Introduction

In viticulture, Nitrogen (N) is not only crucial for vegetative growth, fruit set and resulting yield but also for grape quality and the subsequent wine fermentation processes that depend on nitrogenous compounds of the grape must (Bell & Henschke, 2005; Keller et al., 2015; Verdenal et al., 2021). During the twentieth century, N fertiliser was widely applied in viticulture to ensure maximum yield and optimum grape quality. Excessive use of N fertilisation is associated with a huge risk of N losses to the environment, as observed in many wine growing regions in Germany (Hessian State Agency for Nature Conservation Environment and Geology, 2022). Losses can occur through the emission of the gaseous dinitrogen oxide (N₂O) into the atmosphere or through the leaching of nitrate (NO₃^—) into the hydrosphere. The latter case is particularly relevant in the wine growing regions of Hesse in Germany: due to the solubility of NO₃^—, the extensive use of N fertilisers has increased NO₃^— concentrations in the groundwater of many regions along the Rhine Valley (Figure 1).

NO₃^— contamination of water and ecosystems is a significant issue, because the uptake of NO₃^— into a human body can be harmful, especially for infants. When NO₃^— is transformed into nitrite (NO₂) in the digestion system, NO₂ can block the receptors on the haemoglobin, leading to oxygen deficit, which can be fatal for infants.

In order to reduce NO₃^— pollution and to work towards more sustainable N management, the European Union (EU) has introduced the Water Framework Directive (European Parliament & Council, 2000). In addition to monitoring NO₃^— in vineyards and adjacent water systems, the Water Framework Directive applies measures such as advising winegrowers on N fertilisation and tillage systems. Contrary to common practice which often favours higher N rates, N fertilisation of vines was found to be sufficient at ~ 50 kg ha ^- 1 year ^- 1 to ensure sustainable and optimum yield. The necessary amount, of course, will depend on factors such as soil characteristics (e.g., texture, humus and N content), the grape cultivar, rootstock, and previous season fruit count (Zufferey et al., 2015; Schreiner et al., 2018). If the nitrogen status of the vines is adequate, additional nitrogen fertilisation will not enhance yield and, in addition to pollution through NO₃^— leaching, may instead lead to adverse effects and increased risks. High N status, derived from “overfertilisation” of N, leads to high vigour. This can result in enhanced canopy growth and shading, leading to the limitation of light, which in turn could lead to reduced nitrate reductase activity, decreasing the utilisation of any N that has been taken up (Perez & Kliewer, 1982; Smart et al., 1988). Another effect of increased vegetative growth is higher canopy and grape density, which increases the risk of fungal infections by Botrytis cinerea. However, an increase in vegetative growth also leads to competition between vegetative growth, generative growth or, later in the season, fruit ripening. Affecting not only grape quality (as N is utilised for growth instead of grape ripening) increased vegetative growth can also reduce yield. (Kliewer et al., 1971; Wolf et al., 1988).

While the measures taken under the Water Framework Directive have successfully reduced peak NO_3- concentrations in groundwater over the past two decades (Hessian State Agency for Nature Conservation Environment and Geology, 2022), an undesirable side effect has often been observed: a decline in amino acid (AA) concentrations in the must of wines produced in cooler climates (Löhnertz et al., 2000; Nicolini et al., 2004). The AA concentration of the must is of particular importance in winemaking. Yeast metabolises N for its growth, which is why N is essential for a complete alcoholic fermentation. Yeast assimilable nitrogen (YAN) contains primary amino acids (except proline and hydroxyproline) and ammonium. For a complete fermentation process, must YAN concentrations of 130-160 mg L ^- 1 are required (Bell & Henschke, 2005; Keller et al., 2015). A common practice to ensure the complete fermentation of musts with low YAN concentrations is the addition of mono- or diammonium phosphate (DAP) as a supplement. While the addition of ammonium salts is useful for preventing fermentation from being sluggish or incomplete, grape must with low AA content can result in the production of substandard flavours, as AAs are precursors of desirable aroma components. The addition of ammonium salts to grape juice can reduce the production of aromatic thiols, such as 4-methyl-4-mercaptopentan-2-one, by up to 30 % through Nitrogen Catabolic Repression, or it can even increase the development of the carcinogenic ethyl carbamate (Ough et al., 1991; Subileau et al., 2008). Although there are other additives like yeast hulls, which can provide nitrogenous compounds to the grape must, ensuring sufficient YAN concentration through wise N management in the vineyard is imperative.

For sustainable N management, Nitrogen use efficiency (NUE) is often used to compare how efficiently different N-fertilizers are taken up by the plant. Fertiliser efficiency can be described as “the percentage of fertilizer that is taken up by the plant accounting for background soil N levels” (Congreves et al., 2021). Different NUEs exist in the literature. Here, to determine the effect of N fertilizer on grape berry quality, we use an NUE that describes the efficiency of an N fertiliser in terms of its uptake and translocation to the grape berries. A commonly used N fertiliser in viticulture is calcium ammonium nitrate (CAN). How well the use of CAN will ensure sufficient must YAN concentrations has not been investigated alongside other fertilisers, such as Entec 26®, which supposedly reduces nitrate leaching through the addition of nitrification inhibitors for the slow release of nitrate, or urea ammonium nitrate (UAN), which is applied in the controlled uptake long-term ammonium nutrition (CULTAN) and also releases nitrate slowly due to high concentrated depot fertilisation (Walg, 2023). Despite not being a substitute for soil N application, the foliar fertilisation of grapevine with urea during veraison as well as fertigation have great potential for increasing grape must YAN concentration (Reynolds et al., 2005; Garde-Cerdán et al., 2014; Gutiérrez-Gamboa et al., 2017; Wagner et al., 2023). While some studies do not report any effect on YAN concentrations, they do report an increase in amino acid concentration (Murillo-Peña et al., 2023). A few studies have compared soil and foliar-applied N fertilisation in vineyards, finding that foliar fertilisation with urea is more effective in boosting grape juice ammonium and amino acid concentrations (Hannam et al., 2016; Canoura et al., 2018). These studies did not consider different soil fertilizer forms or application techniques, and were not carried out on the Riesling cultivar or in locations with cool climates.

Little information exists on how the N application method (subsoil, top soil or foliar application) and form of N in the fertiliser (ammonium, nitrate or urea) affect the yield and quality of the berries. To our knowledge, until now, no direct comparison of N fertilisers and application methods has been made on the Riesling grapevine cultivar grown in the Rhine Valley terroir of Hesse in Germany; this variety is of significant relevance, as it not only originates from this area but also represents one of the most extensively cultivated varieties. We therefore carried out a three-year field study to explore the influence of the various N application methods and forms on the N levels, YAN and flavour-active AAs in Riesling berries with a dual focus: enhancing grape quality and establishing a sustainable N management strategy. To our knowledge this is the first study to investigate the effect of applying different N fertilisers (N-forms and application method) to Riesling grapevines in a cool climate on the N content of the grape berries or must YAN content.

Materials and methods

1. Experimental site and plant material

The field experiment was conducted from 2012 until 2014 in a vineyard of the Hochschule Geisenheim University in Germany (49° 59’ 20” N, 7° 55’ 57” E, 147.5 -157.5 m.a.s.l.). Soil Ap horizon (0-30 cm) consisted of 29 % sand, 46 % slit, and 25 % clay; the initial soil nutrients are shown in Table 1. Bt horizon (30-60 cm) comprised 27 % sand, 45 % slit, and 28 % clay. The vineyard had been planted in 2007 with Riesling (Vitis vinifera L. cv. Riesling) grafted onto Selection Oppenheim 4 (Gm 47) rootstock. Canopies were trained according to a vertical shoot positioning. Vine rows where oriented north to south and the vine spacing was 2.10 m × 1.05 m (4535 vines ha ^- 1). Permanent grass cover was maintained in the vine rows, while alternate interrows were kept bare. Vines were trimmed once per season at the beginning of June and pruned once after the season in December (2012 and 2013). The grapevines were managed according to Good Agricultural Practice (Bundesminsterium für Ernährung und Landwirtschaft, 2024).

2. Experimental design and treatments

For the experiment, the vineyard was split into four blocks for a randomised block design. Each block comprised eight cells, each representing one of the eight distinct experimental groups (i.e., the treatments; see below). Each cell functioned as a biological replicate. Each cell contained a total of 48 vines in four rows; i.e., 12 vines per row (Figure S1). To mitigate the impact of edge-related discrepancies, only the central 16 vines were selected for analytical evaluation. All of the experimental groups received 50 kg ha ^- 1 year ^- 1 of N in different N forms (Table 2):

Control treatment: In the control group, no N was applied over the 3 years. The soil nutrient content was recorded before the start of the trial (16 March 2012) as follows: 0.1 % total N, 10 kg ha ^- 1 NO₃^— - N, 44 mg 100 g ^- 1 P₂O₅, and 43 mg 100 g ^- 1 K₂O (Table 1). The last fertiliser to have been applied before the start of the experiment was Mg [SO₄]·H₂O (340 kg ha^– 1) in 2007.

Treatment 2: Calcium ammonium nitrate (CAN, Yara, Germany) was applied three times by hand within-row and inter-row on 18 May 2012 (BBCH 15), 21 May 2013 (BBCH 15), and 30 April 2014 (BBCH 15). The fertiliser contained 26 % N in the form of ammonium nitrate (76 % NH₄NO₃) and 10 % calcium (Ca) in the form of calcium carbonate (24 % CaCO₃).

Treatment 3: CULTAN (Controlled Uptake Long Term Ammonium Nutrition) fertilisation was carried out, modified to include NO₃ ^— in addition to urea and NH₄⁺. Urea ammonium nitrate (UAN; containing a total of 28 % N: 7 % NO₃^— - N, 7 % NH₄⁺ - N and 14 % urea) was diluted in water (7.5 % v/v), and was injected in the interrow via hoses attached to a cultivator that opened up the soil (Figure 2). Application was carried out on 21 May 2012 (BBCH 15), 6 June 2013 (BBCH 55), and 5 May 2014 (BBCH 55).

Treatment 4: Fertigation was carried out via drip irrigation of a CALCINIT™ (Yara, Germany) water solution (15.5 % total N, 26 % calcium oxide, 14.4 % NO₃^— - N, 1.1 % NH₄⁺ - N). One dropper with a discharge rate of 2 L h ^- 1 was installed per vine in the row. The application process involved an initial 30-min pre-run using only water, followed by a 2-hr main run with a Calcinit® solution (1.82 g L ^- 1), and then a final 30-min water flush. This cycle was repeated once a week for a total of 10 times, beginning at growth development stage BBCH 74 in the first year on 2 July 2012 and at the end of flowering (BBCH 69) in the following years on 2 July 2013 and 6 June 2014.

Treatment 5: N was given as urea (CH₄N₂O) via Folur® (Tradecorp, Germany). Folur® was used as foliar fertiliser containing 21 % N and was applied once a week in a total of 10 times with a backpack sprayer as a 2.5 % v/v solution (Folur® + H₂O) starting at growth development stage BBCH 74 in the first year on 2 July 2012 and at the end of flowering (BBCH 69) in the following years on 4 July 2013 and 6 June 2014.

Treatment 6: Liquid fertilisation with UAN (28 % total N) diluted with water (5 % v/v) was carried out manually using a watering can. The fertiliser was thereby applied in bands in the interrow at BBCH 15 on specific dates each year: 18 May 2012, 21 May 2013, and 30 April 2014, respectively. This fertilisation method was a low-concentration alternative to the Controlled Uptake Long-Term Ammonium Nutrition (CULTAN) method.

Treatment 7: N was applied in compost (Veolia, Germany). The N content of the compost (1.4 g N 100 g ^- 1 compost fresh weight (fw)) was determined via lithium digestion and was used for the calculation of the compost application rate. The compost was manually applied at a rate of 150 kg N ha ^- 1 3 years ^-1 on 23 April 2012 (BBCH 13).

Treatment 8: ENTEC® 26 (EuroChem Agro, Germany) comprising a total of 26 % N (7.5 % NO ₃ ^— N and 18.5 % NH₄⁺ - N) and nitrification inhibitors was applied manually at BBCH 15 on the following dates: 18 May 2012, 21 May 2013, and 30 April 2014.

3. Sampling and Analyses

3.1. Soil NO₃^—-N

Over the three study years, soil sampling was carried out twice a year at sprouting (15 May 2012, 2 May 2013, and 8 April 2014) and after leaf fall (9 November 2012, 6 November 2013, and 5 November 2014) at soil depths of 0-30 cm, 31-60 cm, and 61-90 cm. In each cell, four soil samples were taken (90 cm soil depth) and pooled for further analysis. Soil available N (kg NO₃^—-N ha ^- 1) was obtained by extracting plant available N from samples using 0.0125 M calcium chloride (CaCl₂) with shaking (1 h; 20 rpm) in an overhead shaker, subsequent filtration (87 g m^-²) and analysis for NO₃^— in a flow injection analyser (FIAstar 5000, Foss, Germany). The plant-available N was calculated by summation of NO₃^— content from all three soil depths.

3.2. Leaf and berry N content

For the evaluation of grapevine N-status, leaf N content (% dry weight (dw)) was sampled once at the end of flowering (BBCH 69) until harvest on a weekly basis in all 3 years. In each cell, the samples were obtained by randomly sampling 10 basal leaves along with their petioles opposite a grape cluster. For the determination of berry N content (% dw, mg N berry ^- 1), samples were taken at grain size (BBCH 73) on a weekly basis until harvest. A total of 80 berries were sampled per cell on the first sampling date and 50 berries, including the rachis and pedicel, on subsequent sampling dates.

Berries were lyophilised and then finely ground with a mortar. Leaf and berry N were determined by wet incineration. The samples were mixed with a digestion mixture (46 mM Se, 96 mM Li₂SO₄, 18 M H₂SO₄, and 9.8 M H₂O₂) and incinerated. During digestion, the sample was boiled with an excess of sulphuric acid in an open flask to remove the organic components; the organically-bound N was not oxidized and remained in the sample as ammonium sulphate ((NH₄)₂SO₄). A subsequent analysis using a flow injection analyser (Foss, FIAstar 5000 Analyzer) was carried out to determine NH₄⁺ content. Harvesting was carried out manually on 16 and 17 October 2012, 21 and 22 October 2013, and 6 October 2014; the dates were chosen based on a °Brix value of 80 °Oe in the must of the control group and the weather conditions. The inner 16 vines in each cell were used to determine yield-related data and to produce grape must. The harvested berries were processed to form grape must on the day of harvest without being destemmed using a tube press (Willmes, WP 100) with a pressure of 0.2-1.5 bar. A 25 L glass balloon was then filled with the grape must for further processing to produce wine.

3.3. AA content

AA content of the must was determined according to Krause and Löhnertz (2017). The extraction was performed using lithium citrate buffer (2.2n, Sykam, Eresing, Germany) and norleucine (200 nM ml ^- 1) as the internal standard followed by filtration (syringe filter, nylon, 0.45 µm, MS Scientific, Berlin, Germany). Subsequent chromatographic separation was carried out for two hours using an amino acid analyser S433 (Sykam, Eresing, Germany) with a 4.6 x 150 mm LCA K 07/Li cation exchange column (Sykam, Eresing, Germany). The quantification of the AA (arginine, asparagine, aspartic acid, 𝛾-aminobutyric acid, alanine, citrulline, glutamic acid, glutamine, glycine, histidine, lysine, ornithine, proline, serine, threonine, and tyrosine) was accomplished using an automatically post-column ninhydrin derivatisation before photometrical detection at 570-440 nm (UV/VIS-detector) took place. Integration of received peaks was done with the software 'Clarity Amino' (data apex, Prague, Czech Republic).

3.4. Fertiliser efficiency

In our research, having recognised the critical role of N levels in the grape berries (berry N), as opposed to the overall yield of the plant, we focused our analysis on determining how effectively berries can absorb N. This entailed comparing the N levels in grape berries from vines that had received fertiliser (Groups 2-8) to N levels in grape berries from vines that had not received fertiliser (Group 1), relative to the quantity of nitrogen fertiliser utilised (kg N ha ^- 1):

Equation 1: $\mathrm{\Delta } \text{NUE}\text{berry} = \frac{Nf yield-\bar{N}0 yield}{Nfertilier} = \frac{\mathrm{\Delta }Nyield}{Nfertilier}$

where Nf yield = total yield N content of berries from fertilised vines, N0 yield = average total yield N content of berries from unfertilised control, and Nfertiliser = amount of applied fertiliser.

4. Statistical analysis

Data were managed in Microsoft Excel software. Statistical analysis was conducted using R-Studio (Version 2023.09.1). Graphics were produced using R ((R Core Team, 2024) and R-Studio (RStudio Team, 2024)), while leaf and berry N content data were visualised in detailed figures generated by SigmaPlot (Version 13.0). Normal distribution of the data was tested using Shapiro Wilk Test and qqplot. Homogeneity of variance was tested using the Levene Test. To determine the differences of the treatments and possible fluctuations and interactions depending on the vintage, we used a mixed model for the analysis of variance (ANOVA), with treatments (between) and Year (within); the experimental block included is shown in Equation 2. As a post-hoc test, the SidakTest was used (package ‘Emmeans’) when ANOVA showed a p-value < 0.05 for treatment (between).

Equation 2: ${\gamma }_{i,j,k,}=\mathrm{ }\mathrm{\mu }\mathrm{ }~{Fertilier}_i\times \mathrm{ }{Year}_j+{\left(Fertilier\times Year\right)}_{ij}+{Block}_k+{\epsilon }_{ijk}$

We analysed the relationship between berry N (%) and yeast-assimilable nitrogen of the must (YAN, mg L⁻¹) using our mixed model. The three-way interaction could not be reliably estimated due to sparse data and it produced a singular fit. Therefore, we refitted a simpler model without the interaction year : fertiliser. Because the fertilizer treatment did not affect the berry N – YAN slope (berry N × Treatment: F = 1.22, p = 0.30) but vintage did (p < 0.001), the final model for the regression analysis was:

Equation 3: ${\gamma }_{i,j,k,}=\mathrm{Y}\mathrm{A}\mathrm{N}\mathrm{ }~\mathrm{B}\mathrm{e}\mathrm{r}\mathrm{r}\mathrm{y}\mathrm{ }\mathrm{N}\mathrm{ }\times \mathrm{ }{Year}_i+{Fertiliser}_j+ {Block}_k+{\epsilon }_{ijk }$

Type-III ANOVA with Satterthwaite’s approximation provided F-tests, and pairwise comparisons of slopes were obtained with sidak test by emtrends (package emmeans). Principal Component Analysis (PCA) was carried out using R ((R Core Team, 2024) and R-Studio (RStudio Team, 2024)) and all AA and ammonium salts as variables for each year separately. In 2012, cystine was not measured and therefore excluded from the PCA variables for 2012. Data visualisation of the PCA was done using the R-package ‘factoextra’.

Results

1. Soil NO₃^—

Across all treatments, there were no significant differences in plant available N in the soil (expressed as NO₃^— - N) before the initial fertilisation in May 2012 (Figure 3A). In 2012, plant available soil N was generally higher in November at the end of the season (ranging from 41.3 NO₃^— - N ha ^- 1 to 105.3 kg NO₃^—- N ha ^- 1) compared to in spring (ranging from 18.2 kg NO₃^— - N ha ^- 1 to 23.7 kg NO₃^— - N ha ^- 1). At the beginning of the experiment (Figure 3A), no differences in soil NO₃^—N between all the plots were found. The plots fertilised with CULTAN (105.3 kg NO₃^— - N ha ^- 1) showed an increase (p = 0.032; p = 0.0046) in plant available soil N only in November 2012 (Figure 3D) compared to the control (46.3 kg NO₃^— - N ha ^- 1). After 2012, no differences in soil NO₃^—N were found (Figures 3B, 3C, 3E, and 3F).

2. Dynamics of N levels in berries

In general, berry N (expressed as a percentage of dw) (Figure 4D–F) decreased over the season, starting at > 1 % berry N for all treatments at the growth development stage BBCH 74 and dropping to ~0.5 % N at the end of the season. Meanwhile, berry N (expressed as mg N berry ^- 1) (Figure 4A–C) increased with berry growth and ripening over time, indicating N had accumulated in the sink organ. While in 2012 initial N content was > 0.5 mg berry ^- 1 in all treatments, in 2013 and 2014 it was below 0.5 mg berry ^- 1 at growth development stage BBCH 74 in all treatments.

A detailed look at percent N content in berries reveals that at the onset of the measurements in 2012 there were no differences between all the treatments and the control group. Of note, berry N content increased in response to foliar fertilisation with urea across all 3 years and all sampling dates with two exceptions: 18 July 2012 and 25 July 2013. In addition, foliar fertilisation with urea increased the nitrogen content of the berries compared to all other treatments on all dates in 2012, with the exception of the start of measurement (18 August 2012). Only fertilisation using the CULTAN method (which comprised urea, NH4+, and NO3- as nitrogen sources) and ENTEC (primarily containing NH4+ as the nitrogen source) was able to compensate for the berry nitrogen content provided by foliar-applied urea by the end of the 2013 season (Figure 4E). In 2014, CULTAN, ENTEC (NH₄⁺ form) and Fertigation (NO₃^— as main N-from) were the only fertilisers able to reach the same N levels in the berry as foliar fertilised grapevines, but only at the end of the season (Figure 4F).

Fertilisation with the CULTAN procedure increased berry N content on several occasions, notably compared to the control: 18 September 2012 (Figure 4D), 21 August 2013 (Figure 4E), 24 June 2014, 2 July 2014, 10 July 2014, 13 August 2014, 17 September 2014, and 1 October 2014 (Figure 4F). Fertilisation via fertigation increased berry N content compared to the control in 2014 only, from 13 August 2014 and until harvest. Increased berry N content as a result of fertilisation with NH4+ fertiliser (ENTEC) can be seen on 18 September 2013, 2 July 2014, 13 August 2014 and 1 October 2014 (harvest).

N content in berries expressed as mg N berry ^- 1 (Figure 4A–C) showed similar results to percent berry N. Foliar fertilisation with urea increased grape berry N compared to all treatments. CULTAN and fertigation fertilisation procedure increased berry N content late in the season around harvest time. Supplementary data files 1–6 provide more detailed berry N and N per berry data.

3. Dynamics of N levels in leaves

In 2012 and 2013 (Figures 4J and K), leaf N content dropped to below 1.5 % at the end of the season, while in 2014 it exceeded 1.5 % on all sampling dates and in all treatments.

In 2012 and 2013, foliar fertilisation was the only treatment affecting leaf N concentration. Leaf N content was higher than the control from 12 September to 25 September 2012, and again on 9 October 2012 (Figure 4J). In 2013 (Figure 4K) treatment with foliar fertilisation increased leaf nitrogen from 10 July to the 19 September and on the harvest date compared to the unfertilised control. In 2014 (Figure 4L), foliar fertilisation increased leaf N content on the sampling dates of 10, 23 and 31 July, as well as from 7 August to 3 September compared to the control. A rise in leaf N content was also observed for CULTAN fertilisation on 7 and 13 August, as well as for fertilisation with ENTEC® on 13 August 2014 (Figure 4L). Supplementary data files 7–9 provide more detailed leaf N data.

4. Yield

There were no yield differences between all treatments and across all 3 years (Figure 5A–C). In 2012, yield exceeded 16 t ha ^- 1 in all treatments, including the control. In 2013, yield ranged from 11.4 t ha ^- 1 for compost to 13.2 t ha ^- 1 for fertigation. A similar yield range was achieved in 2014, ranging from 11 t ha ^- 1 for liquid fertilisation with bands to 12.7 t ha ^- 1 for the control.

5. Berry N content and dry substance at harvest

Berry N content (% dw, Figure 5D–F) at harvest, which shows the endpoint of the measurements at harvest time, was higher after foliar fertilisation across all the years compared to the control and to the other treatments, except in 2013, when it did not differ from fertigation, CULTAN and ENTEC®. In 2013, the treatments CULTAN, fertigation and ENTEC® showed a trend of increased mean values compared to the control. Moreover, in 2014, these differences were significant. The dry matter of berries did not show any differences compared to the control, but a decreased berry dry substance for vines treated with foliar fertilisation compared to the compost in 2012 and 2014 can be seen in Figure 5G–I. In 2014, berries fertilised with ENTEC® also had significantly lower dry substance than those fertilised with compost.

6. Total N in yield

In 2012, foliar fertilisation increased yield nitrogen (which represents the total nitrogen content in the dry mass of the berries) to 23.7 kg ha ^- 1 and 19 kg ha ^- 1 (Figures 5J and K respectively), compared to the control values of 18.2 kg ha ^- 1 and 14.9 kg ha ^- 1. In 2013, foliar fertilisation increased compared to CAN, Bands and ENTEC® but not to the control. In 2014, no impact of the treatments on yield N was found. The highest mean values were obtained in the fertigation (16.4 kg ha ^- 1) and foliar fertilisation treatments (16.5 kg ha ^- 1).

7. Fertiliser efficiency

Fertiliser efficiency, as calculated using Equation 1, represents the efficiency of the applied N fertiliser in terms of increasing berry N content compared to the unfertilised control (Figure 6). In 2012, the foliar fertilisation-treatment showed the highest fertiliser efficiency with a mean of 11.0 % and compost the lowest with a negative value of -0.98 %; meanwhile, CULTAN, ENTEC, and fertigation showed mean values of 3.2 %, 3.23 %, and 3.64, respectively (Figure 6A).

Fertiliser efficiency of foliar fertilisation (8.1 %) was higher than that of the treatments CAN (-4.64 %), bands (-0.4 %), and compost (-2.9 %) in both 2012 and 2013 (Figure 6A–6B). CULTAN and fertigation fertiliser efficiency was higher than CAN in 2013. In addition, fertigation exceeded compost fertiliser efficiency in 2013 (Figure 6B). In 2014, no differences between the fertilisers in terms of fertiliser efficiency was found, although fertigation and foliar fertilisation reached efficiencies of > 6 %, while the other treatments did not exceed 3 % (Figure 6C).

8. YAN and AA content in the must

The impacts of the treatments on AA concentration (in mg L^- 1) in the must for all 3 years are shown in Figure 7 A–BB. As a result of foliar fertilisation, glycine, lysine, alanine and Ƴ-aminobutyric acid contents increased in 2013 and 2014 compared to the control. In 2013, CULTAN and fertigation increased must YAN and arginine concentrations compared to the control; meanwhile, in 2014, CULTAN increased arginine concentrations but not YAN. In 2014, fertigation also increased the concentrations of arginine, asparagine, glycine, alanine, ammonium salts, Ƴ-aminobutyric acid, and overall YAN in the must compared to the control. The YAN concentration in the must of fertigated grapevines was higher than in that of grapevines fertilised with CAN. In response to foliar fertilisation, tyrosine, ornithine, citrulline, histidine, asparagine, serine, threonine, arginine, glutamine, YAN, and ammonium salt concentrations increased compared to the control across all 3 years. Aspartic acid and asparagine concentration in the must of foliar-fertilised grapevines increased compared to the control in 2012. Overall, YAN concentrations of the must of grapevines fertilised with foliar-applied urea increased compared to all treatments, including the control, across all 3 years, with one exception in 2014: the YAN and arginine concentrations in the musts of fertigated and foliar-fertilised grapevines did not differ. Other quantified AAs, like tryptophan, methionine, valine and leucine, did not differ across the treatments.

9. Regression analysis

The final regression model explained a substantial portion of the variation in YAN (conditional R² ≈ 0.88; marginal R² ≈ 0.88). Berry N (%) was a strong positive predictor of YAN (F = 38.54, df = 1 / 73.2, p < 0.001). Year had an effect (F = 102.9, df = 2 / 80.8, p < 0.047), with mean YAN highest in 2013 and lowest in 2014. Fertiliser treatment had a clear main effect (F = 6.33, df = 7, 80.8, p < 0.001). The interaction between berry N and Year was significant (F = 9.14, df = 2, 81.1, p < 0.001). The calculated slope for 2013 was significantly the steepest (2012: p = 0,03; 2014: p <0.001) with an increase of 793 mg L ^– 1 % berry N ^- 1 compared to 2012 (492 mg L ^– 1 % berry N ^- 1) and 2014 (317 mg L ^– 1 % berry N ^– 1).

10. PCA Analysis

Principal component analysis (Figure 9A–F) for all 3 years was used to separate the treatments in terms of amino acid (AA) concentrations in the grape must at harvest. Principal component 1 (PC1) 67.7 % and PC2 11 % (Figure 9 A & B) explained 78.7 % of the variance for 2012. Most of the AAs contributed highly to this result, while Ƴ-aminobutyric acid, proline, aspartic and glutamic acid contributed the least in 2012. In 2013, PC1 65.4 % and PC2 21 % (Figure C-D) explained 86.4 % of the variance, with glutamic acid, aspartic acid, and cystine contributing the least to the variance, and threonine, serine, glutamine, glycine, citrulline, histidine, arginine, and ammonium salts contributing the most. In 2014, PC1 71.4 % and PC2 13.3 % (Figure E-F) explained 84.7 % of the variance, with phenylalanine, lysine, and aspartic acids contributing the least to the variance and threonine, asparagine, glutamine, alanine, valine, Ƴ-aminobutyric acid, histidine, and arginine contributing the most. Exact PC loadings for all 3 years can be found in Supplementary data files 10, 11, and 12. In all 3 years, the urea foliar fertilisation treatment can be distinguished clearly from all other treatments.

Discussion

This study evaluated the impact of varying nitrogen (N) fertilisation techniques on berry N content and YAN of the grape must in the Riesling grape cultivar, offering insights into optimising grape quality for wine making. A constant increase in leaf N (% dw) in all 3 years was observed when grapevines were fertilised by foliar application with urea. Results from fertigation and CULTAN fertilisation procedures, while less definitive, still demonstrated significant increases in leaf N (% dw). Since berry N content is dependent on grapevine N status (Bell & Henschke, 2005), it is no surprise that the increase in leaf N content resulted in elevated grape berry N content. Grape berries serve as N sinks during grape ripening, suggesting a translocation of leaf N into the grape berries (Conradie et al., 1991).

Our study applied a comparable N application rate (50 kg N ha^- 1 year ^- 1) for all the different N forms and application methods in order to be able to evaluate N efficiency in terms of improving grape must quality. The foliar application of urea increased NUE by 11 % compared to the control, whereas classical soil application of N reached a maximum of 3.6 % for fertigation (NO₃^—), and 3.2 % for the CULTAN procedure (NH₄⁺, NO₃^—, urea) and ENTEC (NH₄⁺) fertiliser in 2012. Since foliar application was found to be more efficient, it may be a more sustainable way of fertilising vines, especially considering the ripening phase of grape berries. It allows for targeted N delivery, which improves uptake by the plant and reduces environmental risks compared to traditional soil methods. It is worth highlighting the fact that CAN and compost fertilisation did not improve grape must quality: although CAN fertiliser efficiency was positive in the first year (2012) at 0.4 %, it decreased to -4.6 % in the second year (2013), and was still negative in 2014 (-0.4 %); meanwhile, compost had negative N efficiency in all 3 years of the experiment. Therefore, a detailed investigation of the application of CAN and compost fertilisation for improving grape must quality should be conducted before its use can be recommended- especially since the whereabouts of N remains unknown and the data on soil NO₃^— suggest that N was leached into the hydrosphere.

Considering that fertiliser efficiency was calculated using the parameters “N content”, “dry substance” and “yield”, with no differences in yield between the treatments and the control but with differences in N content, the parameter that might have influenced the N efficiency results is the dry substance. Considering that the dry substance data showed rather high variance and a tendency for lower dry substance in foliar-fertilised grape berries, the water supply factor most likely had a strong influence on the results of this parameter. In light of this aspect, when evaluating the efficiency of nitrogen fertilisers to improve grape must quality, the focus in future studies should be on the YAN values of the must rather than on total N yield. Grape berry N content (in % and in mg N berry ^- 1) increased in response to foliar fertilisation with urea compared to all treatments, including the control, in 2014. Amino acid (AA) content of the must of grapevines with foliar applied urea significantly increased compared to all treatments and in all years. Despite these results, fertiliser efficiency did not show any differences between the treatments in 2014, which indicates that another factor might have influenced these parameters. Fluctuations in AA content of the different years are highly influenced by vintage because of water supply, temperature, and sunshine duration. In particular, “low plant disposable water in soil” and “high temperatures in August and September” seem to have strong impact on Riesling AA concentrations; this corroborates the significant results for the regression regarding the berry N content to YAN concentration of the different years (2012-2014) found in this study (Löhnertz et al., 2000). In a study by Murillo-Pena et al. (2023), amino acid concentration increased due to the application of foliar-applied urea independently of application time. Although the authors found that environmental factors seem to have great impact on the effectiveness of the treatment, they suggest that higher precipitation led to an increase in effectiveness of the treatment and thus in the amino acid concentration of the must.

In the present study, the regression analysis shows that the relationship between berry N and must YAN is more dependent on vintage than on N fertilization form or application method, suggesting that factors such as water deficit and reduced sunshine duration, and therefore less photosynthetic activity, may have influenced this relationship. In 2014, precipitation exceeded 300 L m^-2 during grape ripening from July till September, and in 2012 and 2013 it did not exceed 200 L m^-2. This indicates that water supply affected the efficiency of N fertilisation (Figures S2–S4).

Jreij et al. (2009) studied the effects of applying both soil N fertilisation and foliar fertilisation on water-deficient Sauvignon blanc vines: total N in the pulp increased more through soil N fertilisation than through foliar, while AA concentration in the berry skin was positively influenced by foliar fertilisation. Although the partitioning of AA or grape berry N content were not analysed in this study, data show that AA content increased in the treatments with foliar-applied urea and, to a lesser extent, with fertigation and CULTAN fertilisation. When processing grape berries to make the must of white wine, their pulp is more important than the skin, in contrast to red wine production, for which maceration – the dissolving and transfer of phenols, like tannins and anthocyanins, from the skin and seeds into the must - is a fundamental practice. In the present study, the concentration of YAN in the must of foliar fertilised grapevines increased in all years compared to all the other treatments, suggesting that no matter the partitioning, this treatment has the highest potential for increasing Riesling YAN concentrations.

During the early growth stages of grapevine, its leaves are an important N sink, responding to the plant’s N uptake (Conradie et al., 1991). When comparing our leaf N concentrations with those indicated in the literature, they fell within the adequate range (2.0-2.3 %; Verdenal et al., 2021) during veraison in all the treatments, including the unfertilised control); this suggests that the soil characteristics (N content, humus, C content, C/N relationship; Table 1) of the vineyard were favourable for adequate grapevine nutrition. This is reinforced by the YAN values of the control, which ranged from 104 to206 mg L ^- 1 over the 3 years, as values of between 130-160 mg L ^- 1 are known to be sufficient for complete alcoholic fermentation (Bell and Henschke, 2005). This might be the reason why soil fertilisation did not strongly increase berry N content and must AA concentration, which is also corroborated by the consensus in the literature that N fertilisation of N-deficient grapevines contributes to increasing grape N and must AA concentration (Verdenal et al., 2021; Bell & Henschke, 2005). The finding in the present study that soil-applied did not improve must AA concentrations, unlike foliar applied N, is corroborated by another study (Hannam et al., 2016). Meanwhile, contrasting results were found for Cabernet-Sauvignon when grown under sufficient N supply: some AAs in the grape berry must decreased in concentration, while proline content increased (Gutiérrez-Gamboa et al., 2017). It is possible that proline concentrations in the must significantly increase in water-stressed grapevines (Canoura et al., 2018). Interestingly, proline concentrations in the must were only increased in the first year for foliar fertilized grapevines. It should be noted that water stress was not measured in our study. Environmental factors, such as water availability, temperature, and sunlight, influence both N utilisation by grapevines and grape quality. This is indicated by the variations in AA levels during the 3 years of the experiment. The application of foliar urea during dry years enhances the quality of grape must. Considering YAN and total N together provides a more comprehensive assessment of fertiliser effectiveness. The PCA reveals a clear distinction between the foliar fertilisation treatment and the control and other treatments in terms of the distribution of amino acid and ammonium salt concentrations in the grape must at harvest time. Similar results have also been found in other experiments which tested foliar fertilisation with urea (Garde-Cerdán, et al.; 2014; Hannam et al., 2016; Canoura et al., 2018).

The results of our three-year field experiment indicate that spraying urea directly onto grapevine leaves enhances must quality, particularly during dry years. This method increases N content in the leaves, grapes, and the must YAN (important for fermentation) without affecting the yield or sugar content (relevant for quality). It also reduces N losses through minimised leaching. Although urea was not tested in soil samples, urea is known to be rapidly hydrolysed in the presence of urease and is thus easily degraded by urease utilizing microbes (Sigurdarson, 2018). In this study, foliar fertilisation with urea was found to have a high potential for increasing amino acid concentrations in the must, which confirms the results of many other studies (Ancín-Azpilicueta et al., 2011; Canoura et al., 2018; Cheng et al., 2021; Garde-Cerdán et al., 2014), in which foliar fertilisation with urea increased YAN concentrations in the must, even with lower application rates and in different climates. While these previous studies found similar results in N-deficient grapevines of other varieties, our study reveals that in Riesling grapevines, YAN concentration in the must can be increased by foliar fertilisation with urea, even when grapevine N-levels are adequate. Further research is necessary to confirm these findings across different grape varieties and regions, and considering the influence of terroir.

It should be noted that N-fertilisation via foliar application of urea has its limits. Higher concentrations of foliar-applied urea can cause leaf scorching or necrosis. This is mainly caused by the accumulation of NH₄⁺ and urea in the leaves as a result of high nitrogen assimilation (Krogmeier et al., 1989; Castro et al., 2022). Concerning environmental factors, it is worth mentioning that research into gaseous emissions in grasslands after N application to the soil and leaves resulted in an increase in N₂O emissions when urea was applied, regardless of whether it was applied to the soil or the leaves (Hube et al., 2022).

Further limitations to foliar urea application as an N fertiliser in viticulture need to be considered. Foliar application cannot be the sole N source in vineyards. Although plant available N is mineralised in a healthy soil containing adequate organic matter, the amount of N for vegetative growth and especially flowering might not be sufficient. If the mineralisation of organic/available N is insufficient at the beginning of the season, lower amounts of soil fertilisation will be necessary at that time. Moreover, if canopy development has not progressed to a certain extent early in the season, foliar fertilisation cannot be applied. Future directions for N management in vineyards could include combining foliar and soil fertilisation using urea; to this end, the timings and applications rates of the co-applications should be studied.

Another negative aspect of nitrogen fertilisation is the formation of the carcinogen ethyl carbamate, which can result from fermentation and wine aging when the precursors of ethyl carbamate, urea or citrulline and ethanol are present. During malolactic fermentation, urea and ornithine are formed by the enzymatic breakdown of arginine by lactic acid bacteria in the arginine deiminase (ADI) pathway. Ornithine is subsequently transformed into citrulline. (Ough, 1976; Ough et al., 1988; Ough et al., 1991; Stevens & Ough, 1993; Azevedo et al., 2002; Zhao et al., 2013). An increase in arginine content in the grape must, as observed in our study for foliar fertilised grapevines, might increase the risk of the formation of the precursor of ethyl carbamate. Additionally, the AA citrulline increased in the grape must of foliar urea fertilised grapevines in two years of this study, indicating an increased risk of the formation of ethyl carbamate.

While our study focused on the enhancement of grape must YAN to ensure a complete fermentation and good quality wine, the use of foliar-applied urea is not limited to grapevines. Insufficient YAN content of apple must has been reported to result in reduced apple cider quality due to incomplete fermentation and formation of hydrogen sulphides (Peck et al., 2016; Ma et al., 2018). The application of foliar applied urea (1 %, 5 times) has been found to increase the YAN content of the apple must of ‘Red Spy’ apples by 229 % and 408 % compared to the control (Cook et al., 2024). While no differences in yield and fruit quality were observed when soil application and foliar-applied urea were compared in a study on Malus domestica cv. Gala grafted on M9, soil application increased NO₃^—-N leaching compared to foliar application (Dong et al., 2005).

Acknowledgements

We would like to thank the Department of General and Organic Viticulture of Hochschule Geisenheim University for the assistance of the overall management of the experimental-vineyard throughout the year. Furthermore, we would like to express our sincere gratitude to all our colleagues from the Department of Soil Science and Plant Nutrition of Hochschule Geisenheim University for their great support and very helpful advice. Special thanks go to Stefan Muskat for his invaluable contributions to the project.

Conclusion

This study demonstrates that, under cool-climate conditions, foliar urea is the most effective nitrogen strategy for improving must YAN and amino acid composition in Riesling, outperforming all soil-based fertilizers tested at an equivalent application rate. The consistently higher leaf and berry N concentrations achieved through foliar application indicate more efficient N uptake and translocation during ripening, whereas the weak or inconsistent responses to soil fertilization suggest that additional soil N provides limited oenological benefit when baseline vine N status is adequate. The robustness of foliar urea across contrasting vintages further highlights its suitability under increasing climatic variability.

Overall, the findings support a shift toward integrated N management in which foliar urea serves as a targeted ripening-stage supplement, complemented by limited early-season soil N inputs. However, potential foliar injury at higher concentrations and the requirement for soil-derived N before canopy development constrain its sole use. Moreover, the observed increases in arginine and citrulline warrant further investigation regarding potential implications for ethyl carbamate formation during winemaking. Future studies should refine optimal soil–foliar application regimes, assess applicability across cultivars and terroirs, and examine environmental controls on nitrogen partitioning to enhance the resilience and sustainability of vineyard N management.

Author contribution

RK conducted the field experiment and analysed the parameters, JG evaluated data statistically and wrote MS with CMG and LD. OL designed study and oversaw experiment and proof read the MS.

Data availability statement

Data are available upon request.

Conflict of interest statement

There is no conflict to declare.