Carbon and nitrogen isotopic variability in foxtail millet (Setaria italica) with watering regime

Rationale Carbonised plant remains are analysed for reconstruction of past climates and agricultural regimes. Several recent studies have used C4 plants to address related questions, and correlations between modern C4 plant δ13C values and rainfall have been found. The millets were important food crops in prehistoric Eurasia, yet little is known about causes of isotopic variation within millet species. Previous research has shown there to be significant isotopic variation between millet accessions. Here we compare isotope ratios from plants grown under different watering regimes. This allows for a consideration of whether or not Setaria italica is a good proxy for environmental reconstruction. Methods We compare stable isotope ratios of Setaria italica plants grown in a controlled environment chamber with different watering regimes. We compare the carbon isotope ratios of leaves and grains, and the nitrogen isotope ratios of grains, from 12 accessions of Setaria italica. Results We find significant isotopic variability between watering regimes. Carbon isotope ratios are positively correlated with water availability, and on average vary by 1.9‰ and 1.7‰ for leaves and grains, respectively. Grain nitrogen isotope ratios also vary with watering regime; however, the highest isotope ratios are found with the 130‐mL watering regime. Conclusions The carbon isotope ratios of Setaria italica are strongly correlated with water availability. However, the correlation is the opposite to that seen in studies of C3 plants. The difference in isotopic ratio due to watering regime is comparable with that seen between different accessions; thus distinguishing between changing varieties of Setaria italica and changing climate is problematic. In terms of grain nitrogen isotope ratios, the highest δ15N values were not associated with the lowest watering regime. Again, δ15N variation is comparable with that which would be expected from an aridity effect or a manuring effect, and thus distinguishing between these factors is probably problematic.


| INTRODUCTION
Increasingly in recent years, stable isotope studies of charred plant remains have been used in archaeological research to answer questions about palaeoclimate and farming practices, as well as to improve our interpretations of human and animal isotope results. [1][2][3][4] Fundamental to this research is a sound understanding of the causes and magnitude of isotopic variation in plants. The causes of plant isotopic variation have been investigated using modern experimental studies, led by both plant scientists and archaeologists. For example, it has been shown that manuring can increase nitrogen isotope ratios by as much as 9‰ in cereals manured with cattle slurry. 5 While most archaeological isotopic research on charred plant material has focused on C 3 plants, most notably wheat and barley, increasing archaeological and isotopic research in China, Central Asia and Eastern Europe has highlighted the importance of millets (a generic term for all small-grained cereals, which are typically found to be C 4 plants) in the archaeological record.
Millets have various advantages over other major food crops in that they have a short growing season, relatively high nutritional value and high water use efficiency, and can grow on poor soil. [6][7][8] Two species of millets are important for Eurasian prehistoric archaeology, foxtail and broomcorn millet (Setaria italica and Panicum miliaceum, respectively). While other C 4 plants were probably available to prehistoric farmers, these species represent the only staple C 4 crops distributed widely across Eurasia 9 and, as such, are easily discernible in palaeodietary isotope studies of human and animal bone collagen. Both foxtail and broomcorn millet were domesticated in China before 5000 BC and spread across Eurasia to Europe by the middle Bronze Age (c. 1500 BC). 10 Millet consumption has been shown both isotopically 11 and archaeobotanically 12 across prehistoric Eurasia. Carbonised millet grains therefore offer an opportunity to study palaeoclimate and farming practices in the past, as well as having the potential to provide baseline information for palaeodietary studies. Given the recent geographic expansion of isotopic archaeological applications, it is now timely to consider further the causes of isotopic variation in millet plants.
In a previous study, 13 we reported on isotopic variation in different Setaria italica accessions grown in a controlled environment chamber. Our reasons for choosing Setaria italica included: its importance to archaeology; its high levels of intraspecific variability plus the recent sequencing of its genome (which facilitates analysis of the functional genetic variation underlying phenotypic variability) [14][15][16] ; and its relatively short life cycle. 8 That study showed significant isotopic variability within single leaves and panicles, and between leaves and panicles within the same plant. Carbon isotope ratios in leaves and grains varied by c. 2‰ between different accessions (a plant or grain sample, variety or population, collected from a particular area and kept in a gene bank for conservation, cultivation and research), while nitrogen isotope ratios in grains varied by c. 6‰. There was an average offset of 0.9‰ between leaf and grain carbon isotope ratios.
Here, we build on this previous research by characterizing carbon and nitrogen isotopic variation in Setaria italica plants subjected to different watering regimes. We grew four plants each of 12 Setaria italica accessions and subjected the plants to four different watering regimes (hereafter 'experimental lines'). Control plants were also grown to characterise intra-line variation due to environmental variation within the growth chamber and genetic variation within the line.

| C 4 photosynthesis and isotope discrimination
There are two major photosynthetic pathways, C 3 and C 4 , which use different methods of uptake of carbon dioxide from the atmosphere. C 4 plants are more efficient in terms of water and nitrogen use than C 3 plants, and have higher light use efficiencies above 25-30°C. 17,18 The majority of the world's plants use the C 3 pathway, but several important crop plants are C 4 including maize, sugar cane, sorghum and the millets. It is well established that multiple environmental and genetic factors affect the carbon isotope ratios of C 3 plants. 19,20 These differences allow the use of carbon isotope ratios of charred plant remains to infer the environmental conditions under which they grew. C 4 plants, however, are thought to be relatively insensitive to environmental factors and show less isotopic variability. 21 Both photosynthetic pathways discriminate against 13 C during the uptake of CO 2 , with C 4 plants discriminating less than C 3 plants.
Isotopic discrimination in C 3 plants is well understood and is largely controlled by the diffusion of CO 2 through the stomata and the action of enzymes. 19,22 Isotopic discrimination in C 4 plants is less well understood, but a theoretical basis has been presented. 22,23 The dissolution and hydration of CO 2 , and CO 2 leakage from bundle sheath cells, as well as the stomatal and enzymatic components, are important. As primary fixation of CO 2 occurs efficiently at lower concentrations than in C 3 plants, C 4 plants are less sensitive to the partial pressure of CO 2 inside the leaf mesophyll and in the atmosphere. Discrimination should increase either through increases in the amount of CO 2 that leaks out of the bundle sheath cell, or in the concentration of the enzyme phosphoenolpyruvate (PEP) carboxylase. 24 There are three subtypes of C 4 photosynthesis, relating to the different enzymes used to release CO 2 in the bundle sheath cells.
Although the reasons are not fully understood, these subtypes show small differences in δ 13 C values. 22,25,26 Setaria italica uses the NADP-ME (NADP-malic-enzyme) subtype, which has the highest δ 13 C values of the three subtypes when they are grown under controlled conditions. 25,26 Early compilations of plant carbon isotopic data showed that the range in C 3 plants was larger than that of C 4 plants, 27 31 and foxtail millet (Setaria italica). 13,32 Isotopic differences have been seen between photosynthetic and non-photosynthetic tissue in C 4 plants. 13,[32][33][34] In terms of different chemical compounds, alkanes and lipids have been shown to have δ 13 C values that are 8-10‰ lower than those of bulk leaf matter in C 4 species, 35 and cellulose δ 13 C values tend to be higher than those of lignin). 36,37 Turning to environmental parameters, studies have shown relationships between C 4 plant isotope ratios and light intensity, [38][39][40] salinity, 41 latitude, 32 altitude 42 and water availability, 38,43,44 although the relationship in each instance is not always simple or linear. 32,45 In order to use C 4 plants to reconstruct past environments and farming practices, we need to understand the isotopic variation  (although studies on individual species often fail to find such relationships 5,71 ). These relationships are believed to relate to higher nitrogen loss in hot, arid environments than in colder, drier environments, which tend to conserve and recycle nitrogen. 49 Nitrogen loss is associated with large fractionations, leaving the remaining soil nitrogen enriched in 15  Where C 4 plant nitrogen isotope ratios may be high due to aridity, distinguishing between C 4 consumption and marine consumption may not be possible on the basis of bulk collagen isotope ratios alone. It is therefore important to understand the extent to which aridity can increase nitrogen isotope ratios in staple C 4 plants.

| MATERIALS AND METHODS
A total of 12 accessions of Setaria italica were analysed in this study, In previous experimental work, randomly chosen grains from each of the 360 accessions were sown and plants grown to maturity, with panicles bagged to prevent cross-pollination. 73 The resulting S1 selfed grain (i.e. the progeny of a plant where the only pollen that could reach the stigma of the flowers was the pollen from the anthers of that same plant) was harvested, and these grains were used in the previous experiment and as part of the wider study of Setaria italica genetic diversity. 13,73 With the exception of one of the control accessions, the grains harvested as part of the initial study were used in the current experiment. The accessions grown here therefore represent seed derived from a second-generation plant (S2  based on a 13-cm diameter pot size and a growing time of 120 days (which in reality proved to be an underestimate).
Watering regime D was originally expected to be 500 mL every 2 days; however, this was reduced to 300 mL as that was the maximum amount that would reasonably fit into each plant pot.
Nevertheless, the regime D plants had an excess of water, and it can be assumed that they were waterlogged at points during the experiment, although the redox potential of the soil was not quantified. The excess water remaining in the tray was discarded before each watering session. Two sets of control plants were also grown, with six replicates each, and watered using watering regime

| RESULTS
The full dataset is given in the supporting information.

| Leaf carbon isotope variation
The δ 13 C leaf results are summarised by line in Table 1  The δ 13 C leaf results are summarised by watering regime in Table 2 and shown in Figure 2A. There are three exceptions -SIT0040, SIT0150 and SIT0586.

| Grain carbon isotope variation
The δ 13 C grain results are summarised by line in Table 3  The δ 13 C grain results are summarised by watering regime in Table 4 and shown in Figure 2B.

| Grain nitrogen isotope variation
The δ 15 N grain results are summarised by line in Table 5 and shown in Figure 1C. Control line SIT0555 (n = 6) has a δ 15 N grain range of  The δ 15 N grain results are summarised by watering regime in Table 6 and shown in Figure 2C.   Comparing the δ 13 C leaf and δ 13 C grain results by watering regime clearly shows that the amount of water given to the plants had a strong effect on the carbon isotope ratios for both leaves and grains.
In fact, the watering regime accounts for over 80% of the variation in δ 13 C values (r s = 0.88 and 0.83 for leaf and grain, respectively). In  theory, therefore, Setaria italica carbon isotope ratios can be used for the reconstruction of water availability in the present and also the past (provided, of course, that the other potential problems are resolved, such as preservation of the primary isotope signal, removal of contamination, and so on). There are, however, two problems which are likely to make this difficult in practice.
First, the mean difference in carbon isotope ratios between watering regime A (50 mL) and watering regime D (300 mL) is only     A second problem is the nature of the correlation between water availability and carbon isotopic ratios in millet. In most C 3 plants, there is a negative relationship between water availability and carbon isotopic ratios, related to water use efficiency (WUE). 22 In this study we found a positive correlation between the carbon isotope ratio and the amount of water given, as was also found by An  the δ 13 C value will decrease with increasing water availability.
However, this pattern is not the case for our samples. The δ 13 C value of a C 4 plant may therefore either increase or decrease with increasing water availability. While there may be scenarios where determining change is the primary aim, in most scenarios the direction of said change towards higher or lower water availability is  probably the purpose of the study. The use of C 4 plants to study water availability in the past and present, therefore, seems to be of limited potential.
While not the aim of this study, we note that the mean difference in δ 13 C values between grains and leaves (1.3‰) is slightly higher than that seen in other studies. 13,32 As noted elsewhere, this pattern has implications for the interpretation of animal and human bone collagen isotope results, particularly where humans and animals eat different parts of the same plant. 13

| Nitrogen isotopic variation
The control lines have δ 15 N grain interquartile ranges that are generally smaller than those of the experimental lines but for three experimental lines (SIT0164, SIT0248 and SIT0560) this is not the case. This pattern indicates that the variation caused by intra-line differences and any variation caused by position in the growth chamber are, in some cases, as big as that caused by the watering regime.
The δ 15 N grain variation within the two control lines is similar when the outliers are excluded and is more substantial in control line SIT0555 when the outliers are included. This pattern is the opposite to would be expected given that control line SIT0555 was grown from S2 selfed seed while control line SIT0560 represents grain from the original accession (i.e. grain derived directly from the germplasm bank). Nevertheless, this pattern indicates that the analysed landrace is not more diverse isotopically than the selfed lines.
Comparing the δ 15 N grain results by watering regime indicates that while the watering regime does have an effect on plant nitrogen isotope ratios, this effect is not as expected. There is not a simple relationship between the nitrogen isotope ratio and the amount of water given, nor do the plants given the lowest amount of water have the highest nitrogen isotope ratios, as would be expected with an aridity effect. [81][82][83] This finding indicates that Setaria italica grain δ 15 N values are not negatively correlated with water availability and, as such, they cannot be used as a palaeoclimate proxy in this way. It follows from this that aridity cannot simply be used to explain high human bone collagen δ 15 N values in populations consuming millet, as while aridity does affect plant δ 15 N values this is not necessarily in a predictable way (e.g. 84,85 ). Rather, the data presented here suggests that, in relation to Setaria italica at least, high δ 15 N values are associated with well-watered (but not over-watered) plants. High nitrogen isotope ratios in both Setaria italica grains and human bone collagen from millet-eating populations may therefore be indicative of optimal water availability rather than aridity.
The within-line δ 15 N variation with watering regime reported here (mean = 2.7‰) is less than the variation seen between 29 different lines in our previous experiment (6‰). 13 This is clearly problematic as, in the case of Setaria italica at least, increases in nitrogen isotope ratios could be related to genetic variation, aridity or manuring, amongst other factors. We would therefore recommend that plant isotope analysis is conducted in conjunction with other studies (such as grain morphometrics, weed seed analysis and other climate proxies) in order to provide a robust understanding of the past.

| CONCLUSIONS
This study has shown that the carbon isotope ratios of Setaria italica are strongly correlated with water availability, but the correlation is the opposite to that seen in studies of C 3 plants. The change in isotopic ratio due to watering regime is comparable with that seen due to change in accession. Thus, distinguishing between changing varieties of Setaria italica and changing climate is problematic. In terms of grain nitrogen isotope ratios, the highest δ 15 N values were not associated with the lowest watering regime, as would be expected if aridity were the cause of these high δ 15 N values. Again, the variation in δ 15 N values is comparable with that expected from an aridity effect or a manuring effect, and thus distinguishing between these factors is likely to be problematic. We suggest that in order to use the stable isotope ratios of archaeological Setaria italica grains to investigate past cultivation practices, these data are best used in conjunction with other lines of evidence.