香京julia种子在线播放

In order to assess the effect of seaweed-fertilisation on barley δ³⁴S values, a field trial was conducted. As the aim of this field trial was to gain information relevant to the reconstruction of past diets and farming practises, the field trial was designed to follow historically documented traditional farming practises as far as possible and practicable (rather than using e.g., present-day seaweed-extracts as fertiliser). The trial was undertaken with barley, as this was one of the earliest crops to be domesticated and has been of particular importance for both human and animal consumption until today (Fuller and Weisskopf, 2020), and historical evidence indicates it was frequently fertilised with seaweed (Fenton, 1997; Martin, 1716; Russell, 1910; Sauvageau, 1920). The traditional barley variety “bere barley” (Hordeum vulgare L.), a hulled lax-eared six-row landrace of barley (Martin et al., 2008), was chosen for this field trial rather than a modern barley variety, making greater similarity to barley varieties from archaeological sites more likely. A review of the historical literature on seaweed-fertilisation practises, and more detailed reasoning behind the field trial design choices are given in a previous publication on the same field trial (Blanz et al., 2019).

The field trial was conducted in 2017 on the Orkney islands (Scotland) at an agronomic experimental site ca. 100 m north of Orkney College and ca. 250 m south of the nearest coastline (58° 59′ N and 2° 57′ W; grid reference HY 456 114; Figure 1). This coastal site was exposed to oceanic-influenced rain and sea spray in windy weathers, which, with an δ³⁴S value of ca. +20 ‰, can elevate δ³⁴S (Richards et al., 2001; Zazzo et al., 2011). As seaweed-fertilisation is also expected to elevate δ³⁴S values in the barley (cf. Gröcke et al., 2021; Szpak et al., 2019), this might obscure fertilisation effects on δ³⁴S. However, proximity to the ocean was likely frequently the case when fertilising crops with seaweed (cf. Fenton, 1997), so that this coastal site location offers a realistic perspective on the extent of the effects of seaweed fertilisation on δ³⁴S. The trial site was used for growing bere barley from 2002 and was likely mainly used for grazing through most of the twentieth century as part of Weyland Farm. The earliest mention of this farm dates back to 1595 (Peterkin, 1820), so that it is likely that the trial site has a long history of being used for grazing and also growing bere barley. The field consisted of an acidic clay loam soil that was previously cultivated and fertilised with a low-level NPK mineral fertiliser (50 kg N/ha). It had not been subjected to other fertilisation-based agronomic field trials before, ensuring soil homogeneity throughout the trial area. A randomised block design was chosen with five replicate plots per fertilisation treatment, each measuring 3 m × 3 m (9 m²) and spaced 1 m apart.

Stranded seaweed of various species (including Laminaria spp., Fucus spp., and Ascophyllum nodosum; all with evidence for historical use as a fertiliser: Fenton, 1997; Hendrick, 1898; Neill, 1970; Russell, 1910; Sauvageau, 1920) was collected from the shore at Newark Bay, Mainland, Orkney (Grid reference: HY 567 041). As historical documents indicate that seaweed was frequently applied composted at the time of seeding (Fenton, 1997; Sauvageau, 1920), the seaweed in this trial was composted for 1.5 months in aerated plastic bags. After ploughing and power-harrowing, the seaweed was applied to the seaweed-treated plots at two rates: 25 t seaweed/ha and 50 t seaweed/ha (wet weight; Figure 2). A third set of plots was fertilised with a conventional mineral 14-14-21 NPK fertiliser (YaraMila MAINCROP 14-14-21; Yara UK Ltd, Belfast, UK) at a rate of 50 kg N/ha. A fourth set of plots (control plots) was left unfertilised. This gave a total of 20 plots: 5 unfertilised, 5 with 25 t seaweed/ha, 5 with 50 t seaweed/ha, and 5 NPK-fertilised. The fertilisers in each plot were mixed into the soil by power-harrowing. The following day (early May 2017), bere barley was sown at a rate of ~16 g/m² with a thousand grain weight of 30.3 g, using a tractor drawn seeder (width 3 m). A Cambridge roller was then used to flatten the soil surface. In early June 2017, an herbicide mixture was applied to all plots. Additional information on the field trial is available in Blanz et al. (2019) and Brown et al. (2020).

In early September 2017 (i.e., after 4 months of growth), the bere barley was harvested from a 1 m × 1 m square at the centre of each 3 m × 3 m plot (to avoid edge effects, fertiliser run-off, and effects from tractor wheels). The barley was dried at 30 °C for 48 h (until constant weight) and weighed for yield evaluation (previously published in Brown et al., 2020). Thousand grain weights (TGWs) were estimated using 20 g grain subsamples using a MARVIN grain analyser (GTA Sensorik GmbH, Germany; i.e., 20 g grain was weighed, and then counted by the analyser). In archaeological contexts, there are frequently only few grains available, and whole grains need to be selected manually. Therefore, archaeological grains are usually weighed individually and measured for their size, but in much smaller numbers. To mirror these measurement conditions, this more detailed, small-scale approach was therefore also imitated in this study in parallel to the present-day agricultural yield evaluation. Fifteen bere barley plants were subsampled from each plot's central 1 m × 1 m square from which 10 individual grains per plot were randomly sub-sampled and archived. Five of these grains were randomly sub-sampled for this study, using random.org (Haahr, 2010). Each grain selected had a large proportion of the inner glume material (lemma and palea, ca. 90%) removed by scalpel, prior to biomass and isotopic measurements. This was undertaken as most hulled barley carbonised plant macrofossils found in the archaeological record in Atlantic Scotland have had the lemma and palea burnt off by the carbonisation process (except in rare circumstances such as the low temperature, reducing thermal environments in some parts of conflagrations; Church, 2002). The length, width and depth of each grain was then measured to the nearest 0.1 mm using the internal graticule of a Leica M80 stereomicroscope under × 7.5–60 magnification and the mass of each grain was measured to the nearest 0.001 g using a Mettler PM480 Delta Range balance. Two of the measured grains were then randomly sub-sampled for isotope analysis, ground individually using an agate mortar and pestle and weighed into tin capsules.

Additionally, 10 g of straw, including both stems and leaves, were taken from the 15 bere barley plants that were subsampled from each plot's central 1 m × 1 m square, homogenised using a spice and nut grinder (Model SG20U, Cuisinart Corp., Greenwich, USA), and sieved to 1 mm using a plastic mesh. The stems and leaves likely had different δ³⁴S values (Tcherkez and Tea, 2013), but were homogenised to gain an overview—future studies should ideally analyse these separately.

For the analysis of the fertilisers, a pooled sample (120 g dry weight) of the composted seaweed as it was at the time of application in May was dried, ground and sieved as described for the straw samples. An aliquot of 1.5 g of the conventional NPK fertiliser was homogenised to a fine powder using a mortar and pestle. The top 0–25 cm of soil were sampled in three random spots in each plot using an augur, both before fertilisation, and after the trial, and pooled within each treatment group. The material was then sieved through a metal 1 mm mesh sieve.

Sulfur stable isotope ratio measurements were undertaken on the ground individual grains (2 grains per plot, equalling 10 grains per treatment), the homogenised straw (one sample each per replicate plot), the soils (before and after the trial), and fertilisers (two replicates each). Sulfur isotope analysis of the samples were performed using a Thermo Scientific™ EA IsoLink IRMS Delta V instrument in the Stable Isotope Biogeochemistry Laboratory (SIBL) at Durham University. Isotopic accuracy was monitored through repeated analysis of in-house standards (e.g., sulfanilamide) and international standards (e.g., IAEA-SO-5, IAEA-SO-6, NBS-127). These analytical standards provided an isotopic range from −31 ‰ to +20.3 ‰ in δ³⁴S. Analytical uncertainty in δ³⁴S for replicate analyses of the international standards was ± 0.2 ‰ (2 σ) and ± 0.3 ‰ (2 σ) for in-house standards and replicate sample analysis. Total sulfur concentrations were obtained as part of the isotopic analysis using sulfanilamide (sulfur = 18.62 wt %).

The results of this study were assessed by one-way ANOVA followed by post-hoc Tukey using RStudio software (R Core Team, 2021). The statistical significance threshold was set at α = 0.05. In a previous study, δ¹³C and δ¹⁵N of crop material from this field trial had already been measured (Blanz et al., 2019), the results of which are also discussed in this study with relation to the newly measured δ³⁴S values.

Total biomass yields (both straw and grains) were highest for barley grown with 50 t seaweed/ha, and lowest for the unfertilised plants, as exemplified in Figure 3. The individual fertilised grains were heavier on average than the unfertilised grains by 10% following 25 t/ha seaweed fertilisation, 15% heavier with 50 t/ha seaweed, and 9% heavier using the NPK mineral fertiliser (based on thousand grain weights measured with present-day agronomic measurement methods using 20 g subsamples; previously published in Brown et al., 2020). Additional yield results (including straw yields) from this trial are reported in detail in Brown et al. (2020).

The grain measurements of randomly selected grains in this study, meant to simulate grain measurements on small numbers of preserved archaeological grains, matched this pattern (see Table 1; Figure 4): all three fertilised sets of grains were demonstrably larger in length, width and depth measurements and heavier than the unfertilised grains. ANOVA followed by post-hoc Tukey indicated significant differences in grain length [F_{(3, 96)} = 16.7, p < 0.001], width [F_{(3, 96)} = 13.1, p < 0.001], depth [F_{(3, 96)} = 16.8, p <0.001], and mass [F_{(3, 96)} = 13.1, p < 0.001] for all comparisons of fertilised grain with unfertilised grain.

Sulfur isotope ratios of straw and grains from the seaweed-fertilised barley plants ranged between +13.2 ‰ and +16.0 ‰, with the 50 t/ha seaweed fertilised plants having on average slightly higher δ³⁴S values compared to those fertilised with 25 t/ha seaweed (by 0.3 ‰ for grains, and 0.8 ‰ for straw). These δ³⁴S values are higher than for both unfertilised and NPK-fertilised barley (range +10.8 ‰ to +13.4 ‰; Figure 5 and Table 2; all results are provided in the Supplementary material). For grains, ANOVA followed by post-hoc Tukey indicated the δ³⁴S differences to be significant [F_{(3, 36)} = 87.5, p < 0.001, n = 10 per treatment] for all comparisons of seaweed-fertilised grain with unfertilised or NPK fertilised grain (all p < 0.001), whereas differences between the two seaweed-treatments were not found to be significant (p = 0.69), and neither were differences in δ³⁴S between NPK fertilised and unfertilised grains (p = 0.09). Similar results were found for straw following a separate ANOVA followed by post-hoc Tukey [F_{(3, 16)} = 25.1, p < 0.001, n = 5 per treatment], with all comparisons of seaweed-fertilised straw with unfertilised or NPK fertilised grain being significantly different (all p < 0.001). Differences between the two seaweed-treatments were not found to be significant (p = 0.29), and neither were differences in δ³⁴S values between NPK fertilised and unfertilised straw (p = 0.37).

On average, straw tended to have higher δ³⁴S values than grains, except for the 25 t/ha seaweed treatment, where the straw results were variable. The NPK fertiliser had the lowest δ³⁴S value (+7.0 ‰) in this study, and the seaweed used as fertiliser had the most positive (+18.4 ‰). Soils had δ³⁴S values between +12.1 ‰ and +13.2 ‰. By comparison, seaweed-fertilised grain and straw had higher δ³⁴S values than the soil after the field trial (an increase of +2.4 ‰ on average), whereas NPK-fertilised and unfertilised grains had lower δ³⁴S values than the soil (a decrease of −0.8 ‰ on average), and unfertilised straw and NPK-fertilised straw had similar values to the soil after the field trial. A comparison to δ¹³C and δ¹⁵N values of bere barley grains from the same field trial (previously published in Blanz et al., 2019) is shown in Figure 6, and indicates a moderate correlation between δ³⁴S and δ¹⁵N [statistically significant with r₍₁₈₎ = 0.67, p = 0.001], but no correlation with δ¹³C [r₍₁₈₎ = 0.06, p = 0.8].

In this field trial, fertilisation with seaweed and NPK fertiliser both led to an increase in grain size. However, increased grain size is not in itself diagnostic of fertilisation: increased fertilisation frequently leads to larger number of tillers and ears, which may then result in a larger overall yield, but smaller individual grain sizes (see Bingham et al., 2007; Shah et al., 2017). Additionally, droughts, for example, can also lead to small grain sizes despite fertilisation, whereas a low seeding density (including growth on edges of fields) can lead to larger amounts of nutrients being available per plant and thus, larger grains despite a lack of fertilisation. Therefore, grain size itself is not sufficient to determine the nutritional status of a plant.

Nevertheless, these results show that fertilisation can increase size and mass of hulled bere barley grains and that an equivalent thousand grain weight (TGW) can be estimated based on the average mass of 25 individual grains multiplied by 1,000 (see Table 1). This estimate compares very well to the TGWs estimated using 20 g grain samples from each plot in this field trial using present-day agronomic standard methods (see Figure 4; Brown et al., 2020). These results indicate that in archaeological contexts, where it is frequently impossible to recover, identify, weigh and count similarly large amounts of complete grains, it should still be possible to arrive at reasonable estimations of TGW using smaller sample sizes (here n = 25), albeit with larger associated uncertainties.

Ferrio et al. (2004) have outlined a method to estimate individual grain weights from carbonised archaeological cereal grains, based on furnace experimentation of modern grains of wheat and barley. By combining the estimated individual weights of carbonised cereal grain assemblages, it is therefore possible to convert the averages of the grain weights of the individual archaeological cereal grains into TGW estimations, allowing direct comparison to the modern agronomic metric of TGW to the archaeological assemblages. This has clear research implications for understanding long-term trajectories of cereal productivity over thousands of years in different parts of the world (cf. Gron et al., 2021).

The δ³⁴S values of the seaweed-fertilised barley were elevated by ~2 ‰ compared to unfertilised barley. This was most likely caused by the elevated δ³⁴S value of the seaweed fertiliser (+18.4 ‰) compared to the soil prior to the trial (+12.7 ‰), as sulfur from decaying seaweed is incorporated into the growing bere barley. The absence of a statistically significant difference between the two seaweed treatments is likely due to the barley only taking up a limited amount of sulfur, and thus more available sulfur did not directly lead to a greater increase in δ³⁴S.

For the NPK fertilised barley, the opposite effect, i.e., a lowering of δ³⁴S compared to the unfertilised barley, was expected due to the low δ³⁴S value of the NPK fertiliser (+7.0 ‰); however, this difference was found to be less pronounced than for the seaweed fertilised barley (see Figure 5). This is likely because the NPK fertiliser application introduced less sulfur to the soil owing to the lower application rate and sulfur content of the mineral fertiliser (ca. 5 kg sulfur ha⁻¹) compared to seaweed fertilisation (ca. 785 kg sulfur ha⁻¹ and 1,570 kg sulfur ha⁻¹ for the treatments of 25 t/ha seaweed and 50 t/ha seaweed, respectively—although we hypothesise that not all of this sulfur would have been bioavailable during the course of this field trial).

The results of this study are similar to those previously reported for Celtic beans (Gröcke et al., 2021), where seaweed fertilisation led to ~10 ‰ higher δ³⁴S values compared to unfertilised beans (i.e., +16.2 ‰ in seaweed-fertilised beans and +6.6 ‰ in unfertilised beans in 2017). The much smaller difference in δ³⁴S between seaweed-fertilised and unfertilised barley in the present study is likely in part due to the comparatively high initial soil δ³⁴S value in the present study of +12.7 ‰. As the soil for the field trial in this study already had relatively high δ³⁴S values, the fertilisation of barley with seaweed likely had a smaller effect on δ³⁴S than it would have had in a soil with low δ³⁴S values. Additionally, the soil had a higher wt% sulfur concentration after the trial compared to Gröcke et al. (2021). Thus, when a sulfur-rich biofertiliser is added to a soil with low background sulfur concentrations (a macronutrient) a plant will more effectively incorporate and fractionate sulfur.

The origin of the higher soil δ³⁴S values in the field trial on Orkney is most likely its close proximity to the ocean, with its turbulent weather causing the frequent occurrence of sea spray and oceanic-influenced rain. This is supported by a δ³⁴S isoscape of Northern Ireland showing areas sheltered from oceanic influence to have lower δ³⁴S values than more exposed areas, and elevated δ³⁴S values in sheep from the Orkney Island of North Ronaldsay (Guiry and Szpak, 2020). Historical seaweed-fertilisation in the field trial area may have also elevated soil δ³⁴S values, as ethnographic and historical descriptions from the sixteenth to early twentieth century indicate seaweed fertilisation was common on Orkney (Fenton, 1997). Past elevated sulfur isotope ratios might have been preserved to some extent in the acidic clay loam soil (see work on clayey peat in Bottrell and Coulson, 2003). If seaweed-fertilisation does indeed increase soil δ³⁴S values in the long-term, this might be useful as a marker for seaweed-fertilisation in non-coastal areas (possibly together with elevated soil δ¹⁵N values; Blanz et al., 2019; Commisso and Nelson, 2010, 2007, 2006; Gröcke et al., 2021). Although we show in this study seaweed-fertilisation can elevate soil δ³⁴S, detecting such a difference in practise might prove difficult. Additional sulfur isotope research on modern coastal environments and islands is required in order to develop more robust interpretations from δ³⁴S of archaeological grains and soils.