Abstract
Adequate yeast assimilable nitrogen (YAN) concentration is necessary for successful wine fermentation; therefore, supplementing musts with nitrogen (N) is a common industry practice. In the Finger Lakes region of New York, Riesling juice (Vitis vinifera L.) often has YAN concentrations below the 140 mg/L considered a practical minimal limit. However, little research exists that has confirmed N requirements for cool-climate cultivars and conditions. To test the influence of juice N concentration on desirable sensory characteristics, a Riesling juice of 20 Brix and 130 mg/L initial YAN was supplemented with diammonium phosphate to increase YAN to 180, 250, and 300 mg/L. Each supplemented fraction and the unaltered control were inoculated with three different Saccharomyces cerevisiae yeast strains: EC1118 and W15 (Lallemand), and Côte des Blancs (Red Star). The control concentration of 130 mg N/L was enough for these three yeast strains to complete fermentation, and further supplementation improved fermentation kinetics only for EC1118. N supplementation affected the concentration of eight of the 10 select volatile compounds analyzed in at least one of the yeast strains. Of these, the concentration of most esters increased with N supplementation, with the exception of ethyl cinnamate, which decreased. Concentrations of the higher alcohols 1-hexanol and 2-phenylethanol decreased, and decanoic acid increased, with increased N. Linalool and 1,1,6-trimethyl-1,2-dihydronaphthalene, two volatiles associated with “varietal character” in Riesling, were not affected. In a preference-ranking test, sensory panelists preferred the unsupplemented control (130 mg N/L) over the supplementation treatments for wines fermented with EC1118. No difference in preference was found for the W15 and Côte des Blancs treatments.
The macronutrients required by yeast to conduct fermentation are relatively abundant in grape berries. Nitrogen (N), however, is a key nutrient that is often below optimum concentrations in many viticultural regions (Gockowiak and Henschke 1992, Butzke 1998, Hagen et al. 2008). Yeast assimilable nitrogen (YAN), the nitrogenous compounds that can be used by yeast during fermentation, is composed of ammonium ions (AMM), primary amino acids (PAN), and some oligopeptides (Monteiro and Bisson 1991, Henschke and Jiranek 1993, Marsit et al. 2015). YAN notably excludes proline, a secondary amino acid that cannot be metabolized under anaerobic conditions, and grape proteins. Because of the chemical diversity of YAN components, most methods quantify AMM and PAN, as those components represent the majority fraction of YAN (Bell and Henschke 2005).
Early research on N requirements of Saccharomyces cerevisiae has established 140 mg N/L as the minimum YAN concentration for the successful completion of most wine fermentations (Butzke 1998, Bell and Henschke 2005). Later works suggest that optimal concentrations range from 200 to 350 mg N/L, depending on factors such as initial sugar concentration, yeast strain, and wine style (Bisson and Butzke 2000, Miller et al. 2007, Torrea et al. 2011, Ugliano et al. 2011). Since YAN deficiencies are known to cause stuck or sluggish fermentations, winemakers routinely supplement musts with diammonium phosphate (DAP), which can affect the production of yeast-derived volatiles, including higher alcohols, esters, fatty acids, sulfur compounds, and organic acids (Bell and Henschke 2005), but the relevant pathways are complex and not completely elucidated. Higher alcohols are proposed to be formed via the Ehrlich pathway, which involves transamination of amino acids (Torrea et al. 2011), or via central carbon metabolism (Hirst and Richter 2016). Esters can be formed by condensation of corresponding alcohol and a coenzyme A-activated acid, catalyzed by an acyltransferase (Torrea et al. 2011). Generally, N additions in N-deficient must will increase the production of ethyl and acetate esters and decrease production of higher alcohols (Hernandez-Orte et al. 2006, Miller et al. 2007). Consequently, N supplementation has been suggested as a tool to optimize expression of wine aroma (Miller et al. 2007, Torrea et al. 2011, Ugliano et al. 2011, Vilanova et al. 2012).
Monoterpenes, key aroma compounds in aromatic cultivars, are present in grapes mainly as nonvolatile glycoconjugates, and are modified to various degrees during fermentation when yeast- and bacteria-derived glycosidase enzymes liberate free monoterpenes such as linalool, geraniol, nerol, and citronellol (Strauss et al. 1986, Swiegers et al. 2005, Parker et al. 2018). Some yeast strains can also synthetize monoterpenes, a reaction favored by high YAN concentrations (Carrau et al. 2005, Vilanova et al. 2012). Yeasts may also convert precursors into the C13 norisoprenoids β-damascenone and α- and β-ionone, although this mechanism is poorly understood (Bell and Henschke 2005). In Riesling, the C13 norisoprenoid 1,1,6-trimeth-yl-1,2-dihydronaphthalene (TDN) reportedly contributes to a petrol aroma (Simpson and Miller 1983, Sacks et al. 2012). As with other C13 norisoprenoids, TDN is formed from glycosylated precursors by a combination of acid- and enzyme-catalyzed reactions (Winterhalter et al. 1990). YAN has been shown to affect the formation of β-ionone and β-damascenone, likely by changing glycosidase activity (Vilanova et al. 2012). Consequently, it can be hypothesized that TDN concentration could also be impacted by YAN concentration.
Most YAN studies have used synthetic media or juices and musts from warm-climate regions, making YAN recommendations for cool-climate winemaking largely conjectural. Further, because YAN analysis requires specialized equipment, DAP additions are frequently made prophylactically, often leading to excessive supplementation. This can have negative consequences on a wine’s organoleptic properties and may result in high levels of residual N (Bell and Henschke 2005). In addition to increasing the possibility of microbial instability, excess N may result in the formation of the known carcinogen ethyl carbamate and of biogenic amines, which can cause deleterious health effects in susceptible individuals (Monteiro et al. 1989, Daudt et al. 1992, Lorenzo et al. 2017).
Riesling is the most widely planted Vitis vinifera cultivar in New York state (www.nass.usda.gov/ny), and YAN concentrations in NY Riesling generally average below 100 mg N/L (Nisbet et al. 2014). Although several studies have characterized the volatile composition of Riesling wines (Simpson and Miller 1983, Komes et al. 2006, Bowen and Reynolds 2012), the effect of must YAN on wine composition has not been investigated. This study focused on optimizing YAN concentration in Finger Lakes Riesling fermentations by evaluating the effect of N levels on fermentation kinetics and the formation of volatile and nonvolatile metabolic products.
Materials and Methods
Grape source
Riesling grapes were hand-harvested at 20 Brix on 24 Oct 2014 at the Cornell University research vineyards in Lansing, NY. The fruit was stored in plastic picking lugs at 3°C and then transported to the Vinification and Brewing Laboratory on 27 Oct 2014, where it was immediately crushed and basket-pressed (Mori press type PZ.82, Impianti Enoligici) at 200 psi. After pressing, a sulfur dioxide (SO2) addition of 50 mg/L was made with potassium metabisulfite (K2S2O5). The juice was settled overnight at ambient conditions, then racked and distributed into 11.4 L carboys for fermentation.
N supplementation
Once the juice was fractioned into individual fermentation vessels, it was supplemented with DAP to create treatments with 180, 250, and 300 mg N/L of total YAN. All fermentations were performed in duplicate.
Yeast selection
Each YAN treatment and the control juice were fermented with three S. cerevisiae yeast strains: EC1118 and W15 (Lallemand) and Côte des Blancs (Red Star). These strains were selected because they are popular for Riesling production in the Finger Lakes and have different manufacturer-identified N demands.
Fermentations and sampling
Fermentations were carried out in 11.4 L carboys, each receiving 10.5 kg juice. Yeast was rehydrated as per the manufacturer’s protocol; in short, GoFerm (Scott Laboratories) was dissolved in 40°C deionized water and added at 0.3 g/kg of juice, contributing 10 mg N/L to the juice YAN. Subsequently, yeast was added at a rate of 0.3 g/kg of juice. After inoculation, each carboy was topped with a three-piece airlock with floating bubbler (Buon 129 Vino Manufacturing). Fermentations took place in a controlled-temperature room held at 18°C. Every 24 hr, duplicate 2 mL aliquots were taken from each fermenter with 5 mL disposable sterile pipettes (Celltreat). The samples were centrifuged at 12,000 rpm for 5 min, and the supernatant was removed and stored at −15°C until analysis. After all fermenters had stopped producing CO2, Clinitest tablets (Bayer Inc) were used to estimate the reducing sugar remaining in the wine. At completion, wines were racked from the lees into 7.5 L (2 gal) carboys, and 60 mg/L SO2 was added. The wines were then moved to a 2°C room for cold stabilization. Once the wines were cold stable, they were racked off the tartrate crystals and manually bottled and screwcapped in 750 mL bottles (Saint-Gobain Packaging). Samples (2 mL) for residual sugar and YAN were taken at the end of fermentation. After cold stabilization, samples were taken for pH, titratable acidity (TA), and organic acids, and at bottling, samples were collected for ethanol analysis.
Analytical methods
The concentrations of glucose and fructose, and of tartaric, malic, citric, and acetic acids were measured by high-performance liquid chromatography (HPLC), as described previously (Castellari et al. 2000), on an Agilent 1100 series HPLC (Agilent Technologies). YAN was determined through the separate enzymatic analysis of AMM and PAN, using a Chemwell 2910 Multianalyzer (Unitech Scientific) to rapidly test samples. AMM concentration was quantified using the glutamate dehydrogenase enzymatic test (Ough 1969). Reagents for this test were supplied as enzymatic kits (Unitech Scientific, Ammonia Extended Range UniTAB, 2007). PAN was determined by the o-phthalaldehyde/N-acetyl-l-cysteine spectrophotometric assay (NOPA) procedure (Dukes and Butzke 1998) (Unitech Scientific, Primary Amino Nitrogen UniTAB, 2007).
Wine volatiles were analyzed by gas chromatography-mass spectrometry (GC-MS). TDN, linalool, and several higher alcohols and esters were isolated from the wine using a solid-phase extraction (SPE) protocol referenced in previous studies (Lopez et al. 2002, Sacks et al. 2012). To 50 mL of sample, 25 μL of internal standard (2-octanol, 0.5 g/L in acetonitrile) was added. Samples were loaded onto SPE cartridges (Merck) containing 200 mg of LiChrolut EN sorbent preconditioned with 4 mL of dichloromethane, 4 mL of methanol, and 4 mL of model wine (consisting of 12% [v/v] ethanol and 5 g/L tartaric acid, adjusted to pH 3.5 using NaOH). Cartridges were placed in a 12-piece manifold (J.T. Baker) to facilitate elution, and dried under N2 for 20 min before analytes were eluted with 1.3 mL dichloromethane.
GC-MS analyses were conducted on an Agilent 6890 gas chromatograph with a split-splitless injector coupled to an Agilent HP 5973 mass-selective detector. Separation was performed using an Agilent DB-5MS column (30 m × 250 mm i.d. × 0.25 μm). The initial oven temperature was 35°C and held for 3 min, then ramped to 200°C at 6°C/min, then to 240°C at 30°C/min, and held at 240°C for 3 min. The GC was operated at a constant flow rate of 1 mL/min. One μL of extract was injected splitless. The injector temperature was 250°C and had a purge time of 1.00 min (purge flow 50 mL/min, inlet pressure 68.9 kPa). The auxiliary channel, set point quadruple, and set point source were 280, 150, and 230, respectively. Helium was used as the carrier gas. Data processing and quantification were performed with Agilent Enhanced MSD ChemStation software (G170EA E.02.00.493).
Identification of linalool was confirmed by comparing the retention time and mass spectra to the authentic standard. For the nine additional compounds analyzed, identification was made through comparison of the Kovats retention index and mass spectra with those of commercial standards. After retention times and mass spectra were determined, compounds were detected by using a selected ion-scan mode to increase sensitivity (Table 1).
Linalool was fully quantified by preparing a calibration curve (n = 4) in duplicate, and in model wine with a concentration range of 8 to 220 mg/L and an r2 of 0.9096. In the absence of a pure TDN standard, the calibration curve was created with naphthalene (n = 4), with a range of 1 to 170 mg/L, and an r2 of 0.9922. The TDN concentration was then determined in naphthalene equivalents. Naphthalene is commonly used as an internal standard for TDN determinations via GC-MS because of the similarity in the chemical structures of the two compounds (Daniel et al. 2009).
A semiquantitative approach was used for esters, higher alcohols, and organic acids. For each treatment, relative response factors (RRFs), defined as the ratio between the peak areas of the analyte and internal standard, were compared with the RRFs of the respective control.
Sensory evaluation
Wines were evaluated for preference by a sensory panel consisting of 31 participants (12 women and 19 men), age 21 to 70 yrs (mean 43.6 yrs). All were healthy members of the local community who consumed white wine at least once a month. Testing was performed in the sensory booths of the Cornell Food Research Lab, in Geneva, NY. The experimental procedure was reviewed and approved by the Cornell University Institutional Review Board. All participants provided written consent and were compensated for their participation.
The panelists executed a preference-ranking test (Lawless and Heymann 2010). They received a flight of four wines containing the control and three supplemented treatments for a given yeast strain, and were asked to rank them according to their preference (1 = most preferred and 4 = least preferred). In each session, three independent flights—one for each yeast strain tested—were evaluated, for a total of 12 wines per session. The duplicate fermentations were prescreened by chemical and sensory evaluation by nine panelists in the lab, and were considered to be the same sample for the purposes of sensory evaluation. In each flight, the order of samples was randomized for each participant. In the prescreening, wines fermented by EC1118 and W15 had low sugar-to-acid ratios, which gave them an excessively acidic taste. Since the focus of the study was to investigate differences in wine aroma, sugar was adjusted by adding fructose, so that the four wines in a flight had the same sugar-to-acid ratio (Table 2).
Statistical analysis
Statistical analysis of wine composition parameters was performed with JMP version 11.2 (SAS Institute). For wine nonvolatile composition, separate one-way analysis of variance (ANOVA) tests were conducted for each yeast strain, using YAN level as fixed effect. When significant differences were encountered, means were compared with the Tukey-Kramer honest significant difference test. We analyzed wine volatile compounds with two-way ANOVAs, with YAN level and yeast strain as fixed effects. For the eight fermentation products, linear regressions of RRFs versus YAN level were conducted for each yeast strain with least-square methods. Statistical analysis of sensory data was performed by using tables to identify critical values of differences among rank sums (Basker 1988).
Results
Juice chemistry
Juice TA as tartaric acid equivalents was 7.98 ± 0.25 g/L, and the pH was 3.14 ± 0.01. The YAN of the initial juice was 122 ± 1 mg N/L, and the unsupplemented control had a YAN concentration of 130 mg N/L after the addition of the rehydration nutrient described above.
Fermentation rates and YAN concentration
N supplementation distinctly affected fermentation kinetics of each yeast strain (Figure 1). EC1118 could ferment wines to dryness in all cases, even at the control concentration of 130 mg N/L. This strain was also the most responsive to DAP supplementation, decreasing fermentation time from 12 to seven days at the highest N levels. However, the fermentation rate did not increase when the initial N was increased from 250 to 300 mg N/L.
Côte des Blancs responded the least to the supplementation: fermentation time decreased only by one day (from 15 to 14 days) as N concentration was increased from 130 to 180 mg N/L, and the 250 and 300 mg N/L treatments both completed fermentation at 14 days (Figure 1). Residual sugar levels decreased slightly as YAN increased: at 130 and 180 mg N/L, residual sugars averaged 2.23 g/L, whereas at 250 and 300 mg N/L, the mean residual sugar concentration was 1.20 g/L. For W15, fermentation rates did not increase with N supplementation, and all wines had residual sugar levels >2.0 g/L.
The consumption patterns of each YAN component also varied by yeast strain (Figures 2 and 3). EC1118 exhausted all AMM present in the media within four days or less. On day 4, PAN was almost entirely consumed—from 93% to 99%, depending on the initial YAN level. After this point, all EC1118 fermentations exhibited a rise in PAN, ranging from 11% to 21% of the initial PAN concentration. Likewise, W15 fermentations showed quick N uptake, with PAN being consumed slightly faster than AMM, and both components were exhausted at day 4. The increase in amino acid concentration at the later fermentation stages also followed a similar pattern, except for the 300 mg N/L treatment, which yielded a slightly larger increase. In that case, the wine showed a final YAN of 31 mg/L, and PAN at 34% of the initial concentration. In contrast, in Côte des Blancs fermentations, AMM was consumed entirely only in the control and 180 mg N/L treatments. PAN was exhausted from the media only in the control fermentations, but later increased to 19% of its initial concentration. This apparently low N uptake caused higher levels of residual YAN in the Côte des Blancs wines. Whereas the control wines had a final YAN concentration of 17 mg N/L, YAN concentration increased to 66 and 105 mg N/L in the 250 and 300 mg N/L treatments, respectively.
Final wine composition
EC1118
Fermentation rates for EC1118 were affected by initial N concentration, but all of these wines finished with less than 0.5 g/L residual sugar (Table 3). Ethanol, pH, and malic acid concentrations were unaffected by initial YAN (Tables 3, 4, and 5). TA and lactic acid concentration increased with increasing YAN, while acetic acid concentration decreased.
W15
Residual sugar concentration varied among the W15 treatments, although no clear trend was apparent (Table 2). Ethanol production was not affected by YAN. The pH decreased and TA increased with higher YAN supplementation levels, and concentrations of organic acids were affected as well; final malic acid and acetic acid concentrations decreased with increased YAN supplementation, whereas the production of lactic acid increased (Tables 3, 4, and 5).
Côte des Blancs
The 250 and 300 mg/L Côte des Blancs treatments were notably lower in residual sugar than the control and 180 mg N/L treatment; ethanol concentration, however, was not affected (Table 2). At higher initial YAN concentrations, the pH increased, as did lactic and acetic acids, but TA and malic acid concentration were not significantly affected by YAN levels (Tables 3, 4, and 5).
In summary, although initial YAN concentrations affected residual sugar for two of the three yeast strains studied, ethanol concentration was not affected in any case. There was a trend of increasing TA as the DAP supplementation increased, and this effect was significant for two of the yeast strains. For all three strains, lactic acid production increased with higher initial YAN concentrations. Production of acetic acid was also affected, although the initial YAN concentration that favored maximum production varied with each yeast strain.
Concentration of volatile compounds
Eight of the 10 volatile compounds analyzed by GC-MS were affected by initial N concentration for at least one of the yeast strains (Table 6). In the two-way ANOVA, a significant interaction was found for YAN level × yeast strain for seven of these compounds, suggesting that their synthesis depends on both the yeast strain and initial YAN level of the juice. In general, ester concentration increased with higher N supplementation (Figures 4 and 5). One compound to note is 3-methyl butyl acetate, which showed increased concentrations at higher N levels across all yeast strains. Ethyl hexanoate was only significantly affected for W15, in which concentrations increased with more DAP supplementation. Ethyl octanoate production was impacted in EC1118 and Côte des Blancs strains, with supplementation treatments having higher ester concentrations than the control. Impacts on ethyl propanoate production varied with yeast strain; for EC1118, the fermentations with 300 mg/L initial YAN gave a significantly higher ethyl propanoate concentration than the other YAN treatments. For Côte des Blancs, the 250 mg/L initial YAN was the concentration that favored maximum production, and no change was observed in W15 fermentations. Ethyl cinnamate exhibited the opposite trend, decreasing in concentration at higher N levels in the W15 and Côte des Blancs fermentations.
Of the two higher alcohols analyzed in this study, 2-phenylethanol consistently decreased with DAP addition. In contrast, 1-hexanol showed a decrease in concentration only in EC1118 fermentations. For the wines fermented with EC1118, all supplemented treatments had higher decanoic acid concentrations than the unsupplemented control. Wines fermented with Côte des Blancs had an increase in decanoic acid concentration with 180 mg/L initial YAN level, but the other treatments and the control did not differ.
Linear regression models were run for the RRFs of the volatile fermentation products as a function of initial YAN concentration, and in some cases, the model accurately represented the relationships among the data (Tables 7 and 8). For instance, 2-phenyl ethanol displayed different rates of decrease for Côte des Blancs, but exhibited fairly similar rates for EC1118 and W15, as can be seen by comparing the slopes and confidence intervals in Figure 4. For 3-methyl butyl acetate, the rates of increase with increasing YAN were different for each yeast strain.
The average TDN concentration in the final wines was 2.7 mg/L, and yeast strain or initial YAN level did not affect TDN levels (Table 9). According to the two-way ANOVA, linalool concentrations in the final wines did not depend on the initial YAN level of the juice (p = 0.5813), but were dependent on yeast strain (p = 0.0052). Côte des Blancs wines had a final linalool concentration of 36.9 ± 3.0 mg/L, higher than that of the wines fermented with EC1118 and W15 at 33.9 ± 1.5 mg/L and 33.2 ± SD 1.1 mg/L, respectively.
Sensory evaluation
A difference in preference was found only in wines fermented with EC1118 (Table 10). The control wine was most preferred overall, followed by the 180 and 250 mg/L YAN treatments, which were ranked equally. The 300 mg/L treatment was least preferred.
Discussion
N consumption
YAN consumption was directly proportional to initial YAN concentration and strongly dependent on yeast strain, as observed in previous studies (Vilanova et al. 2007, Ugliano et al. 2011). For the control fermentations, N consumption ranged from 113 mg N/L (Côte des Blancs) to 124 mg N/L (W15). As the initial YAN increased, all three strains increased consumption; EC1118 had the highest uptake, consuming 280 mg N/L in the 300 mg N/L supplementation treatment, whereas Côte des Blancs had the lowest consumption, and its N uptake peaked at 195 mg/L in the highest supplementation treatment. In most of the fermentations, AMM was observed to be completely exhausted from the media within a few days, in some cases being consumed faster than total PAN (Figures 2 and 3). Ammonium has been proposed to be one of the latest N sources to be assimilated during fermentation, after amino acids such as lysine and asparagine (Crépin et al. 2012). Since we did not measure individual amino acids, we could not confirm if that order of consumption was maintained in our experiment. On the other hand, high-biomass-producing strains such as EC1118 have higher rates of AMM uptake, and higher biomass is linked to faster fermentation rates (Varela et al. 2004, Crépin et al. 2014). This is in line with our findings that the fermentations with the highest AMM consumptions were also the fastest.
In contrast, PAN was either exhausted or reached a minimum (between 3.5 and 32.5 mg N/L) before the trend reversed and concentrations increased to reach a fraction of the initial value by the end of the fermentation (Figure 3). S. cerevisiae has the ability to quickly take up amino acids, storing them in vacuoles as glutamate and further redistributing them for the de novo synthesis of amino acids as required (Crépin et al. 2017). In the late fermentation stages, this intracellular amino acid pool can be released into the fermentation matrix, as higher ethanol concentrations promote increased membrane permeability (Salgueiro et al. 1988). As the yeast population enters the death phase, amino acids continue to be liberated into the media by passive diffusion from the cytoplasm through the first stage of yeast autolysis (Alexandre et al. 2001).
Minimum YAN concentration
YAN supplementation impacted fermentation kinetics of EC1118 the most, reducing fermentation time by five days within the supplementation range studied (130 to 300 mg N/L). W15 fermentation times did not change, and Côte des Blancs supplementation treatments fermented one day faster than did the control. Although there were differences in residual sugar for W15 and Côte des Blancs, ethanol concentration was not affected by YAN for any of the yeast strains (Table 2). This suggests that the YAN concentration of the control juice (130 mg N/L) was sufficient to complete fermentation for the three yeast strains studied. This finding is in line with the accepted literature, which often considers 140 mg/L as the minimum YAN required to complete alcoholic fermentation in a 20 Brix must (Butzke 1998, Bell and Henschke 2005). Since 20 Brix is a typical level of harvest soluble solids in cool-climate regions (Nisbet et al. 2014), the recommendation of 140 mg N/L could be generally applied. For musts with higher sugar levels, Bisson and Butzke (2000) recommended an additional 25 mg N/L of YAN for every 1 Brix increase. However, it has been shown that fermentation completion did not depend on this additional YAN supplementation, even for musts of higher sugar levels (Bely et al. 2003). The addition of N could still have benefits for wine quality, however, by changing volatile composition.
Fermentation performance of the different yeast strains
Bisson and Butkze (2000) note that there are basically two types of yeast fermentation profiles. Strains with a Saccharomyces bayanus background, such as EC1118, are typically strong fermenters displaying short lag phases and high fermentation rates. In contrast, Côte des Blancs and related yeasts are characterized as slow fermenters. Several traits influence a yeast’s fermentation performance, and N metabolism seems to be one of the most important (Treu et al. 2014). EC1118 has high expression of genes regulating N uptake, and this high expression is likely responsible for its high fermentation speed (Treu et al. 2014). This is consistent with the observations in the current study, in which EC1118 had the shortest fermentation times and the highest N consumption. Nevertheless, it could ferment all wines to dryness, even the unsupplemented control.
Although no studies exist that link gene expression to N uptake in W15, its high N consumption agrees with results from previous literature reports showing that this strain has a high N demand (Vilanova et al. 2007). In contrast, a previous experiment that used a synthetic grape juice medium (Vilanova et al. 2007) yielded no gain in fermentation speed with N supplementation. In a study conducted with common commercial strains, Côte des Blancs was both a slow fermenter and was notably the least active in terms of amino acid uptake (Ough et al. 1991). This is similar to the findings in the present work, as Côte des Blancs fermentations had the highest levels of residual YAN.
Changes in nonvolatile composition of wines
DAP additions affect the balance of organic acids and the final amino acid concentration in wines, often leading to a higher TA and lower pH (Vilanova et al. 2007, Torrea et al. 2011, Vilanova et al. 2012). In this study, TA tended to increase with higher DAP additions, but the pH was only marginally affected. The production of acetic acid decreased between control and EC1118 and W15 treatment fermentations, but increased for Côte des Blancs (Table 5). It has been suggested that one of the drivers of acetate production during alcoholic fermentation is the regeneration of nicotinamide adenine dinucleotide (NADH) to maintain cell redox balance in a high osmolar medium (Bely et al. 2003). Increased N availability increases NADH production, which, in turn, reduces the need for acetate production. An excess of N compounds, however, can increase the production of volatile acidity, as seen with Côte des Blancs. This effect has been observed in other studies, but the mechanisms behind it are still unclear (Bely et al. 2003, Hernandez-Orte et al. 2006, Vilanova et al. 2007).
Production of volatile compounds
Although 140 mg N/L is generally recognized as the minimum YAN concentration for fermentation completion, several studies suggest that supplementation to final levels as high as 260 to 500 mg N/L can positively affect wine aroma by increasing ester production and decreasing production of higher alcohols and hydrogen sulfide (Miller et al. 2007, Torrea et al. 2011, Ugliano et al. 2011, Vilanova et al. 2012).
Of the two higher alcohols quantified, 2-phenylethanol decreased with higher DAP in all yeast strains, and 1-hexanol decreased in EC1118, but concentrations were generally lower (although not significantly so) across yeast strains (Figure 4). Higher alcohols can be formed in fermentation by de-carboxylation of α-keto acids derived either by transamination of amino acids via the Ehrlich pathway or through sugar catabolism (Hirst and Richter 2016, Crépin et al. 2017). It has been suggested that regulation varies between these two pathways according to the pool of available amino acids and total YAN concentration of the growth medium, and there is usually a range of YAN in which higher alcohol production is maximal (Mendes-Ferreira et al. 2009, Fairbain et al. 2017). Additionally, higher alcohols are precursors in the production of acetate esters, and several studies have pointed out that the corresponding ester production is positively correlated with assimilable N concentration (Torrea et al. 2011, Mouret et al. 2014, Fairbain et al. 2017). This was the case for 3-methyl butyl acetate in our study, which was directly proportional to YAN for all three yeast strains.
Ethyl esters are produced via esterification of medium-chain fatty acids and ethanol, and are activated by coenzyme A; the way by which N impacts their production remains yet to be elucidated. It is hypothesized that this interaction could be due to the synthesis of acyl-CoA derivatives from α-keto acids. Another hypothesis would be that the increased biomass production in higher-N media increases lipid requirements, thereby inducing more acyl-CoA synthesis. This was confirmed in the present study, where most of the ethyl esters measured were present in higher concentrations in the supplemented fermentations.
Besides quantifying fermentation aromas, we also investigated the concentration of two of the key varietal aromas in Riesling: the monoterpene linalool and the norisoprenoid TDN (Meyers et al. 2013). N could affect the final monoterpene concentration of wines by two different means. The first is by changing the activity of must glycosidases, thereby increasing the hydrolysis of grape precursors (Bell and Henschke 2005). The second, and most likely, is by influencing the de novo synthesis of monoterpenes through a defect in the sterol pathway (Swiegers et al. 2005). In this study, no influence of N supplementation on the final linalool concentration was observed. Nevertheless, the maximum YAN concentration used in our study was 300 mg N/L, while in previous studies reporting an increase in terpene production, supplementation reached levels as high as 450 mg N/L (Carrau et al. 2005, Vilanova et al. 2012). Glycosidase activity can likewise affect the production of TDN precursors, but no differences were found in TDN concentrations among the fermentation treatments. Formation of TDN in wines involves multiple transformations after the initial hydrolysis; thus, TDN tends to appear during bottle storage. Since GC analysis was performed only three months after bottling, it is possible that TDN concentrations had not fully developed.
Sensory evaluation
All wines exhibited differences in volatile and nonvolatile composition following DAP additions, but sensory differences among the treatments were detected only in the wines fermented by EC1118. In that case, the control wine was most preferred by the taste panel, followed by the 180 and 250 mg N/L treatments, and the 300 mg N/L treatment was least preferred. This preference ranking may be due to the increase in acetate esters (Figure 4), which might confer nail polish or bruised-apple aromas, as observed previously (Ugliano et al. 2010, Torrea et al. 2011). Other changes in volatile composition that might have contributed to the lower preference for supplemented wines include the measured decrease in concentration of the alcohol 2-phenylethanol, which has a flowery and rose-like odor. Ethyl cinnamate, which possesses a fruity odor, also decreased in concentration in the supplementation treatments. Additionally, production of decanoic acid, which has a rancid and unpleasant odor, increased with DAP addition. Since there was also no increase in the concentration of varietal aromas such as linalool and TDN, these data suggest no evident benefit of DAP supplementation for the aroma expression of Riesling wine.
Conclusion
This study found that a YAN concentration of 130 mg N/L is enough for the three yeast strains used here to complete fermentation of a 20 Brix must, but DAP supplementation up to 250 mg N/L improved the fermentation kinetics of EC1118. Fermentation speeds of Côte des Blancs and W15 did not increase with the DAP supplementation. YAN consumption was strain-dependent, and increased with N supplementation, ranging from 113 to 281 mg N/L. Côte des Blancs had the lowest YAN consumption, and the 300 mg N/L treatment completed with 105 mg N/L residual YAN. This observation suggests that 300 g N/L is an excessive initial YAN concentration for Côte des Blancs.
YAN concentrations ranging from 130 to 300 mg/L and obtained with DAP additions yielded significant effects on wine volatile and nonvolatile composition. Although relative concentrations of most esters analyzed increased in the supplementation treatments, sensory panelists either showed no clear preference among treatments (i.e., for W15 and Côte des Blancs) or preferred the unsupplemented wine (for EC1118). This finding conflicts with those from previous literature reports suggesting that YAN concentrations higher than 140 mg N/L improve volatile composition and wine quality. Two important varietal compounds for Riesling (TDN and linalool) were quantified, and their concentrations were found to be unrelated to YAN concentration. These results suggest that supplementing musts to above 250 mg N/L is not recommended, given the moderate sugar levels common to Finger Lakes Riesling. Winemakers must avoid overaddition of ammonium N, which might generate microbial instability and a decrease in desired sensory properties.
Acknowledgments
This project was supported by NYSAES Federal Formula Funds and by the New York Wine and Grape Foundation. The authors thank Dr. David C. Manns for his technical support with analytical methods.
Footnotes
By downloading and/or receiving this article, you agree to the Disclaimer of Warranties and Liability. The full statement of the Disclaimers is available at http://www.ajevonline.org/content/proprietary-rights-notice-ajev-online. If you do not agree to the Disclaimers, do not download and/or accept this article.
- Received September 2017.
- Revision received April 2018.
- Revision received September 2018.
- Accepted October 2018.
- Published online April 2019
- ©2019 by the American Society for Enology and Viticulture