The 13 C Discrimination of Crops Identiﬁes Soil Spatial Variability Related to Water Shortage Vulnerability

: Spatial variability of crop growth and yields is the result of many interacting factors. The contribution of the factors to variable yields is often di ﬃ cult to separate. This work studied the relationships between the 13 C discrimination ( ∆ 13 C) of plants and the spatial variability of ﬁeld soil conditions related to impacts of water shortage on crop yield. The 13 C discrimination, the indicator of water shortage in plants, 15 N ( δ 15 N) discrimination, and nitrogen (N) content were determined in grains of winter wheat, spring barley, and pea. The traits were observed at several dozens of grid spots in seven ﬁelds situated in two regions with di ﬀ erent soil and climate conditions between the years 2017 and 2019. The principles of precision agriculture were implemented in some of the studied ﬁelds and years by variable rate nitrogen fertilization. The ∆ 13 C signiﬁcantly correlated with grain yields (correlation coe ﬃ cient from 0.66 to 0.94), with the exception of data from the wetter year 2019 at the site with higher soil water capacity. The e ﬀ ect of drought was demonstrated by statistically signiﬁcant relationships between ∆ 13 C in dry years and soil water capacity ( r from 0.46 to 0.97). The signiﬁcant correlations between ∆ 13 C and N content of seeds and soil water capacity agreed with the expected impact of water shortage on plants. The 13 C discrimination of crop seeds was conﬁrmed as a reliable indicator of soil spatial variability related to water shortage. Stronger relationships were found in variably fertilized areas.


Precision Agriculture
Precision agriculture (PA) is based on direct or indirect identification of the spatial variability of various factors, especially field soil conditions, affecting plant growth and yield formation, yield quality, or plant health. Various farm practices are adjusted with respect to the variability of these factors, historical yields, and actual crop status with the aim of increasing productivity and effectivity of resources without negative impacts on the environment [1][2][3][4]. Spatial variability of crop growth and yields is the result of many interacting factors. The contribution of these factors to variable yields is often difficult to separate. Water is a key determinant of plant growth, and thus variable water content, as the result of soil texture and hydraulic properties, slope, etc., affects growth and yield [5][6][7][8][9][10][11].

Experimental Sites
The experiments were carried out on farm fields in two regions with different soil and climate conditions ( Figure 1 and Table A1).

13 C Discrimination
Duffková et al. [23] identified so-called "recharge zones", i.e., the catchment slope zones having a high sand content and infiltration capacity and being declared as vulnerable to nitrate pollution [24], via the monitoring of crop water stress indices from remote sensing combined with field measurements. In this study, we propose to use crop 13 C discrimination as an indicator of spatially variable water shortage during crop growth. In the process of photosynthesis, C3 species discriminate between the heavier isotope 13 C and the lighter 12 C. Under water stress, plants close their stomata to reduce water consumption through transpiration, in turn leading to different concentrations of CO 2 inside the leaves, and to changes in the 13 C and 12 C isotopic ratio [25]. The discrimination of 13 C in seeds integrates the effects of available water supply and plant water status during growth. The plants that sustain their stomata open, thanks to a greater available water supply or to the genotype's effective water management, have a higher ∆ 13 C. This suggests higher photosynthetic activity and a higher yield [26]. For example, Raimanová et al. found in field experiments that every 100 mm of water in precipitation and irrigation corresponded to an increase of ∆ 13 C by approximately 1% in wheat grains [27]. Additionally, ∆ 13 C of soil and its components have been utilized to study soil spatial variability [28,29]. Furthermore, the discrimination of 15 N, together with 13 C, may provide new information on the impact of abiotic stresses [30].
We hypothesize that under conditions of a high spatial variability of soil conditions and soil water capacity, yield differences are caused primarily by the impact of water stress. The discrimination of 13 C isotope, which integrates water availability conditions during growth, should therefore correlate with both yield and soil water capacity, which determines the available water supply in the root zone. It can be expected that in a wetter locality or in rainier years, these relationships will be weaker. The objectives of study were (1) to determine the relationship between the 13 C discrimination of crop seeds and yield and (2) to relate the 13 C discrimination to the spatial variability of soil water capacity at sites with different soil and climate conditions.

Experimental Sites
The experiments were carried out on farm fields in two regions with different soil and climate conditions ( Figure 1 and Table A1).

Lower Jizera River Region
The four experimental fields, denoted So5, So7, and So9, near the village of Sojovice and Ko1 near Kochánky village in the lower Jizera River region were monitored between the years 2017 and 2019 ( Figure 1a, Figure S1, and Table A1). Pea (Pisum sativum L.) and winter wheat (Triticum aestivum L.) were grown in the years 2017-2019 in Kochánky, while winter wheat and potatoes (Solanum tuberosum L.) were grown in the same years in Sojovice. Winter wheat was grown after pea and wheat (Kochánky) or after potatoes or wheat (Sojovice). The data for three fields in Sojovice were merged for correlation analysis as their soil conditions are similar. The soil and plants were sampled at the same grid points; in 2019, three spots were not sampled in Kochánky due to dividing the field for different crops. The field So7 in Sojovice was divided into two parts, A and B, where irrigated potatoes and non-irrigated wheat were grown alternatively.

Bohemian-Moravian Highlands Region
Three fields with winter wheat (Triticum aestivum L.) and a field with spring barley (Hordeum vulgare L.) were studied between the years 2017 and 2019 in the Bohemian-Moravian Highlands, near the villages of Kojčice (Koj), Svépravice (Sv), and Krasíkovice (Kr) (Figure 1b, Figure S1, and Table A1). Winter wheat was grown after oilseed rape (in 2017 and 2018) or after spring barley (2018). Spring barley (2019) was grown after potatoes. Precision agriculture principles were used in some areas of the field, i.e., mineral N fertilizers were applied in a spatially variable manner according to long-term crop yield potential (2017-2018) and the actual crop status derived from satellite images of Sentinel 2a and 2b (2018) ( Figure S1). Other areas of the field were fertilized uniformly, i.e., without any respect to soil heterogeneity and its effect on crop water and nutrient uptake.
At both sites, the experimental crops were grown without irrigation.

Sampling and Soil Analysis
In the lower Jizera River region, the soil and plants were sampled in the same 21 ( The soil sampled at the grid points was analyzed for soil texture classes by the pipette method [31], which determined the particle size distribution in the topsoil (0-30 cm) and subsoil (30-60 cm) layers at both sites and in the deep subsoil (60-90 cm) in the lower Jizera River region.
A fine particle size fraction (FPSF, %) < 0.01 mm was used for calculating the FWC, the wilting point (WP), and the AWC with simple pedotransfer functions [32]: Agronomy 2020, 10, 1691 The values of water capacity were reduced according to the stone content (>2 mm). The above-ground parts of the plants were sampled from the two areas (0.25 m 2 in the lower Jizera River region or 0.2 m 2 in the Bohemian-Moravian Highlands region) near each grid point shortly before or at maturity (BBCH 87-90). The dry matter weight of grains was determined at both sites and the yields are presented in dry matter.
Water balance was calculated in both regions in each experimental year as the difference between the sum of daily precipitation and the sum of daily Penman's evapotranspiration [33] during the growing season (Figures 2 and 3).

Plant Isotope Analysis
The C isotope content of the seeds was determined using the elementary analyzer EA 3200 (Eurovector, Italy) coupled to the isotopic mass spectrometer Isoprime (GV Instruments, U.K.) at the Crop Research Institute, Prague. Plant material was dried and ground into a fine powder with a ball mill (Retch, Germany).
The value of the carbon isotopic ratio δ 13 C was calculated as the ratio 13 C/ 12 C of a sample and of the PDB (Pee Dee Belemnite) standard: where Rs is the ratio 13 C/ 12 C of the sample and Rb is the 13 C/ 12 C ratio of the PDB. For the description of changes in C isotopes, the value of the 13 C discrimination was used: where δ 13 Ca is the δ 13 C value of air (-8‰) and δ 13 Cp is the measured value of the plant [26]. The content of N (%) and the content of N isotopes 15 N and 14 N were determined. The value of the 15 N isotopic ratio (δ 15 ) was calculated as:

Plant Isotope Analysis
The C isotope content of the seeds was determined using the elementary analyzer EA 3200 (Eurovector, Italy) coupled to the isotopic mass spectrometer Isoprime (GV Instruments, U.K.) at the Crop Research Institute, Prague. Plant material was dried and ground into a fine powder with a ball mill (Retch, Germany).
The value of the carbon isotopic ratio δ 13 C was calculated as the ratio 13 C/ 12 C of a sample and of the PDB (Pee Dee Belemnite) standard: where Rs is the ratio 13 C/ 12 C of the sample and Rb is the 13 C/ 12 C ratio of the PDB. For the description of changes in C isotopes, the value of the 13 C discrimination was used: where δ 13 Ca is the δ 13 C value of air (-8‰) and δ 13 Cp is the measured value of the plant [26]. The content of N (%) and the content of N isotopes 15 N and 14 N were determined. The value of the 15 N isotopic ratio (δ 15 ) was calculated as: Where Rs is the ratio 15 N/ 14 N of the sample and Rb is the 15 N/ 14 N ratio of the standard and the

Plant Isotope Analysis
The C isotope content of the seeds was determined using the elementary analyzer EA 3200 (Eurovector, Italy) coupled to the isotopic mass spectrometer Isoprime (GV Instruments, UK) at the Crop Research Institute, Prague. Plant material was dried and ground into a fine powder with a ball mill (Retch, Germany).
The value of the carbon isotopic ratio δ 13 C was calculated as the ratio 13 C/ 12 C of a sample and of the PDB (Pee Dee Belemnite) standard: where Rs is the ratio 13 C/ 12 C of the sample and Rb is the 13 C/ 12 C ratio of the PDB. For the description of changes in C isotopes, the value of the 13 C discrimination was used: Agronomy 2020, 10, 1691 6 of 15 where δ 13 C a is the δ 13 C value of air (-8% ) and δ 13 C p is the measured value of the plant [26]. The content of N (%) and the content of N isotopes 15 N and 14 N were determined. The value of the 15 N isotopic ratio (δ 15 ) was calculated as: where Rs is the ratio 15 N/ 14 N of the sample and Rb is the 15 N/ 14 N ratio of the standard and the atmosphere air [34].

Statistical Analyses
Statistical data evaluation was performed using Pearson regression analysis (with a significance level of p < 0.05). Statistica 13 (Stat-Soft Inc., Tulsa, OK, USA) software and the R environment [35] were used.

Lower Jizera River Region
The water balance in the experimental years showed a shortage of water for the main part of the growth season of pea and winter wheat ( Figure 2). The precipitation sum from 1 March to 15 July amounted to 196 (42 mm on 29 May), 169 (50 mm on 9 June), and 125 mm in the years 2017, 2018, and 2019, respectively, nearby Brandýs nad Labem climatic station. The precipitation was insufficient to cover the water demand of pea and wheat; thus, the water supply from the soil reserves, which formed during winter and early spring, was an important source for sustaining plant growth. The water capacity of the soil within the root zone influenced the intensity of drought. These conditions strongly differentiated the growth and yields among the monitored points of the sampling grids.
The coefficients of variation (CVs) of the grain yield of pea in 2017 and of wheat in 2018 and 2019 in Kochánky were 55%, 32%, and 34%, respectively. The corresponding CVs of the wheat yields in Sojovice were 49%, 38%, and 50% ( Table 1).
The strong spatial variability in yield corresponded to the variability of the FWC calculated from the soil texture. The FWC ranged from 21.9% to 31.4% in the top 0-30 cm of the soil and from 7.3% to 34.7% in the next 30-90 cm in Kochánky, and from 11.4% to 33.3% in the top 0-30 cm of the soil and from 8.7% to 34.7% in the next 30-90 cm in Sojovice ( Table 1). The variability is equal to a difference of up to approximately 180 mm of water in the root zone (0-90 cm) of winter wheat. Soil moisture in early spring in the experimental fields showed that the soil was not filled to the FWC level; however, the maximum difference in the water content among spots of the same fields in spring reached up to 80-140 mm. The values of the FWC tightly correspond to the calculated values of the available water content.
The calculated FWC correlated significantly with the observed grain yields ( Table 2). Similar relationships were observed between the total harvest biomass. The FWC explained from 58% to 78% of the variability in yield in the experimental years. The analysis showed significant relationships between the calculated FWC, the yield, and the 13 C discrimination of the seeds (∆ 13 C), the indicator of water conditions during growth, at both sites and in all experimental years. The ∆ 13 C showed a strong relationship with yield, i.e., correlation coefficients that ranged between 0.84 and 0.97. Except for one case, the correlation coefficients between 13 C and the FWC were also above 0.80. The nitrogen content of the grains was negatively, mostly significantly related to the FWC, ∆ 13 C, and yield. The 15 N discrimination, δ 15 N, correlated positively with ∆ 13 C, yield, and the FWC, and negatively with the N content of the grains. The corresponding correlations of the AWC were almost identical to the results of the FWC, due to a tight correlation between them. Table 1. Median, mean, and coefficient of variation (CV, %) of the grain yield, the nitrogen (N) content (Ncont) of seeds, 13 C (∆ 13 C) and 15 N (δ 15 N) discrimination, and field (FWC) or available (AWC) water capacity. VAR, fertilized variably; UNI, fertilized uniformly.

Site
Year
In 2017 and especially in 2018, when low sums of precipitation in May and June (73 and 69 mm, respectively) worsened yield formation, the significant relationships between grain yield and the FWC in the top 0-60 cm of the soil with ∆ 13 C were revealed (r = 0.46-0.99; Table 2). In the wetter year 2019, the grain yield was not affected by water deficit and, therefore, the correlations associated with ∆ 13 C were not significant (Table 2). However, a significant dependence of the grain yield on the FWC was proved in all years (r = 0.54-0.84; Table 2). The CV values of the grain yield proved higher spatial variability at sites with a lower FWC (i.e., Kr and Koj) in the very dry year 2018 (27.2% in Kr and 28.2% in Koj) compared to the site with a higher FWC (Sv) in the years 2017 (21.7%) and 2019 (22.0%) ( Table 2), when water deficit was substantially lower compared to the year 2018.
In the years 2017-2018, some areas of the field had mineral N fertilizers applied variably (VAR), according to PA principles, which revealed stronger and significant dependence of the grain yield on the FWC and ∆ 13 C in comparison to the areas that were fertilized uniformly (UNI) ( Table 2). The relationships of the grain yield and ∆ 13 C associated with the FWC were not significant. In the very dry year 2018, variable mineral N doses correlated significantly with ∆ 13 C (Kr r = 0.88, Koj r = 0.62) compared to the year 2017, when winter wheat was grown at the site with a higher FWC (Sv r = 0.27).
Similarly to the lower Jizera River region, the nitrogen content of the grains was negative, mostly significantly related to the FWC, ∆ 13 C, and yield. The 15 N discrimination, δ 15 N, correlated positively with ∆ 13 C, yield, and FWC, and negatively with the N content of the grains, with the exception of the wet year 2019 in Svépravice.

Discussion
The dry weather in the lower Jizera River region contributed, together with a light and permeable soil, to a pronounced manifestation of soil variability in plants in all years. Strong differences in plant height, growth, the time of onset and the severity of water stress, withering, and drying were distinguishable in fields at the scale of ≤1 m. This proves the notion the plant itself is the best indicator of soil conditions and fertility. Our results and a vast amount of data in the literature show that cereals, especially wheat, are deeply rooted and are able to effectively exploit water from zones down to 80-100 cm and even deeper [22,36,37]. As a result, plant growth and yield integrate various soil characteristics with the root zone in interaction with weather and agronomic factors [5]. Furthermore, the effect of soil conditions is intensified by a lower content of available nutrients in soil with higher proportions of sand and stones in comparison with more silty soils.
In the Bohemian-Moravian Highlands region, the years 2017 and especially 2018 were warm and dry, while the higher precipitation in 2019 improved the agro-meteorological water balance, and thus the crop growth and yield formation were determined by other soil-related factors. Spring barley plants (grown in 2019) benefited from the greater amount of precipitation in May (Figure 3), an important period for the formation of yield (number of fertile tillers and grains per year). Spring barley is known to have shallower roots and to be more susceptible to worse soil conditions than wheat, which has a longer growth period and a greater root system [22]. The lack of a relationship between yield or soil water capacity and ∆ 13 C in barley is unlikely attributed to a better ability of barley to withstand water shortage, and the observed significant variability in yield was caused by other spatially variable factors.
Except for spring barley in 2019, the ∆ 13 C explained from 58% to 95% of the yield variability in the lower Jizera River and Bohemian-Moravian Highlands regions, which was always better than relating the yields to the FWC (29%-77%). The better performance of ∆ 13 C could be explained by integrated soil and water conditions of the plants in the sampled areas. Most studies found a significant relationship between the ∆ 13 C of the leaves or seeds and yield under different water supplies (e.g., [27,38]); however, there is a lack of work on the spatial variability of the ∆ 13 C in field crops. According to Clay et al., the ∆ 13 C of soya seeds explained 62% of the total yield variability [11]. However, the authors noted that ∆ 13 C can be used to help assess water stress, provided that N stress is absent. Similarly, Clay et al. suggested that ∆ 13 C provides an index for water stress under non-nutrient-limiting conditions [10]. The data presented herein came from experimental fields fertilized with optimal rates of nutrients; however, visual signs of nutritional problems could be observed in a few spots with sandy and stony soils at the Kochánky and Sojovice sites.
As expected, more results could be found on the spatial variability of soil water-related traits in relation to crop growth and yield. Wong and Asseng found, under Mediterranean climate conditions, linear relationships between grain yield and the plant available water storage capacity (PAWc) of the top 100 cm of the soil profile in a 70-ha field [39]. Unlike our results, the authors found that spatial variability was reduced in low rainfall years when yields across the field were low and the higher soil water storage capacity of some sites was often underutilized. With adequate N, spatial variability increased with seasonal rainfall at sites with higher PAWc, which conserved more water in wet seasons to provide a higher yield response than sites with low PAWc. Under the transitional (maritime-continental) climate conditions of the Czech Republic, the situation is different; soil water supply is always at least partially replenished during the autumn-spring period. Unlike many other authors, Castellini et al. found a negative correlation between PAWc and durum wheat yield under Mediterranean climate conditions, probably because the soil parameters were determined only to a 10-cm depth [8]. Lipiec and Usowicz [40] reported that, under the wetter and colder conditions of central Poland, cereal grain yields were significantly negatively correlated with the sand content in the top-and subsoil in one of three experimental years, while Usowicz and Lipiec reported a positive correlation between cereal yields and topsoil water or subsoil clay contents [9].
The FWC and AWC are not the only traits that determine the water dynamics and availability for plants [8]. For example, the effect of other factors, such as the tracks of farm machines, may influence soil water parameters [41,42]. We observed similar significant correlations between ∆ 13 C, FWC, pea, or wheat yields and the simple index derived from aerial RGB images taken from the air during sugar beet and wheat growth in the years 2015 and 2016 (previous to the experimental ones) in fields in Kochánky [43]. This showed the conservative character of soil traits connected with vulnerability to drought.
The impact of different soil water capacities under low precipitation during growth could be expected. The soil texture, especially the clay content (proportion), also affects nutrient content. The correlations between grain yield and the content of Ca and Mg in the top-and subsoil layers were mostly significant (r = 0.4-0.7) thanks to the relationship between the nutrients and the texture and humus content. Further, Müller proposed that an increasing content of clay functions as a "lubricant" for root growth, while a high content of sharp sand particles can damage root tissues [44]. Little is known about the effect of soil spatial variability on root size and depth. Kirkegaard et al. found an additional 10.5 mm of subsoil water uptake from the 1.35-1.85-m soil layer after the flowering stage, leading to a yield increase of 0.62 t ha −1 in wheat [45]. We observed shallower roots of wheat in zones with sandy soil and a high proportion of stones (soil particles > 2 mm) in comparison to spots where the proportion of silt and clay was higher. Deeper roots provide access to a larger supply of water in the subsoil and vice versa; thus, the impact of soil texture on the root system may contribute to spatial differences, resulting in variable growth, yield, and 13 C discrimination. As a further example, a higher number or better stability of biopores after pre-crop roots or earthworms and other soil organisms in medium and heavy soils may contribute to root growth and penetration into the deeper layers of the subsoil [46].
The experimental sites in the Bohemian-Moravian Highlands were selected because of their colder and wetter climate conditions in comparison to the lower Jizera River region. However, the climate has been changing during recent decades, and shorter or longer periods of low precipitation and above average temperatures, resulting in higher evapotranspiration, are becoming common even in highland regions. Similar results as in the drier and warmer region of the lower Jizera River, in terms of the significant relationships among 13 C discrimination, yield, and soil water capacity, confirm that the water supply has become a key condition for high yields in the region [12,13]. The more frequent occurrence of water shortages will demand new approaches to farm management.
Low soil water capacity and vulnerability to drought inevitably result in an increased loss of nutrients, especially nitrate leaching. The risk of nitrogen losses is enhanced by worsened growth and yield, leading to lower utilization of N from conventionally applied fertilizers and a higher residual nitrate content [5]. Both experimental regions are known for a higher risk of nitrate loss by leaching [16,21,47,48] and both are included in the nitrate-vulnerable areas that encompass 49% of the agricultural soils in the Czech Republic. The adoption of precision agriculture approaches, including precision irrigation [49], respecting the spatial variability of soil water capacity and the vulnerability to drought, may significantly improve the nitrogen efficiency of crops [50].
Furthermore, the results of Dalal et al., i.e., the linear relationship between ∆ 13 C in grain and the nitrogen utilization efficiency in wheat, suggest an interrelated effect of nitrogen nutrition and water availability [51]. We observed mostly significant correlations between the ∆ 13 C of grains or yield and the nitrogen content of grains. This suggests that under water shortages, N is not effectively utilized for grain production, as the nitrogen was not diluted by growth. This was confirmed by a negative correlation between N content and the FWC. The observed negative relationships between N concentration and the δ 15 N of grains agree with the lower utilization of N from mineral fertilizers with a lower δ 15 N signature than N depleted from the soil supply (derived from organic fertilizers and soil organic matter) due to water shortage [52,53]. The lower values of δ 15 N are likely also caused by lower soil N supply from the mineralization of soil organic matter in lighter, sandier, and stonier field spots.

Conclusions
The monitoring of several fields confirmed the working hypothesis about the 13 C discrimination. The impact of water stress on plant yield reliably indicated spatially variable soil conditions related to soil vulnerability by water shortage. The effect of water shortages was confirmed by correlations among yield, 13 C discrimination, and soil water capacity. The hypothesis about weaker relationships of ∆ 13 C under higher precipitation corresponded to results of spring barley in the year 2019. The soil vulnerability to water shortages also indicates a spatially specific higher risk of nitrate leaching. The adoption of precision agriculture approaches may contribute to both the adaptation to climate change-induced water shortage episodes and to reducing the nutrient leaching in spatially variable soils. The results of this study suggest that drought will play a major role in precision nitrogen management due to precipitation fluctuations and the deterioration of the water balance. The 13 C discrimination of crops will help differentiate and identify the main causes of spatial variability as a necessary basis for application of relevant agronomic measures. A proven approach makes it possible to integrate immediately vulnerable zone maps into the precision farming system. Funding: This research was funded by projects of the Ministry of Agriculture of the Czech Republic (grant numbers CRI RO0418 and RO0218) and by the Technology Agency of the Czech Republic (grant number TH02030133).

Acknowledgments:
The authors thank Barbora Henzlová and Věra Schäferlingová (CRI) for their technical support, Jakub Brom for creating NDVI images, and Ondřej Holubík for soil classification.

Conflicts of Interest:
The authors declare no conflict of interest.