Abstract
Chardonnay Musqué vines in Ontario were subjected to hedging (control), basal leaf removal (BLR), or cluster thinning (CT) over a 3-year period. Musts from each treatment were combined in a factorial treatment arrangement with yeast strains VL1, EC1118, and EC1118 + enzyme preparation AR2000. Yield was reduced by 21 to 36% by CT as a result of cluster removal. Fruit maturity was advanced by BLR and CT in reduced berry, must, and wine titratable acidity (TA); increased berry and must pH and Brix; increased must and wine free volatile terpenes (FVT) (CT only); and increased berry, must, and wine potentially volatile terpenes (PVT). VL1 wines were lower in pH and both FVT and PVT (1 of 3 years) than those of EC1118 and had higher TA (1 of 3 years). EC1118 + AR2000 wines did not differ from wines produced solely with EC1118 in terms of TA and FVT but were higher in PVT and pH. CT wines differed from BLR and control wines sensorially, with increased depth of color, higher sweetness and herbaceous/grassy aroma, and reduced tropical fruit aroma. BLR wines had higher tropical fruit and floral aromas, while control wines had greater acidity. Principal component analysis of sensory data suggested that variation in wine sensory profiles was due to both viticultural and enological practices.
One of the universal benchmarks of wine quality lies in the expression of varietal character of grapes used in making the wine. Recent viticultural and enological research has therefore sought ways to maximize varietal character in the vineyard and the winery and to quantify these effects analytically and sensorially.
An important component of the typicity of “aromatic” grape cultivars is the monoterpene aroma compounds, found in varying concentrations in many fruits including grapes. Muscat-flavored cultivars develop relatively high concentrations of these compounds, including geraniol, nerol, and linalool (Gunata et al.1985). Aromatic cultivars such as Riesling, Gewürztraminer, and related cultivars also exhibit floral, perfumy, fruity aromas because of elevated monoterpene concentrations in contrast to a cultivar such as Chardonnay, which has relatively low concentrations (Dimitriadis and Williams 1984). However, a group of Chardonnay clones, called Chardonnay Musqué (clones 77 and 809; MAPA 1994) and popular in the Niagara region of Ontario, is known for floral, muscat-like aromas that distinguish them from other more neutral clones of Chardonnay (Versini et al. 1990). The accumulation of monoterpenes during fruit maturation is likely crucial to the identity and typicity of Chardonnay Musqué. Viticultural or enological treatments that might increase monoterpene concentration of Chardonnay Musqué berries, musts, and wines are therefore of critical interest to the winemaker seeking to maximize muscat character.
Cluster thinning and basal leaf removal are two viticultural practices commonly used to improve fruit composition. Previous work has shown that these practices can be used to alter basic berry composition, particularly maturity indices such as Brix, pH, and titratable acidity (Ollat and Gaudillère 1998, Reynolds et al. 1995, 1996). Monoterpenes and/or glucoconjugates have been observed to be responsive to increases in fruit exposure (Macaulay and Morris 1993, Reynolds and Wardle 1989b) and to leaf removal and cluster thinning (Reynolds and Wardle 1989a, Reynolds et al. 1994a, 1995, 1996, Zoecklein et al. 1998). Free monoterpenes, which contribute directly to wine aroma and flavor, are less responsive to viticultural practices than the glycosylated (bound) forms, which are odorless until hydrolyzed either chemically or enzymatically (Dimitriadis and Williams 1984). Glycosidically-bound terpenes can therefore be considered as a pool of potential aroma and flavor. Changes in monoterpene concentration in berries affected by leaf removal or cluster thinning have been substantial enough to evoke changes in sensory response (Reynolds et al. 1994b, 1995, 1996).
Winery practices such as yeast strain selection and/or enzyme use may also increase the varietal intensity of muscat cultivars. One common target is the β linkage between monoterpenes and their accompanying sugar molecules. Wine yeasts express β-glucosidases, but endogenous yeast glucosidases are generally inactive and/or unstable at the pH of wine and must (Delcroix et al. 1994). Often mentioned are the superior β-glucosidase activities of specific yeast strains (e.g., VL1) that allow them to hydrolyze glycosidically-bound terpenes to their free forms (Laffort et al. 1989). However, the effect of these yeasts on free monoterpene concentrations in wines has been found to be insignificant (Delcroix et al. 1994). Recent bio-technological advances have led to commercial enzyme preparations, usually in the form of pectinases with β-glucosidase “side activity.” Research on the effectiveness of such enzyme additives in freeing FVT from glycosidic precursors has therefore met with both negative (Decroix et al. 1994) and positive (Macaulay and Morris 1993) results. Therefore, impact of yeast strains and enzymes on varietal typicity of wine remains a controversial topic. The contribution of different yeast strains on wine composition and sensory attributes has nonetheless been documented (Antonelli et al. 1999, Reynolds et al. 2001), as has the performance of commercial enzyme preparations that are also marketed as having the ability to increase volatile characteristics of wine through β-glycosidase side activity (Grossmann and Rapp 1988).
A significant issue is the relative magnitude of effects of common viticultural (e.g., cluster thinning and basal leaf removal) and enological practices (e.g., yeast strain selection and/or enzyme use). Although other studies have examined leaf removal and cluster thinning at varying degrees of severity and/or timing (Reynolds and Wardle 1989a, Reynolds et al. 1994a,b, 1995, 1996), this study specifically examined viticultural (leaf removal and cluster thinning) and enological (yeast strain and enzyme use) treatments on the berry and must composition and sensory attributes of Chardonnay Musqué.
Materials and Methods
Vineyard site and experimental design.
The trial was carried out in 1999 to 2002 at Cave Spring Vineyards, Beamsville, Ontario. Ten-year-old Chardonnay Musqué vines on 3309 rootstock, spaced 1.0 x 2.4 m (vine x row) in north:south-oriented rows and trained to a Guyot system (head-trained, cane-pruned) were chosen for study. Vines were typically pruned to 20 nodes per vine (two 8-node canes plus two renewal spurs), and cane prunings were retained and weighed annually to determine vine size. The experiment consisted of six rows (blocks) of Chardonnay Musqué, each consisting of 72 vines. In each row, three 24-vine subsets were randomly assigned as a control or to receive either a cluster thinning (CT) or basal leaf removal (BLR) treatment at veraison. The design was a randomized block with six blocks and 24-vine treatment replicates.
Vineyard treatments.
Cluster thinning and BLR were performed on or just before veraison on 1 Aug 1999, 13 Aug 2001, and 15 Aug 2002. Treatments were not imposed in 2000 because of a severe powdery mildew outbreak throughout the vineyard. Cluster thinning involved the removal of all distal clusters on those shoots that produced more than one cluster. Those vines designated as the BLR treatment had a section of leaves removed around the fruiting zone ~20 cm in width on the east side of the canopy.
Berry sampling and harvest.
Harvest was determined at the discretion of the winery. At harvest, cluster number per vine and yield per vine were recorded using a SB23000 top-loading scale (Mettler Toledo Corp., Columbus, OH). Mean cluster weight was calculated from these data. Crop load (Bravdo et al. 1985) was calculated from the yield:vine size ratio. Fruit from common treatment replicates in blocks 1 and 2, 3 and 4, and 5 and 6 were thereafter combined to provide material for three fermentation replicates. Fruit was stored at 2°C until winemaking began less than 24 hr later.
Four 100- and four 300-berry samples were collected from each treatment replicate each season immediately before harvest, which occurred on: 12 Sept 1999, 20 Sept 2001, and 19 Sept 2002. The 100-berry samples were eventually used to determine berry weight, Brix, titratable acidity (TA), and pH. Berries per cluster were calculated from cluster weight and berry weight data. The 300-berry samples were used to determine the concentration of free (FVT) and potentially volatile terpenes (PVT). All berry samples were stored at −25°C before analysis.
Winemaking treatments.
Grapes were destemmed and crushed, treated with 50 mg/L sulfur dioxide, and given 24-hr pomace contact in 20-L food-grade plastic pails at 2°C before pressing, which was performed in an Idropress basket/bladder press (Enoagricola Rossi, Calzolaro, Italy) at 2.0 bars pressure. Immediately after pressing, two 250-mL must samples were taken from each treatment replicate for subsequent analysis of Brix, pH, TA, FVT, and PVT. Musts from each treatment replicate were subsequently divided into three 20-L aliquots, and each was designated to be fermented with Saccharomyces bayanus EC1118 ± AR2000 or S. cerevisiae VL1. Inoculation rate was 0.25 g/L. All fermentations were carried out in 20-L glass carboys at 16°C and were allowed to proceed until residual sugar levels were between 2.5 and 5.0 g/L as determined using Clinitest tablets (Bayer, Etobicoke, ON). With the exception of wines designated as EC1118 + AR2000, all fermentations were stopped at this time by addition of 75 mg/L SO2 and a decrease in temperature to −2°C. Wines designated as EC1118 + AR2000 were warmed to 10°C, after which the enzyme preparation AR2000 (DSM-Gist-Brocades, Servian, France) was added according to manufacturer’s directions (3 g/hL for 5 weeks). After 5 weeks, wines were relocated to a 20°C facility, and bentonite was applied to all fermentations at a rate of 5 g/hL, after which all wines were racked, cold stabilized, and subsequently reracked. Wines were filtered with 0.5-μ pad filters (Scott Laboratories, Pickering, ON) in series with a 0.22-μ hydrophilic cartridge filter (Millipore, Bedford, MA), and immediately bottled. All wines were stored at 11°C until a sensory analysis 7 to 9 months later.
Berry and must composition.
Berry and must analyses were consistent with those described by Zoecklein et al. (1995). Before analysis, 100-berry and/or must samples were removed from −25°C, placed in a 4°C cooler overnight to allow for thawing, heated at 80°C for 1 hr in a Isotemp 228 water bath (Fisher Scientific, Toronto, ON) to dissolve precipitated tartrates, and subsequently cooled to room temperature. Berry samples were thereafter juiced using an Omega model 500 fruit juicer, allowed to settle, and solids were removed by vacuum aspiration. Brix was determined using a temperature-corrected benchtop Abbé refractometer (model 10450; American Optical Corp., Buffalo, NY). Berry/must TA was determined by titration of a 25-mL sample using a PC titrate autotitrator (Man-Tech Associates Ltd., Guelph, ON) using 0.1 N NaOH. Berry/must pH was obtained using an Accumet model 25 pH/ion meter (Denver Instruments, Denver, CO).
Monoterpene analysis was performed on the berry and must samples from all three vintages using a previous method (Dimitriadis and Williams 1984) as modified by Reynolds and Wardle (1989a). Berry/must samples were removed from −25°C, allowed to thaw overnight at 4°C, and then homogenized in a commercial laboratory blender for 20 sec. Two 100 g berry/must aliquots per sample were adjusted to ~pH 6.7 using 20% NaOH. Samples were subsequently steam distilled (Lurex, Vineland, NJ). The FVT fraction was collected in 10 to 15 min in a 25-mL volumetric flask. The sample was then adjusted to pH 2.0 using 10 mL of 50% phosphoric acid, and the PVT fraction was thereafter collected within 15 to 20 min in a 50-mL volumetric flask. A 10-mL aliquot of each FVT and PVT distillate was transferred to a test tube and reacted with 5 mL of 2% (w/v) vanillin in concentrated sulfuric acid. The test tubes were placed in a 60°C water bath for 20 min to allow for color development, and subsequently allowed to cool at room temperature for 5 min. The absorbance of each sample was measured at 608 nm on a Pharmacia Biotec Ultrospec UV/VIS spectrophotometer (model 1000E; Biochrom Ltd., Cambridge, UK). The terpene concentration of each sample was determined from a calibration curve consisting of linalool standards (0 to 10 mg/L) that had undergone the same colorimetric reaction.
Wine composition.
Wine samples were taken from bottled samples of each treatment replicate at least one year postbottling and analyzed for ethanol, TA, pH, FVT, and PVT. Wine TA and pH were determined in the same manner as for the must. Ethanol concentration was determined using an Agilent series 6890 gas chromatograph (Hewlett-Packard, Mississauga, ON). Distillation of wine samples for FVT and PVT analysis was done in the same manner as the musts. However, because of interference of ethanol on FVT measurements (Dimitriadis and Williams 1984), the 25 mL FVT distillate fraction was diluted with 50 mL of distilled water and passed through a Sep-Pak Plus C18 extraction cartridge (Waters Corp., Milford, MA) (previously conditioned using 5-mL washes of distilled water, methanol, and distilled water) at no more than 5 p.s.i. The column was subsequently washed with 5 mL distilled water, and the aqueous phase was discarded. The FVT fraction was thereafter eluted from the column with 5 mL 100% methanol, and the eluent was subsequently brought up to volume in a 25-mL volumetric flask using distilled water. A 10-mL aliquot was taken for colorimetric analysis as previously described. The calibration standards used for wine FVT analysis were prepared in 20% (v/v) methanol to account for any interference caused by this compound. A series of linalool standard solutions (1.0, 1.5, and 3.5 mg/L), each prepared in 0, 10, 20, 30, 40, and 50% (v/v) ethanol, was likewise treated, and recovery of the standards was very close to 100% (Schlosser 2003; data not shown).
Free SO2 interferes with PVT measurements because of a bleaching effect on the colorimetric reaction (Dimitriadis and Williams 1984). This interference, however, was not observed in linalool standards (1.0, 1.5, 3.5 mg/L) containing <100 mg/L free SO2 and was unchanged between 200 and 800 mg/L free SO2 (Schlosser 2003; data not shown). Since all PVT distillates from wine samples were well below 100 mg/L SO2, such interference was of little consequence. Accordingly, wine PVT concentrations were determined in the same manner as berry and must samples.
Sensory evaluation.
Judges participating in this study were primarily faculty, staff, and students from the Cool Climate Oenology and Viticulture Institute at Brock University. The panels for the evaluation of the 1999, 2001, and 2002 wines consisted of 14, 10, and 10 judges, respectively. Judge training was conducted before formal sensory analysis each year. The first training session used a subset of wines from the vintage intended for analysis for the generation of descriptive terms. Flavor and aroma standards for each of the generated descriptors were presented to the judges in subsequent training sessions and modified until a general consensus was obtained in terms of accuracy and intensity of each standard (Table 1⇓).
Descriptors and standards used for flavor profiling of Chardonnay Musqué wines, 1999, 2001, and 2002. Standards were added to 100 mL neutral white wine unless designated otherwise.
Because of the relatively large number of treatment combinations, a randomized incomplete block design was used for each of the formal sessions. Each session (representing a single set of fermentation replicates) consisted of two groups of six wines (treatments) presented in random order. Each judge evaluated each treatment replicate twice, in one morning and one afternoon session. Wines were rated for the intensity of each aroma and flavor attribute on a 100-mm unstructured scale labeled “low” and “high” at the beginning and end of the scale, respectively. A 30-mL sample of each wine was presented to each judge at room temperature in three-digit coded ISO glasses. Evaluation by judges during each session was performed within separate cubicles illuminated with red light to minimize color differences. Compusense Five software (Compusense, Guelph, ON) was used for all sensory data acquisition. Standards created during the training sessions were available to the judges before each formal session.
The color attribute was rated in a separate room under incandescent light on a 100-mm unstructured scale anchored with “pale yellow” and “gold” at the beginning and end of the scale, respectively. A 10-mL aliquot of each sample was presented in a 60-mm petri dish, on a white surface, and coded with a three-digit number. All nine treatment combinations from the 1999 and 2001 vintages were tasted. In 2002, because of similarities during bench testing among enological treatments within each field treatment, but apparent consistent differences among field treatments within each of the three enological treatments, only three of nine treatment combinations were chosen for sensory analysis (EC1118 treatments).
Statistical analysis.
Data analysis was performed with SAS software (SAS Institute, Inc., Cary, NC). General linear models procedure (PROC GLM) was used for all analysis of variance of field, composition, and sensory data to assess impact of viticultural and enological treatments on yield, berry, must, and wine composition, and wine sensory attributes. Treatment x block x vine was used as an error term for all growth and yield data, while treatment x block x postlength was used as an error term for berry samples. Interactions between viticultural and enological treatments on wine composition and sensory attributes were similarly tested by using treatment (field) x treatment (winery) x block as an estimate of error for compositional variables and treatment (field) x treatment (winery) x block x judge for sensory data. Principal component analysis (PCA) was performed on the correlation matrices from each sensory data set without rotation to explain patterns associated with enological and viticultural treatments. Wine composition data were included in PCA analysis of the 2001 vintage.
Results
Yield components and vine size.
Yield per vine (and per hectare) was unaffected in 1999 and 2001 by BLR but was reduced slightly by BLR in 2002 and decreased in all three vintages by CT (Table 2⇓). Thinning also substantially reduced cluster number per vine in all three vintages compared to the control treatment; BLR vines also had slightly fewer clusters than the control in 2002. These differences in yield are mostly illusory, since clusters were removed from the CT treatment. Reduction in yield because of thinning was about 30% during the trial’s first season and of similar magnitude thereafter. Crop loads (yield/weight of cane prunings) decreased in 2001 in thinned plots. Cluster weight (from 68.1–70.0 g, 56.1–59.6 g, and 60.7–67.6 g for 1999 from 68.1–70.0 g, 56.1–59.6 g, and 60.7–67.6 g for 2001, and 2002, respectively) and berries per cluster (from 52–53, 47–50, and 47–53 for 1999 from 52–53, 47–50, and 47–53 for 2001, and 2002, respectively) did not differ among treatments (data not shown). Slight differences were observed in berry weight each season, with the control having the heaviest berries and CT the lightest; BLR berry weight was also lower than the control in 1999 and 2001 but equal to the control in 2002. A slight vine-size reduction was observed in BLR treatment in 2001 (Table 2⇓) but not in 1999 or 2002 (data not shown).
Effect of basal leaf removal (BLR) and cluster thinning (CT) on yield and vine performance of Chardonnay Musqué, 1999, 2001, and 2002.
Berry, must, and wine composition.
Berries.
Berry samples from CT vines had highest Brix in all three vintages (Table 3⇓). BLR berries had slightly higher soluble solids than control berries in 1999 but did not differ from the control with respect to Brix in either 2001 or 2002. Control berries also had highest TA in two of three vintages (not different from BLR in 2002) and lowest pH in two of three vintages (not different from BLR in 2001). Lowest TA was observed in BLR berries in 1999 and in CT berries in 2001, while CT fruit had highest pH in two of three vintages (not different from BLR in 2002). Berry FVT concentration was not different among the three vineyard treatments in 1999 or 2001 but both BLR and CT increased berry FVT in 2002 (Figure 1⇓). Berry PVT in BLR was higher than control and CT in 1999, while both BLR and CT increased PVT over the control in 2002.
Composition of Chardonnay Musqué berries and musts as affected by basal leaf removal (BLR) and cluster thinning (CT), 1999, 2001, and 2002.
Free volatile terpene (FVT) and potentially volatile terpene (PVT) concentrations in Chardonnay Musqué berries at harvest, as affected by basal leaf removal (BLR), cluster thinning (CT), and control (CTL), 1999, 2001, and 2002.
Musts.
As with the berries, must samples from thinned vines in 1999 and 2002 had highest Brix and in 2001 were higher than the control but equal to BLR (Table 3⇑). Must TA tended to be highest in control in 1999 and 2002, although CT treatment was not significantly different in 1999 and BLR treatment did not differ significantly in 2002. Musts from CT vines had highest pH in all three vintages, but BLR musts did not differ from the control with respect to pH. Must FVT were increased over the control in 1999 and 2001 by both BLR and CT, but must from CT vines had lowest FVT in 2002 (Figure 2⇓). Both BLR and CT increased PVT in musts in 1999, but only CT increased must PVT over the control in 2001, and only BLR was higher than the control in 2002. Must FVT, surprisingly, increased relative to berry FVT in 2001, while must PVT declined relative to the concentration in the berries in all vintages, which was expected, since a high concentration of monoterpenes is resident in the skins (Figures 1⇑, 2⇓).
Free volatile terpene (FVT) and potentially volatile terpene (PVT) concentrations in Chardonnay Musqué musts, as affected by basal leaf removal (BLR), cluster thinning (CT), and control (CTL), 1999, 2001, and 2002.
Wines.
Ethanol was increased in CT wines in 1999 and 2001 (Table 4⇓). Wine TA was decreased relative to the control by CT in 2001. Wine TA was also decreased by CT and BLR in 1999 and by BLR alone in 2002. Wine pH was increased by CT over the control in all 3 years, and reduced by BLR in 2002. Both wine FVT and PVT were highest in CT treatment in 2001, but BLR had little impact on wine FVT; however, wine PVT increased in the BLR treatment relative to the control. There were no viticultural treatment effects on wine FVT or PVT in 1999 or 2002.
Composition of Chardonnay Musqué wines as affected by viticultural and enological treatments, 1999, 2001, and 2002.
EC1118 + AR2000 produced highest ethanol in 2002, but there were no yeast effects in 1999 and 2001 (Table 4⇑). Inoculation in 2001 with VL1 yeast strain produced wines lowest in pH (as well as in 1999), FVT, and PVT, and higher in TA compared with wines inoculated with EC1118 ± AR2000. Among the two treatments produced with EC1118, wines treated with AR2000 exhibited higher pH and higher PVT compared to those without AR2000. There were no enological treatment effects on wine FVT or PVT in 1999 or 2002.
A significant viticulture x enology interaction was observed in wine FVT and PVT in 2001 (Table 5⇓). The means of the treatment combinations suggested that wines produced from CT vines and inoculated with EC1118 had highest FVT, whereas those from CT vines and inoculated with EC1118 + AR2000 had highest PVT. Lowest wine FVT was observed in all VL1 treatments regardless of vineyard treatment, as well as BLR/EC1118 and control/EC1118 + AR2000 treatments. Lowest PVT occurred in the control and BLR treatments inoculated with VL1.
Interactions between viticultural and enological treatments for free volatile terpenes (FVT) and potentially volatile terpenes (PVT) of Chardonnay Musqué wines, 2001.
Wine sensory attributes.
1999 vintage.
Canopy manipulation had moderate impact on wine sensory attributes in the 1999 vintage (Table 6⇓). Highest dried fruit aroma and color was found in CT wines, and highest citrus aroma and dried fruit and lychee flavors were found in BLR wines. A greater number of the 1999 attributes appeared to be affected by yeast strain and/or enzyme use (Table 7⇓). Compared to VL1, wines produced using EC1118 ± AR2000 exhibited higher color, lychee and dried fruit aromas, lower grassy aroma, higher floral, lychee, and dried fruit flavors, as well as lower grassy flavor and acidity. With the exception of a more pronounced lychee flavor in wines treated with AR2000, little difference was found sensorially between wines fermented using EC1118 alone and those to which AR2000 had been added. VL1 wines were characterized by highest citrus and grassy flavors and highest acidity. Interactions between viticultural and enological treatments were significant for dried fruit aroma (p ≤ 0.001) body (p ≤ 0.05), and color (p ≤ 0.0001), suggesting that the increase in dried fruit aroma by CT was confined to the EC1118 treatments and that color was maximized in the CT treatments fermented with EC1118 + AR2000 (data not shown).
Impact of three viticultural treatments on wine sensory attributes of Chardonnay Musqué, 1999.
Impact of three enological treatments on wine sensory attributes of Chardonnay Musqué, 1999.
Principal component analysis explained 65.3% of the variation in the sensory data from the 1999 wines (Figure 3⇓). Distribution of the aroma vectors indicated that factor I was explained by the highly correlated attributes grassy/herbaceous and spicy aroma and flavor, which were inversely correlated with lychee, floral, and dried fruit aromas and flavors. Citrus aroma and flavor, body, and finish were highly correlated and were located positively along factor II, while grassy/herbaceous aroma had a strong negative loading. A clear division of wines according to yeast strain was evident along factor I, with those produced with EC1118 and VL1 positioned on the right and left of factor II, respectively. Those wines produced using VL1 were associated with citrus, grassy/herbaceous, and spicy aroma and flavors. Within those wines produced using VL1, the close location of the control wines to the grassy/herbaceous aroma vector indicated that this treatment was intense in terms of this attribute. The loadings of wines fermented with VL1 and produced from the BLR and CT treatments suggested that they were characterized by grassy/herbaceous and citrus flavors as well as body. EC1118 wines were characterized by lychee, dried fruit, and floral aromas and flavors; however, the EC1118/control was strongly weighted on factor II and was thus associated with citrus flavor and finish. Four of six treatments from CT or BLR plots, two of which were produced using AR2000, were positioned farther to the right along factor I, indicating high dried fruit and lychee aromas and flavors as well as floral character. The EC1118/CT treatment was also highly loaded on factor II suggesting a lack of finish and body. The three EC1118 + AR2000 wines were distinguished from those treatments produced from EC1118 alone by a higher loading on factor I, thus suggesting a greater perception of floral and dried fruit characteristics of finish and deeper color intensity.
Principal component analysis of Chardonnay Musqué wines from nine treatment combinations of three viticultural and three enological treatments, 1999. Lowercase variables are aroma attributes and up-percase variables are retronasal and taste attributes. Eigenvectors representing citrus, dried fruit, floral, lychee, and spicy aromas; astringency; and color have been omitted for clarity.
2001 vintage.
In the attributes examined in the 2001 vintage, only tropical fruit, herbaceous/grassy aroma, sweetness, and color varied substantially among viticultural treatments (Table 8⇓). Wines from CT fruit ranked highest in intensity of color, sweetness, and herbaceous/grassy aroma, while control fruit was lowest in these attributes and highest in tropical fruit aroma. Yeast strain or application of AR2000 failed to elicit any differences in wine color, aroma, flavor, or mouthfeel, with the exception of sweetness, which was highest in VL1 wines (Table 9⇓).
Impact of three viticultural treatments on wine sensory attributes of Chardonnay Musqué, 2001.
Impact of three enological treatments on wine sensory attributes of Chardonnay Musqué, 2001.
Principal component analysis of the 2001 wines accounted for 60.7% of the variation in data (Figure 4⇓). Factor I was explained by highly correlated attributes color, herbaceous/grassy, mineral, and lychee aromas, lychee/muscat flavor, and sensation of alcohol, which were inversely correlated with apple/pear, citrus, floral, and tropical fruit aromas. Factor II explained variation among wines in terms of tropical fruit and apple/pear flavor, acidity and sweetness, and TA, which were inversely correlated with pH, FVT, and PVT concentrations. A clear division of wines by PCA according to viticultural treatment was again observed among 2001 wines, with CT treatments strongly loaded on factor I and located to the left of factor II, characterized primarily by mineral, lychee/muscat, and herbaceous/grassy characteristics. Both BLR and control treatments were located to the right of factor II. The VL1/BLR and EC1118 + AR2000/control were heavily loaded on factor I. Delineation between control and BLR treatments was evident, with those wines tending to be located above and below factor I, respectively. Control wines were characterized by tropical fruit and floral aroma characteristics and a high perception of acidity, while BLR wines exhibited citrus and apple/pear characteristics and high FVT and PVT. General but less clear patterns among enological treatments were also seen, whereby VL1 wines tended to be positioned above factor I while five of six EC1118 and EC1118 + AR2000 wines were positioned below factor I. Overall, wines produced from VL1 were not easily characterized and were associated with lychee and grassy characteristics (VL1/CT) or citrus, floral, and tropical fruit. Wines from EC1118 ± AR2000 were likewise not easily characterized, with two (CT) associated with mineral, herbaceous, and lychee characteristics, while others were defined by high PVT and FVT (BLR) or floral, tropical fruit, and citrus characteristics (controls). EC1118 ± AR2000/BLR wines were farthest from the center, suggesting that these treatments were most intense in the eigenvectors in that quadrant.
Principal component analysis of Chardonnay Musqué wines from nine treatment combinations of three viticultural and three enological treatments, 2001. Lowercase variables are aroma attributes and up-percase variables are retronasal and taste attributes (sensory) or compositional (FVT and PVT). Eigenvectors representing apple/pear, lychee aromas; apple/pear and tropical fruit flavors; sweetness, acidity, alcohol, and color; and berry TA and pH were omitted for clarity.
2002 vintage.
Wines made from the canopy treatments were generally easily separable based on means of aroma and flavor intensities. Control wines were characterized by high acidity, astringency, and both earthy/mushroom and grassy aromas and flavors. These wines also showed the lowest apple/pear, muscat, and floral/perfumy aroma and flavor intensities and the lowest overall impression on judges (Table 10⇓). Wines made from BLR vines, in general, had the highest intensities of the most desirable variables, and scored highest for floral/perfumy aroma and overall impression on judges. These wines also had high grassy aroma and the lowest earthy/mushroom aroma. Wines made from CT vines also had high apple/pear and floral/perfumy aromas and a high overall impression on judges. They also had lowest intensities of both grassy aroma and earthy/mushroom flavor intensities.
Impact of three viticultural treatments on wine sensory attributes of Chardonnay Musqué, 2002.
Discussion
Impact of BLR and CT on yield components and vine size.
Cluster thinning substantially reduced the number of clusters per vine compared to control and BLR treatments and was undoubtedly the main contributing factor to reduced crop load and the ~21 to 36% reduction in yield per vine. The lack of substantial yield differences between BLR and control treatments is in agreement with literature (Reynolds and Wardle 1989a), although some studies have reported small increases in yield in the year following BLR, which they attributed to enhanced node fruitfulness (Reynolds et al. 1996).
The BLR treatment reduced berry weight in 2 of 3 years. This has been previously reported and has been attributed to either to increased berry temperature and/or sunlight exposure (Bergqvist et al. 2001, Reynolds and Wardle 1989a) and/or a reduction in assimilate supply associated with the loss in leaf area (Ollat and Gaudillère 1998). Grenache berry mass was observed to decline with increased sun exposure, due possibly to an effect on berry cell division and/or elongation as well as berry transpiration and dehydration (Bergqvist et al. 2001). Reynolds and Wardle (1989b) observed that fully exposed Gewürztraminer berries exhibited lower weights than those either shaded or partially exposed to sunlight. Others have speculated that loss in assimilate supply caused reduced Cabernet Sauvignon berry weights, since the diminished leaf area was associated with declines in berry growth rates (Ollat and Gaudillère 1998). In our study, vine size across all treatments (0.8 kg/m row) suggested overvigorous vines and crop loads (mean 4.3) were concomitantly low (Bravdo et al. 1985); therefore, it appears doubtful that lack of assimilate supply was the reason for the reduction in berry weight in our study.
Thinning did not increase berry weight in this study and decreased it in one season. Lack of treatment effect was somewhat expected; thinning at veraison tends to elicit minimal response in terms of yield compensation (Ferree et al. 2002). Timing of thinning notwithstanding, crop level reduction may fail to elicit any increases in berry weight under conditions where nonthinned vines are not source-limited (Kaps and Cahoon 1989). In one report, cluster thinning of three V. vinifera cultivars in Washington did not affect berry weight (Keller et al. 2005). In our study, vine size across all treatments suggested excess vigor, and their crop loads suggested that they were undercropped; it is not surprising that berry weight did not increase in response to CT as has been observed in other thinning studies (Bravdo et al. 1985, Kliewer and Weaver 1971, Reynolds et al. 1994a).
Leaf removal decreased vine size slightly in one season, consistent with other studies (Reynolds and Wardle 1989a). Normally, leaves removed during this cultural practice are considered as superfluous with respect to their dried matter contributions to the vine. However, the BLR treatment, despite being applied within the parameters of good commercial practice, may nonetheless have lowered the leaf surface area to a point whereby vine size was reduced. BLR in this study may have thus reduced assimilate supply to the shoot tips, ultimately lessening vegetative growth. Among supportive evidence of this, Kaps and Cahoon (1992) observed that vegetative growth of Seyval vines exhibited an increasing linear response in relation to leaf surface area and speculated that vegetative growth was highly dependant on assimilate supply even more so than berry growth. Alternatively, leaf removal can cause an alteration to assimilate distribution patterns, whereby carbohydrates normally moving acropetally alter their pattern of movement toward the site of leaf removal (Quinlan and Weaver 1970).
Impact of BLR and CT on berry, must, and wine composition.
Compared to the control, berries from the CT treatment were higher in Brix and pH and lower in TA. Similar increases in Brix and pH and decreases in TA because of reduction in crop level have been well documented (Kaps and Cahoon 1989, Kliewer and Weaver 1971, Reynolds et al. 1994a). In some circumstances increases in Brix have resulted from thinning vines carrying high crop loads (Kliewer and Weaver 1971), but in some cases thinning has had little effect on berry composition (Keller et al. 2005, Reynolds et al. 1994a). Thinning usually increases Brix because of a reduced number of sinks and resultant faster rates of solute accumulation. Such advancement in soluble solids accumulation is thus believed to be the result of greater leaf area:fruit weight ratios in thinned vines, which increase assimilate supply on a per berry basis. Cabernet Sauvignon vines thinned at two levels attained 22.5 Brix earlier than did a nonthinned control (Bravdo et al. 1985). Increase in assimilates to the berry may therefore result in earlier occurrence of and greater rate of hexose accumulation.
Thinning also decreased TA (2 of 3 years) and increased pH (3 of 3 years), in some cases to a greater magnitude than BLR. Kliewer and Weaver (1971) similarly observed an increase in TA and a reduction in pH in Tokay grapevines carrying high crop levels, the result of a slower rate of maturation among vines with low leaf area:fruit weight ratios (i.e., overcropped vines). On the other hand, no effects of thinning were found on TA or pH in three cultivars in Washington (Keller et al. 2005) or in Seyval vines in Ohio (Kaps and Cahoon 1989). In our study, during fruit maturation the CT vines were roughly a week ahead of the control and BLR treatments with respect to Brix, TA, and pH, suggesting that nonthinned vines were high enough in crop level to elicit a response (Schlosser 2003). Alternatively, since CT also increased the rate of pH increase, it is possible that an increased K+ concentration into fewer sinks may have increased pH and reduced TA within the CT treatment, consistent with observations that K+ concentration is often positively and negatively correlated with pH and tartaric acid concentration, respectively (Bravdo et al. 1985).
BLR also decreased TA and increased pH compared with the control in 2 of 3 years. Decreases in TA (Ollat and Gaudillère 1998) without impact on Brix but with reductions in pH (Reynolds et al. 1996) have been observed from application of fruit zone leaf removal. Reduction in TA though leaf removal is generally ascribed to increased light and temperature associated with improvement to the canopy microclimate, with concomitant enhancement of malic enzyme. Bergqvist et al. (2001) speculated that fruit zone leaf removal might have increased Grenache berry temperatures beyond ranges suggested to be optimal for malic acid accumulation to the point where degradation was favored.
Berry FVT was little influenced by either BLR or CT (1 of 3 years). Other studies have likewise found little impact of viticultural practices (cluster exposure, leaf removal, or cluster thinning) on berry FVT (Reynolds and Wardle 1989b, Reynolds et al. 1994a). Berry PVT, however, was increased by BLR in 2 of 3 years, consistent with other work (Macaulay and Morris 1993, Reynolds et al. 1995, 1996, Zoecklein et al. 1998). A reduction in the number of superfluous sinks (i.e., old leaves) and enhanced canopy microclimate are cited as two possible explantions for an increase in berry monoterpene concentrations in leaf-pulled vines (Bureau et al. 2000, Reynolds and Wardle 1989a), although a concentration effect resulting from reduced berry size in leaf removal treatments has also been speculated (Reynolds et al. 1996). Both explanations could account for the increased PVT concentration observed in the BLR treatment in this study. In the CT treatment, an increased rate of berry maturation was the likely basis for enhanced PVT (Bureau et al. 2000), especially as CT was often advanced over the control by approximately one week in Brix, TA, and pH throughout fruit maturation (Schlosser 2003).
Trends in must and wine TA, pH, and PVT relative to BLR and CT were consistent with those in berries. There were some inconsistencies in treatment effects between berry and must, however. For Brix and FVT, trends for berries were not consistent with those for musts and wines. This inconsistency might be simply the result of sample size, since 100 berries and ~25 kg of fruit were the basis of the berry and that must composition data, respectively. Berry FVT varied little among treatments in 1999 and 2001, but must FVT was increased by both BLR and CT. In 2001, only CT increased wine FVT. In 2002, BLR and CT increased berry FVT, but musts from CT vines had lower FVT than control or BLR treatments. Wine FVT and PVT were unresponsive in 1999 and 2002. Such variation among berry, must, and wine samples is expected, since berry skins contain a substantial proportion of most chemical constituents compared with flesh. Specifically, FVT exhibit variations in composition between the berry skin, pulp, and juice fractions both qualitatively and quantitatively (Vazquez et al. 2002). Removal of skins during pressing may have created different trends in must and wine composition from those observed previously in the berry. Despite variation in treatment effects among berry, must, and wine samples, CT produced wines higher in pH, FVT, and PVT and lower in TA (one season only) than wines produced from the BLR or control treatments.
Impact of yeast and enzyme on wine composition.
Wines fermented with EC1118 were higher in pH (3 of 3 years) and FVT and PVT concentration (one year only) and lower in TA (one year only) than those produced using VL1. Variations in TA among wines produced using different yeast strains are often explained by observed differences in metabolic activity. Antonelli et al. (1999) observed that S. bayanus strains synthesized malic acid, while S. cerevisiae strains generally degraded it. Reynolds et al. (2001) reported differences in metabolism of citric, lactic, malic, and succinic acids by several Saccharomyces strains in fermenting Riesling musts. Results from this trial suggest VL1 could have synthesized more organic acids than EC1118, or was less efficient at metabolizing them, either case likely contributing to reduced pH in VL1 wines.
Yeast strains have the potential to alter terpene concentration, and thus wine sensory properties, through biosynthesis of terpenes, biotransformation of free terpenes, and/or hydrolysis of terpene glycosides. In this study, the higher FVT concentration in the EC1118 fermentations in one season may have been the result of increased synthesis of terpenes during fermentation. Production and/or release of terpenes by Saccharomyces was reported (Zea et al. 1995), whereby geraniol, linalool, α-terpineol, (E)-neroldiol, and (Z)-neroldiol accumulated within the yeast cell during fermentation. The increased FVT in the EC1118 wines in one year of this study may therefore have been caused by higher β-glucosidase activity by the yeast, hydrolyzing PVT into their free forms at a faster rate than VL1. There is ample evidence for this in literature. Varying degrees of β-glucosidase activity were found among six Saccharomyces strains (Mateo and Di Stefano 1997) and in seven non-Saccharomyces species (McMahon et al. 1999). At the other extreme, some studies have suggested that Saccharomyces exhibit little to no β-d-glucosidase activity (Rosi et. al. 1994) and there is limited β-glycosidase activity in the S. cerevisiae VL1 strain (McMahon et al. 1999). Some authors have stressed temporal limitations: one study found commercial yeast strain β-glucosidase activity was maintained at 90% of its optimal activity for 3 weeks following fermentation (Grossmann and Rapp 1988), whereas another study found that activity in three Saccharomyces strains was most evident only during the exponential growth phase of fermentation (Delcroix et al. 1994). Others have found β-glucosidases in association with yeasts, but have emphasized their limitations because of inactivity at pH ranges, temperatures, and ethanol concentrations commonly present in wine production (Delcroix et al. 1994, Mateo and Di Stefano 1997, Rosi et al. 1994). A 95% loss was found in β-glucosidase activity from an optimum pH of 6.0 when exposed to pH 2.8 for 90 min (Delcroix et al. 1994). An 85 to 90% loss of activity was similarly observed at temperatures commonly used in white wine production (18 to 20°C) compared with an optimal activity at 50°C. Little inhibition as a result of ethanol concentration was found by Delcroix et al. (1994); however, Mateo and Di Stefano (1997) observed a sharp decrease in enzyme activity at ethanol concentrations commonly found in wine.
Addition of AR2000 (which is claimed to have β-glucosidase side activity) increased PVT in one season but had little effect on FVT compared with wines produced solely using EC1118. These results are contrary to other studies where additions of Aspergillus niger enzyme preparations increased FVT and decreased PVT (Vazquez et al. 2002). As with yeast-derived enzymes, our results may reflect other studies that saw inhibition of β-glucosidase enzyme activity under pH, temperature, and ethanol conditions commonly used in wine production (Delcroix et al. 1994, Mateo and Di Stefano 1997, Rosi et al. 1994).
The significant enological treatment x viticultural treatment interactions observed in one year of this study suggest that FVT and PVT may be favored by certain combinations of yeast strain/enzyme and viticultural treatment combinations. Wines produced with EC1118, regardless of viticultural treatment, generally resulted in substantially higher wine FVT. However, among EC1118 fermentations, those produced from CT treatments were higher in FVT than those from BLR and control treatments. The fact that very little difference in FVT was observed between the BLR and the CT treatments before fermentation suggests that the elevated wine FVT was not directly due to higher must FVT from CT treatments. Instead, FVT concentration was either increased or decreased, respectively, depending on whether CT or BLR fruit was used in combination with EC1118. Zoecklein et al. (1997) observed that individual monoterpene concentrations either increased (α-terpineol, hotrienol oxide, and t-pyran linalool oxide) or decreased (geraniol and nerol) from the must to the finished wine and that the magnitude of change depended upon yeast strain, with Fermiblanc, VL1, and D47 resulting in higher total free monoterpenes than EC1118. Such variation is undoubtedly a complex set of interactions that may involve varying abilities of yeast strains in β-glucosidase activities plus differing abilities to biosynthesize and biotransform individual monoterpenes. Results here, however, indicate that neither viticultural practices nor yeast strain/enzyme selection necessarily result in higher berry, must, or wine FVT or PVT concentration, but that specific combinations of field and winery practices ultimately affect final concentrations in the wine.
Impact of treatments on wine sensory attributes.
Viticultural effects on wine sensory attributes were clearly seasonal in nature, with moderate effects in 1999 and greater effects in 2001. In 2002, wines made from the canopy treatments were easily separable. These viticultural effects on wine aroma and flavor are generally consistent with previous studies in other regions. Fruit-zone leaf removal has enhanced wine aroma and/or flavors in several aromatic Vitis vinifera cultivars (Reynolds et al. 1995, 1996). Muscat aroma of wines made from sun-exposed clusters of Golden Muscat was also increased (Macaulay and Morris 1993). Crop-level reduction has also proved to have a positive impact on wine sensory characteristics (Bravdo et al. 1985, Reynolds et al. 1994b). In situations where initial vine size is quite high, crop reduction increased vegetal aromas in wines and reduced fruity components (Chapman et al. 2004). Additionally, pooled FVT and PVT values from 1999, 2001, and 2002 were significantly correlated with floral and muscat aroma (Roberts 2003), suggesting that, although individual vintages may not have been conclusive, monoterpenes and sensory attributes were related overall.
Yeast strain and enzyme use had much greater effects than viticulture in 1999. EC1118 ± AR2000 increased lychee and dried fruit aroma, floral, lychee, and dried fruit flavors, and color compared with wines produced using VL1. The PCA suggested that yeast strain/enzyme selection and viticultural treatment both had significant impacts on wine sensory attributes. Most EC1118 wines, regardless of viticultural treatment, were characterized as high in floral, lychee, and dried fruit flavor, whereas VL1 wines were described primarily by grassy/herbaceous and spicy characteristics. The treatment combinations having greatest intensity in floral, lychee, and dried fruit characteristics were those in which either CT or BLR were employed in the vineyard and/or AR2000 was used in winemaking. In 2001 yeast strain had little impact. Wines produced from VL1 were not readily defined. Wines from EC1118 ± AR2000 were also difficult to generalize. Likewise, in 2002, viticultural treatments had clear sensory impacts but enological treatments did not. The maturity levels of the fruit at harvest, which averaged 24.4 (1999), 22.3 (2001), and 20.4 Brix (2002), suggest that season, maturity level, and perhaps other characteristics may impact the magnitude of effect of viticultural and enological practices on cultivars such as Chardonnay Musqué.
Conclusions
Fruit maturity of Chardonnay Musqué in Ontario was advanced by basal leaf removal (BLR) and cluster thinning (CT) in terms of reduced berry, must, and wine TA; increased berry and must pH and Brix; increased must and wine FVT (CT only); and increased berry, must, and wine PVT. Wines fermented with VL1 were substantially lower in pH, FVT, and PVT than those of EC1118 and had higher TA. Wines fermented with EC1118 + postfermentation enzyme AR2000 did not differ from those produced solely with EC1118 in terms of TA and FVT but were higher in PVT and pH. Although season-dependent, CT wines generally differed from BLR and control wines sensorially in increased color depth, higher sweetness and herbaceous/grassy aroma, and reduced tropical fruit aroma. BLR wines had higher citrus, lychee, tropical fruit, and floral aromas and flavors, while control wines had greater acidity. Principal component analysis of sensory data suggested that variation in wine sensory profiles was due to both viticultural and enological practices. This research confirms that viticultural practices, yeast strain selection, and enzyme use have the potential to create profound differences in white wine sensory properties. The correlation of chemical variables with some sensory attributes implied that alteration in wine chemical composition by yeast strains may cause some of the perceived differences among treatments. Further detailed analysis of the wines and volatile aroma composition would provide clearer insight.
Footnotes
Acknowledgments: The authors thank Angelo Pavan and Kevin Latter, Cave Spring Estate Winery, for their cooperation. Efforts of the many sensory panelists are also acknowledged.
Sigrid Gertsen-Briand, with Lallemand Inc., provided yeasts and Gist-Brocades provided AR2000. Financial assistance from the Natural Sciences and Engineering Research Council and the National Research Council of Canada is hereby acknowledged.
This paper represents work included in the MSc thesis of J.W. Schlosser (2003) and the undergraduate theses of R. Power (2000) and R.W. Roberts (2003).
- Received May 2005.
- Revision received October 2005.
- Revision received July 2006.
- Copyright © 2007 by the American Society for Enology and Viticulture
Literature Cited
Vol 58 Issue 1
Jump to section
Related Articles
Cited By...
More from this TOC section
Similar Articles
