Must Composition and Nitrogen Uptake in a Long-term Trial as Affected by Timing of Nitrogen Fertilization in a Cool-Climate Riesling Vineyard

Field experiment.

The trial was conducted in a vineyard of Riesling grapevines on 5C rootstock established in 1977. The site was located in Rheingau, Germany (50°N, 8°E), and consisted of loamy sand with 1.3% humus and pH 7.5. The underground soil texture was tertiary sea sand with a soil depth of more than 8 m. The percentage of the gravel in soil was 11.6%; particle size distribution in fine soil was 34.6% sand, 40.8% silt, and 24.7% clay. Field capacity was high with an available water capacity of 280 mm and a wilting point at 9 vol% (per definition measured at soil suction of pF 4.2, the base 10 logarithm of the water potential in cm). Weather data were recorded from the German Meteorological Service, Geisenheim, situated 5 kilometers from the experimental vineyard (Table 1⇓).

Vines were spaced 1.3 m by 1.9 m (vine by row), trained to one vertically positioned cane, and spur-pruned to 10–12 buds/m² at the beginning of the trial, to 8 buds/m² in 1993, and to 5 buds/m² since 1999. Slope was 10° southeast. Since 1987 there was permanent green cover in every second row. The vineyard has been fertilized since 1985 with different quantities of N (ammonium nitrate) at different application times. Treatments were 0/0, 0/30, 30/0, 0/60, 30/30, 60/0, 90/0, 60/30, and 30/60, the first number indicating N fertilization (kg N/ha) on budbreak and the second number indicating fertilization (kg N/ha) at fruit set. There was no 90 kg N/ha fruit-set treatment (0/90), because in the present climatic condition a late high N fertilization is neither recommended (Schaller 1991) nor commercial practice. Further yearly fertilization addition was 30 kg P/ha, 120 kg K/ha, and 8 kg Mg/ha. Fertilizer was annually applied in granular form distributed regularly over the vineyard; organic material was not applied. Pest control and other vineyard operations were consistent with commercially accepted practice; vineyards were not irrigated as it is generally not allowed in Germany. N treatments were completely randomized; each fertilization treatment consisted of four replicates of 48 vines (four rows of 12 vines each). Vines of the two rows in the middle, except the plants at the end of the replicates, were used for investigation.

Measurements.

All treatments were harvested on the same date. Grapes of each replication were weighed for an estimation of yield, destemmed, and crushed. The crushed grapes were pressed into musts and allowed to settle for 24 hr. Must sample was frozen at −20°C until subsequent analysis. (Due to an acetic acid rot in 1999, a considerable thinning was necessary, so that no yield data were available.) Soil was sampled in the middle of the cultivated row at budbreak (~27 Apr), at fruit set (~23 Jun), at veraison (~25 Aug), at 15 Brix (~12 Sep), and after harvest (~15 Nov) for determination of NO₃-N. Soil sampling was conducted in the middle of the rows to soil depth fractions of 0–30 and 30–60 cm; the budbreak and fruit-set samples were taken before the fertilizer treatments were applied. In 1989 and 1999 samples were analyzed for pH (CaCl₂), P, K, Mg (CAL-method), and organic C (conductometric). Cane pruning weight was determined at the end of the trial in 1999; cane perimeter was measured in 1998 at the smaller side at the bottom of 10 randomly selected canes per replicate. For the estimation of the average leaf size in 1998, on three canes of two vines per replicate the length of the leaf central vein (LLCV) was measured and calculated to leaf area (LA) with the equation LA = 1.18 LLCV² - 3.07 LLCV - 26.85 (Schultz 1992). Cane pruning weight and leaf size were only taken in one treatment per N amount level (0/0, 30/0, 0/60, 90/0). Leaf samples, opposite the basal clusters of the first bunch of all treatments, were taken at fruit set, veraison, and harvest. Leaves were washed and dried at 100°C. Leaf and juice samples were analyzed for N, P, K, Mg, Ca, Fe, Zn, Mn, and Cu after a Kjeldahl digestion with sulfuric acid and hydrogen peroxide. Methods for leaf, soil, and juice analysis are described in Schaller (2000). Analysis was carried out in duplicate.

Chlorophyll in leaves was measured by means of a Hydro-N-Tester hand photometer (Hydro, Oslo, Norway). This nondestructive method calculated the transmission at 650 nm of 30 replicates to a dimensionless value x which was calculated to chlorophyll concentration using the equation chlorophyll [mg/g FW] = 0.0047 x + 0.0758 (Rupp et al. 1999).

Free amino acids in must were determined in 1996 by HPLC. Samples were extracted with sulfosalicylic acid, derivatized with dansyl chloride and separated on an RP-18 column (Linsenmeier et al. 2004).

Statistical analysis.

Data of duplicated analysis were combined for further statistical analysis. To calculate the significance of means of the four field replicates, an ANOVA was conducted using Fischer’s test. With the total data, a two-way ANOVA was conducted. The portion of variance was calculated by dividing the sum of squares of the effect through the total sum of squares. The effect of the fertilizer N amount was tested by calculating the determination coefficient r² according to a linear regression with total annual kg N/ha as the independent variable. For a better comparison, both portion of variance and determination coefficient were expressed as percent values. The significance of the timing effect was evaluated by a three-way ANOVA using vintage, N amount, and timing as effects. The effect “timing” consisted of two groups: the budbreak treatments 30/0, 60/0, and 90/0 and the fruit-set treatments 0/30, 0/60, and 30/60 (as 0/90 was not established). To test the timing effect (exclusively), the other treatments were excluded in the three-way ANOVA.

Soil.

Organic material in the soil at the beginning of the trial was ~1.3%, which was equivalent to 0.6% organic C. At the end of the investigation, the organic material had increased compared with the content measured five years after the beginning of the trial (Table 2⇓). Whereas K in the soil increased by 50% in the last 10 years of the investigation, changes in P, Mg, and pH were low. Neither the timing nor the amount of nitrogen fertilization affected OM content or the other chemical soil parameters (Table 2⇓).

Seasonal variations of NO₃-N in the soil were clearly affected by the timing of fertilization (Table 3⇓). Fertilization at budbreak resulted in the highest N supply at fruit set, whereas fertilization at fruit set increased NO₃-N in the soil mainly at veraison. In the 30/0 and 90/0 treatments, the amount of mineral N in the soil was considerably higher than fertilization amounts. As these are mean values, within vintages notable differences between the quantities of applied N fertilizer and resulting NO₃-N in the soil could be found. On average, fertilization at bud-break resulted in higher soil N supply than fertilization at fruit set, but this result was significant only in 90 kg N/ ha treatments. Budbreak fertilization tended to increase the postharvest amounts of NO₃-N in soil.

Canes and leaves.

After a period of four years in which no differences in vegetative performance could be detected among the treatments, the unfertilized vines differed from those of the other treatments visually, that is, in growth, foliage surface, and leaf color. At the end of the trial, pruning weight was reduced by 20% and leaf size by 18% in the unfertilized control compared with the N treatments (Table 4⇓). Chlorophyll in control leaves was reduced by 10% compared with the 30 kg N/ha treatment, by 12% (60 kg N/ha) and by 17% (90 kg N/ha), respectively.

The concentration of N in leaves showed great annual differences. Therefore, the effect of the vintages clearly exceeded the fertilization effect. Nevertheless, highly significant treatment differences were noted (Table 5⇓). Independent of sampling time, the correlation between N concentration in leaves and the amount of nitrogen fertilization was highly significant. The timing of fertilization only showed an effect for treatments with 90 kg N/ha, where budbreak application resulted in higher N concentrations in leaves. Within treatments of 30 and 60 kg N/ha, no treatment effect could be noticed: even more, none of the 30 kg N/ha treatments differed from 60 kg N/ha treatments.

In addition to N concentration, only Mg and Ca concentration in leaves reacted positively on fertilization, whereas P, K, and Cu decreased with higher amounts of fertilized N (Table 6⇓). Especially P was strongly affected by fertilization, which accounted for ~60% of the variance. Furthermore a timing effect was noted with generally lowered P concentration in leaves because of N fertilization at budbreak. There was an interaction between the factors “timing” and “fertilization amount.” In the highly fertilized treatment (90 kg N/ha), P concentration in leaves was higher in the 90/0 and 60/30 treatments. The concentration of K tended to follow the observations made for P in leaves. Mg and Ca also reacted similarly: although fertilized with 30 kg N/ha, application at bud-break resulted in higher concentrations of Mg and Ca in leaves. In treatments fertilized with 60 or 90 kg N/ha, the highest concentration of these minerals was found with later fertilization at fruit set.

Grapes and must.

The effect of fertilization on grape yield was highly significant. The control yielded 16% less than the fertilized treatments. Treatments fertilized with 90 kg N/ha showed lower yield than treatments fertilized with 30 and 60 kg N/ha, respectively. Nevertheless, within the vintages, often no differences in yield could be detected (not shown). Grape yield also showed a significant average timing effect. Fertilization at budbreak resulted in 3% (30 kg N/ha treatments) to 5% (90 kg N/ha treatments) lowered yield compared with fertilization at fruit set (Table 7⇓). Total soluble solids (TSS) and titratable acid (TA) declined with higher N fertilization, whereas must pH was higher in fertilized treatments. There was an interaction of fertilization and timing on TSS: in the 30/0 treatments, TSS increased. Moderate (60 kg N/ha) and high (90 kg N/ha) fertilization showed a reversed timing effect: TSS was decreased by fertilization at bud-break compared with fertilization at fruit set. The interaction between fertilization amount and timing tended to influence TA in reverse to TSS: fertilization at budbreak decreased TA by the low N amount and increased TA by the moderate and high N amounts. Must pH was not affected by time of fertilization.

Among all nutrients in must, only N showed an increase due to N fertilization (Table 8⇓). An enormous fluctuation in annual N concentrations in must was observed. This vintage effect exceeded the fertilization effect. Whereas the variance of N concentration was affected by fertilization to 2%, vintage explained 85% of the variance. On average, unfertilized vines resulted in significantly lowered N concentrations in must. A timing effect of fertilization was found in 30 and 90 kg N/ ha treatments with lowered N concentration in the treatment fertilized at budbreak. Lowered Mg and Ca concentrations in must were also observed due to earlier fertilization.

The dominant amino acids in the must were arginine and glutamine, which combined constituted 60% of total amino acid N (Table 9⇓). High fertilizer effects were found for these two amino acids. But minor compounds such as alanine, histidine, glycine, γ-amino butyric acid, and glutamine acids, which contributed to 11% of the amino acids, were also affected by N fertilization. Pro-line, which accounted for 8% of total amino acid N, was not affected by fertilization. There was a tendency for ty-rosine and tryptophan to decline with N fertilization. As a result of the different impact of fertilization on amino acids, their profile differed slightly because of N supply: the lack of N fertilization resulted in a higher percentage of amino acids not or negatively affected by fertilization (proline, leucine, tryptophan). There was no apparent effect of timing of N fertilization on amino acid concentration in must (Table 10⇓).

Soil analysis.

Organic material increased during the 15 years of the trial (Table 2⇑). This increase was not caused by fertilization, as organic material in the unfertilized control also increased at the end of the trial and organic material was not applied to the vineyards. Levels of P, K, and Mg were all regarded as sufficient or even too high for the cultivation of grapevines (Schaller 2000). Plants given fertilizer N often show an increased uptake of N from the soil compared with those not given N (Jenkinson et al. 1985). One reason for this phenomenon is that fertilizer N can induce N mineralization in soil (Ruppel and Makswitat 1998). This interaction between fertilizer N and soil N is called “priming effect” or added nitrogen interaction (ANI). In this experiment a priming effect on soil NO₃-N was also found in 30/0 and 90 kg N/ha treatments, which has to be considered when comparing these treatments with the others (Table 3⇑). As the main uptake of nitrogen in vines is after bloom (Löhnertz 1991), under present weather condition, the earlier fertilizing date at budbreak seems to be physiologically more reasonable, as peak N supply coincides with peak N uptake. On the other hand, late fertilization in cool-climate regions is critical because the leaching of the rest of the NO₃ in soil after leaf fall is associated with an elevated level of NO₃ in groundwater (Schaller 1991, Vos et al. 2004). The higher NO₃ concentrations in the soil at fruit set compared with budbreak treatments was not significant in this trial.

Leaves.

Treatments with 30 and 60 kg N/ha did not differ in their N concentration in leaves (Table 5⇑). That matches the soil analyses, where, on average, NO₃-N concentrations found in the 30 and 60 kg N/ha treatments were similar, which is probably the result of higher mineralization due to higher amounts of organic matter in the 30 kg N/ha treatments compared with the 60 kg N/ha treatment soils (Table 2⇑). Nevertheless, the correlation between the amount of N fertilization and N concentration in leaves was highly significant, which has regularly been detected by most investigations (Conradie 2001b, Hilbert et al. 2003).

Apart from N, Mg concentration reacted positively on fertilization (Table 6⇑). Corresponding to other studies (Bucher 1969, Brechbühler and Meyer 1988), P and K concentrations in leaves decreased with nitrogen fertilization, whereas Conradie (2001a) did not find these results. Especially P was strongly affected by fertilization, which accounted for nearly 60% of the variance. According to Bell (1991), the decrease of P concentration with N fertilization is the result of a nonspecific interaction between N and P. Increasing N supply stimulates vegetative growth, which induces a decreased concentration of P by a dilution effect. In contrast, Conradie and Saayman (1989) attribute the fertilizer effect on P concentration to increased pH in the rhizosphere due to nitrate nutrition, which results in a minor uptake of P. Considering that P was much more affected by fertilization than K, Mg, and even N concentration, nonspecific interactions cannot explain the fertilization effect on P concentration in leaves in this trial.

Budbreak fertilization here increased N concentration in leaves in the 90 kg N/ha treatment (Table 5⇑). As NO₃ in the soil was elevated even in the 90 kg N/ha treatment, a timing effect could not be clearly concluded: an effect of the amount of N supply is also possible. Other investigations conducted in warmer regions either found a tendency toward higher N concentrations in leaves when N application was at budbreak (Goldspink and Gordon 1991, Bettiga and West 1991), did not find a timing effect (Christensen et al. 1994, Conradie 2001b), or found higher N concentrations in petioles when fertilization was at fruit set compared with budbreak (Peacock et al. 1991). However, in an investigation in a cool-climate region, vines fertilized at budbreak showed elevated N uptake in leaves compared with later fertilization (Vos et al. 2004).

Grapes and must.

At a sufficient N status, the additional application of N does not increase yield (Löhnertz 1991). In this trial, the optimum N amount was achieved with 30 kg N/ha. Budbreak fertilization compared to fruit set here resulted in lowered yields (Table 7⇑). Elsewhere, authors have reported an “obscure difference” where the budbreak timing yielded less (Peacock et al. 1991). Other investigations found the highest grape yield was due to N fertilization at budbreak compared with fruit set (Goldspink and Gordon 1991), but mostly no significant difference between the timing treatments was reported (Bettiga and West 1991, Dukes et al. 2001, Christensen et al. 1994, Brechbühler and Meyer 1988, Vos et al. 2004). The timing of N application can affect grape yield by two main physiological mechanisms: first, grapes are a strong sink in phase II (after fruit set) and IV (at veraison) (Conradie 1980, 1991). In this trial, however, N supply in this time was not elevated by fruit-set application compared with budbreak fertilization. So, in this case, a better nutrition of grapes can be excluded as cause for increased yield. Furthermore, high N fertilization decreases fruit set and favors shoot growth (Satorius et a. 1952), which was also the case in the present trial (Table 4⇑).

A higher amount of total soluble solids (TSS) due to a lack of N fertilization, as it was found in this trial, is often reported (Kliewer et al. 1991, Delas et al. 1991, Hilbert et al. 2003), but the underlying physiological process is unclear. Several secondary effects of fertilization play an opposite role. Considering earlier leaf fall for unfertilized vines, an accelerated maturation could be assumed and is confirmed by Scienza and Düring (1980), who found delayed leaf aging due to N fertilization. Leaf shading in fertilized treatments also could have influenced TSS in must negatively. Moreover, higher sugar concentrations in must can be caused by lowered yield. Conradie (2001b) detected higher total soluble solids as a result of N deficiency, but no correlation to yield. Conversely, N fertilization can also result in higher TSS (Bucher 1969, Fox 1995). These discrepancies can be explained by reduced leaf size and chlorophyll amount in leaves in unfertilized treatments (Table 4⇑), which also could have led to minor TSS production because of reduced photosynthesis. In fact, not yield, but leaf:fruit ratio influences sugar storage in grapes. As both yield and leaf surface were reduced by nitrogen deficiency (Tables 4⇑, 7⇑), this ratio was not as much affected by fertilization as yield alone, explaining that TSS was only slightly correlated to yield. In this trial, TSS in 60 and 90 kg N/ha fruit-set treatments were higher than in budbreak treatments. TSS was also elevated in the unfertilized control. Thus, the timing effect on TSS can be explained with differences in N supply because fruit-set N treatments showed lower average NO₃ in the soil. The minerals Mg and Ca also increased in fruit-set treatments, and N in must was higher in fruit-set than in budbreak treatments (Table 8⇑). Considering lower N concentrations in leaves and lower average NO₃ in the soil in fruit-set treatments, this reverse effect could not be affected by higher N supply. It seems more likely that grape maturity was positively affected by fruit-set timing or lower N supply. The timing effect on must composition has rarely been reported. In general, TSS and N in must have not been affected (Bettiga and West 1991, Dukes et al. 1991, Peacock et al. 1991, Christensen et al. 1994, Vos et al. 2004), perhaps because a short-term trial of a few years is not sufficient to induce fertilizing timing effects in fruits of woody plants. In a nine-year trial, Conradie (2001a, 2001b) also found the lowest sugar concentrations in budbreak treatments, but compared to posthar-vest application and a split application (budbreak, fruit set, postharvest). N in must was not affected by timing. Nevertheless in a one-year trial, the highest N concentration in must was in the fruit-set treatment (Goldspink and Gordon 1991). One study reported that total N in grapes was not significantly affected by the time of N application (Vos et al. 2004). Using labeled N15 in a one-year trial, however, the authors showed that vines fertilized at budbreak contained less fertilizer N and allocated a greater fraction of the fertilizer N to grapes than after later application.

Amino acid concentration in must is reported to be a good parameter for the N status of vines (Löhnertz et al. 1998). Higher amino N due to N fertilization (Table 9⇑, 10⇑) is often reported (Kliewer et al. 1991, Löhnertz et al. 1998, Conradie 2001b, Hilbert et al. 2003), but a group of amino acids attributed to 25% of total amino N was not affected by fertilization. Proline and tryptophane belonged to this group (Table 10⇑). One study reported that N fertilization did not influence proline in must (Conradie 2001b), whereas another found elevated pro-line concentration in grapes in fertilized trials (Hilbert et al. 2003). According to Kliewer and Ough (1970) proline increases with leaf:fruit ratio, which may be why fertilizer trials not affecting leaf:fruit ratio, as in the present trial, do not show any fertilizer influence on proline. Furthermore, proline is known as a stress indicator in plants (Schaller 2005). For the absolute concentration of proline this could not be confirmed here, but the proline:arginine ratio indeed was elevated due to a lack of N supply.

Low amounts of yeast assimilable nitrogen (YAN) in must can result in sluggish fermentation and can cause residual sugar (Spayd et al. 1995). Nitrogen concentrations of 70–270 mg N/L YAN are reported to be a minimum for low risk fermentation (Bell and Henschke 2005). However, Sablayrolles (1996) did not find that sluggish fermentation is related to assimilable nitrogen in must. The assimilable amino acids (total amino acids minus secondary amino acids, proline, hydroxyproline) constitute a major source of YAN for yeast (Conradie 2001b, Bell and Henschke 2005). In 1996 the critical value cited above was achieved by all fertilization levels, but often in unfertilized treatments assimilable amino N was under 150 mg N/L (Linsenmeier et al. 2004). Furthermore the aroma of Riesling wine is strongly affected by aroma compounds, formed during the fermentation, which are mainly influenced by amino acids in must (Rapp and Versini 1996, Linsenmeier et al. 2006). Even if the influence of fertilization on grape yield was of no relevance and TSS was increased by a lack of N supply, fertilization of 30 to 60 kg N/ha is recommended for its influence on wine quality.