Impact of Hybrid Fertilization on Winter Triticale Yield and Its Stability Based on SVD Analysis

Abstract

Nitrogen fertilization is extensively applied in agricultural activities to improve food production. However, the applied dose of nitrogen is often higher than that required for the desired productivity level of a given crop. Thus, research on methods that could increase the uptake of nitrogen supplied with fertilizers by plants is of high significance. One way to achieve this goal is to employ a hybrid fertilization technique (a combination of the application of solid fertilizers in the first dose with foliar application of liquid fertilizers in the second and third doses), using reduced doses of nitrogen fertilizers as well as fertilizers enriched with 10% sulfur in the form of thiosulfate. The aim of our study was to assess the productivity resulting from different fertilization treatments and the stability of the resulting yield based on interactions between the method of fertilizer application and environmental conditions. To determine interaction patterns, an additive main effects and multiplicative interaction model was employed. The key finding is that sulfur-enriched fertilizers can significantly increase yield, but at the expense of reduced stability. However, yield stability was more strongly related to meteorological conditions. Understanding of such interactions can help increase the efficiency of selection and accuracy of recommendations for new technologies of crop cultivation.

1. Introduction

Nitrogen (N), widely used in agricultural fertilizers, is a critical macronutrient influencing the growth and productivity of cereal crops, including winter triticale [1,2]. It directly affects tillering, leaf area development, photosynthetic efficiency, and, ultimately, biomass accumulation and grain yield. Nitrogen is an essential element involved in protein synthesis and enzyme activity, and has a key role in determining the number of grains per spike and thousand-grain weight [3]. The level of nitrogen accumulation in grains is related to the nitrogen provided in the fertilizers applied, with grain protein concentration increasing with increasing N fertilizer input [4,5]. Sulfur is crucial for protein synthesis and nitrogen assimilation. It is a building block of key amino acids and enzymes, and its deficiency leads to poor nitrogen utilization [6]. Numerous works [7,8,9,10] have indicated a positive impact of sulfur on the metabolism of nitrogen in plants. The proportion of nitrogen to sulfur, the N:S ratio, in selected parts of plants is one of the best and commonly used indicators of a plant’s sulfur requirements. Under sulfur-deficient conditions, plants are unable to fully utilize even high nitrogen doses, which consequently reduces yield and deteriorates its quality (e.g., lower protein content). A narrow N:S ratio in plant tissues indicates more efficient nitrogen uptake [11].

The effectiveness of fertilization can be enhanced through application of fertilizers at appropriate doses, by optimizing the timing of their application [12], and by improving soil conditions to prevent limitations in nutrient uptake and transport from the soil to plant roots [13]. A thorough understanding of crop-specific nutrient requirements, together with the influence of various biotic and abiotic factors [14] on nutrient availability and utilization, is expected to mitigate adverse environmental impacts associated with extensive use of nitrogen fertilizers [15]. A main idea behind sustainable agriculture is to reduce the use of toxic pesticides and herbicides, as well as mineral fertilizers. Particular attention must be paid to restricting the use of mineral fertilizers, especially nitrogen fertilizers, as their improper application can lead to serious environmental problems, including soil degradation, water pollution, emission of nitrogen gas to the atmosphere, and effects on biodiversity [16,17,18]. It has been estimated that on a global scale, only 47% of nitrogen from mineral fertilizers is used by plants [19,20], while the rest is lost, contributing to the emission of greenhouse gases and water eutrophication [21,22]. The efficiency of use of the components of applied fertilizers, such as nitrogen, is one of the main indicators allowing assessment of the level of sustainability of agricultural production in the context of cost-effectiveness and environmental impact [3,23]. The main challenges for sustainable agriculture are to introduce appropriate changes in the type of fertilizer, its application, and reductions in the use of nitrogen fertilizers [24].

Intensification of agricultural production, both crop and animal, in response to increasing demand for food cannot be realized without providing the soil with mineral components, mainly nitrogen [19,25,26]. It has been estimated that by 2050, global demand for food will double, aggravating environmental problems [27,28]. In view of the necessary increase in agricultural production and the threats that may entail, continuous efforts must be taken to alleviate the negative effects of agricultural procedures on the entire agroecosystem, including soil fertility, and biotic and abiotic stresses [29,30], while enhancing the uptake of nitrogen from the applied fertilizers [31].

Winter triticale is the crop used mostly in livestock production as it is the main component of feedstuff. Although much attention has been devoted to optimization of its cultivation, its efficient production is still challenging, in particular in regions plagued by droughts in the period of intensive plant growth and those with soil poor in bioavailable species of mineral components. Providing correct nutrients to plants, taking into account the right proportion of ingredients, is vital for the quality and quantity of production, as well as the optimum use of particular components. In crop production, timing the supply of nitrogen fertilizers should be adjusted to the needs of plants in particular phases of vegetation, especially in critical phases when demand for this element is enhanced [32,33,34]. Of particular importance is the proper delivery of nitrogen and sulfur, as these elements affect growth and improve the quality and size of the yield [35]. Meteorological conditions throughout the vegetative period, in particular, low temperatures in the early stages of plant growth and nonuniform distribution of precipitation, are the main factors responsible for poor yield stability, and climatic conditions often determine the sustainability of agricultural systems development [36,37]. With increasing temperature and decreasing amounts of precipitation, the sustainable development of agricultural systems becomes more difficult to achieve. Soil degradation processes, such as decreasing organic matter and soil erosion, in general accelerate with increasing temperature [38]. Climate changes over the last decade have resulted in instabilities in crop yields globally [39,40,41].

Stability yield is the ability of a specific crop to maintain consistent performance across a range of different environments and over time, showing minimal variability in its yield despite changes in growing conditions [42,43,44]. Development of sustainable agriculture requires a broad range of research experiments on fertilization methods to identify these methods, improving the stability of crop yields under different environmental conditions. Such an identification is possible by analyzing interactions between the method of fertilizer application and variable environmental and meteorological conditions. A better understanding of these interactions can enhance the accuracy of recommendations for new crop cultivation technologies [45]. Results from multiple studies may be particularly valuable in agriculture, especially for identifying appropriate fertilization methods that ensure optimal yield under variable weather conditions. To evaluate interaction patterns, additive main effects and multiplicative interaction (AMMI) models have been introduced [45]. This analytical tool combines analysis of variance with singular value decomposition (SVD) to better understand and structure relationships between crop yields and environmental conditions [46,47]. This model applies SVD to second-order interaction effects in a two-way analysis of variance [42]. Visualization through SVD provides a deeper and more distinct evaluation of the effects of environmental conditions and fertilizer variants on yield stability, thereby facilitating the identification of the most stable fertilization strategies and clarifying the impact of specific environmental conditions on yield stability. Canonical variate analysis (CVA) is a useful tool for interpreting the relationships between parameters describing the nutritional status of plants at critical stages of yield formation [48]. It can be applied to various types of data matrices, such as a matrix of interaction effects [49], the interaction matrix [50], the matrix of residuals [51], or matrices indicating differences between experimental factors [50]. Consequently, the canonical approach can be employed in any linear model, particularly in multivariate analysis of variance, which involves the examination of specific correlation relationships [52]. Moreover, taking into account the complexity of the systems of agricultural production, it is necessary to launch multidisciplinary studies aimed at improving soil fertility and water resource management, and enhancing sustainable development over longer time perspectives.

To achieve this aim, we analyzed the impact of the relation between a hybrid method of application of nitrogen fertilizers (i.e., a method consisting of combining the application of solid fertilizers in the first dose with foliar application of liquid fertilizers in the second and third doses) and meteorological conditions (variant x year interaction) on winter triticale yield, using tools of dataset analysis, and modifying the approach of studying the distribution according to SVD with an addition of hierarchical clustering. This approach enables the selection of optimal fertilization variants tailored to specific environmental conditions, leading to relatively predictable yields under increasingly variable weather conditions. Extending the two-way analysis of variance with the proposed methods enhances the capacity to select the most effective fertilization variants, taking into account application practices, fertilizer composition, and environmental factors. Moreover, this analysis can facilitate the identification of fertilizer groups that balance productivity with stability and are best suited for particular agronomic zones.

The assumed research hypothesis was that the use of a hybrid method of nitrogen fertilizer application in three consecutive crop seasons differing in the meteorological conditions leads to a lower stability of winter triticale yield.

The research questions are as follows:

• Does the use of a hybrid method of nitrogen fertilizer application lead to changes in the stability of winter triticale yield?
• Does winter triticale yield stability depend on changes in meteorological conditions in particular crop seasons?
• Does enrichment of a nitrogen fertilizer with sulfur affect the stability of winter triticale yield?

2. Materials and Methods

2.1. Experimental Field Characteristics

The experiments were performed in three successive cropping seasons, 2018/2019, 2019/2020, and 2020/2021, on a farm in the Lubuskie Voivodeship, Poland, in Gądków Wielki (52°14′ N, 14°58′ E). In each season, the soil on the farm was characterized by a slightly acidic pH of 6.1, phosphorus ranging from 68 to 75 mg P kg⁻¹ of soil, potassium ranging from 150 to 177.6 mg K kg⁻¹, and bioavailable magnesium ranging from 37 to 80 mg Mg kg⁻¹. The content of sulfur (as SO₄) was low, ranging from 9 to 12 mg kg^–1 of soil. The content of mineral nitrogen in the soil layer of 0–90 cm (Nmin) before the start of vegetation was high, and ranged from 109 to 116 kg ha⁻¹. The preceding crop in each season was winter wheat. The field experiment with winter triticale of the Tadeus variety was arranged as a single-factor design with five treatments, where the experimental factor was the fertilization strategy. Each treatment consisted of the application of fertilizers with different combinations of nitrogen dose reduction and sulfur addition, and treatments were replicated in four experimental plots.

A hybrid method was used for fertilization that consisted of combining the application of solid fertilizers in the first dose with foliar application of liquid fertilizers in the second and third doses. The fertilizers were applied three times, coincident with the developmental stages of winter triticale: BBCH 20/21, BBCH 32, and BBCH 49–50. In the first application, ammonium nitrate (AN) containing 70 kg N ha^–1 was applied. The control treatment (AN) consisted of nitrogen fertilizer containing ammonium nitrate (34%), with a total amount of 180 kg N ha^–1 applied in three doses. Two treatments, denoted as 1N20 and 1N25, consisted of second and third fertilizer doses reduced by 20 and 25%, respectively, relative to the dose applied in the control plot. The second and third doses were applied on leaves using a liquid form of the fertilizer containing 330 g N L^–1. To evaluate the effects of sulfur, two treatments, denoted as 2N20 + S and 2N25 + S, consisted of a nitrogen fertilizer enriched with sulfur as thiophosphate, containing 100 g S L^–1. A summary of the five treatments is shown in

Table 1. Nitrogen fertilization scheme for winter triticale.

Nitrogen Dose	Treatment Abbreviation
	AN	1N20	1N25	2N20 + S	2N25 + S
First application [kg N ha–1]	70	70	70	70	70
Second application [kg N ha–1]	60	37	32.5	37	32.5
Third application [kg N ha–1]	50	37	32.5	37	32.5
Total nitrogen dose [kg N ha–1]	180	144	137	144	137

Abbreviations: AN—ammonium nitrate, 1N and 2N + S—liquid fertilizers applied to foliage were the source of nitrogen or nitrogen with sulfur, respectively, in the second and third rates of fertilization; 20 and 25—nitrogen dose reduced by 20% or 25% compared to the control variant.

. The total dose of nitrogen was estimated on the basis of the expected yield of 8t·ha^–1 and unit uptake of this nutrient, consistent with the directive issued in the Journal of Laws of February 12th, 2020, item 1339. All other agronomic management practices, including but not limited to tillage and plant protection, were implemented uniformly across all experimental plots in accordance with locally recommended guidelines, ensuring that any observed effects could be attributed solely to the applied fertilization treatments.

At the BBCH55 stage of triticale growth, the biomass of the above-ground parts of the plants was measured based on whole-plant samples collected from a 0.25 m² area in each treatment plot. The total biomass was weighted, then 25 plants chosen at random were separated into 3 fractions: leaves, stems, and spikes. The fractions were dried and weighted, and expressed in tons per hectare (t ha^–1). The plant material was prepared for laboratory determination of nutrient contents. About 100 g subsamples of stems, leaves, and spikes were taken and ground in a stainless-steel grinder. The nitrogen and sulfur contents were determined using a Vario MAX cube CNS elemental analyzer (Elementar Analysensysteme GmbH, Langenselbold, Germany). The triticale grain yield in each year of study was determined from sampling in an area of 50 m², while the area of the whole study plot was 80 m². In the full maturity stage (BBCH 92), plant samples were collected from each study plot in a 0.25 m² area to determine two elements of the yield: number of grains per spike (NGS) and thousand-grain weight (TGW).

2.2. Chemical Analysis of the Soil and Plant Material

The plant samples (straw, spiklets, leaves) were dried at 55 °C, ground with a stainless-steel grinder. Next, the plant material was prepared for laboratory determination of nutrient contents. The nitrogen (N) and sulfur (S) contents were determined with a Vario MAX cube CNS elemental analyzer (Elementar Analysensysteme GmbH, Langenselbold, Germany). The accumulation of N and S in the leaves, straw, and spikes of triticale at the heading stage was calculated by multiplying the organ biomass and the respective nutrient content (N or S).

Soil samples for chemical analyses were collected from the arable horizon after harvesting the forecrop. The content of available forms of phosphorus and potassium in the soil was determined based on the Egner–Riehm method (DL), whereas that of magnesium was determined according to Schachtschabel. The concentration of P was determined by the colorimetric method using a spectrophotometer Specord 40 (Analitik Jena, Jena, Germany). Concentrations of K and Mg in the collected extracts were analyzed by atomic absorption spectrophotometry (SpectrAA 55B, Varian, Australia). The NH₄-N and NO₃-N were determined in field-fresh soil samples. Twenty grams of soil samples were shaken for 1 h with 100 mL of a 0.01 M CaCl₂ solution (soil/solution ratio 5:1; w/v). Concentrations of NH₄-N and NO₃-N were determined with the colorimetric method using flow injection analysis (FIAstar 5000, FOSS Analytical AB, Höganäs, Sweden). The Corg. was determined by combustion analysis using a Vario Max analyzer (Elementar Analysensysteme GmbH, Hanau, Germany). Chemical analyses of the plant material were based on the SSSA methodological guide [53,54].

2.3. Meteorological Conditions

The variation in meteorological conditions over the study period is characterized in

Table 2. Average daily air temperature and monthly sum of precipitation (location—Gadkow Wielki, Poland). (Source: Institute of Meteorology and Water Management—National Research Institute in Poznan).

Month	Average Daily Air Temperature [°C]				Monthly Sum of Precipitation [mm]
	2018/2019	2019/2020	2020/2021	Long-Term	2018/2019	2019/2020	2020/2021	Long-Term
September	16.2	14.9	15.2	13.8	29.8	44.5	53.5	41.7
October	11.2	11.2	11.1	8.9	25.2	32.8	43.0	34.0
November	4.9	6.3	6.7	3.7	8.1	25.7	10.9	34.7
December	3.9	4.2	2.7	0.4	45.0	28.0	31.2	40.4
January	0.9	4.1	0.1	−0.8	44.5	29.1	54.2	32.6
February	4.1	5.9	−0.2	0.2	19.5	70.9	36.1	27.3
March	6.8	5.1	4.3	3.6	46.1	27.6	29.0	37.8
April	10.6	10.3	6.5	8.8	7.2	8.2	23.2	31.3
May	12.4	12.3	12.4	14.0	59.6	37.5	56.6	49.5
June	22.8	18.8	20.2	16.6	34.0	26.6	96.0	56.8
July	19.6	19.0	20.9	18.9	63.8	38.5	40.4	75.7

. To compare the level and distribution of precipitation and temperatures, data from each study year were compared to data from 1991 to 2020. A comparison of the total precipitation revealed an increase of 2% in the season 2020/2021, and decreases of 17% and 20% in 2018/2019 and 2019/2020, respectively. An increase in total monthly precipitation of 2.5 times was noted in February 2020. Average daily air temperatures were usually higher than the long-term average, with significantly higher average daily air temperatures, by about 5 °C, noted in January and February 2020.

2.4. Statistical Analysis

2.4.1. Model of Analysis

Data were analyzed using a two-way analysis of variance to assess interactions between fertilization treatment and year of study on triticale yield. The model for the sample $y_{ijk}$ is defined for the $i$-th fertilization ($i=1,\dots ,I,\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}I=5$), $j$-th year ($j=1,\dots ,J,\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}J=3$), and $k$-th replication ($k=1,\dots ,K,\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{e}K=4$) as follows:

$$y_{ijk}=\mu +{\alpha }_i+{\beta }_j+{\left(\alpha \beta \right)}_{ij}+e_{ijk}$$

(1)

where $\mu$ is the overall mean, ${\alpha }_i$ represents the main effects of the fertilization variant, ${\beta }_j$ represents the main effects of year of study, ${\left(\alpha \beta \right)}_{ij}$ represents the second-order interaction effects for fertilization variant and year of study, and $e_{ijk}$ are random errors [55]. The two-way analysis was used in order to take into account the variation resulting from the type of fertilization and meteorological conditions in the years of study. Moreover, the same mode of analysis was used to evaluate yield stability and relations between winter triticale yield and other factors studied. The two-way analysis of variance was also conducted for thousand-grain weight, biomass, as well as nitrogen and sulfur content, and the N:S ratio in plant organs.

2.4.2. Singular Value Decomposition

Second-order interaction effects were estimated based on relationships between years of study and fertilization treatments, and could be used to explain yield stability. Singular value decomposition is a method that allows visualization of multidimensional results of a second-order interaction effect matrix $\mathsf{\Theta }$ [51], where the $i$,$j$ element is provided by

$${(\widehat{\alpha \beta })}_{ij}=\overline{y}-{\overline{y}}_i-{\overline{y}}_j+{\overline{y}}_{ij}$$

(2)

where $\overline{y}$ is the overall mean, ${\overline{y}}_i$ is the mean of treatment $i$, ${\overline{y}}_j$ is the mean of year $j$ and ${\overline{y}}_{ij}$ is the mean of the repetitions of fertilization variant $i$ in the year $j$. The matrix of second-order interaction effects $\mathsf{\Theta }$ was decomposed [56], enabling the production of a biplot, a graphical presentation of the multidimensional matrix in a two-dimensional orthogonal space to compare the stability of winter triticale yield across treatments and years of study. The biplot illustrates the differences between some linear functions of the treatment $\times$ year interaction effects. A shorter distance from the point corresponding to a given treatment or year to the point of the plot origin is indicative of greater yield stability.

The singular value decomposition provided three matrices $\mathsf{\Theta }=\mathbf{U}\mathsf{\Sigma }{\mathbf{V}}^{\mathrm{T}}$, where $\mathbf{U}$ and $\mathbf{V}$ were orthogonal matrices [56]. Following the decomposition, the matrix of second-order interaction effects can also be presented in the form

$$\mathsf{\Theta }=\sum _{i=1}^r\sqrt{{\lambda }_i}{\mathit{u}}_i{\mathit{v}}_i^T$$

(3)

where $r$ is the rank of matrix $\mathsf{\Theta }$, $\sqrt{{\lambda }_i}$ is the square root of the eigenvalue of $\mathsf{\Theta }{\mathsf{\Theta }}^{\mathrm{T}}$ or ${\mathsf{\Theta }}^{\mathrm{T}}\mathsf{\Theta }$ matrix, called singular value, ${\mathbf{u}}_i$ is the eigenvector of $\mathsf{\Theta }{\mathsf{\Theta }}^{\mathrm{T}}$ and the i-th column of matrix $\mathbf{U}$, and ${\mathbf{v}}_i$ is the eigenvector of ${\mathsf{\Theta }}^{\mathrm{T}}\mathsf{\Theta }$ matrix and the i-th column of matrix $\mathbf{V}$. The $\mathsf{\Theta }$ matrix can also be written as the product of two matrices, $\mathsf{\Theta }=\mathbf{R}\mathbf{C}$, where the $\mathbf{R}$ matrix represents row coordinates on the biplot and the $\mathbf{C}$ matrix represents column coordinates on the biplot. The columns of the $\mathbf{R}$ matrix can be expressed as $\sqrt[4]{{\mathsf{\lambda }}_i}{\mathbf{u}}_i$ and the columns of $\mathbf{C}$ as $\sqrt[4]{{\mathsf{\lambda }}_i}{\mathbf{v}}_i$. The $\sqrt{{\lambda }_i}$ singular values were used to scale orthogonal vectors obtained from $\mathbf{U}$ and $\mathbf{V}$ matrices, which represent the coordinates of rows and columns, where the rows correspond to fertilizer variants and columns to years of research, and the biplot illustrates relationships between these factors.

2.4.3. Hierarchical Clustering

The dendrograms visualize the distances between fertilizer variants and years of study, considered separately, and reveal similarities and differences between groups in a dataset, named clusters. Each cluster consists of points representing second-order interaction effects, considered separately for fertilizer variants and years of study. To evaluate similarities between clusters, the single linkage method was used, which defines inter-cluster distance as the shortest distance between any two points from different clusters. A dendrogram is a graphical representation of the clustering process in the form of a binary tree, where leaves represent individual data points, nodes correspond to merged clusters, and the vertical axis indicates distances between the points [57]. To create a dendrogram, variables are transformed into Euclidean space and scaled. For fertilizer variants, vectors of the $\mathbf{U}$ matrix (3) were scaled by consecutive eigenvalues as follows $\sqrt{{\mathsf{\lambda }}_i}{\mathbf{u}}_i$. In this case, the consecutive vectors of $\mathbf{V}$ matrix were not scaled. Similarly, to calculate Euclidean distances between years, consecutive vectors of the $\mathbf{V}$ matrix were scaled by their respective singular values, while consecutive vectors of the $\mathbf{U}$ matrix were not scaled. The matrix of second-order interaction effects was also determined and visualized for the N to S ratio for stems.

2.4.4. Multidimensional Analysis

In order to take into account relationships between winter triticale yield and the analyzed parameters (number of grains per spike, thousand-grain weight, biomass of stems, leaves, and spikes, as well as nitrogen and sulfur contents in individual organs), multidimensional tools were used for comparative analyses. Hence, model (1) for parameters describing one experimental object, i.e., a specific fertilization variant in a particular year of study, can then be presented in the form

$${\mathbf{y}}_{ijk}=\mathsf{\mu }+{\mathsf{\alpha }}_i+{\mathsf{\beta }}_j+{\mathsf{\gamma }}_{ij}+{\mathbf{e}}_{ijk},$$

(4)

where ${\mathbf{y}}_{ijk}$ are result vectors; $\mathsf{\mu }$ is the vector of general means; ${\mathsf{\alpha }}_i$ are vectors of fertilization variant effects; ${\mathsf{\beta }}_j$ are vectors of year effects; ${\mathsf{\gamma }}_{ij}$ are vectors of interaction effects; and ${\mathbf{e}}_{ijk}$ are vectors of errors. Finally, model (4) can be described in matrix form

$$\mathbf{Y}={\mathbf{1}}_N{\mathsf{\mu }}^{\prime }+{\mathbf{X}}_1{\mathsf{\Xi }}_1+{\mathbf{X}}_2{\mathsf{\Xi }}_2+{\mathbf{X}}_{12}{\mathsf{\Xi }}_{12}+\mathbf{U},$$

(5)

where $\mathbf{Y}={\left[{\mathbf{y}}_{111}{,\dots ,\mathbf{y}}_{IJK}\right]}^{\prime }$ is the result matrix; $N=I·J·K$; ${\mathsf{\Xi }}_1=$ ${\left[{\mathsf{\alpha }}_1{,\dots ,\mathsf{\alpha }}_I\right]}^{\prime }$; ${\mathsf{\Xi }}_2=$ ${\left[{\mathsf{\beta }}_1{,\dots ,\mathsf{\beta }}_J\right]}^{\prime }$; ${\mathsf{\Xi }}_{12}=$ ${\left[{\mathsf{\gamma }}_1{,\dots ,\mathsf{\gamma }}_{IJ}\right]}^{\prime }$; ${\mathbf{X}}_1$, ${\mathbf{X}}_2$, and ${\mathbf{X}}_{12}$ are design matrices; and $\mathbf{U}{=\left[{\mathbf{e}}_{111}{,\dots ,\mathbf{e}}_{IJK}\right]}^{\prime }$ [50]. Canonical variate analysis made it possible to visualize parameters related to yield on different experimental objects in an orthogonal space. The analysis is based on the matrix [58]

$${\widehat{\mathbf{C}\mathsf{\Xi }}}_{12}=\mathbf{C}{\left({\mathbf{X}}_{12}^{\prime }{\mathsf{{\rm X}}}_{12}\right)}^{-1}{\mathbf{X}}_{12}^{\prime }\mathbf{Y}.$$

(6)

Using the ${\mathbf{C}=\mathbf{I}}_{IJ}-{\left(IJ\right)}^{-1}{\mathbf{1}}_{IJ}{\mathbf{1}}_{IJ}^{\prime }$ matrix, the differences between average parameters for a given fertilization variant in a given year of study and the general means are analyzed. To study the relationships between the experimental object, canonical variate analysis based on matrix transformation ${\widehat{\mathbf{C}\mathsf{\Xi }}}_{12}$ by singular value decomposition [48] and the linear combination method were used [52,57].

2.4.5. Data Analysis

Analysis of variance and multiple regression were processed using STATISTICA software (v13.3, StatSoft, Cracow, Poland). Visualizations were generated using the following Python 3.13.2 and packages: matplotlib [59], numpy [60], pandas [61], and scipy [62]. Hierarchical clustering was performed using R 4.4.1 and its packages: dplyr [63], stats [64], and tidyr [65].

3. Results

3.1. Winter Triticale Grain Yield

Results of the variance analysis of triticale yield are shown in

Table 3. Two-way analysis of variance for winter triticale yield with year × treatment interaction study.

Source of Variation	Degrees of Freedom	Sum of Squares	Mean Squares	F-Statistic	Empirical Significance Level (p-Value)
Intercept	1	4724.6	4724.6	23128.8	<0.001
Year	2	38.13	19.06	93.32	<0.001
Treatment	4	19.35	4.84	23.69	<0.001
Interaction	8	14.44	1.80	8.84	<0.001
Error	45	9.19	0.204
Total	59	81.11

. Elements of crop structure (number of grains per spike and thousand-grain weight), biomass of stems, leaves, and spikes, and nitrogen and sulfur contents in individual organs, are presented in Tables S1–S14 and post hoc tests in Tables S15–S28. Yield significantly depended on treatments, years of study, and second-order interaction effects. The two-way analysis of variance for biomass of particular parts of triticale plants in the period of spike formation revealed relations between biomass changes and experimental treatments, shown in Tables S6–S8. Moreover, the biomass of stems and spikes was dependent on the year of the study, as was the content of nitrogen (as shown in Tables S9–S11). The content of sulfur depended on the interaction between treatments and years, shown in Tables S12–S14. The N:S ratios varied according to the year and to the interaction of the year and treatment, as shown in Tables S3–S5.

Analysis of the combined effects of second-order interactions with those of principal factors calculated from Equation (2) showed that the highest yields, over 11 t∙ha⁻¹, were obtained using hybrid applications of nitrogen fertilizer enriched with sulfur in 2020 (

Table 4. Means ± standard deviation of grain yield with year × treatment interaction study (t∙ha⁻¹).

Treatment	Research Year
	2019	2020	2021
AN	7.62 ab ± 0.30	8.96 cd ± 0.62	7.50 a ± 0.38
1N20	7.53 a ± 0.19	9.56 d ± 0.29	8.35 abc ± 0.16
1N25	8.77 bcd ± 0.59	8.98 cd ± 0.39	8.96 cd ± 0.59
2N20 + S	8.68 bcd ± 0.55	11.43 d ± 0.32	8.21 abc ± 0.44
2N25 + S	8.90 cd ± 0.66	11.08 d ± 0.59	8.58 abcd ± 0.25

Different lowercase letters denote significant differences (p < 0.05); Abbreviations: AN—ammonium nitrate, 1N and 2N + S—liquid fertilizers applied to foliage were the source of nitrogen or nitrogen with sulfur, respectively, in the second and third rates of fertilization; 20 and 25—nitrogen dose reduced by 20% or 25% compared to the control variant.

). The difference between the highest and lowest yield was close to 4 t∙ha⁻¹ (Table 4).

As shown in Table 3, the effects of second-order interaction had significant impacts on triticale yield. Higher values indicate lower yield stability, whereas lower values correspond to higher stability. Values not significantly different from zero indicate that the treatment × year interaction included in Equation (1) has a non-significant second-order effect, which is interpreted as high yield stability.

Table 5. Second-order interaction effects between years and treatments on grain yield (t∙ha⁻¹).

Treatment	Research Year
	2019	2020	2021
AN	0.17	−0.20	0.02
1N20	−0.37 *	−0.05	0.42 *
1N25	0.44 *	−1.05 **	0.61 **
2N20 + S	−0.19	0.86 **	−0.68 **
2N25 + S	−0.05	0.43 *	−0.38 *

* significance level α = 0.05; ** significance level α = 0.01; Abbreviations: AN—ammonium nitrate, 1N and 2N + S—liquid fertilizers applied to foliage were the source of nitrogen or nitrogen with sulfur, respectively, in the second and third rates of fertilization; 20 and 25—nitrogen dose reduced by 20% or 25% compared to the control variant.

presents the values of the second-order interaction effects, with the highest value, 0.86 t∙ha⁻¹, observed for the variant 2N20 + S in 2020, and the lowest, −1.05 t∙ha⁻¹, for 1N25 in 2020. The range of changes is close to 1.9 t∙ha⁻¹, which points to a relatively high variability in yield.

In order to determine which variants of fertilization or which years differentiate the interaction between these factors (Table 5), the effects of second-order interactions were analyzed using the singular value decomposition introduced in Equation (3). Based on values of the first two components, the yield of winter triticale fertilized with ammonium nitrate (AN) was characterized by the highest stability (Figure 1). Consequently, second-order interaction effects for AN were the lowest (Table 5), which is reflected by the shortest distance from the point of the plot origin (Figure 1). Among the remaining fertilization variants, a moderate level of yield stability, close to values obtained for AN, was obtained for 1N20 and 2N25 + S, as indicated by the next shortest distances from the point of origin of the plot (Figure 1). Analyzing similarities between fertilizer variants, it was observed that the yield stabilities of winter triticale for fertilizers enriched with sulfur, 2N20 + S and 2N25 + S, were grouped into a single cluster based on the shortest distance from the plot origin (Figure 2a), indicating the highest similarity. However, this cluster was distinct from the result obtained for the 1N25 variant (Figure 2a). Effects of second-order interactions in 2020 for variants 2N20 + S and 2N25 + S resulted in relatively high values (Table 5). In the same year, the yield difference between sulfur-enriched fertilization variants and the control variant was markedly larger than in 2019 and 2021. On average, the yield obtained after the application of sulfur-enriched variants exceeded that of the control variant by approximately 20% (Table 4), which was most likely driven by favorable meteorological conditions. In 2020, groundwater resources were enhanced due to intensive precipitation in February (Table 2), and no periods of low temperatures were noted in the first phase of growth.

Based on hierarchical clustering, years 2019 and 2021 were the most similar in terms of yield stability and are clustered together in the dendrogram (Figure 2b), indicating that increased groundwater resources had a greater impact on yield than low temperatures during the first phase of growth.

Distances between clusters corresponding to methods of fertilization were smaller than those representing years (Figure 2a,b), indicating that the method of application of fertilizer had a greater impact than the year of study.

3.2. Relationship Between Yield and Plant Characteristics

An important objective of the study was to determine which factors had the greatest influence on triticale grain yield. We began with the transformation of the original variables into the orthogonal space to analyze mutual relations between the factors. Results of the canonical variate analysis based on relationships between yield, elements of crop structure (number of grains per spike, thousand-grain weight), biomass of stems, leaves, and spikes, and nitrogen and sulfur contents of individual organs, according to Equations (5) and (6), are visualized in Figure 3. Thousand-grain weight, number of grains per spike, and yield were factors that most differentiated the experimental objects, i.e., plots for each variant of treatment in a given year of the study.

Contents and uptake of nitrogen and sulfur in triticale plants in the spike-formation stage differed in different organs of the plants Tables S29–S44. Treatment, year, and interactions between them significantly differentiated the N:S ratio in individual organs (Tables S3–S5). A comparison of intergroup variability with intragroup variability showed that sulfur contents differentiated the experimental objects to a lesser extent (Figure 4), which was related to a limited migration of this element in the plant. As the approximated percentage of variance explained by the first two canonical coordinates was not high (30%), relationships between yield and parameters characterizing the individual organs were analyzed separately to establish the significance of their impact.

Multiple regression analysis (Tables S15–S28) was applied to evaluate the dependence of triticale grain yield on elements of yield structure (number of grains per spike and thousand-grain weight), biomass in the spike formation stage, and contents of nitrogen and sulfur in stems, leaves, and spikes (

Table 6. Associations between yield of triticale and analyzed variables.

Independent Variables	Coefficient of Regression	Standard Error	t-Statistic	p-Value
Intercept	−32.84	11.41	−2.879	0.006
TGW	0.024	0.033	0.726	0.471
NGS	−0.038	0.035	−1.081	0.285
B–St	0.426	0.172	2.468	0.017
B–Sp	0.669	0.422	1.588	0.119
B–L	−0.286	0.342	−0.836	0.407
N%–St	−10.72	6.539	−1.639	0.108
N%–Sp	−6.818	4.17	−1.635	0.109
N%–L	−3.748	1.66	−2.258	0.029
S%–St	117.8	64.95	1.814	0.076
S%–Sp	122.5	66.16	1.851	0.071
S%–L	31.95	17.16	1.862	0.069
N:S–St	1.692	0.733	2.309	0.026
N:S–Sp	0.92	0.641	1.435	0.158
N:S–L	0.48	0.524	0.916	0.365

Abbreviations: TGW—thousand grain weight; NGS—number of grains per spike; B–St—biomass of stems; B–Sp—biomass of spike; B–L—biomass of leaves; N%–St—nitrogen content in stems; N%–Sp—nitrogen content in spike; N%–L—nitrogen content in leaves; S%–St—sulfur content in stems; S%–Sp—sulfur content in spikes; S%—sulfur content in leaves; N:S–St—nitrogen to sulfur ratio in stems; N:S–Sp—nitrogen to sulfur ratio in spikes; N:S–L—nitrogen to sulfur ratio in leaves.

). The model accounted for approximately 63.3% of the variance in perception scores (R² = 0.633), with an R² of 0.519. On the basis of the test statistic value F(14.45) = 5.54, the empirical level of significance in the evaluation of model adequacy was lower than 0.05, while the standard error of estimation was 0.81332. In contrast, the foliar application of a sulfur-enriched fertilizer may contribute to maintaining the stability of this ratio in these organs. Analysis of the impact of contents of nitrogen and sulfur in particular parts of triticale plants has shown that a much better indicator of the size and variability of yield is the ratio of N to S. Analysis of the impact of N:S ratio in particular parts of triticale plants in the stage of spikes formation has revealed a significant relation between N:S ratio in stems and triticale grain yield (Table 6).

3.3. The N:S Ratio in Stems

Among the N:S ratios determined in triticale plants, the stems, leaves, and spikes, the most significant impact on yield was found in the stems (Table 6), and was a more adequate indicator of yield variability (Figure 3). According to the variance analysis of biomass, no significant second-order interaction was found between treatments and years of study (Tables S6–S8).

Results of canonical variate analysis based on comparison of parameters describing yield and biomass of stems of triticale plants versus different fertilization variants are shown in Figure 4. The highest triticale grain yield was found, first of all, to be related to high N:S values in stems in 2020, and then with biomass (Figure 4a). Percentage contents of sulfur were similar for all variants of fertilization (Tables S12–S14). The year of experiment was a factor that strongly differentiated the yield of triticale and the studied parameters of stems (Figure 4b). Experimental results obtained for the year 2020 were more scattered than those for the years 2019 and 2021, indicating that the poor stability of winter triticale yield may be mainly related to variability in the N:S ratio in stems and their biomass.

Results of analysis of variance have shown that effects of second-order interactions have a significant impact on the N:S ratio in all triticale plant parts analyzed (Tables S3–S5). Based on the biplot, the shortest and longest distances from the point of the plot origin indicate that the greatest stability of the N:S ratio in the stems was observed for fertilization variant 2N25 + S (Figure 5), while the lowest was for variant 1N25. Comparing the stability of the N:S ratio for different fertilization treatments, the most similar stability was observed for AN, 2N25 + S, and 1N20 (Figure 6a). The points corresponding to these treatments were the closest and thus formed a single cluster, in contrast to points corresponding to 1N25 and 2N20 + S variants appearing at much greater distances (Figure 6a). Stability of the N:S ratio was most similar for the years 2019 and 2021, whereas for 2020, the second-order interaction effects were the highest. In the dendrogram, 2019 and 2021 were clustered together as points corresponding to them were at shorter distances to the point of origin, while the point corresponding to 2020 was further from the origin, which illustrated its distinctiveness.

3.4. Additive Main Effects and Multiplicative Interaction Model

According to the plot illustrating the relationship between yield stability and mean yield of winter triticale, higher average annual yields were obtained for variants of fertilizers enriched with sulfur; however, yield stability for them was lower (Table 4 and Table 5), as evidenced by the distance of individual data points from the zero horizontal line. In particular, the yield of winter triticale obtained after application of fertilizers enriched with sulfur (2N20 + S and 2N25 + S) indicated a greater adaptability of this crop to weather conditions observed in 2020 (Figure 7). In general, variants with nitrogen fertilizer enriched with sulfur applied in the second and third doses on leaves resulted in higher yields across all years of study, compared to the effects of other variants of fertilization (Figure 7). Although all foliar treatments, 1N20, 1N25, 2N20 + S, and 2N25 + S, led to higher mean triticale yields, yields obtained for variants providing higher productivity were less stable than those of the control variant (AN).

4. Discussion

The study demonstrated that sulfur-enriched hybrid fertilization increased the grain yield of winter triticale, but simultaneously reduced yield stability compared to the conventional ammonium nitrate (AN) treatment. Meteorological conditions had a strong modifying effect, with the most favorable experimental year, i.e., 2020, amplifying yield differences between treatments, particularly for sulfur-enriched variants. Although hybrid fertilization improved yield potential across experimental years, the control variant AN showed the highest stability, confirming that yield level and stability do not necessarily coincide. These outcomes highlight a clear productivity–stability trade-off that must be considered when designing fertilization strategies.

The additive main effects and multiplicative interaction model used in our study indicated the possibility of increasing yield through reduction in the yield gap, the difference between the yield that can be obtained under optimum plant growth conditions and the yield actually obtained after the implementation of changes in the fertilization method and the properly chosen type of fertilizer. Environmental stress is unavoidable for plants growing in the natural environment and is reflected in variations in yield stability. Variabilities in meteorological conditions, including temperatures lower or higher than optimum, water insufficiency, excess of water, salinity, and level of pollution, have significant effects on the growth and development of plants and on yield [66,67]. In Poland, the main factors having an impact on yield variability are meteorological conditions, including the distribution of precipitation in the vegetation season, and temperature, type of mineral fertilizers used, proper soil conditions for cultivation, type of plant protection measures used, and choice of variety and date of sowing. In the climate of Poland, heat stress is most often accompanied by drought, which contributes to yield loss. Meteorological conditions in the years of study were characterized by a high variability of precipitation and temperatures relative to long-term averages. Total precipitation in the three seasons was 382 mm, 369 mm, and 474 mm, respectively. The 2019/2020 season was characterized by lower precipitation in autumn and winter (in particular in December), while in February the precipitation was higher than in other years, which considerably increased water retention in the soil (Table 2). In 2021, the precipitation level was the highest, but a period of the lowest temperatures was observed. Throughout the crop vegetation season, the most dangerous conditions are too low (ground frost) or too high temperatures. The related heat stress may lead to yield losses through hampering plant growth and development of grain structure elements, in particular in the plant development stages from spike formation to watery maturity of grains, leading to flower sterility and rejection of grain buds [68]. For the evaluation of the effects of thermal conditions on the vegetation of winter crops, the most important are autumn and early spring (resumed vegetation) [68]. High temperatures in the stages of spike formation and blooming lead to shortening of these stages, so in a decrease in the number of grains in the spikes [69]. The minimum temperature determines the time of onset of physiological processes leading to the appearance of the first leaf and the beginning of CO₂ assimilation [19]. Low temperature in the first stages of plant growth probably hampered metabolic processes in plants, affecting the dynamics of the uptake of nutrients from the soil, thus also the size of the yield.

Grain yields were high relative to average values for Poland [70] and varied from 7.5 t·ha^–1 to 11.43 t·ha^–1. The wide range of yield variation points to a very poor yield stability and a strong dependence on meteorological conditions. Grain yield in 2019 and 2021 was 17% lower than that in 2020. Triticale, similarly to wheat, has a high productivity potential, and with sufficient water and mineral nutrients, should bring a high yield. Although the yield range between treatments reached 1.9 t·ha⁻¹ (Figure 1, Table 5), such variability is relevant in practical terms. For a typical production level of 6–8 t·ha⁻¹, it represents approximately 25–30% of total yield, which may influence profitability at the farm scale. In favorable seasons, sulfur-enriched hybrid fertilization provided a clear yield advantage, while in less favorable conditions, it involved a higher risk of reduced yield, reflecting the observed decrease in yield stability. At the same time, the hybrid approach required lower nitrogen input compared with conventional fertilization, which may lower production costs and reduce environmental load. For farmers, this means that the choice of fertilization strategy involves balancing yield potential, yield stability, and input intensity. In years with favorable weather, hybrid fertilization may improve profitability per hectare, while in less favorable seasons, the economic outcome will depend on the extent to which reduced nitrogen use compensates for increased yield variability. In this context, the observed differences should be interpreted not only agronomically, but also in relation to farmers’ risk tolerance and sustainability goals.

Taking into account repeated periods of drought in Poland, a change in fertilizer application method and choice of appropriate fertilizer may be an alternative to standard procedures and may be beneficial in the context of sustainable agriculture. Understanding the relationship between the level of nitrogen fertilizer supply and meteorological conditions in the years of study may help in the choice of the nitrogen fertilization method, ensuring the lowest use of fertilizer and maximizing both profitability and sustainability. Enrichment of nitrogen fertilizers with thiosulfate and natural plant extracts has brought about an increase in triticale grain yield by 12 to 25% [12].

Another important factor affecting yield size is ineffective management of nitrogen supplementation [71]. In the middle stages of plant development, the best choice is foliar application of nitrogen fertilizer, as in later developmental stages, the availability of nutrients from the soil is reduced because of the aging of root systems and accompanied water insufficiency in the soil [72]. Foliar application of nitrogen fertilizer at later stages is also beneficial for environmental reasons as it reduces the risk of nitrogen loss due to denitrification and nitrogen washout [73,74]. Implementation of the hybrid method of nitrogen fertilizer application should be more effective as it supplies nutrients at critical developmental stages [12]. Moreover, the proposed method is more environmentally friendly because it permits more effective use of nitrogen and reduces its loss [75,76]. According to Robertson and Vitousek [23], a lack of synchronization between the release of bioavailable nitrogen species and the dynamics of their uptake contributes to the loss of nitrogen from agroecosystems. Gaj et al. [77], in an experiment on winter triticale, showed that the correct choice of fertilizer and the hybrid method of its application stimulated microbiological and enzymatic activity of soil, which resulted in enhanced availability of mineral nutrients and thus, in increased yield by 16–24% relative to effects of conventional fertilization. Use of nitrogen from fertilizers depends on many factors, particularly on the dose and method of fertilization application, soil humidity, and temperature [12]. The results of Gaj et al. [12] on triticale also showed that the tested group of fertilizers ensured a higher nitrogen use efficiency. Moreover, the hybrid method of fertilization generated the carbon trace lower on average by 30% relative to that of conventional fertilization [78]. Although hybrid fertilization resulted in reduced yield stability, it may still offer environmental advantages, particularly through lowering carbon emissions and enhancing nitrogen-use efficiency. Usually, nitrogen fertilizers are used in solid form, while foliar application of liquid fertilizers is an additional type of fertilization used to supplement plants with magnesium, microelements, or nitrogen in a solution of urea [79].

In general, grain yield is analyzed as a function of two components: number of grains and individual grain mass [80,81]. In our experiment, a comparison of intergroup and intragroup variabilities of the analyzed variables showed that thousand-grain weight was significantly differentiated by the variation in nitrogen fertilizer doses. Analysis of variation in canonical variables showed no positive correlation between grain yield, mass of thousand grains, or number of grains per spike. With increasing triticale grain yield, thousand-grain weight decreased. No positive correlation was found between grain yield and the number of grains per ear and thousand-grain weight. The highest thousand-grain weight was found in 2019, and the highest number of grains per spike in 2021, while the greatest triticale grain yield was obtained in 2020 (Figure 3). No significant differences were observed in grain mass between plants from plots fertilized with ammonium nitrate and those on which the hybrid method of fertilization was applied.

4.1. Winter Triticale Grain Yield Versus Its Stability

Productivity of winter triticale is a function of its adaptability, whereas yield stability is a statistical measure of the variant x year interaction [82]. Multi-year field trials are the most important, but at the same time, the most expensive part of fertilization techniques evaluation in the process of their commercialization. Our results point to benefits following from the decomposition of the data matrix under study and the visualization of empirical departures from the line describing differences in relations between winter triticale yield and the variant of fertilization. Selection of fertilization variants, as well as the choice of hybrids in relation to the environment, is based on their yields, but also on other important agronomic traits [55].

Investigation of relationships between stability of triticale yield and variants of fertilization and year of study can be compared to typical analyses of the choice of genotype to suit a given environment [46,47]. The significance of yield stability was analyzed by testing the hypotheses concerning particular effects of second-order interactions. This approach is similar to that in which datasets for each environment type are grouped in uniform clusters used in the Eberhart and Russell model [44,83]. Furthermore, understanding of interactions between fertilization variant and year of study [84] is crucial for developing fertilization techniques allowing limitation of the use of nitrogen fertilizers and maintaining high yield in different environmental conditions. When second-order interaction effects are insignificant, yield performance remains relatively consistent, regardless of specific conditions, allowing its reliable predictions [85]. However, significant interactions of the fertilization variant by year of study can lead to variations in yields depending on different conditions whose avoidance requires the selection of appropriate fertilization methods. Ma et al. [86] have shown the way genotype should be chosen in order to identify optimum maize species suitable for particular types of environment, ensuring relatively high yield and stability. Based on the results of maize yield, the hierarchical clustering was performed. However, in our work, the application of additional scaling of variables in orthogonal space permitted a comparison of the effects of second-order interactions. This comparison permitted drawing more reliable conclusions on the stability of triticale yield in relation to different variants of fertilization and years of study.

Enrichment of nitrogen fertilizers with sulfur does not guarantee yield stability. Often, a preferred approach has been the application of greater amounts of fertilizer to enhance yield and grain quality [87]. Chen et al. [88] observed that in smallholder farms growing maize, often an excess of nitrogen fertilizers was used; however, it has not led to higher yield. In order to curb the negative effects of nitrogen fertilization, sustainable crop management practices have been proposed [89], based on the use of nitrogen fertilizers enriched with additional ingredients like magnesium [90] or biostimulators, which enhance the use of fertilizer components by plants [91,92]. Requirements related to food safety and sustainable agriculture have induced the need for optimization of hitherto used methods. Tamagno et al. [93] studied the possibilities of adjustment of triticale and wheat production to these requirements, and established that, taking into account yield size and stability, cultivation of triticale under a wide range of growing conditions, including N-limited environments, will permit optimization of nitrogen fertilizer use.

4.2. Relations Between the Nutritional Status of Triticale at the Stage of Spike Formation and Yield

Plant growth and yield depend on the contents of a number of nutrients; however, to evaluate the nutritional status, it is not enough to know the contents of each individual nutrient—relations between them must also be taken into account. In our study, analyses were made taking into account ratios of nitrogen to sulfur contents in leaves, spikes, and stems of winter triticale plants at the stage of spike formation. In this way, it is possible to eliminate the impact of dry mass variability. It has been illustrated that decomposition of the data matrix in analysis and visualization of empirical departure from differences [50] in the determination of the yield of winter triticale depending on the variant of fertilization and years of study can be useful.

The experimental factor (nitrogen fertilization) was found to have no effect on N:S ratios in spikes and leaves. The stability of triticale yield (Figure 1 and Figure 2) has not been related to the N:S ratio in the analyzed plant parts in the spike formation phase (Figure 5 and Figure 6). Analysis of relationships between N:S ratios, particularly in plant parts at the stage of spike formation and components of yield structure, has shown that changes in N:S ratios in leaves and spikes had the greatest impact on thousand-grain weight. A natural source of nitrogen for forming grains is the vegetative parts of plants. It is estimated that 70–80% of the nitrogen in grains is remobilized from reserves accumulated earlier in vegetative organs [94,95]. It was observed that a reduction in nitrogen fertilization contributed to changes in the stability of the N:S ratio in the stems. In contrast, the foliar application of a sulfur-enriched fertilizer may contribute to maintaining the stability of this ratio in these organs. From the viewpoint of environmental protection, a higher coefficient of nitrogen remobilization is desirable, as it is associated with a greater plant ability to uptake this element from the soil and lower loss of this element [96]. Another argument for taking care to ensure the optimum N:S ratio is the fact that a deficiency of sulfur decreases the effectiveness of the use of nitrogen supplied from fertilizers [97]. The optimum N:S ratio should vary from 15 to 10:1, depending on plant species [98]. The proportion of contents of nitrogen to sulfur determines yield as well as qualitative parameters of usable yield [99,100]. Sulfur affects the degree of nitrogen usage, takes part in many biochemical and physiological processes in plants, in the biosynthesis of proteins, carbohydrates, and fats, and is indispensable for the synthesis of chlorophyll [101,102].

4.3. Relations Between Yield and Biomass of Triticale at the Stage of Spike Formation

In addition to knowledge of the contents of nutrients, biomass is a good indicator for the prediction of yield. Usually, the biomass of above-ground parts of plants is determined as it can be related to particular developmental stages of plants [68]. Crop yield is determined by the plant’s ability to accumulate and effectively distribute carbon among individual plant parts throughout the entire period of vegetation (generative and vegetative growth) [103]. Analysis of the relation between biomass at the BBCH 55 stage and triticale grain yield (Table 6) showed a significant relationship between yield and biomass of spikes in the stage of spike formation, accounting for 38% of the yield. Information on biomass at early developmental stages permits the prediction of final yield with a high probability. Moreover, when the plant growth is inhibited by a deficiency of nitrogen or other elements, there is a possibility of additional fertilization.

4.4. Implications for Sustainable Fertilization Strategies

Overall, our results indicate that sulfur-enriched hybrid fertilization substantially increased grain yield potential, although this improvement comes with a reduction in inter-annual stability. Environmental conditions played a decisive role in shaping productivity outcomes, with favorable moisture and temperature patterns in 2020 amplifying treatment effects far more strongly than in the remaining two experimental years. The N:S ratio in stems reflected nutrient balance dynamics and suggested that higher yield potential may increase the risk of temporary nitrogen–sulfur imbalance, particularly under rapid biomass accumulation. By integrating agronomic field data with SVD-based interaction analysis, CVA, and hierarchical clustering, we were able to disentangle how fertilization strategy and climate variability jointly drive production outcomes in triticale. Taken together, these findings demonstrate a clear trade-off between yield maximization and yield stability, highlighting the importance of matching fertilization strategy to environmental risk, nutrient balance, and sustainability goals.

5. Conclusions

The three-year study on triticale grain yield showed that the use of a hybrid fertilization method with a nitrogen fertilizer enriched in sulfur had a positive effect on yield. Yield increased by 9.5% to 27.5% relative to the yield obtained from the control plot on which standard fertilization with ammonium nitrate (AN) was applied. Interactions between experimental factors (differing applications of nitrogen fertilizer) and meteorological conditions in the years of study were highly significant. Enrichment of nitrogen fertilizer with sulfur contributed significantly to a reduction in yield stability, but stability was more strongly impacted by meteorological conditions. Introduction of foliar application of fertilizers enriched with sulfur can lead to increased yield, but it does not guarantee yield stability. These findings indicate that fertilization strategies can enhance yield while simultaneously improving nitrogen-use efficiency and reducing nitrogen input, which aligns with the goals of sustainable crop production. At the same time, the observed trade-off between higher productivity and reduced year-to-year stability provides a practical basis for selecting fertilization strategies that support sustainable and environmentally responsible triticale cultivation.