Effects of marine biofertilisation on Celtic bean carbon, nitrogen and sulphur isotopes: Implications for reconstructing past diet and farming practices

Rationale The application of fertilisers to crops can be monitored and assessed using stable isotope ratios. However, the application of marine biofertilisers (e.g., fish, macroalgae/seaweed) on crop stable isotope ratios has been rarely studied, despite widespread archaeological and historical evidence for the use of marine resources as a soil amendment. Methods A heritage variety of Celtic bean, similar in size and shape to archaeobotanical macrofossils of Vicia faba L., was grown in three 1 × 0.5 m outdoor plots under three soil conditions: natural soil (control); natural soil mixed with macroalgae (seaweed); and 15 cm of natural soil placed on a layer of fish carcasses (Atlantic cod). These experiments were performed over two growing seasons in the same plots. At the end of each growing season, the plants were sampled, measured and analysed for carbon, nitrogen and sulphur stable isotope ratios (δ13C, δ15N, δ34S). Results The bean plants freely uptake the newly bioavailable nutrients (nitrogen and sulphur) and incorporate a marine isotopic ratio into all tissues. The bean δ15N values ranged between 0.8‰ and 1.0‰ in the control experiment compared with 2‰ to 3‰ in the macroalgae crop and 8‰ to 17‰ in the cod fish experiment. Their δ34S values ranged between 5‰ and 7‰ in the control compared with 15‰ to 16‰ in the macroalgae crop and 9‰ to 12‰ in the cod fish crop. The beans became more 13C‐depleted (δ13C values: 1–1.5‰ lower) due to crop management practices. Conclusions Humans and animals consuming plants grown with marine biofertilisers will incorporate a marine signature. Isotopic enrichment in nitrogen and sulphur using marine resources has significant implications when reconstructing diets and farming practices in archaeological populations.


| INTRODUCTION
Evidence for deliberate soil amendment strategies, or the use of crop fertilisers, has been identified amongst the earliest farming communities in many areas of the world. 1,2 However, the continued use of synthetic and chemical fertilisers in modern environments is under increasing scrutiny due to climate and human-induced changes in the Earth System. 3 Consequently, greater emphasis is now being placed on the use of organic 'traditional' fertilisers such as animal manure or marine resources such as seaweed or fish. 4 The effect of different fertilisers can be traced in archaeobotanical remains of crops using stable isotope ratios, although current research has predominantly focused on the impact of animal manure, 5 despite evidence for the use of other biofertilisers. 6 Hereafter, we use the term "biofertilisation" to categorise the use of natural biological materials for fertilisation compared with processed, synthetic or chemical fertilisers. Animal manure is likely to have been the most widely used form of soil amendment in both prehistoric and historic periods. 7 However, a wide range of biofertilisers were potentially available including domestic refuse, human faeces or 'nightsoil', ashes, turves, animal products (bone, blood, hooves) and, of particular relevance here, marine resources such as shell sand, macroalgae (i.e., seaweed) and fish. 6,7 There is widespread archaeological and historical evidence for the use of seaweed as a fertiliser, especially throughout the medieval period in northern Europe along the Atlantic coast. [7][8][9][10][11][12] Eighteenth and nineteenth century farmers living near the English coast would apply seaweed (either in a fresh-state or burnt/ashed) and fish to their crops, often mixed with household waste. [13][14][15] In comparison, direct evidence for the use of fish (e.g., fish heads, innards) as a fertiliser is less prevalent in the archaeological record, although a range of historical sources suggest its use throughout the medieval and post-medieval periods. 6,7,[16][17][18][19][20] Fish waste, alongside other refuse, has also been identified as a component of anthrosols, many of which developed around rural settlements and urban centres with the expansion in fish consumption during the medieval period. 6,[21][22][23] Fish appears to have been a particularly important fertiliser in North America, 24 being used by Native Americans and European settlers in areas such as the Plymouth Colony. 25 The same was true in nineteenth century New England, where a reported 27,000-37,800 lbs (13.5-18.9 tons) of fish were applied to an acre under cultivation in Marshfield, Massachusetts. 26 Direct evidence for the use of fish as fertiliser has been identified in a preserved Native American field at Cape Cod. 27 Modern studies have demonstrated the benefits of amendment with fish-based compost to various plants, an application with commercial potential, and processed fish remains are an organic alternative to chemical fertilisers. 28 In comparison, whilst there is a long history of using marine bird guano in South America, this form of fertiliser did not become widely used until the nineteenth and twentieth centuries in the Old World. 29,30 Taken together, it is evident that the benefits of marine resources as soil amendments have long been recognised.
Isotopic analysis of archaeobotanical material has been shown to be a powerful method for investigating past agricultural practices and land-use patterns. [31][32][33][34][35][36][37] Field experiments have indicated that the use of animal manure from terrestrial herbivores increases nitrogen isotope ratios (δ 15 N values) by up to 10‰ in cereals 35,38,39 and 3‰ in legumes. [40][41][42] In legumes, however, very intensive manuring is required to affect δ 15 N values, since legumes are nitrogen-fixing through the use of symbiotic bacteria in root nodules: therefore, they typically exhibit plant δ 15 N values near to that of nitrogen in air (0‰). [43][44][45] On the other hand, carbon isotope ratios (δ 13 C values) are thought to be minimally affected by manuring, 46 with crop δ 13 C values primarily interpreted on crop water management practices 47 or cultivation in different fields/soils. 31,34 However, other environmental factors (e.g., salinity, light intensity, temperature and nitrogen availability) can also affect plant δ 13 C values. 48,49 The relationship between plant δ 13 C values and manuring is not clearly understood and/or demonstrated, with both 13 C enrichment and depletion being shown in experimental studies. 30,35,46,50 Plants uptake sulphur from the soil, normally in the form of sulphate (SO 4 2− ), as well as from the atmosphere (SO 2 ). Sulphur isotope analysis of archaeobotanical remains can be undertaken as part of a multi-isotope approach together with δ 13 C and δ 15 N, and has the potential to provide information on the use of different soils/areas for cultivation as well as crop management practices. 51 The application of sulphur isotope ratios (δ 34  2 | MATERIALS AND METHODS

| Experimental design
The overall research aim of experimental archaeology is to aid interpretation of the material remains from the human past. 64 This general approach does not seek to precisely observe, quantify and model experimental data, such as the data produced from modern agronomic research, but rather allows the researcher to gain new interpretive insights from the experimental process. Within the established theory, method and practice of experimental archaeology, we chose to conduct simulation experiments. 65 These were based upon our working assumption that pulses (e.g., Celtic bean) were routinely grown using a form of 'garden cultivation' in small amended plots, based on archaeological evidence from across European history and prehistory. 66 Celtic bean (Vicia faba L.) was cultivated in three 1 × 0.5 m outdoor plots at the Botanic Garden, Durham University, in the summer of 2017 and 2018 ( Figure 1). The plots were surrounded with wire netting to prevent rabbits from entering and disturbing the plots.
The variety selected for cultivation was Celtic Black broad bean, a heritage landrace of broad bean. This variety produces small, rounded seeds which are comparable with Vicia faba var. minor and morphologically similar to prehistoric and later finds of V. faba. 67,68 Twenty beans were planted directly in each plot during April of each year after soil amendment with the biofertilisers. All beans were sowed individually and spaced accordingly to avoid over-crowding.
The experimental area used in the Botanic Garden has not been previously used for any other research (e.g., no fertilisation or liming).
One plot acted as a control (Plot 1c), and the other two plots were amended with naturally harvested biofertilisers: macroalgae (Plot 2m) and filleted Atlantic cod (Plot 3ac were then randomly sub-sampled from the total bean population for each plot, using a true random number generator: https://www. random.org. 70 Five pods were then chosen randomly from the population of the pods previously sub-sampled for ten beans: for example, for Plot 1c in 2017, seven pods were randomly subsampled to produce the ten bean samples and the seven pods were then in turn sub-sampled to generate the five pods. Five samples of stems and leaves were then taken from the five plants matching the pods. By doing this, the isotope measurements from the randomly sub-sampled beans could be compared with their corresponding plant parts. At the end of the experiment in each year, four random soil samples were taken from each plot. The top 2 cm of surface soil was removed, and the next 5 cm of soil sampled. The soil sample was placed in a drying oven at 40 C for 48 h. The soil was lightly compressed and passed through a 1 mm stainless steel sieve: the <1 mm size fraction was ground into a fine powder and analysed for bulk stable isotope values. Sulphur isotope analysis of the samples was performed in SIBL using an ECS 4010 elemental analyser connected to a Delta V Plus isotope ratio mass spectrometer. Isotopic accuracy was monitored through repeated analyses of in-house standards (e.g., sulphanilamide) and international standards (e.g., IAEA-SO-5, IAEA-SO-6, NBS-127).

| Stable isotope analysis
These analytical standards provided an isotopic range from −31‰ to 20.3‰ in δ 34 S. The analytical uncertainty in δ 34 S for replicate analyses of the international standards was <0.2‰ (2 sd) and <0.3‰ (2 sd) for in-house standards and replicate sample analysis. Total sulphur data were obtained as part of the isotopic analysis using sulphanilamide (sulphur = 18.62%).
The sample weights varied depending on the material type in order to obtain an SO 2 intensity of >800 mV for mass 64. Bulk leaf and bean tissue required >9 mg of sample, whereas bulk stem and pod material required >30 mg to achieve this mV criterion. Vanadium pentoxide (V 2 O 5 ) is often used as an oxidant additive when performing sulphur isotope analysis; however, we have tested a range of V 2 O 5 suppliers and noticed an appreciable sulphur blank when weighing more than 5 mg of V 2 O 5 (Gröcke, unpublished data).
Therefore, in SIBL we have opted to use tungstic oxide (WO 3 ) as an additive: it has fewer health and safety issues and no measurable sulphur up to 100 mg (Gröcke, unpublished data). The "macro" oxygen setting on the Costech elemental analyser was used for all sulphur isotope analyses (standards and samples).

| Celtic bean biomass
The biomass analysis of the Celtic bean plants is summarised in Table 1 (all raw data are provided in the supporting information).
Where appropriate, a two-tailed Student's t-test was used to assess statistical significance between the amended (e.g., Plot 2m, Plot 3ac) and the control plants. Over the two experimental years

| Celtic bean stable isotopes
A summary of the average Celtic bean plant tissue stable isotope ratios from 2017 and 2018 is presented in Table 2. All individual isotopic data is available in the supporting information. All stable isotope data are presented in Figure 2 (Table 3).
Both biofertilisers are 13

| Marine biofertilisation effects on growth
There was variability in germination success rates between plots, with a higher success rate from the control plot than from the amended plots. We suggest that this is a function of the increased ability of the amended soils to contain pests, such as slugs and snails, due to F I G U R E 2 Box and whisker plot of the stable isotope data for the experimental plots in 2017. c = control, m = macroalgae, ac = Atlantic cod. Each graph column represents different components of the Celtic bean plants that were analysed (see Table 2

| Carbon isotope discrimination in Celtic beans
Each plant component (e.g., stem, leaf, pod and seed) has a different purpose and therefore has biochemical reactions that would naturally differentiate the uptake of 13 C versus 12 C (idem, nitrogen). The average δ 13 C values of each component are summarised in Table 2 and graphically presented in Figures 2 and 3. Each plant component is The δ 13 C value of plants is predominantly controlled by the isotopic composition of CO 2 and photosynthetic pathway (C 3 , C 4 , CAM). 71 Since carbon within plant tissues ultimately derives from photosynthesis, 71 it was predicted that the different biofertilisers  become more 13 C-enriched due to stomatal conductance. 46,48,71,72 The relationship between δ 13 C and water availability in temperate environments (as opposed to semi-arid environments) where water is not a limiting factor on plant growth has been little studied and the relationship is not clear. Changes in the δ 13 C value could reflect a number of factors: (1) soil composition (i.e., increased organic matter increasing water retention); (2) plant biomass/dry matter production; and (3)  An illustrative plot of only the bean δ 13 C results is presented in Despite this, our results demonstrate that biofertilisation with marine resources has the potential to have an effect on δ 13 C values.

| Nitrogen isotope discrimination in Celtic beans
It has previously been shown that the nitrogen-fixing beans can produce higher δ 15 N values under significant soil amendment (e.g., manure) 40,42 ; this has also been recorded in amino acid δ 15 N values. 41 Only extremely intensive (i.e., >80-100 t/ha) manuring caused 15 N-enrichment (up to 3‰), probably due to a reduction in the proportion of nitrogen obtained via N 2 -fixation in favour of nitrogen uptake from the 15 N-enriched soil. 42 In the case of this study, the macroalgae applied to soils had a similar δ 15 N value to manure used in a previous experiment by Treasure et al. 42 However, the Atlantic cod δ 15 N values are much higher (Atlantic cod muscle tissue = 15.5‰) (see Table 3).
As presence of sulphur, nitrogen uptake has been shown to increase in crops. [80][81][82] Therefore, by using a biofertiliser with high sulphur content (e.g., macroalgae > 2 wt % S, marine fish > 1 wt % S), nitrogen uptake efficiency would increase. In addition, uptake of nitrogen from the soil, as opposed to atmospheric N 2 , will yield a reduction in N 2fixation rates. High soil nitrogen availability has been widely demonstrated to suppress N 2 -fixation rates in legumes. 75 The δ 15 N values of the Celtic beans in Plot 3ac ranged from 8‰ up to 18‰, with an average of 15.2‰: very similar to the Atlantic cod, 15.5‰ ( Figure 6). This degree of scatter may in fact relate to the randomness of the cod remains at 15 cm, and therefore changes in the presence or amount of muscle tissue (e.g., proteins/nitrogen) available to the root system (see Figure 6)

| Sulphur isotope discrimination in Celtic beans
Sulphur plays an important role in the legume-rhizobia system of nitrogen-fixation in plants. 81 The addition of marine biofertilisers with F I G U R E 4 Δ values for carbon, nitrogen and sulphur isotope ratios from each of the plant components (see text for description). A positive Δ value means that there is enrichment of 13 C, 15 N and 34 S, respectively, in that system compared with soil (background) values. pc = plant component. See Figure 2 for descriptions [Color figure can be viewed at wileyonlinelibrary.com] elevated sulphur content to the experiment had significant effects on nitrogen assimilation. In fact, sulphur uptake in the Celtic beans probably mimicked that of nitrogen in that it was efficiently absorbed and transferred into the plant. The beans showed the greatest isotopic shift of the order of 6‰ to 10‰ compared with beans from Plot 1c (Figures 2 and 3). In fact, these plants have shifted from a terrestrial δ 34 S signal (Plot 1c) to a marine value in just one growing season.
Sulphur isotope fractionation, Δ 34 S (Δ 34 S = δ 34 S pc -δ 34 S soil ), can be employed to assess the degree of change from a background soil to the biofertiliser source ( Figure 4). There is a clear shift towards greater fractionation in Plot 2m and Plot 3ac. It is greatest in Plot 2m as the δ 34 S value of the macroalgae (18‰) is more positive than that of Atlantic cod muscle tissue (15‰) (Figure 4). Due to its importance in forming essential amino acids (cysteine and methionine), 83 it is likely that the sulphur would have been rapidly incorporated into the plant and deposited in the protein-rich beans. light of this study they may be reinterpreted to indicate biofertilisation using fish remains (e.g., Plot 3ac, Figure 6); of course, this will depend on how scattered/dense the fish remains are in the crop soil.
A method to distinguish marine from freshwater biofertilisation would be difficult using just δ 13 C and δ 15 N values (see sections 4.2 and 4.3), but in the case of δ 34 S it is more feasible, although this may not be possible when the sample from freshwater fish is enriched in 34 S (e.g., seawater sulphate) due to bacterial sulphate reduction. 88,89 Although archaeological communities living by the ocean may not rely on marine resources for dietary consumption, they may potentially use seaweed for biofertilisation. In this scenario, the human population may record elevated δ 15 N (see Figure 6) and δ 34 S values (see Figure 7) that would suggest the presence of marine resources in their diet, but the δ 13 C values may still be quite low (see Figure 5).
Idealised stable isotope plots for δ 13 C versus δ 15 N are depicted in Figure 8. Whether freshwater or seawater fish were used as a biofertiliser it would be difficult to differentiate them just using δ 13 C and δ 15 N, but either way they increase the δ 15 N value by at least a trophic level. However, when combining δ 34 S with δ 15 N the effect of marine biofertilisation (e.g., macroalgae, fish) is more apparent (Figures 9 and 10 and not just animal manures. 5 In addition, different sources of manure (e.g., human, sheep, pigs) are likely to cause isotopic variability and future experimentation will be required to assess the extent of this.

| CONCLUSIONS
An experimental study was performed on Celtic bean crops grown in marine-amended soils over two consecutive years, and subsequently analysed for carbon, nitrogen and sulphur isotope ratios. The soils were amended with marine biofertilisers, macroalgae and Atlantic cod.