration results. A correction factor of 0.31‰ was applied to all measured δ15N values to account for charring following the recommendations of Nitsch et al. (2015).

The Δ13C (carbon discrimination) of cereal grains from different time-periods (until 2010) can be assessed using data from the AIRCO2_LOESS data calibrator (Ferrio et al. 2005). The calibrator accounts for past variations in δ13C in the atmosphere and estimates the isotopic composition of CO2 in the atmosphere at the time the plants were growing (age of the sample). As the AIRCO2_LOESS data calibrator only supports data for time-periods until 2010, the Δ13C values for the modern cereals were calculated using the isocalcR package in R (Mathias and Hudiburg 2022).

Data analysis

The δ15N values of the pre-treated fossil cereals were compared to the non-treated using a pairwise t-test to check for differences. In addition, the δ15N values of modern single-grain and 10-grain bulk samples were compared for differences using the same test. The δ15N values of the single fossil grains were tested for correlation with age using linear regression. All tests were run in R (R Core Team 2015), code and results are given in ESM 2.

Results

Fossil cereals

The t-test of the two treatments of fossil grain δ15N values show a p-value of 0.93 (see ESM 2 for results), thus there is no statistically significant difference between the pre-treated and the non-treated grains, and the treatment status is not shown in the figures. A full list of cereals and pre-treatment is given in ESM 1 Table 1.3.

The results show variations in fossil grain δ15N values both within sites and from site to site for both cereal types (Figs. 2, 3, ESM 1 Table 1.3). For Hordeum, there is a slight increase in δ15N values through time and there is a statistically significant correlation between age and δ15N values (correlation = 0.2897446, p-value = 0.02737, r2 = 0.08395), also indicated in the boxplot (Fig. 4). Generally, the δ15N values of cereals from the LN-EBA sites are above 3‰, but for five sites some of the δ15N values are below this threshold (range = 1.21–8.93, see Table 4). Two sites dated to LBA-PRIA have cereals with δ15N values below 3‰, and five above 6‰. Three of the seven sites dated to LBA-PRIA, have measurements of both H. nudum and H. vulgare. At one site, the δ15N value of H. nudum (13.6‰ δ15N) is much higher than those of H. vulgare (around 6‰). From VP-MA there are measurements for both Avena and H. vulgare, and the range in δ15N values are high for both cereals. At one site Avena has δ15N values below 3‰, the other sites all show cereal δ15N values above 3‰. When comparing the threshold lines of manuring intensity after Fraser et al. (2011) and Bogaard et al. (2013), low manuring intensity is indicated for LN-EBA, medium intensity in LBA-PRIA and medium to high in VP-MA.

Fig. 2
figure 2

Plot showing the δ15N values of fossil Hordeum ordered by age with the modern results in the right-hand panel for comparison. Sites are placed along a relative time scale, oldest to the left, youngest to the right. Numbers refer to sample names given in Table 2. Bars represent 2σ standard deviations of the measurements for cereal grains with multiple measurements (see ESM 1 Table 1.3 for full list of measurements). Dashed lines represent the threshold levels of manuring intensity (high, medium and low/none) after Fraser et al. (2011) and Bogaard et al. (2013)

Full size image
Fig. 3
figure 3

Plot showing the δ15N values of fossil Avena ordered by age with the modern results in the right-hand panel for comparison. Sites are placed along a relative time scale, oldest to the left, youngest to the right. Numbers refer to sample names given in Table 2. Bars represent 2σ standard deviations of the measurements for cereal grains with multiple measurements (ESM 1 Table 1.3 for full list of measurements). Dashed lines represent the threshold levels of manuring intensity (high, medium and low/none) after Fraser et al. (2011) and Bogaard et al. (2013)

Full size image
Fig. 4
figure 4

Boxplot of δ15N values for fossil Hordeum grouped after time-periods. Bars represent the 2σ standard deviations of all measurements in each time group, dots are measurements that fall outside the normal distribution within each time-period. Time-period abbreviations from Table 1. Dashed lines represent the threshold levels of manuring intensity (high, medium and low/none) after Fraser et al. (2011) and Bogaard et al. (2013)

Full size image
Table 4 Mean, range, and variance in δ15N values (in ‰) for the different cereal types through time. See Table 1 for time-periods
Full size table

The investigated sites are located at different distances to the shoreline, and a tendency of higher δ15N values at sites within 300 m distance compared to sites at longer distances is indicated in Fig. 5.

Fig. 5
figure 5

Boxplot of δ15N values for fossil Hordeum grouped after time-period and distance to the sea. Bars represent the 2σ standard deviations of all measurements in each time group, dots are measurements that fall outside the normal distribution within each time-period. Time-period abbreviations from Table 1. Dashed lines represent the threshold levels of manuring intensity (high, medium and low/none) after Fraser et al. (2011) and Bogaard et al. (2013)

Full size image

The Δ13C results (Figs. 6, 7) show that the cereals from most sites have Δ13C values higher than 17.5‰ in all time-periods, but with large variations. At two sites in LBA-PRIA, H. nudum has Δ13C values below 17.5‰, and four sites have Δ13C values between 17.5 and 18.5‰. In VP-MA, four sites have Hordeum Δ13C values between 17.5 and 18.5‰, one site has Avena Δ13C values close to 17‰ and one site has Avena Δ13C values between 17.5 and 18.5‰. For Hordeum, a decreasing trend from LN-EBA to LBA-PRIA, and then stable, or increasing values in VP-MA is indicated. There is a weak correlation between Δ13C values and age for the complete dataset (using only Hordeum), and age does not seem to explain variations in Δ13C values well. However, when looking at the period from LN-EBA to PRIA and VP-MA separately, there is a good correlation and age seems to explain the changes in Δ13C values well. All models are statistically significant, for results see ESM 2.

Fig. 6
figure 6

Plot showing the estimated Δ13C values from the δ13C-measurements for Hordeum. Sites are placed along a relative time scale, oldest to the left, youngest to the right. Numbers refer to sample names given in Table 2. Bars represent 2σ standard deviations of the measurements for cereal grains with multiple measurements (see ESM 1 Table 1.3 for full list of measurements). Dashed horizontal lines represents threshold level indicating good and dry growing conditions for barley after Wallace et al. (2013)

Full size image
Fig. 7
figure 7

Plot showing the estimated Δ13C values from the δ13C-measurements for Avena. Sites are placed along a relative time scale, oldest to the left, youngest to the right. Numbers refer to sample names given in Table 2. Bars represent 2σ standard deviations of the measurements for cereal grains with multiple measurements (see ESM 1 Table 1.3 for full list of measurements). Dashed horizontal lines represents threshold level indicating good and dry growing conditions for barley after Wallace et al. (2013)

Full size image

Modern cereals

The results of the t-test of the δ15N values from the single-grain and 10-grain bulk sample measurements of the modern cereal types and varieties, show that there is no statistically significant difference (see ESM 2). Thus, the results were pooled for the different cereal varieties in Figs. 2, 5. The modern Hordeum δ15N values from Mogstad are close to 3‰ δ15N, and the δ15N value of Avena is below 3‰. At Havrå (only Avena) the values are slightly higher than 6‰ δ15N. A full list of measurements of the modern cereals is given in ESM 1 Table 1.5.

Discussion

Limitations of the data

The number of fossil cereal grains from each site is limited, and bulk analysis using assemblages of 10 or more grains was not possible. Thus, single-grain analysis was chosen. Measuring single grains will emphasise the difference between the grains and can indicate the probability of grains from the same context coming from different fields or types of fertilisation regime. Previous work has pointed out potential drawbacks with measuring single grains, as grains from different parts of a spikelet can show large differences in δ15N values (Bogaard et al. 2007; Heaton et al. 2009; Nitsch et al. 2015). There may also be differences between plants, as well as variations within the cereal assemblage (e.g. Kanstrup et al. 2014). Nevertheless, recent studies have shown that single-grain analysis can give an indication of levels/intensity of manuring if present (Kanstrup et al. 2014; Gron et al. 2017; Treasure et al. 2019). We assume that this is the case for our study. In addition, this could give important information regarding agricultural practices that would be lost when analysing cereal assemblages of several grains. The variability in the results of our study may be related to differences in pedological settings between sites or in age (Fraser et al. 2013a). The observed variation is probably real in the sense that some cereal plants may have received more fertiliser than others or were affected by other environmental factors, and it highlights the variability of the material (see discussion below). Following this, we think the single grain analysis is important to understand the variation in datasets that is made invisible in mean values of larger samples.

The two modern test sites differ regarding manuring levels; at Mogstad 20t/ha was added in 2020, whereas at Havrå > 70t/ha was added (Table 3). The cereals from the modern experiments are from just one growing season (2020), which in Norway was a challenging year with a chilly spring, early summer drought and a wet late summer and autumn. At Mogstad, the weather conditions with early summer drought may have caused a low effect of the manuring at this site, especially for Hordeum, as high enough manuring levels early in the growing season is important for Hordeum in ecological farming (Solemdal et al. 2021). Thus, the δ15N values from Mogstad might be at the lower end of what would be expected of an ecological farming regime. However, the two test-sites were probably equally affected by the weather conditions and the relationships between the two management regimes as captured by our results are probably reliable. The high rainfall in the later part of 2020 may have led to lower δ15N values (Handley et al. 1999; Szpak 2014) and could also be reflected in the high Δ13C values for Hordeum at Mogstad. The manuring level at our test site Mogstad is comparable to sites with low annual levels (15–20 t/ha of manure) of manuring and the manuring level at Havrå is well above the high levels of more than 35 t/ha per year in test plots of Hordeum from other regions (Fraser et al. 2011; Bogaard et al. 2013). There are hardly any studies involving Avena in manuring experiments, but Kolmanič et al. (2022) show how application of 30 t/ha of cattle manure every 3 year in crop rotation leads to δ15N values in Avena of around 6‰ (ranging from ca. 4.8–8‰). This means that our modern test sites do not indicate that the use of 3 and 6‰ δ15N as limits for identification of manuring levels is wrong for our region.

The effects of ploughing and tilling on δ15N values is considered negligible, whereas burning/shifting cultivation is thought to potentially cause increased δ15N values in the period shortly after burning, especially if the nutrient level of the soil is high (Szpak 2014). At two of the sites, both dated to LN-EBA, there are signs of burning in the form of identified clearance layers and plant macrofossils that substantiate this in the sediment record (sites 12 and 13, see Halvorsen and Hjelle 2017). The measured δ15N values at these two sites are contrary; the results from site 13 indicates little or no manuring whereas those from site 12 indicate manuring. It thus seems that burning did not affect the δ15N values at site 13, whereas burning may have caused elevated δ15N values at site 12, probably due to good soil properties at this site. The analysed cereals at both sites are from clearance layers, so there could be other factors influencing the results and leading to higher δ15N values at site 12, for example burning in combination with manuring from grazing livestock. At site 13, the analysed cereals and the charcoal interpreted as a clearance layer may represent two different activity phases. The clearance could be a later activity than the one represented by the analysed cereal grains, explaining the low δ15N values.

Possibility of manuring in early management systems

From the isotopic signature of red deer and wild boar found by Rosvold et al. (2010), it is possible to infer a δ15N value around 0‰ for the natural vegetation in western Norway. Thus, δ15N values above this probably reflect amelioration of the soils, possibly through manuring of the cereal fields. This means that the threshold levels of manuring intensity (high, medium and low/none) after Fraser et al. (2011) and Bogaard et al. (2013) might be applicable for our region. In the youngest time-periods (VP-MA), cereals (except one grain of Avena) show δ15N values above 3‰, and a few grains also higher than 6‰ (Figs. 2, 3). This is the case both for Avena and H. vulgare and could be a sign of medium to intensely manured cereal fields. The bulk of the cereal grains from the LBA-PRIA show comparable δ15N values, although with a few exceptions. This suggests a well-developed management system where fertilisation provided the basis for permanent cultivation, which is also indicated in pollen and macrofossil data from the area (Halvorsen and Hjelle 2017). This indicates that the cultivation systems of the Viking Age and the Middle Ages have their roots in the LBA/PRIA management systems.

Enough fertiliser to keep the soil organic might have been difficult to come by. The use of plaggen soil, where peat/organic material is added to soil for cultivation (Blume and Leinweber 2004) might have been practiced at this time. Based on the presence of pollen and spores reflecting forest vegetation, Kvamme (1982) postulated plaggen soil formation in western Norway in the Viking Age, a practice also suggested from in the coast of western Norway based on macrosporangia of Selaginella (Hjelle and Halvorsen 2013). Micromorphological studies have documented the addition of plant material to enhance soils since the Bronze Age (Sageidet 2009). In the LBA-PRIA, accumulation of thick deposits took place at several sites in western Norway (Diinhoff 1999). This build-up of material would not be possible without adding organic material to the soil. Thus, it is reasonable to assume that plaggen formation was part of the agricultural system in the PRIA, and probably also in the LBA.

The older time-periods (LN-EBA), show large variations in the isotope data, and generally the δ15N values are lower than in later periods (Fig. 4). However, most sites have δ15N values above 3‰, and are within the range of or higher than the ecologically grown modern cereals at Mogstad. There has probably been variation in land-use practices in both space and time within the LN-EBA, where intensively managed fields close to the settlements received livestock manure and household waste, and a more extensive management practice with fallow periods took place in areas further afield (cf. Viklund et al. 2013). Additionally, variations in δ15N values within sites could be related to manure availability, and uneven spreading of manure on the fields. If manure was spread in lumps, some plants could receive more (or less) manure and thus nutrients than a neighbouring plant in the same field. Additional explanations can arise from the archaeological contexts (i.e. houses): the sampled cereals could have come from different cultivation years, and perhaps there was variation in the amounts of manure between years. The cereals found in different contexts may also reflect different uses of the grains, with cereals cultivated for different purposes having different demands or availability of manure. This variation could potentially be evaluated together with archaeological data from each site, to gain additional information about activity and social life.

The sea contains higher proportions of the heavier isotopes of C and N compared to air, and salt spray is known to lead to elevated δ15N values (Heaton 1987; Britton et al. 2008), although in areas with high rainfall this effect may be somewhat diluted (Heaton 1987). Several of our sites are situated close to the coastline, but mainly sheltered from the outer coast and the prevailing westerlies, and with moderate to high rainfall. Salt spray may therefore not have been of major importance. However, the proximity to the sea means kelp and other seaweeds would be easily available, and the use of such material as fertilising agents is known from coastal areas (Kaland 1979; Bakels 1997; Guttmann et al. 2005). Kelp and other seaweeds are enriched in 13C and 15N (Schulting et al. 2017; Sharp 2017), and it has been shown that use of seaweed, alone or mixed with manure, or the use of fish remains (cf. Gröcke et al. 2021), increases the δ15N values of cereals. As plants get their CO2 from the air, the δ13C values, however, remain unchanged (Blanz et al. 2019). In Fig. 7 higher δ15N values at the sites closest to the sea is indicated for all time-periods. However, it is important to remember that the figure does not consider topographic differences, aspect, or soil properties and that it is not known whether cereal plots were placed in sheltered areas away from the sea or not. However, the figure indicates the possibility that seaweed was used at all sites situated along the coast. A combination of livestock manure mixed with seaweed has probably been used for as long as cultivation has taken place in western Norway.

As the δ15N values indicate that manuring has taken place in areas in the region from the LN-EBA, a fully developed farming system was probably established in western Norway at that time. Pollen analysis indicates that cereal cultivation was introduced to the region already during the EN, but it was not until ca. 2400/2300 bce that the agrarian economy was fully established (Hjelle et al. 2006, 2018; Prøsch-Danielsen et al. 2018; Bergsvik et al. 2020, 2021). Also, from this time macrofossil remains of charred cereals are found in the region. Although with a fully established farming system in LN-EBA, the increase in δ15N values from LBA-PRIA and onwards indicated in our data suggests that manuring became a more regular part of the management system from then on. This probably reflects the importance of farms as an economic factor in the society in the Iron Age (Solberg 2000; Myhre 2004). Open farmed landscape characterised western Norway in this period (Hjelle et al. 2018), and could suggest a high degree of contact between well-established farming societies. Our data are limited to one site for the LBA, but high consistency in the macrofossil records (Halvorsen and Hjelle 2017) and a marked increase in cultivated fields in the region (Hjelle et al. 2018), suggest that the same practices go back to the LBA. This indicates that the development in western Norway is comparable to southern Scandinavia and that the practice of manuring is a result of the development of three-aisled houses in this period (Grabowski 2011; Kanstrup et al. 2011).

Water availability

Previous studies have used Δ13C values to elucidate water management in dry areas (Araus et al. 1997, 2014; Ferrio et al. 2005; Aguilera et al. 2012). For barley (H. vulgare and H. nudum) a Δ13C value exceeding 18.5‰ is considered to indicate cereals grown under conditions with plenty of water, and values below 17.5‰ Δ13C to indicate drought (Wallace et al. 2013). Currently, there have not been any studies regarding Δ13C values in Avena in relation to water status, thus only Hordeum Δ13C values are considered in our interpretations. Today, most areas in western Norway are not restricted in water availability. From our data this seems also to be the case in prehistory. From the PRIA a more unstable climate with fluctuating precipitation (including dry periods) has been shown (De Jong et al. 2009). In this period, cereals from two sites display Δ13C values possibly indicating dry conditions.

At site 4, two H. nudum and one H. vulgare grains have Δ13C values below 17.5‰, which indicates drought/water stress. This is also the case for one H. nudum grain at site 2, and one H. vulgare grain at site 7. These sites also have Hordeum with Δ13C values either above 18.5‰ or between 17.5 and 18.5‰, which could reflect the variability of the field where the cereals were grown, where some areas were drier than others. However, it is probably more likely that they represent different years where some years have been drier at the time of grain filling than others.

The modern Hordeum from Mogstad shows ∆13C values well above 18.5‰ in a year that had early summer drought and high levels of precipitation in the late summer (and autumn). This perhaps indicates that the later part of the growing season influenced the ∆13C values more than the early parts. From this, one could postulate that the fossil grains of our study were grown under drier conditions than at Mogstad. However, precipitation is not the only factor that can potentially influence ∆13C values (see Flohr et al. 2019 and references therein) and finding one explanation for all sites through all time-periods is not possible within the framework of this study.

Conclusion

This study has shown indications of low to moderate levels of manuring in western and north-western Norway from the LN and EBA (2300–1200 bce), by comparing δ15N values of charred, fossil cereals with measurements of modern cereals grown under known manuring regimes. This implies that people had knowledge of the positive effects of manuring of crops from the time of the expansion in arable farming in western Norway in the LN. Site-to-site variation in the LN and EBA is evident, and not all sites display elevated δ15N values. This suggests that manuring was prioritised in some places and not in others or a limited availability of livestock manure. Many of our sites are situated along the coast, and it is likely that the use of marine resources to amend soil quality was a widespread practice. From the PRIA onwards (after 500 bce), and probably some hundred years earlier, there is clear evidence of manuring in all investigated areas. Although livestock manure probably is the main source of the elevated δ15N values of the cereals at our sites in this period, utilisation of marine resources (kelp, other seaweeds and fish) was possibly common. This practice, established at latest in the PRIA, was still in use in the MA and is probably comparable to the traditional management systems continuing into the early twentieth century.

For future studies it would be beneficial to use larger assemblages of cereal grains to lower the grain-to-grain variability and investigate if this corroborates the findings of the present study. This, however, assume that high quantities of fossil grains are available. The data from the modern-day plots are from only one growing season and it would be interesting to repeat the measurements from these test plots over several years to verify the results.