Introduction

Stable isotope analysis is becoming increasingly widespread in archaeobotanical research as a powerful method for investigating agricultural practices and land use patterns (Aguilera et al. 2008; Bogaard et al. 2013; Fraser et al. 2013; Heaton et al. 2009; Kanstrup et al. 2014; Masi et al. 2014; Vaiglova et al. 2014). Plant stable carbon (δ13C) and nitrogen (δ15N) isotope ratios vary in relation to a range of environmental factors (Dawson et al. 2002) including agricultural practices, and these variations can be analysed in archaeobotanical remains (Fiorentino et al. 2015; Szpak 2014). In particular, recent research has been directed towards the identification of manuring in cereals and pulses (Bogaard et al. 2007, 2013; Fraser et al. 2011; Kanstrup et al. 2011, 2014). Since manuring is closely linked to intensive land use patterns, it is an important area of research into past crop husbandry regimes (Bogaard 2012).

Field experiments and farm studies indicate that animal manuring increases cereal δ15N values (≥6 ‰) (Bogaard et al. 2007; Bol et al. 2005; Fraser et al. 2011; Kanstrup et al. 2011). In comparison, manuring has minimal effect on δ15N values in legumes, except where very intensively applied over a long duration (Fraser et al. 2011; Styring et al. 2014). There is a requirement to further examine the relationship between manuring intensity and legume δ15N values (Styring et al. 2014). In terms of palaeodietary reconstructions, it is important to assess the impact of manuring on legume δ15N values as a source of 15N enrichment in animal and human tissues (Bogaard et al. 2007; Fraser et al. 2011; Hedges and Reynard 2007).

Plant δ13C values have been primarily applied to investigate crop water management practices (Fiorentino et al. 2015). However, other environmental factors that alter δ13C need to be considered (Stokes et al. 2011). The relationship between plant δ13C values and manuring is not clearly understood with both positive and negative shifts observed in δ13C (Kanstrup et al. 2011; Senbayram et al. 2008; Szpak et al. 2012b; Wallace et al. 2013).

The aim of this study is to analyse the influence of manuring on δ13C and δ15N values in the legume, Celtic Black broad bean (Vicia faba L.). V. faba is frequently recovered in archaeobotanical assemblages across the Near East and Europe from the Neolithic and Bronze Age onwards (see Colledge and Conolly 2007; Stika and Heiss 2013). A recent study identified extensive evidence for V. faba in Britain in the Neolithic-Iron Age, totalling over 70 archaeobotanical assemblages containing Celtic Bean, and suggested that, in some areas, it may have been cultivated on a small ‘garden’ scale in intensively managed plots treated with animal manure (Treasure 2014). In order to accurately assess the relationship between δ13C and δ15N values in V. faba and manuring, it is necessary to undertake field experiments where the rates of manure application can be quantified.

Methodological background

Nitrogen isotopes (δ15N)

Plant δ15N values reflect the net effect of a range of factors, including the form of nitrogen acquired (NH4 +, NO3 −, N2) and the method of nitrogen assimilation (uptake of soil nitrogen, fixation of atmospheric N2) (Evans 2001; Högberg 1997).

Pulses are harvested from leguminous plants which can assimilate nitrogen through fixation of atmospheric N2 via symbiotic bacteria (rhizobia) in the roots (Franche et al. 2009; Howard and Rees 1996). As N2 fixation involves minimal fractionation, legumes dependent on fixation as a source of nitrogen have δ15N values typically around 0 ‰, reflecting atmospheric N2 (i.e., δ15Nair = 0 ‰) (Kohl and Shearer 1980; Shearer and Kohl 1986; Virginia and Delwiche 1982). A range of factors, however, can influence N2 fixation, in particular, soil nitrogen availability (Liu et al. 2011). In soil N-rich environments, N2 fixation is inhibited and legumes preferentially take up nitrogen from the soil (Ledgard et al. 1996; Peoples et al. 2009; Vinther 1998). The nitrogen isotopic composition of legumes taking up nitrogen from soil (rather than N2 fixation) will reflect the δ15N value of the soil nitrogen (Andrews et al. 2011). As manuring increases soil mineral nitrogen and δ15N values (Choi et al. 2003; Simpson et al. 1999; Watzka et al. 2006), it has the potential to increase δ15N values in legumes above 0 ‰.

In cereals (non-N2-fixing plants), animal manure significantly increases δ15N values (up to +10 ‰) due to the uptake of 15N-enriched soil (Bogaard et al. 2007; Bol et al. 2005; Fraser et al. 2011; Kanstrup et al. 2011, 2012). In comparison, a recent study suggests that only very intensive animal manure application, in excess of >20–35 t/ha, alters δ15N in V. faba due to the preferential fixation of atmospheric N2 (Fraser et al. 2011; Styring et al. 2014). The largest increase in V. faba δ15N values (+2.2 ± 1.4 ‰) was observed in a farm study in Evvia, Greece, where very intensive manuring creates artificial ‘dung-soil’ (Fraser et al. 2011). However, the rate of manure application (t/ha) was not measured in the Evvia farm study.

Carbon isotopes (δ13C)

During photosynthesis, C3 plants (cereals, legumes) discriminate against 13C due to the preferential use of 12C by the enzyme RuBisCO during carbon fixation (Lloyd and Farquhar 1994; O’Leary 1981). The amount of 13C discrimination is closely linked with stomatal conductance (intrinsic water use efficiency) (Farquhar and Sharkey 1982; Farquhar et al. 1989; O’Leary 1988). Restricted water availability increases stomatal closure and reduces 13C discrimination (Chaves 1991; Farquhar et al. 1989). In comparison, during high water availability, stomata are open, increasing 13C discrimination (Chaves 1991; Farquhar et al. 1989). Studies have identified a causal link between δ13C values in plants and water availability in greenhouse and farm studies (Araus et al. 1997; Farquhar and Richards 1984; Flohr et al. 2011; Wallace et al. 2013).

However, a wide range of other environmental factors can alter δ13C values in plants including salinity, light intensity, temperature and nitrogen availability (Condon et al. 1992; Gröcke 1998, 2002; Heaton 1999; O’Leary 1981; Tieszen 1991). The relationship between plant δ13C values and manuring is complex with both positive and negative relationships observed (Kanstrup et al. 2011; Senbayram et al. 2008; Szpak et al. 2012b; Wallace et al. 2013). At present, there is no published data which specifically focuses on the impact of manuring on δ13C values in V. faba.

Materials and methods

Experimental design

Celtic Black broad beans (V. faba L.) were cultivated in three 1-m2 outdoor plots at Durham University Botanic Gardens between May and September 2013. The Celtic Black broad beans used are a heritage variety which produces small-rounded seeds which are morphologically similar to prehistoric finds of V. faba. One plot acted as a control, and two plots were treated with manure and decomposed leaf litter (Table 1). All the available plants were harvested and the plant height, number of ripe/un-ripe pods, number of beans and dimensions of each bean were recorded.

Table 1 Field treatments for the plots used in this study
Full size table

One plant from each plot was randomly selected for isotopic analysis and air-dried. Samples of bean cotyledons, bean testae, pods, leaves and stem were analysed. All of the pods and beans available for each selected plant were analysed. The cotyledon was sampled separately, as the testa is rarely preserved in archaeobotanical remains of V. faba. The position of each pod on the plant and each bean within individual pods was recorded. This detailed sampling methodology enables analysis of within-plant δ13C and δ15N variation. Dried soil and manure samples were sieved at <1 mm. All samples were homogenised in an agate pestle and mortar.

Stable isotope analysis

Stable isotope measurements were performed at Durham University using a Costech Elemental Analyser (ECS 4010) coupled to a Thermo Finnigan Delta V Advantage isotope mass spectrometer. Carbon isotope ratios are Craig-corrected for 17O contribution and reported in standard delta (δ) notation in per mil (‰) relative to VPDB. Nitrogen isotope ratios are reported relative to AIR. Data accuracy is monitored through routine analyses of in-house standards, which are stringently calibrated against international standards (e.g., USGS 40, USGS 24, IAEA 600, IAEA N1, IAEA N2). Analytical uncertainty for δ13C and δ15N measurements is typically ±0.1 ‰ for replicate analyses of the international standards and typically <0.2 ‰ on replicate sample analysis. Total organic carbon and nitrogen data was obtained as part of the isotopic analysis using an internal standard (i.e., glutamic acid, 40.82 % C and 9.52 % N).

Results

Biomass analysis

The results of the biomass analysis are presented in Table 2. Six plants from the midden plot suffered insect damage and did not produce any pods. Plants in the amended plot were taller than in the control plot. In comparison, the heights of plants in the control and midden plots are identical (though this may be due to insect damage in the midden plants). Pod and bean yield was highest in plants in the amended plot. The comparative results for the midden plot are significantly lower, though, as noted above, this may be due to insect damage. Bean dimensions increased (particularly length) in the amended and midden plots.

Table 2 Biomass analysis of the Celtic Black broad beans from this study
Full size table

Stable isotope analyses

Mean δ13C and δ15N values for the samples analysed are presented in Table 3. Figure 1 presents the δ13C and δ15N values for all the plant and soil samples analysed. The supplementary data includes the results for each sample in addition to C/N atomic ratio, %C and %N results.

Table 3 δ15N and δ13C analysis of the Celtic Black broad beans from this study
Full size table
Fig. 1
figure 1

δ15N and δ13C values for each sample (excluding the manure samples). Note the different scale on the soil graph

Full size image

Soil and manure analyses

Mean soil δ15N values in the amended (5.5 ± 0.4 ‰) and midden (8.1 ± 1.7 ‰) plots are significantly higher than in the control plot (4.6 ± 0.2 ‰). Mean δ15N values for the farmyard manure were 7.7 ± 0.3 ‰ and, for the sheep manure, 7.5 ± 0.2 ‰. In the amended and midden plots, mean δ13C values are lower than in the control plot.

Plant analyses

Mean δ15N values in the control samples varied between −1 and 0.7 ‰. Mean δ15N values in cotyledons were 1.5 ± 0.2 ‰ in the amended samples and 2.6 ± 0.3 ‰ midden samples. Pods, stems and leaves were 15N enriched in the amended and midden samples. There is small variation in δ15N values between cotyledons sampled from the same pod and plant. δ15N variation between leaves, stems and pods was typically ≤1 ‰, with the largest offset (1.7 ‰) between the midden pod and midden stem samples (see Electronic supplementary material). Testa δ15N values are significantly lower than cotyledon δ15N values. Manuring intensity and δ15N values are positively correlated.

Mean δ13C values in the amended and midden samples are similar to the control samples, with small increases observed in some samples of up to 1.7 ‰. Within-plant δ13C variation between leaves, stems and pods was minimal. There is only a small variation in δ13C values between cotyledons sampled from the same pod and plant.

Discussion

Soil and manure δ13C and δ15N variability

In agreement with previous studies, manuring increased soil δ15N values (Bol et al. 2005; Kanstrup et al. 2011; Senbayram et al. 2008; Watzka et al. 2006). The mean δ15N values for the farmyard manure (7.7 ± 0.3 ‰) and sheep manure (7.5 ± 0.2 ‰) are high for animal manure (cf. Szpak 2014). Mean δ13C values were lower in amended and midden plots as animal manure is 13C-depleted (Bol et al. 2005). Previous studies have observed decreases in δ13C values in manured soils (Gerzabek et al. 1997; Senbayram et al. 2008).

Plant δ13C variability

Plant δ13C values were minimally affected by manuring, displaying only small increases, typically around +1 ‰. These results are consistent with previous studies which indicate that manuring may be a source of small variation in 13C values (Senbayram et al. 2008; Szpak et al. 2012b; Wallace et al. 2013). Small increases in δ13C values in response to nitrogen fertilisation observed in previous studies may be related to increased plant biomass which can limit stomatal conductance causing less 13C discrimination (Jenkinson et al. 1995; Serret et al. 2008).

Plant δ15N variability

The results presented in this study indicate that very intensive animal manuring (>70 t/ha) can increase δ15N values (+3 ‰) in V. faba. There is little δ15N variation between plant tissues, with the exception of testa samples, comparable with results from previous studies (López-Bellido et al. 2010; Nebiyu et al. 2014). Control sample δ15N values varied between 0 ± 1 ‰ and are consistent with fixation of atmospheric N2 in V. faba (Fan et al. 2006; López-Bellido et al. 2011; Nebiyu et al. 2014; Tryderman et al. 2004; Unkovich 2013). In contrast, the elevated δ15N values in the amended and midden samples indicate preferential uptake of soil mineral nitrogen in comparison to atmospheric N2 fixation. The δ15N values in the midden plot are higher than in the amended plot due to the application of pure manure compared to farmyard manure (i.e., a mixture of straw and manure).

The results presented here agree with a previous study which indicated that only very intensive manuring can significantly alter δ15N values in V. faba due to the preferential fixation of atmospheric N2 (Fraser et al. 2011; Styring et al. 2014). Fraser et al. (2011) only observed a large increase in δ15N values for V. faba in a farm study at Evvia, Greece, where very intensive manuring (sheep/goat dung) creates an artificial dung-soil. The quantity of manure applied (t/ha) in the Evvia farm study could not be measured, though the soil conditions appear to be similar to the midden plot in this study.

The relationship between N2 fixation and animal manuring in legumes is an area which requires further research. It is not clear why δ15N values in legumes are only altered by very intensive manuring as it requires less energy to take up soil mineral nitrogen compared to fixing atmospheric N2 (Andrews et al. 2009). A possible explanation may be due to ability of certain varieties of V. faba to continue atmospheric N2 fixation in the presence of high soil mineral nitrogen (Köpke and Nemecek 2010). Recently, Szpak et al. (2014) observed large 15N enrichment (+16 ‰) in a legume (common garden bean, Phaseolus vulgaris L.) amended with seabird guano which has a very high δ15N value (> +20 ‰) compared with animal manure (Szpak et al. 2012a; Szpak 2014). This indicates that manures which are high in plant available nitrogen and have high δ15N values can significantly enrich 15N in legumes and suppress N2 fixation compared to animal manure (Szpak et al. 2014). In comparison, a recent study demonstrated that animal manure increased N2 fixation in peas (Pisum sativum L.; Jannoura et al. 2014).

Archaeological implications

The results of this study suggest that intensive manuring of V. faba may be identifiable in archaeobotanical remains using nitrogen isotope analysis (cf. Fraser et al. 2011). The high intensity of manuring required to effect the nitrogen isotopic composition of pulses indicates that pulse δ15N values can reflect the scale of cultivation. In recent farming contexts, intensity of manuring is closely correlated with the scale of cultivation, with smaller plots receiving intensive manure application (Bogaard et al. 2000; Jones 2005). In Evvia, Greece, V. faba is cultivated in small infield (garden) areas, some of which are very intensively manured, creating an artificial dung-soil (Jones et al. 1999). Similarly, in Asturias, Spain, V. faba is cultivated in small intensively manured plots that are rotated with cereals (Charles et al. 2002). It is suggested that high δ15N values in archaeobotanical remains of V. faba may indicate small-scale cultivation with very intensive manuring. The results of this study should be viewed as preliminary in character, and further research is currently ongoing to explore δ13C and δ15N variability between plants cultivated in the same plot. For example, Wallace et al. (2013) have demonstrated large variation in δ13C values in V. faba cultivated under similar conditions. Variation in plant isotope values is expected in traditional farming regimes where growing conditions can be variable (Wallace et al. 2013)

In terms of palaeodietary reconstructions, measurement of plant δ15N values is necessary in order to accurately reconstruct baseline data and interpret δ15N values in animal and human tissues (Casey and Post 2011; Hedges and Reynard 2007). In particular, Fraser et al. (2013) have demonstrated that the plant component of diets can be assessed with greater accuracy through direct measurement of archaeological plant δ15N values. The impact of manuring on cereal δ15N values (up to 10 ‰) is significantly higher than the Celtic Black broad beans analysed in this study. Despite this, the enrichment in δ15N between the control and the midden samples (~3 ‰) is equivalent to the trophic-level effect (3–5 ‰, Bocherens and Drucker 2003) and, hence, could subsequently cause 15N enrichment in animal and human tissues. This is particularly significant as palaeodietary studies typically consider legumes to have δ15N values around 0 ‰ (DeNiro and Epstein 1981; van Klinken et al. 2002).

In manured cereals, there is a large offset in δ15N values between the grain and chaff (2.4 ± 0.8 ‰), suggesting that the use of chaff for animal fodder and grain for human consumption will result in significantly different nitrogen isotopic signatures between animals and humans (Bogaard et al. 2007; Fraser et al. 2011). In comparison, the results of this study indicate a comparatively small offset in δ15N values of manured V. faba between different plant components. This is significant as ethnographic evidence indicates that V. faba was used as a source of fodder, either the seeds, chaff or whole plants (Forbes 1996; Halstead 2014).

Conclusion

The results of this study indicate that highly intensive animal manuring can increase δ15N values in legumes. Celtic Black broad beans (V. faba) displayed significantly higher δ15N values in intensively manured plots. In comparison, manure minimally affected δ13C values, indicating that manuring may only be a source of small variation in δ13C values. In terms of palaeodietary reconstructions, manuring increased δ15N values on a scale equivalent to a single step in trophic level. Based on the experimental results presented here, it is suggested that high δ15N values in archaeobotanical remains of V. faba should be attributed to small-scale cultivation with very intensive manuring.