Hydroxycinnamic acids as affected by different fertilization of Rebula grapevines

2.1 Plant cultivation

The experiment was set up using grapevines (Vitis vinifera L. cv. Rebula) grafted on SO4 rootstock, that were planted in the year 2007 in 21 L plastic pots filled with soil collected and sieved to pass a 20-mm screen from the surface layer (ca. 0–10 cm) of a typical vineyard of Goriška Brda in Slovenia. The calcareous soil (33% CaCO₃) had a clay loam texture (20.1% sand, 47.2% silt, and 32.7% clay) with normal electrical conductivity (0.80 dS m⁻¹), good cation exchange capacity (24.5 cmol kg⁻¹), high pH (8.0 in water), and relatively low level of organic matter (1.9% C). The concentration of extractable-P (Olsen procedure) was deficient (8 mg kg⁻¹). The contents of NH₄NO₃-extractable K and Mg and EDTA-extractable Fe and Zn were low to marginal (174, 43, 143, and 3.9 mg kg⁻¹, respectively).

Pots were maintained outdoor, covered with nets in order to avoid hail damages, and irrigated with a drip irrigation system during the summer time (ca. 1.5 L every 2 d) in order to avoid water stress. Vines were trained vertically and annual pruning retained one shoot per vine bearing 5–6 buds. On average, four canes had developed during the summer season and after first sampling (in June) only one cluster per cane was retained (four clusters per vine as maximum crop load).

2.2 Experimental design

The experiment was performed during three consecutive growing seasons (2009–2011) in Goriška Brda, within the Western Slovenia wine-growing district, an area characterized by a typical sub-Mediterranean climate with frequent dry periods in summer and with an average annual rainfall of 1200 mm (Rusjan et al., 2006). Nine treatments (along with the control–untreated vines, Table 1) were applied in a randomized design with three replicates and four vines within a replication. The rate of pre-plant mineral fertilization (50 N, 40 P, and 116 K in kg ha⁻¹) was decided based on the soil analysis that was performed before grapevine planting and according to recommendations for yearly side-dressing (Vršič and Lešnik, 2005), and was added in all pots except the control treatment. Nitrogen as ammonium sulfate (20.6% N; Italy) was applied each year in the spring, while phosphorus as superphosphate and potassium as potassium sulfate were applied in 2008 only. In addition, agrochemicals containing Mg (Bittersalz), Fe (‘Foliacon Fe', Fe complexed to amino acids), and Zn (water-soluble Zn complexed by carboxylic acids) were added each year before bloom (in May), individually or combined. Mg, Fe, and Zn applications were calculated according to producer recommendations.

2.3 Leaf and grape sampling and sample-preparation steps

Sampling and phenology dates in the years 2009–2011 are presented in Table 2. Leaves (2–3 leaves per vine of each plant) were sampled at berry set (end of June, leaf opposite to the cluster) and at véraison (beginning of August, mid-shoot leaves) (Haefs et al., 2002; Sinskey, 2009). All grape clusters (in 2009, 2010, and 2011) were collected at harvest (end of August–beginning of September). Each sample was put in its own clean plastic bag and transported to the laboratory for further analysis.

After sampling, the blades were separated from the petioles. For grape analysis, ca. 20–25 undamaged berries (retained with pedicels) were randomly selected on all clusters from each replicate. Both petioles and berry samples were first washed with tap water and then with deionized water in order to remove dust and other residues from the surface. All samples were than oven-dried at 105°C (Hoenig, 2001) to a constant weight (berries firstly at 60°C to avoid juice drainage from the berries). Pedicels were removed from berries after drying.

Another sample of 50 berries was retained and 1 g of potassium metabisulfite (Esseco, San Martino di Trecate, Italy) was added to inhibit polyphenol oxidase after berry crushing. Juice was collected and stored in a dark flask at –20°C with 1 g of ascorbic acid (Laffort Enologie, Bordeaux, France) in order to reduce any quinine formation until phenol determination (Vrhovšek, 1998).

2.4 Wet digestion of petioles and whole grape berries

After homogenization and grinding (Mixer mill MM 400, Retsch, Haan, Germany), petioles (0.3 g) and grapes (0.5 g) were digested overnight with nitric acid (3 mL and 5 mL, respectively; 65% HNO₃, Suprapure, Merck, Darmstadt, Germany) and hydrogen peroxide (four times with 0.5 mL 30% H₂O₂; Suprapure, Merck, Darmstadt, Germany) in the polytetrafluoroethylene (PTFE) beakers (covered with a lid) on a sand bath (Harry Gestigkeit, Düsseldorf, Germany) according to Hoenig et al. (1998) without the HF step. The dry residues were re-dissolved in 0.5 mL of HNO₃ with 1–2 min heating, and filled to an appropriate final volume (20 mL for petiole samples and 25 mL for grape samples) with double deionized water in polypropylene centrifuge tubes. All analyses were performed in duplicate. Potassium, Mg, Fe, and Zn concentrations were determined with atomic absorption spectrophotometry (SpectrAA-10, Varian, Victoria, Australia) in an air-acetylene flame at the following wavelengths (nm): 766.5 (K), 285.2 (Mg), 248.3 (Fe), and 213.9 (Zn). Appropriate quality controls [standard reference materials (SRMs) from National Institute of Standards & Technology–NIST (SRMs 1572-citrus, 1547-peach, 1573 and 1573a-tomato leaves; Gaithersburg MD, U.S.] were performed for each set of measurements.

2.5 HPLC–DAD quantification of grape phenolic compounds

Aliquots of juice samples were filtered (0.45 μm, Chromafil Xtra PTFE 45/25, Macherey-Nagel, Düren, Germany) and injected directly onto the high performance liquid chromatograph (HPLC) to determine trans-caftaric acid (trans-CTA) and total hydroxycinnamic acids [HCAs, sum of trans-CTA, trans-coutaric (CoTA), and trans-fertaric (FTA) acids; Mozetič et al., 2006. An Agilent 1100 series system (Agilent Technologies, Waldbronn, Germany) with an auto injector (20 µL injection volume) and a diode array UV-Vis detector (DAD G1315A), recording at 280, 320, and 530 nm, were used to detect and to quantify the phenolic compounds. The column, a PFP-2 100 A Luna (250 mm × 4.60 mm; Phenomenex, Torrance, USA), 5 µm particle size, was kept at 25°C. A constant flow rate of 1 mL min⁻¹ was used with two solvents: solvent A, 2.2% ((v/v)) aqueous formic acid (≈ 98% p.a.; Fluka, Buchs, Switzerland); solvent B, methanol (HPLC grade; Sigma-Aldrich, Steinheim, Germany). The following linear gradient was used: in 24 min from 12% to 33% B, hold for 5 min at 100% B to wash the column and then return to the initial conditions to re-equilibrate for 10 min. The identification of compounds was achieved by comparing retention times and their UV-Vis spectra at 280, 320, and 365 nm, as well as by the addition of an external standard (trans-caftaric acid, ≥ 97.0%; Sigma-Aldrich, Steinheim, Germany), which was used also for the quantification of total HCAs content (Mozetič et al., 2006) in the years 2009 and 2011.

2.6 Statistical analysis

Data analysis was conducted with STATGRAPHICS Centurion XVI software package (Statpoint Technologies, Warrenton, Virginia, U.S.). A general linear model (GLM) was applied in order to ascertain the significance of nutrient treatment (fixed factor), growing season (random factor), and their interaction on the element concentrations in petioles and grape berries. Means were separated according to Student–Newman–Keuls's test (P < 5%). The relationships between elements (K, Mg, Fe, and Zn) and between elements and HCAs were analyzed by simple correlation analysis (determination of the Pearson`s correlation coefficient).

3.1 Mineral concentrations of Rebula grapevine as affected by fertilization treatment and year

Leaf analysis is widely recognized as the most reliable way to ascertain the grapevine nutritional status (Fregoni, 1998; Sinskey, 2009), but also grape berry mineral composition must be considered equally important, since nutrients play a fundamental role in fruit development and they are involved in wine chemical composition and quality (Etchebarne et al., 2009). The season effects on almost all elements studied (K, Mg, Fe, Zn) were highly significant (Tables 3 and 4), and Fe in petioles at berry set was the only variable not influenced. Similar results were found for treatments with significant effects on almost all parameters except on Fe concentration in whole grape berries. In addition, for Mg (at berry set) and K and Fe concentrations in the petioles (at both phenophases) there was also a significant effect of the year × treatment interaction, suggesting that the effect of the fertilization treatment on these elements was different among seasons (2009–2011), as can be observed in Table 5. On the contrary, the Zn concentration in grape berries and in the petioles at both sampling times seemed to be influenced by both year and fertilization treatment, without any interactive effect.

3.2 Effect of soil NPK fertilization on K concentrations of leaves and grapes

The average K concentration of petioles of untreated vines was 27 g kg⁻¹ in the years 2009–2011 regardless of sampling times (Table 3). The K concentration was much higher in 2009 at both berry set and véraison compared to 2010 and 2011. Looking in detail at interaction effects of season and treatment, in the year 2009 the K concentration of the petioles of the treatments, to which K was added (#2–10), was 70% higher than in untreated vines, while the difference diminished in the following years (Table 5). Also in grape berries, the K concentration of grapes was significantly higher in 2009 as an average of all treatments, but the differences between untreated and NPK-treated vines was more similar among seasons (no interaction effects).

The addition of K to the soil resulted in higher K concentrations of leaf petioles and grape berries. In any case, the statistical differences in K concentration of petioles between control and NPK-treated vines (#2–10) decreased in 2011 most probably as a consequence of lowered K availability in the soil of NPK-treated vines due to yearly removal for vine growth and grape harvesting (Brataševec, 2013). The K concentration of petioles of control vines sampled at two phenophases was found within an optimal range (Fregoni, 1998; Sinskey, 2009). The K concentration of the petioles was enhanced in NPK-treated vines (#2–10) reaching high values at both sampling times in 2009, while in the following season the values fitted within the optimal range, too. Water rate was the same for all the treatments, thus, differences ascribed to water supply cannot be expected. Therefore, it is likely that greater K uptake in 2009 in NPK treated vines was due to higher soil K availability as a consequence of K fertilization.

3.3 Effect of soil NPK and Mg fertilization on Mg concentrations of leaves and grapes

The Mg concentration was approximately two-times higher in the leaf petioles of untreated (control) vines at both phenophases than of treated vines (Tables 3 and 5). During the experiment, the petiole concentration of Mg increased approx. 1.6-fold from berry set to véraison. In grape berries, the Mg concentration was nearly 20% higher in the control vines than in K-treated vines in all three years.

Opposite to K, the Mg concentration was lower than optimal in the petioles as compared with values reported in the literature (Bigot et al., 2009; Fregoni, 1998; Sinskey, 2009) and in all years basal leaf chlorosis was assessed in all treatments but not in control vines. Marschner (1995) reported that the addition of NPK yields a lower Mg concentration of petioles and this could be explained in terms of an antagonistic relationship with K. In agreement with this, the correlation between Mg and K in our experiment revealed to be statistically significant (r_max = –0.85, P < 0.1%). Another aspect important to be taken into account is related to root absorption, since Stefanini et al. (1994) already reported that SO4 rootstock poorly takes up Mg from the soil. The wide range of K concentrations of petioles (lowered by 40% from 2009 to 2010) could possibly support the high variations in Mg, if only the K–Mg antagonism were responsible for the lower Mg uptake (approximately 20% variation from 2009 to 2010 and 2011; P < 0.1%). As reported by Bergmann (1992), nitrogen can also diminish the Mg concentration of the plant; this could explain the differences in leaf Mg concentration between the control and NPK-treated vines and the trends observed over the three years.

According to Capps and Wolf (2000), Haefs et al. (2002), Stefanini et al. (1994), and Gluhić et al. (2009), foliar or soil application of MgSO₄ could be very effective in enhancing Mg levels in grapevine blades and petioles. On the other hand, no measurable effects could be obtained depending on experimental conditions (e.g., application rate, distribution timing and frequency, soil type).

In petioles at berry set the K : Mg ratio was 3.0, 2.8, and 1.9 times higher in NPK-treated than in untreated vines in the years 2009, 2010, and 2011, respectively (Brataševec, 2013). The same trend was shown also for petioles at véraison and berries at harvest, but the magnitude of difference for untreated vines was lower.

3.4 Effect of NPK and Fe fertilization on Fe concentrations of leaves and grapes

The Fe concentration was not consistently affected by soil application of NPK (#2; Table 4). In all three years, the Fe concentration in petioles sampled at berry set was significantly enhanced by foliar fertilization with ‘Foliacon Fe' (treatments #4, 5, and 6; Table 5). At véraison, the differences among treatments were reduced as compared to berry set. Looking at the year × treatment interaction, a higher Fe concentration was still observed in 2010 and 2011, when the same element was added throughout foliar fertilization (0.45% ‘Foliacon Fe'; #6). During the vegetative period in years 2009–2011, Fe concentrations of petioles of most treatments were low-marginal (< 25 mg kg⁻¹) in agreement with Fregoni (1998) and Bigot et al. (2009). However, in the treatments in which Fe was applied by foliar spraying of ‘Foliacon Fe' (#4–6), the Fe concentrations were statistically higher (P < 0.1%) and close to the reference range. In the three-year experiment no differences were obtained in petiole Fe concentrations between control and NPK-treated vines (not including vines of treatments #4–6, in which Fe was applied with foliar spraying), showing that application of NPK to the soil had no influence on Fe uptake. The results obtained at berry set in 2010 and 2011 by foliar spraying (#4–6) suggest that Fe in the commercial ‘Foliacon Fe' solution could penetrate the leaves significantly enhancing foliar Fe concentration, while the soil application of Fe (in spite of producers recommendations for vine soil applications) was not profitable for a significant increase of the same nutrient in véraison/fruit set petioles or in berries at harvest.

3.5 Zinc as affected by NPK coupled with Mg and Fe fertilization

In contrast to Fe, the Zn concentration of petioles sampled at véraison was significantly increased by NPK treatment alone and in combination with foliar or soil-applied Mg and Fe (41 vs. 69 mg kg⁻¹ in the control vs. treatments #3–9; Table 4). Parallel results were highlighted in grapes with 37% higher Zn concentrations in treatments #3–9 than in the control vines, thus, Mg and Fe fertilization provided positive evidence for higher Zn concentrations both of petioles and grapes. In addition, positive effects of Zn foliar fertilization (#10) could be seen in petioles at berry set (Table 4), although totally diminished at véraison. The lowest values were shown for petioles sampled in 2010 and the highest in 2011. In contrast to K, Mg, and Fe, no significant effects of the year × treatment interaction were found in all three samples (petioles at both phenophases and grapes).

In all treatments, Zn concentrations of vines (Table 4) showed values within the reference range of 26–150 mg kg⁻¹ (Sinskey, 2009). In contrast to Fe, Zn uptake seemed to be affected by the addition of NPK (#2) and Mg and Fe fertilization (treatments #3–9; P < 0.1% in 2010 and 2011) in petioles at véraison and, what is more interesting, in the grape berries (Table 4). Aciksoz et al. (2011) and Kutman et al. (2011) noticed that high soil N application could elevate Zn (and Fe) concentrations of wheat grains, while Peuke (2009) reported a negative influence of increased N-fertilization rate on Zn concentration of grapevines leaves. Moreover, Díaz et al. (2010) found that the application of Fe-chelate fertilizers and synthetic vivianite in one to three-years-old potted grapevines had no effect on Zn concentration or even a reduction was shown in the first year of application.

3.6 Changes in K, Mg, Fe, and Zn leaf concentrations over time

The concentrations of N, K, and P in vine are highest early in the season and then decrease as growth continues, while the concentrations of Mg of leaves increase or remain the same during vine growth (Mullins et al., 1992). In our experiment, the concentrations of Mg and Zn in leaf petioles increased by approximately 60% (on average of 3 years) from berry set to véraison. In contrast, the concentrations of K and Fe were more or less stable during the vegetative period (the average ratio between berry set and véraison was approximately 1.0 for both elements). In agreement with Christensen (1984) and Peuke (2009), the element concentrations depend on the grapevine cultivar. In our study, petiole K levels did not decline between berry set and véraison, as observed by Christensen (1984) for a wide range of cultivars. On the other hand, Peuke (2009) did not find a clear trend for K concentration and did not observe Zn enhancement in the leaves of Vitis vinifera L. cv. Riesling during the vegetative period. In agreement with Peuke (2009), the time of collecting samples for leaf analysis must be taken into account for viticulture practice, especially when the concentrations of elements change (e.g., for Mg, Zn) during the vegetative period.

3.7 Correlations between elements in petioles and grape berries

Many positive and negative correlations for K, Mg, Fe, and Zn between the petioles at berry set and véraison and grape berries at harvest were obtained during the three growth seasons comparing the average annual content of each element at defined sampling time regardless the treatment. Significant correlations (P < 5%) were found between petioles at berry set and petioles at véraison for K (r = 0.81, 0.53, and 0.73 in 2009, 2010, and 2011, respectively) and Mg concentrations (r ≥ 0.86 in all three years), and between petioles at véraison and grape berries (K: r = 0.52, 0.65, and 0.71 in 2009, 2010, and 2011, respectively; Mg: r = 0.81 and 0.65 in 2009 and 2010, respectively). As expected, negative correlations between K and Mg were found in the petioles collected at the same sampling time (at berry set: r = –0.81 and –0.62 in 2009 and 2011, respectively; at véraison: r = –0.85 and –0.64 in 2009 and 2010, respectively). No detectable correlations were found in the case of grapes.

For micro-elements, not many correlations between concentrations of the petioles and the grape berries were shown, occasionally only in one season, if any. A positive relationship (P < 5%) was found between the Zn concentration of petioles at berry set and of grape berries at harvest in the years 2009 and 2011 (r = 0.54), and between Fe in petioles at berry set and at véraison (r = 0.78 and 0.51 in 2010 and 2011, respectively).

In addition, some correlations between macro and micro-elements were evident between Mg (or K) and Zn. Significant negative correlations were found in all three years between Mg and Zn in the petioles at véraison (r = –0.5, –0.64, and –0.69 in 2009, 2010, and 2011, respectively) and between Mg in the petioles and Zn in the whole grape berries (r = –0.68, –0.75, and –0.55 in 2009, 2010, and 2011, respectively). In grape berries and petioles a synergism was found between Zn and K concentration at véraison (grape: r ≥ 0.62 in 2010 and 2011; petioles: r = 0.73; 0.57 in 2009 and 2010, respectively).

Significant positive correlations observed for K and Mg between leaf petioles and grape berries and between petioles sampled at two different vegetative stages are in agreement with a high mobility of these elements in plants (Bergmann, 1992). The positive correlation of K and Mg concentrations between grape berries and petioles confirmed the remobilization/transfer of K and Mg from mature leaves to the berries during ripening via phloem (Coombe, 1992; Etchebarne et al., 2009). Opposite to K and Mg, Fe and Zn are plant-immobile elements (Bergmann, 1992) and they need to be continuously absorbed by the plant during the vegetative growth. This may explain why the correlations of Fe and Zn concentrations between plant samples were poor, and if any, they were obtained in one season only. An exception was the correlation between Zn concentrations of grape berries and petioles sampled at berry set in 2009 and 2011 and between Fe in petioles of two different phenophases. In any case, an antagonism between these two micro-elements (Alloway, 1995) was not observed.

Beside the known antagonism between K and Mg (Marschner, 1995), which was showed also in our experiment, interesting correlations were found between K or Mg and Zn. Potassium concentrations of grapes and petioles sampled at véraison positively correlated with Zn in the same plant sample, in disagreement with Bergmann (1992) who reported that high concentrations of K could inhibit the uptake of Zn. A positive relationship was found between Mg and Zn in both plant organs, leaves and grapes. Best to our knowledge, these relationships have not been described yet.

3.8 Hydroxycinnamic acids (HCAs) concentrations of grape juice

Tartaric esters of HCAs represent the main phenols in white grapes and wines when made both with and without pomace contact (Singleton et al., 1986). Total HCAs (sum of trans-CTA, CoTA, and FTA) and trans-CTA concentrations were measured in grape juice of our experiment in 2009 and 2011. The levels of HCAs in grapes depend on many factors such as grape variety, growth conditions, climate, etc. (Rentzsch et al., 2009). By working with pots, all the factors were controlled; thus, differences observed can be ascribed to fertilization treatments. Trans-CTA is the predominant HCA in white grapes, which accounts for ca. 85% of the total HCA in Rebula grape juice (Mozetič et al., 2006). The grapes with higher proportions of trans-CTA (80–90%) in the juice have high oxidation potential, leading to browning reactions and negatively influencing the aroma and the color of the juice (Sapis et al., 1983).

The total HCAs concentration was 1.3-fold (in 2009) and 1.5-fold (in 2011) higher for the grapes of the control vines than of grapes harvested from NPK-treated (#2–10) vines (100 vs. 75 mg L⁻¹ and 82 vs. 53 mg L⁻¹ on the average in 2009 and 2011, respectively). The same trend was observed also for trans-CTA (87 vs. 63 mg L⁻¹ and 71 vs. 48 mg L⁻¹ on average in 2009 and 2011, respectively; Table 6). The percentage of individual HCAs did not differ between two years.

A negative correlation between trans-CTA in grape juice and K in whole grape berries was found in 2009 and 2011 (r = –0.69, P < 1% and r = –0.58, P < 5% in 2009 and 2011, respectively; Brataševec, 2013), while between trans-CTA and K in the petioles only in 2009 (r = –0.54, P < 5% and r = –0.64, P < 1% at berry set and véraison, respectively). In addition, a positive correlation was obtained between trans-CTA in grape juice and Mg in grape berries (r = 0.69, P < 1% in 2009) and in the petioles at both sampling times and in both years (r_max = –0.72, P < 5%). Only one significant correlation between trans-CTA and Fe in grape berries at harvest was found (r = 0.52, P < 5% in 2009).

In our experiment, a negative correlation between K in grape berries/petioles and the trans-CTA and a positive with Mg was obtained in both years. There is much information about the relationship between nutrients and anthocyanins (Awad and de Jager, 2002; Delgado et al., 2004; Hilbert et al., 2003), while for hydroxycinnamic acids, to our knowledge, data are still missing. Within the growth season (2009 or 2011), the concentration of trans-CTA and total HCAs was much higher in the control vines in comparison to the NPK-treated ones, but the percentage of individual HCAs did not differ between two years. The results of our experiment provide interesting outcomes, since NPK fertilization (treatments #2–9) was profitable for a reduction of trans-CTA in juice (25% in 2009 and 35% in 2011). In white wines, lower concentrations of HCA are desirable since there is a lower probability of browning reactions that could compromise the overall quality of wines (Romeyer et al., 1983; Sapis et al., 1983). In addition, Fe and Zn did not affect the concentration of HCA in juice, once again highlighting the leader importance of macro-elements in mineral nutrition and grape quality.