The Effect of Granular Commercial Fertilizers Containing Elemental Sulfur on Wheat Yield under Mediterranean Conditions

The demand to develop fertilizers with higher sulfur use efficiency has intensified over the last decade, since sulfur deficiency in crops has become more widespread. The aim of this study was to investigate whether fertilizers enriched with 2% elemental sulfur (ES) via a binding material of organic nature improve yield when compared to the corresponding conventional ones. Under the scanning electron microscope, the granules of the ES-containing fertilizer were found to be covered by a layer of crystal-like particles, the width of which was found to be up to 60 μm. Such a layer could not be found on the corresponding conventional fertilizer granules. Several fertilization schemes with or without incorporated ES were tested in various durum wheat varieties, cultivated in commercial fields. The P-Olsen content of each commercial field was found to be correlated with the corresponding relative change in the yields (YF/YFBES) with a strong positive relationship. The content of 8 ppm of available soil phosphorus was a turning point. At higher values the incorporation of ES in the fertilization scheme resulted in higher yield, while at lower values it resulted in lower yield, compared with the conventional one. The experimental field trials that established following a randomized block design, were separated in two groups: One with P-Olsen ranging between 18–22 ppm and the other between 12–15 ppm, the results of which corroborated the aforementioned finding. The use of ES in all portions of fertilization schemes provided higher relative yields. The coexistence of ES with sulfate in the granule was more efficient in terms of yield, when compared to the granule enriched with ES alone under the same fertilization scheme and agronomic practice. The application of fertilizer mixtures containing the urease inhibitor N-(n-butyl) thiophosphoric triamide (NBPT), ES and ammonium sulfate resulted in even higher relative yields. Yield followed a positive linear relationship with the number of heads per square meter. In this correlation, the P-Olsen content separated the results of the two groups of blocks, where the applied linear trend line in each group presented the same slope.


Introduction
Sulfur (S) deficiency in crops has become a common agronomic issue in many countries, mainly due to reduced atmospheric inputs and the use of sulfur free high-analysis fertilizers (i.e., fertilizer

Commercial Field Trials
In the wheat growing season of 2015-2016, several crops with durum wheat were established in different areas of central and northern Greece (Groups A1 and A2). Sowing and harvesting took place mid November 2015 and end June 2016 respectively. In 2016-2017, more crops were established in the same or other fields (Group A3). Sowing and harvesting took place mid November 2016 and end June 2017 respectively. All fields were of commercial use. Each production field was divided into two equal areas; one of them was subject to conventional fertilization treatment according to the local agricultural practices (F-trial), while the other one received the corresponding FBES-trial. The fertilization scheme and the agronomic details per trial (i.e., area in ha, sowing rate, cultivar, fertilizer type and rate) are provided in Table 1, along with the percentage relative yield. Soil fertility of each field was determined prior to crop establishment. In order to ensure comparable soil conditions between the corresponding F-and FBES-trials, each area was arbitrarily divided into plots 15 m × 7 m (105 square meters) each [14]. All perimetric plots were excluded, while the internal plots were grouped into five successive groups. Within each group of plots, one composite sample per group (0.5 kg) was collected at the depth of 0-20 cm, these samples were pooled together and a final composite sample per trial (2.5 kg) was formed and analyzed. Soil analysis data and yields are provided in Table 2. Only the fields with comparable F-and FBES-trials with regard to soil fertility are further analyzed and discussed in this study. Table 1. Fertilization schemes per trial. FT, field trial; C, field trial with the conventional fertilization (F-)treatment, S, the corresponding field trial with incorporated elemental sulfur in the fertilization scheme (FBES-treatment); F1, initial fertilization along with sowing; F2a, first additional fertilization; F2b, second additional fertilization; ∆x/x, percentage relative change of grain yield; 46-0-0, urea; 40-0-0, urea plus ammonium sulfate; 34.5-0-0, ammonium nitrate; /ES, fertilizer granules with incorporated 2% elemental sulfur.

Experimental Field Trials
Based on the experience gained during the seasons 2014-2015 and 2015-2016, two experiments with durum wheat crops (cv. Simeto) were established in the experimental fields of the Agricultural University of Athens in Aliartos at Viotia county during the season 2016-2017, based on a randomized block design. Each plot was 18 square meters and 1 m apart from the next ones. In each plot of the provided area for experimentation, the P-Olsen content was determined and two areas of almost 0.2 ha each were located, one with P-Olsen values ranging between 18-22 ppm (Group B1) and the other between 12-15 ppm (Group B2). Soil samples were collected at depth of 0-20 cm prior to the crop establishment and fertilizer application. The plot that received ammonium phosphate (16-20-0-13S) as initial fertilizer (starter) and urea (U; 46-0-0) as additional fertilizer (topdressing) served as the reference crop. The compatibility of urea-FBES (U + 2% ES; U/ES) or urea ammonium sulfate (UAS + 2% ES; UAS/ES) along with the corresponding ones containing the urease inhibitor NBPT (U/NBPT) or urea ammonium sulfate (UAS/NBPT) was also studied in this experiment. Each combination was repeated three times.

The Nature of the FBES Granules
The FBES granules tested in this research (16-20- mixed initially with ES in the form of dust at a percentage of 2% to 4% (w/w, ES to fertilizer granules). The mixture passed through a shower of fine droplets consisting of a 1:1 mixture of molasses and glycerol, which acted as the binding system, in such a way that the whole surface was exposed. The binder was added at a percentage of 0.4% to 1.2% (w/w, binder to fertilizer granules). Then, the mixture was led to a mixer, where the sticky ES dust was evenly attached onto the sticky fertilizer granules, thus forming the final FBES product [9].

Soil Analysis
After the soil samples were collected, were air-dried and ground to 2 mm prior to analysis. The particle size analyses were conducted using the hydrometer method, with a 2-h reading for the clay content [15]. The CaCO 3 equivalent percentage was estimated using a digital calcimeter [16]. Soil pH was measured in a 1:1 soil: Distilled water (w-v) suspension [17] and electrical conductivity (ECe) in saturation extract. Soil organic matter (SOM) was determined using the Walkley-Black wet digestion method [18] and the available P (P-Olsen) according to the Olsen method [19]. Exchangeable K (K exch ) was determined using the ammonium acetate extraction method [20]. Extractable Fe, Mn, Cu and Zn were determined with diethylene-triamine-pentaaceticacid (DTPA) [21]. Many studies have shown that DTPA is the most widely used extractant for the determination of metal availability in soils [22,23]. The soil humic substances content (HS) was determined according to Jones (1999) [24]. The main soil properties of each field trial are presented in Table 2.

Head Measurements
In the experimental field trials harvesting took place on 1-3 July 2017. Per plot, all stems were collected and were placed into bags. In the laboratory, each stem was separated from its head, and the heads were weighed. Three sets of a hundred heads each were weighed, and the seeds per set were separated, collected and weighed.

Scanning Electron Microscopy
Cross sections of granules were made using razor blades and mounted on SEM pedestal using conductive silver epoxy paint (EMS, Hatfield, USA). Samples were then sputter coated with gold (Denton Vacuum Desk II Sputter coater, Denton Vacuum, Moorestown, USA) and observed with a JSM-5410 scanning electron microscope (JEOL, Tokyo, Japan) at a working distance of 40 mm at 10 kV.

Statistical Analysis
Yield levels were handled as continuous variables. A linear mixed effects (LME) model [25] was used to assess the statistical significance of the covariates on the response variable, taking into account the within-group variability that exists for the repeated measurements setting. Analysis of variance (ANOVA) was performed on the LME model to determine which of the model covariates (i.e., Phosphorus level, Initial Fertilization and Additional Fertilization) had a significant effect on the average value of yield. All higher-order interactions were tried. Tests were declared significant when p < 0.05. All analyses were performed using the R statistical software (R Development Core Team, 2018) [26].
Under the SEM, granules of the ES-containing fertilizer were found to be covered with a layer of loose material containing fine crystal-like particles, obviously representing the ES powder layer attached via the binder ( Figure 1A,B). The width of the layer was found to be up to 60 µm. Such a layer could not be found on granules of the conventional fertilizer ( Figure 1C). Under the SEM, granules of the ES-containing fertilizer were found to be covered with a layer of loose material containing fine crystal-like particles, obviously representing the ES powder layer attached via the binder ( Figure 1A,B). The width of the layer was found to be up to 60 μm. Such a layer could not be found on granules of the conventional fertilizer ( Figure 1C).

The Effect of the Fertilization Schemes on Yield
Five fields of Group A1 contained no calcium carbonate or traces of it, with pH ranging from 6.20 to 7.74. Four of them characterized by adequate fertility levels, and the FBES-trials presented a relative yield increase of 8.1%, 18.5%, 26.9%, 28.5%, compared to F-trials. In the field FT_07 of Group A1 that was characterized by very low soil initial phosphorus content coupled with low potassium content, the yield of the FBES-trial presented a relative yield decrease (−10.9%) (Tables 1,2).
Five fields of Group A2 contained moderate or high calcium carbonate, with pH ranging from 7.96 to 8.20. In three of them, those with adequate fertility levels, the FBES-trials relative yield increased by 2.7%, 4.2%, 7.8%, respectively. The other two cases of this category presented relative decreases in the yield (−3.4%, −26.8%). The rhizosoil of these cases was characterized by very low initial phosphorus content coupled with marginal iron content, while the rhizosoil of the latter one (−26.8%) additionally presented low concentration of humic substances along with very low sand content (Tables 1,2). Table 1. Fertilization schemes per trial. FT, field trial; C, field trial with the conventional fertilization (F−)treatment, S, the corresponding field trial with incorporated elemental sulfur in the fertilization scheme (FBES−treatment); F1, initial fertilization along with sowing; F2a, first additional fertilization; F2b, second additional fertilization; Δx/x, percentage relative change of grain yield; 46-0-0, urea; 40-0-0, urea plus ammonium sulfate; 34.5-0-0, ammonium nitrate; /ES, fertilizer granules with incorporated 2% elemental sulfur.

Group
Field Trial Area ha

The Effect of the Fertilization Schemes on Yield
Five fields of Group A1 contained no calcium carbonate or traces of it, with pH ranging from 6.20 to 7.74. Four of them characterized by adequate fertility levels, and the FBES-trials presented a relative yield increase of 8.1%, 18.5%, 26.9%, 28.5%, compared to F-trials. In the field FT_07 of Group A1 that was characterized by very low soil initial phosphorus content coupled with low potassium content, the yield of the FBES-trial presented a relative yield decrease (−10.9%) (Tables 1 and 2).
Five fields of Group A2 contained moderate or high calcium carbonate, with pH ranging from 7.96 to 8.20. In three of them, those with adequate fertility levels, the FBES-trials relative yield increased by 2.7%, 4.2%, 7.8%, respectively. The other two cases of this category presented relative decreases in the yield (−3.4%, −26.8%). The rhizosoil of these cases was characterized by very low initial phosphorus content coupled with marginal iron content, while the rhizosoil of the latter one (−26.8%) additionally presented low concentration of humic substances along with very low sand content (Tables 1 and 2).
Most of the five fields of Group A3 contained significant amounts of calcium carbonate, with pH ranging from 7.74 to 7.88. In two of them, characterized by adequate fertility levels, the yields of the FBES-trials presented relative yield increases (3.2%, 9.2%). In the fields that characterized by very low soil initial phosphorus content (FT_4, FT_3, FT_5), the yield of the FBES-trial presented a relative yield decrease (−5.9%, −13.0%, −18.0%) ( Table 2).

P-Olsen as a Yield-Limiting Factor
The P-Olsen content of each field was correlated with the corresponding relative change in the yields (Y F /Y FBES ) and a strong relationship was revealed. This held true for both wheat growing seasons (Figure 2A-C). The season 2015-2016 provided good yields, while the season 2016-2017 provided poor ones. Despite this fact, the relative change due to the application of the elemental sulfur seemed not to be influenced by the Y F level. As a next step the data of both years were combined

The effect of ES and NBPT Containing Fertilizers on Yield
The results of experimental field trials supported the aforementioned finding. The corresponding treatments within Group B1 with P-Olsen values ranging between 18-22 ppm (Table  3 and Figure 3) provided higher yields compared with those of Group B2 with P-Olsen values ranging between 12-15 ppm, i.e., less phosphorus by −47% to −50%; (Table 3 and Figure 3). Within each group, the same rank was obtained, i.e., U < UAS < U/ES < UAS/ES. The use of incorporated ES along with the initial fertilization provided statistically higher yields in most cases. This held true within both groups B1 and B2.

Treatment
In Group B2, i.e., when P-Olsen was marginal (12-15 ppm), the situation altered. In this case the rank was U < UAS < U/ES < UAS/ES < U/NBPT < U/ES/NBPT < UAS/NBPT < UAS/ES/NBPT. Again, UAS provided higher yields compared to U alone. Incorporation of NBPT provided higher yields compared with those of incorporated ES, while UAS/ES/NBPT provided the highest possible yields under the circumstances. Again, the incorporation of elemental sulfur in the initial fertilization provided even higher yields.
In all cases the incorporation of ES in the fertilizer granule provided higher yields than the fertilizer granule alone. Moreover, the 50:50 combination of UAS/ES and UAS/NBPT (UAS/ES/NBPT) provided the highest yields. It is noteworthy that in such a combination actually 1% ES and half the rate of the urease inhibitor were applied to the plot with the additional fertilization.
ANOVA resulted in a model which shows that the linear effects of Phosphorus level (P-Olsen), Initial fertilization (IF) and Additional Fertilization (AF), as well as the pairwise interaction of Phosphorus level with Initial fertilization have a statistically significant effect on the yields' levels (Tables 4 and 5).

Yield and Head Characteristics
Yield presented a positive linear relationship with head mass per square meter ( Figure 4A), with the same slope for both Groups B1 and B2. This held true for the correlation of yield with the number of heads per square meter ( Figure 4B), however in this correlation Group B1 was separated from Group B2. The mass per head presented a negative linear relationship with the number of heads per square meter ( Figure 4C), and again Group B1 was separated from Group B2. Such a negative linear relationship was also revealed when mass per head was correlated with the ratio of seed mass per head mass ( Figure 4D). Again, the two groups were separated, however the slope was not steep.
Yield presented a positive linear relationship with head mass per square meter ( Figure 4A), with the same slope for both Groups B1 and B2. This held true for the correlation of yield with the number of heads per square meter ( Figure 4B), however in this correlation Group B1 was separated from Group B2. The mass per head presented a negative linear relationship with the number of heads per square meter ( Figure 4C), and again Group B1 was separated from Group B2. Such a negative linear relationship was also revealed when mass per head was correlated with the ratio of seed mass per head mass ( Figure 4D). Again, the two groups were separated, however the slope was not steep.

The Role of Soil Available P on the Relative Yield
The results of the relative yields were contradictory, as some FBES-trials provided lower yields compared to the corresponding conventional ones. It was further shown that soil available phosphorus was a strong limiting factor and the P-Olsen value of 8 ppm was established as the turning point. Above this threshold the relative yield was increased.
Sulfate-S is the available S source to the plants, while ES as S source is not readily available to plants. The conversion of ES to sulfate requires the action of microorganisms [27]. Recent studies showed that S-oxidizing bacteria activity depends on the variations in soil organic C and nutrient availability [28]. In a recent study, it was concluded that in soils with pH > 6.65, along with high S and organic matter content, the ES oxidation rate is higher compared to the other soils [29]. According to our results, it is speculated that the low organic matter content and the low available phosphorus are limiting factors for the activity of S-oxidizing rhizosoil microbes, the growth of which is boosted by the added ES with the fertilizer granule. We have shown that the application of FBES fertilizers increases the total numbers of microbial populations in the rhizosphere [30]. In the

The Role of Soil Available P on the Relative Yield
The results of the relative yields were contradictory, as some FBES-trials provided lower yields compared to the corresponding conventional ones. It was further shown that soil available phosphorus was a strong limiting factor and the P-Olsen value of 8 ppm was established as the turning point. Above this threshold the relative yield was increased.
Sulfate-S is the available S source to the plants, while ES as S source is not readily available to plants. The conversion of ES to sulfate requires the action of microorganisms [27]. Recent studies showed that S-oxidizing bacteria activity depends on the variations in soil organic C and nutrient availability [28]. In a recent study, it was concluded that in soils with pH > 6.65, along with high S and organic matter content, the ES oxidation rate is higher compared to the other soils [29]. According to our results, it is speculated that the low organic matter content and the low available phosphorus are limiting factors for the activity of S-oxidizing rhizosoil microbes, the growth of which is boosted by the added ES with the fertilizer granule. We have shown that the application of FBES fertilizers increases the total numbers of microbial populations in the rhizosphere [30]. In the initial experiment [13], the 20-10-10/ES that was applied at sowing contained both ES and sulfate, i.e., both readily available and non-available S to plants, thus supporting crop growth with sulfate for an extended period. The role of soil microbes in the biogeochemical S cycle and S supply of plants has been reviewed; fungi and bacteria release S from sulfate-esters using sulfatases [31,32], while the interactions of roots with soil microorganisms, especially non-symbiotic plant growth promoting rhizobacteria, in relation to nutrient availability, as well as the mechanisms that are associated with plant growth promotion have been discussed [33]. We have found that a portion of the existing microorganisms in the rhizosphere presented arylsulfatase activity [30], thus releasing sulfate bound in the soil organic matter, the percentage of which increased with the addition of ES-enriched fertilizer. Moreover, the fertilization with FBES at sowing affected the iron fractions of the rhizosoil towards iron mobilization, thus providing more iron to the crop, which apart from the iron nutrition fortified the crop's sulfur nutrition, too [14].
On the other hand, phosphorus (P) is one of the most limiting essential nutrients for crop production. The dynamic processes determining P availability in the soil and in the rhizosphere, P mobilization, uptake, and utilization by plants have been reviewed [34,35]. Especially, the response of wheat grain yield to phosphorus has been reported by a number of researchers, as wheat crops have high phosphorus requirement, mostly during the early growth stages [36][37][38][39]. Although yield limiting factors are complex, the effect of soil available P was studied further in experimental plots under a randomized block design. Already existing spatial information of the experimental site [40] was used to guide sampling to representative plots, referred to in the literature as directed, smart or targeted sampling [41]. In accordance with Friesen (1996) [42], our results suggested that the presence of adequate P for the needs of the S-oxidizing bacteria may be a critical factor for the ES use efficiency. It is speculated that under low soil available P, the increased numbers of microbial populations, due to the addition of ES in the fertilization scheme were stronger P consumers compared to the wheat crop during the growing season, leaving less than adequate P for the crop's needs, thus possibly explaining the depressed yields. More research is needed to fully explain the effect of soil available P on yield under the circumstances.

The Efficacy of ES Fertilization Schemes
ES is used either as granules, or in the form of fine powder, the micronized ES. Micronized ES is preferable compared with ES granules, because the transformation to sulfate-S by the microorganisms is quick. On the other hand, the application of fine powder in large scale agriculture is a serious problem, because the handling of dust is involved. To overcome the problem, several technologies towards successful co-formulation of micronized ES with different materials have been arisen through the years, in order to produce granules applicable to large scale agriculture, including monoammonium phosphate, diammonium phosphate, and triple superphosphate, containing micronized ES without or with AS [43,44]. Scanning electron migrographs of 20-10-10/ES (2%) granules clearly show that elemental sulfur forms a dense coating layer on the surface, which is achieved via the sticky mixture of molasses plus glycerol. The product 20-10-10/ES produced dust. This held true for 16-20-0/ES(2%), too. The dust was estimated to be in the order of 4%, a minor fraction of which was ES (data not shown). On the other hand, U/ES(2%) and UAS/ES(2%) were indeed dustless.
The factors that influence the oxidation of elemental sulfur in soils have been discussed [27], while a "concept of the negative locality effect on ES oxidation" has been developed [8,44]. Briefly, when the fertilizer granule disintegrates it releases micronized ES particles to the soil. Due to the hydrophobic nature of the ES, the very fine ES particles create clusters and localize around the applied granule site. Clustering decreases contact between localized ES particles and soil, which in turn decreases the colonization of soil S-oxidizing bacteria on the surface of the ES particles. Moreover, the released micronized ES particles can coalesce to form larger aggregates, which in turn can further decrease ES oxidation [42,45]. It seems that the FBES technology diminishes the negative locality effect on ES oxidation. The binder utilized by the FBES technology is also a positive factor, because the binder around the ES and the granule particles can be consumed by the microorganisms as well.
The fertilizer type 16-20-0 was used and tested against 16-20-0/ES as NP starter fertilizer, which was combined with U or UAS, applied in spring as topdressing added N (and S). U and UAS are of common use and both water soluble. N from urea is available to the plant only after hydrolysis by the urease enzyme to ammonium carbonate, which can significantly increase soil pH around the applied urea granule sites (can be as high as [8][9], due to hydroxyl anion production. AS provides N and S, both critical plant nutrients. Compared with urea, AS may have some potential agronomic and environmental benefits [44]. One approach to enhance the N efficiency of urea is to partially substitute AS for urea in the mixture. The dilution effect of AS-N should be considered because ammonia volatilization is greater with increased urea concentration and/or rate of application [46]. An increase in crop yield should be expected by mixing AS and urea compared with urea alone. Another reason to mix AS and urea is to supply S along with N. Moreover, AS reacts with calcium carbonate to precipitate calcium sulfate. Sulfur-coated urea (SCU) as a source of S was used in wheat crop and it was found that 5% of SCU supplied 50% of the S wheat requirements and at the same time increased N recovery efficiency by 60.3% over prilled urea [47]. Our results showed that the incorporation of ES improved the performance of U or UAS, suggesting that the granules enriched with ES were fortified, while their performance was influenced by the soil available phosphorus.

The Efficacy of ES and NBPT Fertilization Schemes
The stability of urea in the soil is affected by the enzyme urease, which is released by the soil microbial population or is derived from the decomposition of organic matter [11]. The incorporation of urease inhibitors into urea granules delays urea hydrolysis and urea is available for plant acquisition. N-(n-butyl) thiophosphoric triamide (NBPT) is the urease inhibitor that is often formulated into granular urea fertilizers. This compound is a structural analog of urea and inhibits urease activity by forming stable complexes with urease [11,48,49]. The use of granules co-enriched with ES and NBPT onto the same granule in a field experiment has been reported [50]. According to the authors, the inhibitor and ES were coated onto the fertilizers by Summit-Quinphos Ltd. (New Zealand) and applied directly onto the pasture of the respective plots by hand. No information was provided on the nature of the coating. In the present study, we tested the efficacy of a 50:50 mixture of NBPT containing granules and ES containing granules, using commercially available fertilizers produced by the same company. The 50:50 ratio was arbitrarily chosen. Such combination provided even higher yields compared to the yields provided by the application of the corresponding fertilizers alone and the soil available phosphorus influenced the performance of the mixture. According to the FBES technology, the core fertilizer can be enriched up to 4% with ES. To our knowledge, the least percentage of ES in NP fertilizer granules reported in the literature is 5% [10]. In this research, granules enriched with 2% ES were used, i.e., the minimum ES enrichment. Therefore, in the 50:50 mixture, more or less 1% of ES was applied, i.e. half of the amount, and the same holds true for NBPT.
In summary, UAS provided sulfate-S for immediate use, ES as a coating in its majority provided sulfate-S after oxidation, NBPT positively contributed to ammonium use efficiency provided by both U and AS constituents of the fertilizer granule, while the minor amount of the binding materials probably fed the microorganisms around the granule. A portion of these microorganisms were arylsulfatase producing ones, which implies that organic S was released as sulfate-S, thus comprising a third pool of sulfate-S. The more the available P the more effective the UAS/ES/NBPT mixture was. The efficacy of this activity was enhanced when FBES (16-20-0/ES) was used as starter fertilizer at sowing, instead of the corresponding 16-20-0 conventional fertilizer.

Conclusions
The use of UAS in spring applied fertilization scheme enriched with 2% ES and the urease inhibitor NBPT in the granules, significantly increased the yield of durum wheat crops, and in combination with the use of 2% ES as an ingredient of the starter fertilizer, it boosted the yield to even higher scores. The soil available phosphorus at P-Olsen value of 8 ppm seemed to be the threshold, below which the performance of the aforementioned combination was strongly depressed.