Food Research International 34 (2001) 483±499
www.elsevier.com/locate/foodres
Vini®cation eects on the sensory, colour and GC pro®les of Pinot noir wines from British Columbia $
B. Girard a,*, D. Yuksel a, M.A. Cli a, P. Delaquis a, A.G. Reynolds b a
Agriculture and Agri-Food Canada, Paci®c Agri-Food Research Centre, 4200 Highway 97S, Summerland, BC, Canada V0H 1Z0 Cool Climate Oenology and Viticulture Institute, Brock University, 500 Glenridge Avenue, St-Catharines, ON, Canada L2S 3A1
b
Received 8 September 2000; accepted 30 September 2000
Abstract Analytical and sensory pro®les were obtained for Pinot noir wines fermented with two yeasts (Lalvin EC-1118, P; Lalvin L-2056, L) and four vini®cation techniques: high (T), ambient, cold and modi®ed-cold temperature (M) fermentations. Wines were analysed for titratable acidity, pH, ethanol, colour, total phenolics, ¯avonols, tartaric esters, anthocyanins and volatile constituents. Sensory descriptive analysis was conducted using 12 judges to evaluate colour, aroma, ¯avour, body and overall quality. Wines from the M vini®cation had the most red colour, body, tropical fruit and spicy aromas and had the least vegetal character. Wines from the T vini®cation were the most vegetal. Increased ester concentration (mainly acetates) and total volatile compounds were found in wines from the M vini®cation with yeast P in comparison to the other vini®cation methods and yeast L. Principal component analyses distinguished four distinct groups of wines based on 33 main volatile constituents. Partial least square regressions revealed strong relationships between sensory and instrumental colour measurements in terms of red colour intensity (r=0.88) and red hue (r=0.87). Crown Copyright # 2001 Published by Elsevier Science Ltd. All rights reserved. Keywords: Pinot noir; Descriptive analysis; Principal component analysis; Principal component similarity analysis; Partial least squares regression
1. Introduction The British Columbia (BC) wine industry is expanding and has undergone a dramatic increase in the availability of wines made from varieties such as Pinot noir. Over the past 5 years, the acreage of Pinot noir has nearly doubled from 250 to 462 acres (1994±1999) and makes up 24% of red varieties grown in BC. It is generally well accepted that the character of Pinot noir wines will vary with the regional eects of soil, climate, cultural practices, vini®cation procedures and storage. While regional sensory characteristics have been identi®ed for young wines from CoÃte de Beaune (freshly squeezed strawberries), CoÃte de Nuit (sun-warmed strawberries), Chalonnaise (ink), California (plum jam) and for older wines from Burgundy (violets to game; Robinson, 1986), the character of Pinot noir wines from the newer wine growing regions, such as the Paci®c Northwest and BC, are less well documented and the * Corresponding author. Tel.: +1-250-494-6390; fax: +1-250-4940755. E-mail address:
[email protected] (B. Girard). $ For the Department of Agriculture and Agri-Food, Government of Canada, Minister of Public Works and Government Services, Canada.
eects of cultural practices and vini®cation methods require further characterization. In these expanding wine growing regions, fruit composition has been shown to vary with sun exposure (Price, Breen, Valladao, & Watson, 1995), water-stress (Reynolds & Naylor, 1994), clonal selection (Price, Lombard, & Watson, 1988; Watson, Lombard, Price, McDaniel, & Heatherbell, 1988) and canopy management methods (Reynolds, Price, Wardle, & Watson, 1994). Reynolds et al. showed that wines can be signi®cantly less vegetative and grassy when produced from vines with limited shoot densities (10 shoots/m) and restricted crops (half). Other research on Pinot noir has focussed on the identi®cation of the volatile compounds (Brander, Kepner, & Webb, 1980; MirandaLopez, Libbey, Watson, & McDaniel, 1992a,b) and the eect of the fermentation temperature on the ®nal composition of red wines. Ough and Amerine (1961) reported that the fermentation of Pinot noir at 21.1 or 26.7 C gave superior wines than at 11.7 C. Girard, Kopp, Reynolds, and Cli (1997) compared three vini®cation methods to assess their eect on chemical composition, sensory characteristics and headspace volatile composition of BC Pinot noir wine. They
0963-9969/01/$ - see front matter Crown Copyright # 2001 Published by Elsevier Science Ltd. All rights reserved. PII: S0963-9969(00)00177-0
484
B. Girard et al. / Food Research International 34 (2001) 483±499
reported that a two-stage pre-fermentation treatment involving high temperature extraction and a 15 C vini®cation with bentonite produced wines with higher anthocyanin content and a more intense fruity aroma and ¯avour. The ester content of wines vini®ed using this modi®ed method was markedly increased. More research is necessary to investigate various enological factors such as dierent yeasts strains and explore alternate fermentation procedures for Pinot noir. Relating wine sensory and physicochemical variables is a complex task that can be facilitated by the use of multivariate techniques. Partial least squares regression (PLS) was ®rst used to investigate the relationships between sensory and physical and chemical variables in vegetables by Martens (1985, 1986). PLS is a soft modelling technique used to compare two sets of data by seeking out latent variables common to both data sets (Martens & Martens, 1986). Principal component similarity analysis (PCS) is another technique that was developed by Nakai (1992) as an unsupervised classi®cation method. It combines the Principal component analysis (PCA) and pattern similarity computations. PCS was applied to gas chromatographic data to classify mangos (Vodovotz, Arteaga, & Nakai, 1993), microbial defects in milk (Horimoto, Lee, & Nakai, 1997) and high performance liquid chromatography (HPLC) data of cheddar cheese to identify quality defects (Furtula, Nakai, Amantea & Laleye, 1994a,b). These multivariate methodologies may be combined to provide useful interpretations of large datasets from Pinot noir wine assessment. The purpose of this work was to pro®le the sensory and analytical characteristics of BC Pinot noir wines on the basis of four vini®cation methods with two dierent yeasts using descriptive analysis, chemical measurements and volatile constituents, and to assess the relationships among sensory and analytical pro®les using PCA, PCS and PLS. 2. Materials and methods 2.1. Vini®cation protocol Pinot noir (Vitis vinefera L.) grown at Boucherie Mountain Vineyards, Westbank, BC (Girard et al., 1997) were transported to the Paci®c Agri-Food Research Centre and stored at 1±2 C for 16 h. Grapes were macerated using a Garolla-type crusher-destemmer; 35 l of must were then placed in each of 24 primary fermenters and treated with potassium metabisul®te (30 mg/l SO2 equivalent). Soluble solids content, pH and titratable acidity of the must were 23.00.8 Brix, 3.210.05, and 9.10.6 g tartaric acid/l, respectively. The experimental design consisted of four vini®cation methods, two yeast strains and three replicates.
The four vini®cation methods were high-temperature (30 C, H), ambient-temperature (20 C, A), cold-temperature (15 C, C) and modi®ed-cold-temperature (15 C, M) fermentations. For C, A and H vini®cations, the methods were identical except for the temperature of fermentation. Inoculation was done with one of two yeasts (P, Saccharomyces bayanus, Lalvin EC-1118; and L, Saccharomyces cerevisiae, Lalvin L-2056). The yeasts were obtained from Lallemand Inc. (Montreal, Canada) and were prepared according to the manufacturer's recommendations. Fermentation caps were punched down twice daily and fermentations were monitored for temperature and residual sugar content by refractometry. For the M vini®cation, must was gravity ®ltered to separate the free run juice (73±75% wt) and the marc (pomace, 25±27% wt). The sediments in the juice were allowed to settle overnight at 1±2 C and racked. The free run juice was brought up to 15 C before inoculation with either yeasts P or L. The marc was frozen at 20 C for later use. When the fermentable sugars had decreased to 10±12 Brix, the marc was thawed overnight and added to the active fermentation. Upon completion of the alcoholic fermentations as indicated by hydrometer readings <0% sugar, wine musts were pressed in a hydraulic rack and cloth press maintained at 13 800 kPa (2000 psi) and transferred to 20-l glass carboys. Wines were stored at 0 C for cold stabilization, racked at 1 week and 1 month after pressing, and protein-®ned with bentonite at 1g/l. Wines were then provided with 50 mg/l SO2 and stored at 15 C in glass carboys for 6 months prior to bottling. All analyses reported in this manuscript were performed after 2 years in the bottle. 2.2. Standard chemical analysis Soluble solids content ( Brix), titratable acidity (TA), and pH were assessed according to Amerine and Ough (1980). TA was measured with a Brinkman autotitrator (Metrohm, Herisau, Switzerland) and pH on a Fisher 825MP pH metre. Ethanol was determined using a Hewlett-Packard 5700 gas chromatograph (GC; Hewlett-Packard, Avondale, PA) equipped with a Porapak Q-100 column (3.2 mm2 m). Other conditions of operation included: N2 carrier gas ¯ow rate 20 ml/min; oven temperature 175 C; injector temperature 200 C; and detector temperature 200 C. Wine samples (2 ml) were pipetted into 100 ml volumetric ¯asks and made up to volume with 0.2% isopropanol as the internal standard. Injection volume was 1 ml. 2.3. Colour measurement CIELAB L* (lightness), a* (redness) and b* (yellowness) tri-stimulus values were obtained with a scanning spectrophotometer using a colour determination and
B. Girard et al. / Food Research International 34 (2001) 483±499
matching program (Beckman Model DU 640B, Fullerton, California). The hue and chroma (saturation) attributes of chromaticity were calculated using the formula tan 1 (b*/a*) and [(a*)2+(b*)2]1/2, respectively. Wine colour was also measured on samples diluted 1:10 with pH 3.5 buer (0.1 M citric acid with 0.2 M Na2HPO4). The absorbance of the samples was read at 420, 520 and 700 nm. The colour density and tint were calculated as described by Wrolstad (1976) using the equations [(A520 A700)+(A420 A700)] and [(A520 A700)/(A420 A700)], respectively. 2.4. Phenolics analysis A modi®ed version of the Glories' method (Glories, 1979; Romani, Mancini, Tatti & Vincieri, 1996) was used to estimate the phenolic content of the wines. Samples were diluted 1:10 with 10% ethanol. The method consisted of placing 0.25 ml of sample or standard in a test tube and adding 0.25 ml of 0.1% HCl in 95% ethanol and 4.55 ml of 2% HCl. The solution was mixed and absorbances (A) were recorded at 280, 320, 360 and 520 nm with a spectrophotometer after 15 min. A280 was used to estimate total phenolic content, A320 to estimate tartaric esters, A360 to estimate ¯avonols, and A520 to estimate anthocyanins. Gallic acid in 10% ethanol was used as the standard for total phenolics. Caeic acid in 10% ethanol was used as the standard for tartaric esters. Quercetin in 95% ethanol was used as the standard for ¯avonols. Malvidin-3-glucoside in 10% ethanol was used as the standard for anthocyanins. All standards were obtained from Sigma Chemical Co. (St-Louis, MO, USA), except for malvidin-3-glucoside which was obtained from Extrasynthese (Genay Cedex, France). 2.5. Analysis of volatile constituents Wine samples (20 ml) were preheated at 40 C for 5 min in a 25-ml fritless sparger connected to a Tekmar LSC 2000 purge and trap unit (Tekmar Corp., Cincinnati, OH). Using puri®ed helium at a purging rate of 40 ml/min, volatile compounds were collected for 15 min on a preconditioned glass trap (4 id180 mm length) packed with TenaxTA (60/80 mesh, 100 mg). After a 25 min dry purge, the volatile compounds were desorbed at 200 C for 8 min and cryofocussed onto a 1-m deactivated fused silica capillary column immersed in liquid nitrogen. Separation was performed on a Supelcowax 10 capillary column (60 m0.25 mm id0.25 mm ®lm thickness) in a Hewlett Packard 5890-5970 gas chromatograph-mass selective detector (GC-MSD) system. The end of the GC column was connected to a variable splitter set to deliver equal ¯ow (1:1, v/v) to the MSD and the olfactory detector port. Column pressure was set at 207 kPa (30 psi). Oven temperature was held initially at 35 C for 10 min and then increased by 3 C/
485
min to 220 C. Quanti®cation was calculated relative to the standard cyclohexanone (4.24 mg/l). Spectra were recorded at an ion source temperature of 280 C, ionizing energy of 70 eV, and scan range of 25±250 amu at a rate of 2.6 scan/s. Identi®cation of compounds was performed with a HP G1034C MS Chem Station containing a HP G1035A Wiley (138.1) probability based matching (PBM) library and con®rmed with retention data of available authentic compounds. Retention indices were calculated using a degree-two polynomial spline with discontinuous ®rst derivatives at knots 3 and 7, and discontinuous second derivatives at knots 3, 5, 7 and 10 (Girard, 1996). 2.6. Preliminary sensory testing Preliminary bench testing revealed no dierences between the replications (data not shown), therefore only one replication was used for further sensory analyses. To con®rm that dierences among the treatments (C, T, M, A) were perceptible, dierence testing was conducted prior to descriptive analysis. Triangle tests were used to evaluate dierences in aroma and ¯avour between the treatments for the six possible combinations [C-A, C-T, C-M, M-T, M-A, A-T] for yeast P. Twenty judges from the Paci®c Agri-Food Research Centre (PARC) tested the wines in individual tasting booths for a total of four sessions, once in the morning and once in the afternoon over two days. At each session, they evaluated a set of three triangle tests for aroma dierences and a set of three triangle test for ¯avour dierences. Judges were asked to smell (or taste) the samples and identify the odd sample. Wines for the aroma and ¯avour evaluations (30 ml samples) were served at room temperature in 250-ml tulip shaped black glasses coded with three digit random numbers. Since triangle tests con®rmed perceivable aroma and ¯avour dierences (Table 1) for yeast P, dierence testing was not pursued on yeast L. Instead, descriptive analysis was used and conducted on all wines. 2.7. Descriptive analysis Descriptive analysis was used to pro®le eight Pinot noir wines (two yeasts four vini®cations in duplicate). Twelve judges (PARC employees) participated in a round table discussion session. They were given representative wines and tentative references standards. These standards were prepared from attributes reported in the literature (Cli & Dever, 1996; Guinard & Cli, 1987; Reynolds, Yerle, Watson, Price, & Wardle, 1996). Judges were asked to evaluate the samples, identify possible sensory attributes, and provide feedback as to the appropriateness of the references. Following this session, references were modi®ed and re®ned as required. A second round table discussion was used to con®rm the ®nal list of attributes
486
B. Girard et al. / Food Research International 34 (2001) 483±499
(visual, aroma and ¯avour) and corresponding standards (Table 2) for evaluation of the wines. At this session, judges were introduced to the scoring methodology and had an opportunity to practice the evaluation process. Wines were evaluated in duplicate according to a randomized complete block design. For the aroma and ¯avour evaluations, 30 ml samples of wine were served at room temperature in 250 ml tulip shaped black glasses. For the visual evaluation, 25 ml samples were presented in 250-ml INAO approved crystal wine glasses. All glasses were labelled with three digit random numbers and presented in random order. Judges evaluated the wines in individual tasting booths for a total of four sessions, held once in the morning and once in the afternoon over two days. At each session, judges conducted an aroma, ¯avour and visual assessment on four samples. They evaluated four samples for aroma, four samples for ¯avour and four samples for colour. Judges ®rst smelled the wines and evaluated the aroma attributes on one scorecard. They then tasted the wines and evaluated the ¯avour on another scorecard. This was followed by the evaluation of colour on yet another scorecard. Each scorecard consisted of a dierent set of 3-digit random numbers to ensure independence of the visual, aroma and ¯avour evaluations. On each scorecard, the magnitude of the attributes were identi®ed by placing a mark on an unstructured 10 cm line scale labelled from low to high. Colour attributes (intensity, hue) were evaluated under natural light against a white background. While colour intensity was evaluated for low to high as described above, colour hue was evaluated on a similar 10 cm scale anchored with orange-red, brick-red and red-red at 1, 5 and 9 cm, respectively. All evaluations were quanti®ed using a digitizing pad by measuring the distance of the judge's evaluation from the origin (0 cm).
Guymon, 1948). Descriptive data was analysed using analysis of variance (ANOVA) to determine the eects of wine (W), judge (J) and replication (R) and the twoway interactions [WJ, JR, RW]. Attributes which were not signi®cantly dierent among the wines were dropped from further analyses. Means were plotted using cobweb diagrams (Cli & Wild, 1990). PCA was conducted on the mean sensory scores without rotation. ANOVA and PCA were performed using SAS (SAS Institute Cary, NC, USA). PLS was conducted using GENSTAT (Genstat 4, Numerical Algorithms Group Rothamstead, UK) and PCS was performed by means of an in-house program written at the University of British Columbia, Vancouver, Canada (Vodovotz et al., 1993).
2.8. Statistical analysis
Table 2 Composition of reference standards used for the evaluation of Pinot noir winesa
Signi®cant dierences for the triangle tests were determined using statistical tables (Roessler, Warren & Table 1 Results from triangle tests (n=20) on the aroma and ¯avour of Pinot noir wines fermented using yeast P and dierent vini®cation treatments Paira
Aroma
Flavour
C-A C-T C-M M-T M-A T-A
10 13** 12* 16*** 7 13**
7 7 9 13** 10 10
a C, cold temperature (15 C) fermentation; A, ambient temperature (20 C) fermentation; T, high temperature (30 C) fermentation; M, modi®ed-cold-fermentation (15 C). *P<0.05. **P<0.01. ***P<0.001.
3. Results and discussion 3.1. Sensory analysis The results of the ANOVA carried out on the sensory attributes of Pinot noir (Table 3) indicated that wines and judges were a signi®cant source of variation (P<0.001) for all attributes. Judge variation has been reported in the literature (Guinard & Cli, 1987; Lee, Luna-GuzmaÂn, Chang, Barrett, & Guinard, 1999) and is primarily due to judge physiological and scoring dierences. Sensitivity of judges to the attributes may also vary. The lack of signi®cance (P<0.05) of the replication (R) term and interaction JW was noteworthy. These terms re¯ected panel reproducibly and consistency, and were important for eective pro®ling. Individual judge reproducibility (JR) was also satisfactory although this
Attribute Aroma Berry
Composition
Cherry Currant/cassis Plum/prune Vegetal Spicy Tropical fruit
Equal portions of strawberry, blackberry, and raspberry references Strawberry: 29.5g strawberry jam Blackberry: 350 g syrup+170 g blackberry juice Raspberry: 70 g raspberry juice 350 ml cherry syrup, 170 g cherry drink 30 g of syrup +360 g ribina 200 g prune/plum puree, 70 g of plum juice 200 g of green bean juice, grass 0.15 g cloves+0.15 g pepper 350 g of tropical fruit juice, 1 ml isoamyl acetate
Taste Acid Astringency Bitter
4 g citric acid 1 g aluminum sulfate 0.75 g caeine
a
Ingredients were added to 500 ml of a composite Pinot noir wine.
B. Girard et al. / Food Research International 34 (2001) 483±499
term was signi®cant for two attributes (red colour, spicy aroma) due to the lack of reproducibility by one judge. Having adequately documented panel performance, only the wine treatment eects are discussed further. Ten attributes were signi®cantly dierent (P<0.05) among the wines (red colour, red hue, currant, vegetal, spicy aroma, tropical fruit aroma, spicy ¯avour, astringency, body, overall quality). Because overall quality had a preference `judgement' component rather than being rated as a perceived intensity, this attribute was not considered appropriate for further descriptive pro®ling. Mean wine scores for the dierent vini®cations using yeasts P and L are shown in Fig. 1(a) and (b). For both yeasts, M vini®cation treatment produced wines with the highest body, tropical fruit aroma, currant aroma, spicy aroma/¯avour and lowest vegetal character. These wines also had the brighest red colour associated with the red-red scale region. The eects of the M treatment were more pronounced with yeast P compared to yeast L for body characteristic and tropical
Table 3 Analysis of variance of 17 sensory attributes scored for Pinot noir wines Attribute
F-ratio Judge (J)
Appearance Red colour Red hue
Wine (W)
Replicate JW JR (R)
7.82*** 192.44*** 0.59 7.21*** 89.10*** 0.84
WR
1.25 1.18
2.42* 1.13
7.83*** 1.69
Aroma Berry Cherry Tropical fruit Currant Plum Vegetal Spicy
4.69*** 5.17*** 5.38*** 3.28*** 3.27*** 12.4*** 9.17***
0.25 1.44 2.74* 3.04** 1.13 5.33*** 3.75***
0.64 0.67 1.49 0.7 1.19 0 0.7
0.74 0.91 1.07 0.81 0.7 1.48* 1.2
0.7 0.57 0.64 1.6 1.14 1.34 2.82**
0.77 0.81 1.25 0.46 0.66 1.06 0.62
Flavour Fruity Vegetal Spicy
5.13*** 15.34*** 9.34***
1.66 1.58 3.26**
0.61 1.5 0.05
1.12 1.75 0.73
1.48 1.57 1.16
0.64 1.65 0.58
Mouth feel Astringency Body
7.10*** 8.01***
2.21* 1.77 8.13*** 0.02
1.09 1.31
1.08 1.76
2.16 0.76
Taste Acidity Bitter
5.17*** 7.66***
0.9 0.93
0.08 0.62
0.89 0.81
1.13 0.79
1.4 0.99
3.70**
2.06
0.95
0.73
0.54
Overall quality Quality 11.18*** *P<0.05. **P<0.01. ***P<0.001.
487
fruit aroma. Yeast L was the most `red-red' on the hue scale and had highest currant aroma. Wines made from the other treatments (A, T, C) were more similar to one another, the high temperature (T) vini®cation producing wines with the greatest vegetal aromas for yeast P. The low temperature (C) vini®cation with yeast L produced wines characterized with the most `orange-red' in colour. In contrast, yeast P wines had an orange-red hue for both the L and T vini®cations. The eects of vini®cation treatments as well as the relationships among the attributes are shown in a PCA plot [Fig. 2(a)] representing 80% of the variation in the data set with 72 and 8% explained by PC1 and PC2, respectively. An additional 8% was explained by the third PC3 (data not shown). All attributes were loaded in the positive direction on PC1 except vegetal aroma which was found in the negative region of the PC1 axis. Tropical fruit aroma and currant aroma also had high loadings on PC2 in the positive and negative directions, respectively. High positive correlations were observed between spicy aroma and ¯avour (r=0.84) and between red colour and hue (r=0.97). In contrast, a negative correlation was observed between currant aroma and vegetal aroma (r=0.72). The position of the wines on the same PCA plot [Fig. 2(a)] provided a graphic representation of the wines' sensory character which was consistent with the cobweb diagrams [Fig. 1(a) and (b)]. The two wines fermented using A protocol were centrally located. Wines vini®ed using the M fermentation were situated to the right of the plot and were generally associated with a greater body, red colour and spicy character. Wines fermented with yeast P were located high in the plane and characterized by a tropical fruit aroma. Wines with more currant aroma were fermented with yeast L and found in the lower plane. In contrast, wines fermented at a high temperature (T) were located to the left side and were characterized by the presence of a vegetal aroma, a more `orange-red' colour and a weaker contribution in tropical fruit, currant and spicy aromas. Yeast P provided wines with more vegetative character than that fermented with yeast L. In general, wines vini®ed from yeast P were found higher in the plane [Fig. 2(a)] and to the left whereas wines vini®ed from yeast L are located lower in the plane and to the right. Wines from low temperature fermentations (C) were most similar to those subjected to A, particularly those vini®ed with yeast L. The character of the wine fermented at a low temperature with yeast P had slightly more vegetal and tropical fruit aromas, as seen the higher loading of both PC1 and PC2. 3.2. Chemical analysis The PCA plot of the mean chemical parameters (Table 4) for all vini®cation treatments is shown in
488
B. Girard et al. / Food Research International 34 (2001) 483±499
Fig. 1. Mean sensory pro®le of signi®cant attributes of Pinot noir wines from four dierent vini®cations with yeasts (a) P and (b) L.
B. Girard et al. / Food Research International 34 (2001) 483±499
Fig. 2(b). PC1 and PC2 accounted for 39 and 21% of the total variance, respectively. Wines were separated in terms of fermentation methods as well as the type of yeasts. Wines vini®ed using the M fermentations were located to the left on the plot and characterized mostly by colour parameters such as hue, a* (redness), b* (yellowness), chroma, colour density on PC1. Wines fermented with yeast P had high negative loadings for ethanol content and high pH on the PC2. In contrast, wines made from H and C vini®cations using yeast P had high lightness
489
and tint values on PC1. Similarly wines with the same yeast from C vini®cation had high ethanol content on the PC2. Lowest ¯avonols, tartaric esters, total phenols, total acids and antocyanins contents were found in wines vini®ed with yeast P using C and M vini®cations. In contrast, Pinot noir wines fermented at a high temperature (H) as well as those fermented under the M protocol with yeast L had higher values for these parameters in comparison to the other wines. In general, wines vini®ed from yeast L are located higher in the plane
Fig. 2. Principal component analysis of mean (a) sensory and (b) chemical data for Pinot noir wines. P and L refer to yeasts and A, C, M and T refer to vini®cation methods. Only statistically signi®cant attributes were used.
490
Table 4 Chemical composition of Pinot noir wines from four dierent vini®cations with yeasts P (EC-1118) and L (L-2056)a Fermentationb
P L
A
P L
C
P L
M
P L a b
pH
TA (g/L)
Ethanol (vol%)
Colour
L*
a*
b*
Hue
Chroma
Total phenolics (mg gallic acid/l, 280 nm)
Tartaric esters (mg caeic acid/l, 320 nm)
Flavonols (mg quercetin/ l, 360 nm)
Anthocyanins (mg malvidin-3glucoside/l, 520 nm)
Colour density [(A420-A700)+ (A520-A700)]
Tint [(A420-A700)/ (A520-A700)]
3.37 (0.05) 3.35 (0.05)
6.99 (0.09) 7.13 (0.09)
13.04 (0.18) 12.90 (0.26)
91.17 (3.20) 91.49 (0.92)
10.22 (3.11) 10.18 (0.96)
4.10 (1.67) 3.32 (0.04)
2.43 (0.26) 2.95 (0.28)
11.02 (3.51) 10.71 (0.92)
439.56 (19.49) 478.42 (7.17)
84.57 (5.37) 96.51 (5.87)
44.77 (6.99) 56.60 (8.32)
74.95 (11.84) 75.58 (6.04)
2.87 (1.09) 2.78 (0.27)
0.74 (0.02) 0.74 (0.02)
3.27 (0.06) 3.26 (0.07)
7.26 (0.11) 7.38 (0.09)
13.11 (0.22) 13.17 (0.30)
90.63 (1.23) 89.18 (1.43)
10.57 (1.39) 12.12 (1.50)
3.26 (0.43) 3.42 (0.14)
3.20 (0.76) 3.43 (0.31)
11.08 (1.24) 12.59 (1.48)
385.53 (26.73) 367.72 (32.55)
84.17 (5.92) 80.58 (11.61)
47.52 (5.27) 46.88 (10.93)
63.28 (6.27) 61.39 (7.56)
3.02 (0.31) 3.41 (0.44)
0.74 (0.04) 0.70 (0.02)
3.34 (0.02) 3.30 (0.11)
7.13 (0.16) 7.21 (0.16)
13.49 (0.34) 13.39 (0.25)
92.60 (0.78) 90.81 (0.90)
8.19 (0.91) 10.21 (0.99)
3.08 (0.16) 3.94 (0.37)
2.54 (0.43) 2.47 (0.40)
8.76 (0.80) 10.95 (0.92)
331.08 (30.48) 354.97 (28.45)
71.42 (4.49) 78.33 (1.47)
43.08 (3.12) 48.47 (4.60)
61.90 (2.56) 63.90 (6.41)
2.40 (0.19) 2.90 (0.25)
0.79 (0.04) 0.76 (0.01)
3.49 (0.02) 3.59 (0.07)
6.77 (0.06) 6.62 (0.13)
13.52 (0.19) 14.10 (0.50)
90.03 (1.09) 87.02 (1.79)
10.99 (1.15) 13.42 (1.48)
3.44 (0.14) 4.06 (0.40)
3.09 (0.24) 3.21 (0.21)
11.51 (1.13) 14.02 (1.51)
298.29 (22.90) 444.49 (22.69)
61.67 (4.00) 70.43 (6.07)
33.58 (1.58) 49.63 (6.69)
66.16 (3.38) 74.33 (8.85)
3.17 (0.28) 4.00 (0.46)
0.74 (0.01) 0.73 ( 0.01)
Average of three replicates. Standard deviations in parentheses. C, cold temperature (15 C) fermentation; A, ambient temperature (20 C) fermentation; T, high temperature (30 C) fermentation; M, modi®ed-cold-fermentation (15 C).
B. Girard et al. / Food Research International 34 (2001) 483±499
T
Yeast
B. Girard et al. / Food Research International 34 (2001) 483±499
[Fig. 2(b)] and to the left whereas wines vini®ed from yeast P are located lower in the plane and to the right. PLS regression analysis was performed in order to relate sensory and physicochemical data sets. First, PLS was carried out on all the signi®cant sensory and chemical variables for the eight vini®cation treatments. Using the outcome of this analysis, selected chemical and sensory variables (including all colour parameters) from both data sets were chosen for further PLS analysis. Only one factor was validated; this factor explained 78% of the colour measurements (x variables: L, a*, b*, hue, chroma, colour density and tint) and 74% of the sensory attributes (y variables: red colour and red hue). Red hue and red colour (sensory terms) were mostly explained by colour measurements such as hue, colour density, chroma and a* (redness) on the PC1 which was consistent with the ®ndings from PCA (®gure not shown). The correlation between predicted and measured red colour (sensory) and red hue (sensory) using PLS were 0.88 and 0.87, respectively [Fig. 3(a) and (b)]. The prediction of sensory colour attributes was found to be reasonably accurate with the parameters from instrumental colour measurements. The root mean square error of prediction (RMSEP) were 0.97 (red colour) and 0.95 (red hue), respectively, which implies that sensory colour parameters can be instrumentally predicted within approximately one point on the 10-cm line scale used in this experiment. The error of prediction was comparable to that of a trained sensory panel. Fermentation temperature substantially aects red wine style (Amerine, 1955; Amerine & Ough, 1957; Daudt & Ough, 1973; du Plessis, 1983; Gao, Girard, Mazza, & Reynolds, 1997; Ough, 1964, 1966a,b; Ough & Amerine, 1960, 1961, 1967; Ough, Amerine, & Sparks, 1969; Ough & Groat, 1978). Fermentation carried out at elevated temperature in the range of 25±35 C generally increases phenolic extraction of red wines. However, wine colour intensity is not only determined by anthocyanin content but other factors such as the type of anthocyanin and other polyphenolic compounds, pH and co-pigmentation. During the enological process, extracted monomeric anthocyanins tend to polymerize. Fermentation temperature was found to be the critical factor in the generation of polymeric anthocyanins which were directly related to the darkness of the Pinot noir wine (Gao et al.). In this study, the brightest (less dark) red wine was obtained by vini®cation M. The initial fermentation temperature of this treatment was similar to the cold fermentation C at 15 C. Half way through the fermentation, the marc that had initially been frozen was thawed and added to the must. This allowed for an extraction of the phenolic components from the marc. Even though the time period of solid-toliquid contact for extraction was shorter, a desirable colour intensity was obtained. During the freeze-thaw step, the cellular structure of the skin was likely disrupted
491
which would facilitate the extraction and diusion of phenolic compounds and other constituents. Since wines from vini®cation M were brighter, it appeared that polymerization of anthocyanins was more limited. HPLC studies will be required to further investigate these differences. Yeasts dierences were noted for colour, ¯avour and chemical characteristics but these variations were smaller than that noted for vini®cation methods. 3.3. Pattern of volatile constituents All the volatiles analyzed in Pinot noir wines from dierent vini®cation treatments are shown in Table 5. A total of 70 constituents were identi®ed in the Pinot noir wines, including 32 esters (four coeluting), 20 alcohols, ®ve aldehydes, ®ve ketones (two coeluting), ®ve acids, two ethers and one sulphur-containing compound. Alcohols and esters were the largest groups and made up more than 95% of the recovered volatiles. The majority of volatile compounds were identi®ed in all vini®cation methods for two dierent yeasts, and previously reported in Pinot noir wines (Girard et al., 1997; Miranda-Lopez et al., 1992a, b). Both the vini®cation process and yeast strain aected the amount of alcohols, esters and total volatile compounds in wines. Fig. 4(a) and (b) shows the relative concentrations of these compounds for the four vini®cation processes with yeasts P and L in Pinot noir wines. For yeast P, high temperature (T) vini®cation had lower concentrations of esters, alcohols and total volatile compounds. Modi®ed cold (M) vini®cation had highest esters, alcohols and total volatile compounds followed by cold (C) vini®cation and ambient temperature (A) vini®cations. For yeast L, vini®cation process T also had lower concentrations of volatile compounds. The dierences in volatile concentrations among the other three vini®cation methods were not as pronounced. Low temperature fermentation such as treatments C and M elevated ester and alcohol production. These results are consistent with previous ®ndings (Girard et al., 1997). Higher levels of 3-methylbutyl acetate explained the stronger perception of tropical aroma. Other fruity ¯avours and aroma like blackcurrant may be due to esters such as ethyl butanoate as well as monoterpenes such as benzyl alcohol and 2-phenyl ethanol (Latrasse, 1991; Latrasse, Rigaud, & Sarris, 1982). These terpenes along with geraniol, a-ionone, 3-ethoxy ethanol and benzaldehyde can contribute to spicy and cherry ¯avour (Berger, 1991; Kwan & Kowalski, 1980; Kwan, Kowalski, & Skogerboe, 1979; Miranda-Lopez et al., 1992a, b; Schmid & Grosch, 1986). Dierences in grassy vegetal aroma notes were likely associated with variations in hexanol and other C6 compounds. PCA was applied to the gas chromatography data to reveal details of the relationships between volatile compounds as they are aected by vini®cation and yeast
492
B. Girard et al. / Food Research International 34 (2001) 483±499
treatments. All replicate measurements were included in the model in order to conserve the PCA matrix structure. First, PCA was carried out on the vini®cation methods with all volatile compounds. In subsequent iterations, volatile compounds with high noise (low loading values) were removed and 33 volatile compounds were eventually chosen for further analysis. These compounds were labelled with a star symbol in Table 5. The use of 33 volatile compounds improved the explained variation in the data. Seventy percent of the variation was explained
by three PCs. The ®rst two PCs accounted 28 and 21% of the total variance, respectively. Another 16% was explained by PC3. Fig. 5(a) shows the relations between volatile compounds on the ®rst two PCs and Fig. 5(b) shows the position of the wines in relation to the volatile compounds. Wines from C and A fermentations were located in the lower right quadrant approaching the centre; these two groups were mostly segregated although they remained in close proximity to each other. Wines fermented at high temperature (T) were situated
Fig. 3. Measured and predicted sensory (a) red colour intensity and (b) red hue for Pinot noir wines. P and L refer to yeasts and A, C, M and T refer to vini®cation methods.
Table 5 Relative concentrations (x102 mg/L) of volatile constituents in Pinot noir wines from 4 dierent vini®cations with yeasts P (EC-1118) and L (L-2056) A: ambient (20 C)
Compounda
C: cold (15 C)
T: high (30 C)
M: modi®ed (15 C)
L
P
L
P
L
P
L
Esters Ethyl acetate* Ethyl propanoate* Ethyl 2-methylpropanoate* Propyl acetate (& diacetyl) 2-Methylpropyl acetate Ethyl butanoate Ethyl 2-methylbutanoate* Ethyl 3-methylbutanoate (& 2,3-pentandione)* 3-Methyl-1-butyl acetate (& ethyl pentanoate)* Ethyl hexanoate Hexyl acetate* Ethyl 5-hexenoate Ethyl 3-hexenoate Ethyl heptanoate* Ethyl lactate (& ethyl 2-hexenoate)* Heptyl acetate* Methyl octanoate Ethyl octanoate Ethyl 7-octenoate Ethyl 3-hydroxy-butanoate Ethyl decanoate 3-Methylbutyl octanoate* Diethyl succinate* Ethyl dec-9-enoate* 1,3-Propanediol diacetate 2-Phenylethyl acetate Ethyl dodecanoate Diethyl malate*
890 954 952 954 991 1016 1030 1046 1096 1220 1251 1260 1287 1319 1334 1359 1374 1428 1474 1514 1641 1659 1685 1694 1751 1831 1845 2041
4206.8400.8 403.634.7 913.199.7 106.845.5 139.834.5 897.2124.6 280.570.7 462.064.8 2273.4406.1 3204.6546.1 137.821.5 23.414.5 7.42.6 39.89.1 424.139.9 3.10.3 14.81.6 3981.6467.3 3.70.8 4.70.2 905.5148.1 25.81.2 205.933.6 76.813.0 9.32.2 6.85.9 25.25.1 49.99.4
3903.1453.4 339.69.9 947.828.4 65.32.2 211.918.0 837.437.4 249.110.9 458.712.7 2935.6113.8 2712.1357.6 189.210.8 8.42.7 7.90.7 42.46.8 372.810.6 2.60.4 10.71.0 2983.3277.1 3.70.0 4.81.5 707.469.7 20.91.6 171.517.6 87.714.6 6.82.8 8.78.0 20.55.2 39.618.6
3865.4416.6 489.657.7 727.262.4 146.139.0 147.28.9 1264.5102.2 197.41.1 373.022.3 2964.7163.4 3998.3205.8 178.810.2 24.612.7 8.71.6 65.19.5 411.436.3 5.30.3 15.81.3 4079.4191.7 5.51.3 4.31.6 745.054.2 24.22.9 163.417.9 107.817.0 6.03.1 5.00.9 14.02.8 16.411.1
4568.71063.0 421.457.4 831.3178.2 126.153.2 228.659.2 1084.5112.3 217.544.2 369.441.0 3188.5795.4 3239.5324.5 228.359.7 14.21.5 5.31.1 72.011.9 288.415.4 4.71.7 10.21.2 3136.5203.5 7.00.3 5.21.4 730.9104.7 21.54.1 142.230.1 105.915.6 5.90.9 3.70.7 14.65.5 23.77.9
5204.62704.9 381.8239.9 940.3790.8 161.5105.9 106.7127.8 492.7349.5 137.596.3 254.1203.9 888.01105.2 1422.31050.0 36.254.6 5.38.1 5.67.2 17.013.3 364.7413.5 1.72.2 6.65.0 2131.01543.1 4.63.3 3.45.6 733.0481.6 18.911.1 136.4154.6 29.645.2 9.617.5 5.92.3 21.217.7 28.474.8
4481.91559.1 302.062.5 1003.1228.5 98.029.1 165.133.4 436.975.7 154.861.4 274.866.3 1342.0233.5 1221.7221.4 48.85.3 5.92.6 6.61.2 17.13.7 348.256.6 2.40.2 4.80.5 1699.5151.8 3.80.7 4.11.3 498.815.0 13.11.9 131.120.3 41.68.3 10.75.9 4.11.6 18.60.8 47.024.1
4267.01177.2 337.656.9 313.843.0 239.57.5 282.042.9 1355.2202.8 79.25.7 172.213.8 7105.9467.3 3999.3721.6 626.775.3 37.924.4 4.62.8 53.210.0 301.249.6 16.11.1 15.11.2 4640.2230.7 14.75.6 10.15.0 890.660.7 20.32.3 99.613.5 183.718.5 33.110.9 10.710.0 21.13.4 63.828.8
5476.2876.6 301.231.2 475.5319.4 175.144.2 198.117.7 1140.7131.3 81.515.7 143.225.9 2988.11095.1 3692.8374.4 272.6130.5 52.624.0 6.82.9 54.01.2 135.49.2 5.02.1 11.84.8 3214.4607.1 25.912.5 6.11.5 731.681.2 19.09.5 86.07.8 100.624.7 8.34.0 9.65.1 22.33.0 21.37.4
Alcohols 2-Methyl-1-propanol 1-Butanol* 3-Methyl-1-butanol* 1-Pentanol 2-Ethyl 1-butanol 4-Methyl-1-pentanol* 2-Heptanol* 3-Methyl-1-pentanol* 1-Hexanol* (Z)-3-Hexen-1-ol (E)-3-Hexen-1-ol
1082 1114 1202 1229 1292 1301 1306 1313 1343 1351 1370
260.218.5 17.93.4 25789.6888.4 137.538.4 8.04.9 21.77.7 12.66.5 40.17.7 1297.844.7 48.67.2 11.91.4
287.438.2 10.01.6 26609.1779.0 21.013.8 5.71.4 21.96.2 16.03.2 32.95.0 1419.286.7 55.31.7 12.81.8
214.018.0 15.11.9 23910.9880.1 82.056.3 10.11.8 19.73.2 9.72.6 44.35.6 1693.8136.5 58.62.0 13.50.7
418.0251.1 14.43.6 24867.7487.5 127.222.6 3.20.4 18.96.3 13.85.4 34.95.1 1538.4211.6 53.39.3 10.70.9
321.1162.5 34.013.0 18611.616497.3 49.455.4 4.24.6 10.57.7 9.75.8 12.37.4 984.9896.1 37.731.5 5.65.2
241.685.8 17.44.6 19877.23054.4 53.71.6 4.00.5 7.41.3 6.71.6 8.91.4 928.835.5 37.45.1 5.70.6
164.982.2 28.58.1 20046.31566.5 47.93.8 15.79.0 12.65.0 15.210.7 19.45.6 1883.9222.6 60.34.2 16.30.5
193.538.0 16.25.4 19224.61480.2 102.060.1 12.75.9 10.23.3 20.28.0 19.34.0 2256.8397.1 73.221.9 13.73.2 493
(continued on next page)
B. Girard et al. / Food Research International 34 (2001) 483±499
RIwaxb P
494
Table 5 (continued) A: ambient (20 C)
Compounda
C: cold (15 C)
T: high (30 C)
M: modi®ed (15 C)
L
P
L
P
L
P
L
1-Heptanol* threo-2,3-Butanediol meso-2,3-butanediol 1,2-Propanediol Geraniol* Benzyl alcohol* 2-Phenyl ethanol*
1442 1551 1586 1603 1862 1889 1924
45.46.2 4648.1741.7 1604.6274.1 324.048.0 9.44.8 7.10.5 537.9130.1
34.95.0 3916.61729.4 1431.3622.0 331.2159.2 17.92.0 6.91.5 477.8118.5
75.89.6 1417.91181.6 1133.7989.9 85.893.1 7.23.1 5.31.0 267.638.8
51.06.3 3423.2884.0 1270.0215.5 284.429.7 9.04.6 5.30.4 285.384.2
16.411.9 3110.210106.3 1067.83775.3 223.2963.7 6.03.5 8.216.5 230.5492.8
14.22.0 6225.33387.4 2265.71315.2 543.9365.4 7.44.0 10.85.0 390.090.4
137.356.3 6908.91822.8 2556.7678.3 605.1169.3 10.22.4 9.80.8 252.854.8
54.122.9 6209.63278.0 2255.51326.1 535.3318.8 12.72.5 8.32.5 193.427.3
Aldehydes Acetaldehyde Nonanal 2-Furanaldehyde Decanal Benzaldehyde (& ethyl nonanoate)
674 1380 1458 1490 1523
123.818.5 8.85.2 5.51.3 19.29.7 31.35.9
97.131.7 15.611.1 9.12.8 22.59.5 44.58.7
102.446.2 16.30.8 8.31.5 18.12.5 35.93.1
141.842.8 17.17.7 5.91.2 35.920.5 50.420.5
69.773.4 21.310.9 12.56.9 24.815.5 29.725.4
83.313.9 10.72.1 6.41.6 19.44.1 24.21.4
87.24.3 9.15.2 9.61.1 25.216.9 55.216.8
83.727.9 9.01.0 9.24.0 31.417.6 50.415.9
Ketones 3-Hydroxy-2-butanone* 6-Methyl-5-hepten-2-one* a-Ionone* Dihydro-2(3H)-furanone*
1273 1325 1520 1652
51.825.2 5.94.7 2.51.9 92.38.3
49.837.1 11.12.1 4.40.1 70.518.5
28.34.6 6.20.8 3.20.7 52.37.2
54.38.0 7.72.2 3.70.3 62.06.1
18.526.4 4.72.5 2.93.1 75.2160.7
18.17.2 5.53.1 3.20.7 101.651.6
57.220.5 5.11.1 1.81.1 68.021.0
101.321.5 6.10.8 1.30.3 76.232.8
Acids Acetic acid (and 2-methylbutyl hexanoate) 2-Methylpropanoic acid* Butanoic acid Hexanoic acid* Octanoic acid*
1447 1571 1679 1856 2052
151.44.6 74.716.4 34.78.1 31.63.0 24.22.7
240.933.5 73.013.7 28.19.6 28.96.6 21.04.5
107.625.3 36.915.4 19.44.5 31.67.6 16.44.4
187.036.3 66.510.4 78.187.4 26.58.9 14.05.9
135.9205.6 69.8142.0 20.040.2 14.022.9 10.715.4
208.937.2 97.439.6 27.511.2 16.75.4 12.72.5
256.251.4 64.414.8 14.22.7 62.220.4 30.78.4
250.4115.3 63.728.8 17.54.3 40.56.5 13.72.2
Ethers Ethoxy-ethene 3-Ethoxy-1-propanol*
659 1364
27.00.0 27.14.0
27.00.0 5.10.8
27.00.0 36.59.2
27.00.0 7.61.5
27.027.0 25.45.8
27.00.0 5.32.3
27.00.0 106.425.9
27.00.0 14.24.8
Sulfur 3-(Methylthio)-1-propanol*
1730
20.74.4
14.76.7
7.02.5
7.94.1
8.318.7
14.73.9
5.30.7
3.70.6
a b
Selected volatiles for PCA and PCS analyses are marked with an asterisk (*). Retention index relative to a column with a polyethyene glycol phase (Supelcowax 10).
B. Girard et al. / Food Research International 34 (2001) 483±499
RIwaxb P
B. Girard et al. / Food Research International 34 (2001) 483±499
at the upper right quadrant of the plot and were characterized by the increased concentrations of several esters, several linear and branched alcohols and a sulfur compound. Wines vini®ed using the M fermentation were located on the left side of the graph. This was the only fermentation treatment where a segregation between yeast strain emerged. These dierences were based on variations in concentration of several linear alcohols, several acetates, one acid and one aromatic alcohol.
495
The applications of PCS analysis to the 33 selected volatile compounds con®rmed the results obtained by PCA and showed a similar segregation pattern for the Pinot noir wines (Fig. 6). PCS allowed a more comprehensive interpretation by including several additional contributing principal components (e.g. six PCs explaining 86% of the variation in the data) in a two dimensional format. This was accomplished by comparing PCs of each sample to a ``reference'' via linear regression and
Fig. 4. In¯uence of vini®cation methods on the classes of volatile aroma compounds in Pinot noir wines with yeasts (a) P and (b) L. Vini®cation methods within volatile classes with dierent letters are signi®cantly dierent at P<0.05.
496
B. Girard et al. / Food Research International 34 (2001) 483±499
plotting the resultant slope (deviations) against the coecient of determination (similarities/dissimilarities). Means of individual volatile compounds from the Pinot noir wine vini®ed using ambient temperature (A) with yeast P was designated as a reference (APavg) since these conditions correspond to a relatively common practice in the wine industry. Fig. 6 shows that there were at least four distinct groups of wines. The vini®cation treatments T, A, C and M were well separated. Fermentations C and A were adjacent to each other. Furthermore, the
yeasts dierences between L and P were also evident for the M vini®cation. According to the literature, the greatest dierences in production of aroma compounds correspond to dierent yeast species, while strains of the same species tend not to dier signi®cantly. Our results on fermentation conditions such as low temperature and extent of solids contact time which in¯uenced aroma volatile production by yeasts L and P corroborate with previous ®ndings (Antonelli, Castellari, Zambonelli, & Carnacini, 1999; Cabrera, Moreno, Ortega, &
Fig. 5. Principal component (a) loadings and (b) scores of selected volatile constituents for Pinot noir wines. The symbols correspond to volatile classes: &, esters; *, alcohols; ^, ketones; , acids; ?, ether. P and L refer to yeasts; A, C, M and T refer to vini®cation methods; 1, 2, 3, refer to replicates.
B. Girard et al. / Food Research International 34 (2001) 483±499
497
Fig. 6. Principal component similarity analysis of selected volatile constituents for Pinot noir wines. P and L refer to yeasts; A, C, M and T refer to vini®cation methods; 1, 2, 3, refer to replicates.
Medina, 1988; Daudt & Ough, 1973; Klingshirn, Liu, & Gallander, 1987). 4. Conclusion Dierences between Pinot noir wines vini®ed with distinct yeasts using several vini®cation treatments were characterized by sensory and physicochemical analyses. This information on dierent vini®cation processes based on temperature and yeast parameters documents and extends some of the latitude that winemakers can aord for fermenting and formulating wines of targeted styles. In general, wines vini®ed using a modi®ed cold temperature (M) vini®cation with yeast P had a brighter red colour (red-red), strong fruity (currant and tropical fruit) and spicy aroma while wines from high temperature fermentation with yeast P had strong vegetal aroma and a darker red colour (orange-red). The dierences between wines from vini®cations using yeast L were less pronounced. The main dierences between the two yeasts were notable for M and T vini®cations. Wines from M vini®cation with yeast P had higher tropical fruit aroma in comparison to wines made with yeast L of the same vini®cation. Furthermore, wines from T vini®cation with yeast P had more vegetal aroma in comparison to the wines with yeast L of the same vini®cation. Visual sensory descriptors (red colour intensity and red hue) were predicted from instrumental and chemical colour measurement with high correlations and relatively low RMSEP values. GC/MS analysis showed that wines from M vini®cation with yeast P had
higher concentrations of esters and total volatile compounds as compared to the other treatments. The PCA and PCS analyses with 33 volatile constituents classi®ed wines according to vini®cation method but yeast dierences were noted only for M vini®cation. PLS regression and PCS analysis were complementary and useful in studying complex relationships between sensory and chemical data and to classify wines on the similarity patterns of volatile compounds, respectively. Acknowledgements The authors wish to thank Tom G. Kopp and Virginie Dimier-Vallet for their technical assistance.
References Amerine, M. A. (1955). Further studies on controlled fermentations. American Journal of Enology, 6, 1±16. Amerine, M. A., & Ough, C. S. (1957). Studies on controlled fermentations. III. American Journal of Enology, 8, 18±30. Amerine, M. A., & Roessler, E. B. (1980). Wines: their sensory evaluation. New York: W.H. Freeman. Antonelli, A., Castellari, L., Zambonelli, C., & Carnacini, A. (1999). Yeast in¯uence on volatile composition of wines. Journal of Agricultural and Food Chemistry, 47, 1139±1144. Berger, R. G. (1991). Fruits I. In H. Maarse, Volatile Compounds in Foods and Beverages (pp. 283±304). New York: Marcel Dekker. Brander, F.C., Kepner, R.E., & Webb, A.D. (1980). Identi®cation of some volatile compounds of wine in Vitis Vinifera cultivar Pinot noir. American Journal of Enology and Viticulture, 31, 69±75. Cabrera, M. J., Moreno, J., Ortega, J. M., & Medina, M. (1988).
498
B. Girard et al. / Food Research International 34 (2001) 483±499
Formation of ethanol, higher alcohols, esters, and terpenes by ®ve yeast strains in musts from Pedro XimeÂnez grapes in various degrees of ripeness. American Journal of Enology and Viticulture, 39(4), 283± 287. Cli, M., & Wild, J. (1990). Development of SAS programs for sensory graphics. Journal of Sensory Studies, 4, 249±260. Cli, M. A., & Dever, M. C. (1996). Sensory and compositional pro®les of British Columbia Chardonnay and Pinot noir wines. Food Research International, 29, 317±323. Daudt, C. E., & Ough, C. S. (1973). Variations in some volatile acetate esters formed during grape juice fermentation. Eects of fermentation temperature, SO2, yeast strain, and grape variety. American Journal of Enology and Viticulture, 24, 130±135. du Plessis, C. S. (1983). In¯uence de la temperature d'eÂlaboration et de conservation. Bulletin de l'Oce International de la Vigne et Vin, 624, 104±115. Furtula, V., Nakai, S., Amantea, G. F., & Laleye, L. (1994a). Reverse phase HPLC analysis of cheese samples aged by a fast ripening process. Journal of Food Science, 59, 528±532, 567. Furtula, V., Nakai, S., Amantea, G. F., & Laleye, L. (1994b). Reverse phase HPLC analysis of reference cheddar cheese samples for assessing accelerated cheese ripening. Journal of Food Science, 59, 533±538. Gao, L., Girard, B., Mazza, G., & Reynolds, A. G. (1997). Changes in anthocyanins and color characteristics of Pinot noir wines during dierent vini®cation processes. Journal of Agricultural and Food Chemistry, 45, 2003±2008. Girard, B. (1996). Retention index calculation using KovaÂts constant model for linear temperature-programmed gas chromatography. Journal of Chromatography, A721, 279±288. Girard, B., Kopp, G. T., Reynolds, A. G., & Cli, M. (1997). In¯uence of vini®cation treatments on aroma constituents and sensory descriptors of pinot noir wines. American Journal of Enology and Viticulture, 48, 198±206. Glories, Y. (1979). Recherches sur la matieÂre colorante des vins rouges. Bulletin de la SocieÂte Chimique de France, 9, 2649±2655. Guinard, J. X., & Cli, M. (1987). Descriptive analysis of Pinot noir wines from Carneros, Napa, and Sonoma. American Journal of Enology and Viticulture, 39, 223±227. Horimoto, Y., Lee, K., & Nakai, S. (1997). Classi®cation of microbial defects in milk using a dynamic headspace gas chromatograph and computer-aided data processing. 1. Principal Component Similarity Analysis. Journal of Agricultural and Food Chemistry, 45, 733±742. Klingshirn, L., Liu, J., & Gallander, J. (1987). Higher alcohol formation in wines as related to the particle size pro®les of juice insoluble solids. American Journal of Enology and Viticulture, 38, 207±209. Kwan, W., & Kowalski, B. R. (1980). Pattern recognition analysis of gas chromatographic data. Geographic classi®cation of wines of Vitis vinifera cv. Pinot noir from France and the United States. Journal of Agricultural and Food Chemistry, 28, 356±359. Kwan, W., Kowalski, B. R., & Skogerboe, B. (1979). Pattern recognition analysis of elemental data. Geographic classi®cation of wines of Vitis vinifera cv. Pinot noir from France and the United States. Journal of Agricultural and Food Chemistry, 27, 1321±1326. Latrasse, A. (1991). Fruits III. In H. Maarse, Volatile compounds in Foods and Beverages (pp. 329±387). New York: Marcel Dekker. Latrasse, A., Rigaud, J., & Sarris, J. (1982). L'aroÃme du cassis (Ribes nigrum L.). Odeur principale et notes secondaires. Science de l'Alimentation, 2, 145±162. Lee, S.-Y., Luna-GuzmaÂn, I., Chang, S., Barrett, D. M., & Guinard, J. X. (1999). Relating descriptive analysis and instrumental texture data of processed diced tomatoes. Food Quality and Preference, 10, 447±455. Martens, M. (1985). Sensory and chemical quality criteria for white cabbage studied by multivariate data analysis. Lebensmittel Wissenschaft und Technologie, 18, 100±104.
Martens, M. (1986). Sensory and chemical/physical quality criteria of frozen peas studied by multivariate analysis. Journal of Food Science, 51, 599±603. Martens, M., & Martens, H. (1986). Partial least squares regression. In J. R. Piggott, Statistical procedures in food research (pp. 293± 359). London, UK: Elsevier Applied Science. Miranda-Lopez, R., Libbey, L. M., Watson, B. T., & McDaniel, M. R. (1992a). Identi®cation of additional odor-active compounds in Pinot noir wines. American Journal of Enology and Viticulture, 43, 90±92. Miranda-Lopez, R., Libbey, L. M., Watson, B. T., & McDaniel, M. R. (1992b). Odour analysis of Pinot noir wines from grapes of dierent maturities by gas chromatography-olfactometry technique (Osme). Journal of Food Science, 57, 985±993. Nakai, S. (1992). Importance of protein functionality in improving food quality: role of hydrophobic interaction. Comments on Agricultural and Food Chemistry, 2, 339±387. Ough, C. S. (1964). Fermentation rate of grape juice. I. Eects of temperature and composition on white juice fermentation rates. American Journal of Enology and Viticulture, 15, 167±177. Ough, C. S. (1966a). Fermentation rates of grape juice. II. Eect of initial Brix, pH, and fermentation temperature. American Journal of Enology and Viticulture, 17, 20±25. Ough, C. S. (1966b). Fermentation rates of grape juice. III. Eects of initial ethyl alcohol, pH, and fermentation temperature. American Journal of Enology and Viticulture, 17, 74±81. Ough, C. S., & Amerine, M. A. (1960). Studies on controlled fermentation. V. Eects on color, composition, and quality of red wines. American Journal of Enology and Viticulture, 12, 9±19. Ough, C. S., & Amerine, M. A. (1961). Studies with controlled fermentation. VI. Eects of temperature and handling on rates, composition, and quality of wines. American Journal of Enology and Viticulture, 12, 117±128. Ough, C. S., & Amerine, M. A. (1967). Studies with controlled fermentation. X. Eect of fermentation temperature on some volatile compounds in wine. American Journal of Enology and Viticulture, 18, 157±164. Ough, C. S., & Groat, M. (1978). Particle nature, yeast strain, and temperature interactions on the fermentation rates of grape juice. Applied Environmental Microbiology, 35, 881±885. Ough, C. S., Amerine, M. A., & Sparks, T. C. (1969). Studies with controlled fermentations. XI. Fermentation temperature eects on acidity and pH. American Journal of Enology and Viticulture, 20, 127±139. Price, S. F., Lombard, P. B., & Watson, B. T. (1988). Pinor noir clones and their eects on cluster morphology and grape composition. In R. E. Smart, R. J. Thornton, S. B. Rodriquez, & J. E. Young, Proceedings of Second International Symposium on Cool Climate Viticulture and Enology (pp. 279±281). Auckland NZ: New Zealand Society for Viticulture and Oenology. Price, S. F., Breen, P. J., Valladao, M., & Watson, B. T. (1995). Cluster sun exposure and quercetin in Pinot noir grapes and wine. American Journal of Enology and Viticulture, 46, 187±194. Reynolds, A. G., & Naylor, A. P. (1994). `Pinot noit' and `Riesling' grapevines respond to water stress duration and soil water-holding capacity. Horticultural Science, 29, 1505±1510. Reynolds, A. G., Price, S. F., Wardle, D. A., & Watson, B. T. (1994). Fruit environment and crop level eects on Pinot noir. I. Vine performance and fruit composition in British Columbia. American Journal of Enology and Viticulture, 45, 452±459. Reynolds, A. G., Yerle, S., Watson, B., Price, S. F., & Wardle, D. A. (1996). Fruit environment and crop level eects on Pinot noir. III. Composition and descriptive analysis of Oregon and British Columbia wines. American Journal of Enology and Viticulture, 47, 329±339. Robinson, J. (1986). Vines, grapes and wines. New York: Alfred A. Knopf.
B. Girard et al. / Food Research International 34 (2001) 483±499 Roessler, E. B., Warren, J., & Guymon, J. F. (1948). Signi®cance in triangle taste tests. Food Research, 13, 503±505. Romani, A., Mancini, P., Tatti, S., & Vincieri, F. F. (1996). Polyphenols and polysaccharides in Tuscan grapes and wines. Italian Journal of Food Science, 1, 13±24. Schmid, W., & Grosch, W. (1986). Quantitative Analyse ¯uchtiger Aromastoe mit hohen Aromawerten in Sauerkirschen (Prunus cerasus L.), Susskirschen (Prunus avium L.) und Kirschkon®turen. Zeitschrift fuer Lebensmittel Untersuchung und-Forschung, 183, 39± 44. Vodovotz, Y., Arteaga, G. E., & Nakai, S. (1993). Principal compo-
499
nent similarity analysis for classi®cation and its application to GC data of mango. Food Research International, 26, 355±363. Watson, B., Lombard, P.Price S., McDaniel, M., & Heatherbell, D. (1988). Evaluation of Pinot noir clones in Oregon. In R. E. Smart, R. J. Rodriquez, S. B. Rodriquez, & J. E. Young, Proceedings of Second International Symposium on Cool Climate Viticulture and Enology (pp. 276±278). Auckland NZ: New Zealand Society for Viticulture and Oenology. Wrolstad, R. E. (1976). Color and pigment analyses in fruit products. Agricultural Experiment Station Bulletin 624. Corvalis, OR: Oregon State University.