Article pubs.acs.org/JAFC
Influence of Fermentation Temperature, Yeast Strain, and Grape Juice on the Aroma Chemistry and Sensory Profile of Sauvignon Blanc Wines Rebecca C. Deed,*,†,‡ Bruno Fedrizzi,† and Richard C. Gardner‡ †
School of Chemical Sciences, University of Auckland, Auckland, New Zealand School of Biological Sciences, University of Auckland, Auckland, New Zealand
‡
ABSTRACT: Sauvignon blanc wine, balanced by herbaceous and tropical aromas, is fermented at low temperatures (10−15 °C). Anecdotal accounts from winemakers suggest that cold fermentations produce and retain more “fruity” aroma compounds; nonetheless, studies have not confirmed why low temperatures are optimal for Sauvignon blanc. Thirty-two aroma compounds were quantitated from two Marlborough Sauvignon blanc juices fermented at 12.5 and 25 °C, using Saccharomyces cerevisiae strains EC1118, L-1528, M2, and X5. Fourteen compounds were responsible for driving differences in aroma chemistry. The 12.5 °C-fermented wines had lower 3-mercaptohexan-1-ol (3MH) and higher alcohols but increased fruity acetate esters. However, a sensory panel did not find a significant difference between fruitiness in 75% of wine pairs based on fermentation temperature, in spite of chemical differences. For wine pairs with significant differences (25%), the 25 °C-fermented wines were fruitier than the 12.5 °C-fermented wines, with high fruitiness associated with 3MH. We propose that the benefits of low fermentation temperatures are not derived from increased fruitiness but a better balance between fruitiness and greenness. Even so, since 75% of wines showed no significant difference, higher fermentation temperatures could be utilized without detriment, lowering costs for the wine industry. KEYWORDS: aroma, fermentation temperature, Sauvignon blanc, Saccharomyces cerevisiae, sensory evaluation
■
INTRODUCTION Aroma and flavor drive the quality and popularity of Sauvignon blanc wine, with the best new world examples characterized by the balance between tropical, fruity, and passionfruit aromas and fresh, ripe, and herbaceous green characters.1 Low fermentation temperatures (10−15 °C) are commonly used by winemakers to produce white wines and are standard practice for producing Sauvignon blanc,2 primarily because low temperatures are believed to increase the production and retention of desirable secondary metabolites, such as fruity acetate esters, while decreasing undesirable compounds, such as higher alcohols imparting solvent-like characters.3−6 Since temperature has a significant effect on wine aroma, it is important to consider the role of low fermentation temperatures in producing Sauvignon blanc wines since quality for this variety is partially dependent on the balance between fruity and green aromas. All Saccharomyces cerevisiae-derived or -modified aroma compounds are influenced by fermentation temperature,7 including key odorants contributing to Sauvignon blanc varietal character, such as polyfunctional mercaptans (volatile thiols) 3mercaptohexan-1-ol (3MH), 3-mercaptohexyl acetate (3MHA), and 4-mercapto-4-methylpentan-2-one (4MMP), imparting intense tropical and musky grapefruit/passionfruit skin, passionfruit pulp, and cat urine characters.8 Other yeast-derived compounds of importance for the balance between fruity and green notes in Sauvignon blanc include fruity acetate and ethyl esters, higher alcohols (particularly C6-alcohols, contributing fresh apple and cut grass characters),9 fatty acids, and volatile sulfur compounds (VSCs) such as dimethyl sulfide, responsible for asparagus aromas.10 More positive green aromas, © XXXX American Chemical Society
including red capsicum and snow pea, are derived from alkylmethoxypyrazine compounds in the grape,11 which are not thought to be influenced by fermentation temperature.12 However, the known effect of temperature on the production of 3MH13 and the interaction between 3MH and alkylmethoxypyrazines14 could affect the perception of positive green characters (fresh herbaceous) versus negative green characters (under-ripe green), which occur when the concentration of alkyl-methoxypyrazines is very high. Achieving the right balance between fruity and green characters is further complicated by the enhancing and masking effect between different compounds. For example, fruity esters produced by yeast at low temperatures can have a masking effect on Sauvignon blanc varietal character.15 Additionally, alkylmethoxypyrazines can mask the tropical passionfruit aroma of 3MHA.16 The fruity and green sensory descriptors associated with 3MH and 2-isobutyl-3-methoxypyrazine (IBMP) shift to higher or lower intensities depending on their concentrations.14 Furthermore, the S-enantiomer of 3MH can impart positive green characters to wines, such as tomato leaf, alongside the fruity notes of the R-enantiomer.17 It is known that low temperatures are crucial for retaining varietal characters after fermentation,18 but the importance of maintaining low temperatures throughout fermentation to produce the characteristic new world style of Sauvignon blanc Received: Revised: Accepted: Published: A
July 13, 2017 September 15, 2017 September 19, 2017 September 19, 2017 DOI: 10.1021/acs.jafc.7b03229 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry
sensory analysis. Wines were filtered to remove solids and yeast lees through a 3 mm2 metal filter, followed by Whatman 3MM Chr filter paper cones. Argon gas was used throughout this process to minimize oxidation. Wines were stored in polypropylene 70 mL plastic specimen containers at −20 °C for WineScan and aroma quantitation or used immediately for sensory analysis. WineScan Analysis. A WineScan FT 120 spectrometer (Foss, Denmark), with Fourier Transform InfraRed (FTIR) spectrometry capability, was used to assess a range of critical quality control parameters from wine samples via scanning across the full infrared spectrum. Quantitative values were obtained for ethanol, residual sugar, pH, total acidity, volatile acidity, malic acid, citric acid, and lactic acid. Volatile Aroma Compound Analysis. Volatile thiols 3MH, 3MHA, and 4MMP were extracted from pooled wines in duplicate using the method developed by Tominaga et al.19 with specific modifications detailed in Patel et al.20 Esters, higher alcohols, and fatty acids were extracted using liquid/liquid extraction with 1:1 diethyl ether/n-hexane.21 Gas chromatography−mass spectrometry (GC-MS) analysis was used to quantitate all volatile compounds using the peak area of ions corresponding to each compound and internal standard using MSD ChemStation. Concentrations were determined using a calibration curve constructed with model wine (water containing 12% v/ v ethanol and 5 g L−1 L-tartaric acid, pH 3.2) spiked with increasing concentrations of the compounds analyzed. Sensory Analysis. A panel consisting of 11 University of Auckland Wine Science Postgraduate Diploma students (five females and six males) was trained in four 2 h sessions via a ranking test to quantitate the fruity and green characters in Sauvignon blanc wines. All students were familiar with a range of Sauvignon blanc wines as they had undergone previous sensory training. The use of participants was approved by the University of Auckland Human Participant Ethics Committee (UAHPEC) (reference 2009/444). Eight starting reference standards22 were mixed together to produce one core fruity standard and one core green standard for training. The fruity standard was made from four fruit-driven standards and the green by combining four green-driven standards (Table 1). Once these core standards were produced, they
has not been investigated. The aim of this study was to determine whether low temperatures throughout fermentation are required to achieve an optimal balance between fruity and green varietal characters in Sauvignon blanc and whether these sensorial parameters can be correlated to the chemical data to pinpoint any compounds associated with high fruitiness/low greenness.
■
MATERIALS AND METHODS
Chemicals. Aroma volatiles were quantitated using calibration curves generated with pure chemical standards. The following compounds were purchased from Sigma-Aldrich-Fluka: diethyl malate, 3-ethoxy-1-propanol, ethyl-2-furoate, ethyl hexanoate, ethyl isobutanoate, ethyl isovalerate, ethyl 2-methylbutanoate, isobutyl acetate, 2phenylethyl acetate, methionol, 3-methyl-2-buten-1-ol, and (E)-3hexenol. Acros Organics was the source of diethyl succinate, ethyl butanoate, ethyl decanoate, ethyl lactate, ethyl octanoate, hexanoic acid, 1-hexanol, hexyl acetate, 2-methoxy-3-isobutylpyrazine, octanoic acid, phenylethanol, and decanoic acid. Benzyl alcohol was obtained from AppliChem; 3MH and 4MMP from Interchim; and 3MHA from Oxford Chemicals. (Z)-3-Hexenol and DL-3-octanol were purchased from Lancaster; (Z)-3-hexenol acetate from Alfa Aestar; isoamyl acetate from Ajax Chemicals; isoamyl alcohol from Panreac Quimica; and isobutanol from Scharlau. Other chemicals used in this study include absolute ethanol (Ajax Finechem); bacteriological agar, bentonite, butylated hydroxyanisole, L-cysteine hydrochloride hydrate, dimethyl dicarbonate, DOWEX Cl−-form resin, 4-hydroxymercuribenzoic acid sodium salt, peptone, L-tartaric acid, and yeast extract (all Sigma-Aldrich-Fluka); diethyl ether and sodium sulfate anhydrous (both Scharlau); dichloromethane (Merck); D-glucose (Merck); n-hexane (Burdick & Jackson); potassium metabisulfite (AK Scientific); and tris(hydroxymethyl)aminomethane (Applichem). Yeast Strains. Four S. cerevisiae strains were used in this study. Commercial wine yeast strains Lalvin EC1118, used globally at a wide range of temperatures, and Zymaflore X5, used to produce white wines with intense varietal character, were selected because they are commonly used for Marlborough Sauvignon blanc. Enoferm M2, used widely for red and white wine fermentations, was selected based on previous work with this yeast at low fermentation temperatures. One Chilean winery isolate, L-1528, was chosen due to prior screening of fermentative ability at low temperature. Culture Medium, Fermentation Conditions, and Postfermentation Processing of Wines. Fermentations were conducted using two different machine-harvested Sauvignon blanc grape juices from the 2008 vintage: BLANW (19.5 °Brix, total acidity 7.1 g L−1, pH 3.16, and malic acid 4 g L−1) (Saint Clair Family Estate Wines New Zealand) and M1016 (22.2 °Brix, total acidity 6 g L−1, pH 3.28, and malic acid 3.7 g L−1) (Pernod Ricard New Zealand). The yeast assimilable nitrogen (YAN) in the M1016 juice was 190 mg L−1 but was not measured in the BLANW juice. Juices were obtained from vineyards located in the Wairau Valley, Marlborough, New Zealand and collected in 2 L plastic containers after potassium metabisulfite (PMS) addition (approximately 50 ppm), pressing, and cold settling. Juices were frozen at −20 °C at the stage where they would be inoculated by the winemaker and transported to Auckland. Juice was thawed and sterilized via overnight incubation at 25 °C with 200 μL L−1 dimethyl dicarbonate. Aliquots of 10, 50, and 100 μL of uninoculated juice were plated onto yeast− peptone−dextrose medium (YPD) to confirm successful sterilization. Yeast starter cultures were propagated in liquid YPD and incubated overnight at 28 °C, with orbital shaking at 150 rpm. Fermentations were inoculated with 1 × 106 cells mL−1 in triplicate in 750 mL wine bottles containing 700 mL of juice and fitted with airlocks. Fermentations were kept in temperature-controlled rooms and monitored daily via weight loss, representing CO2 release. Additions of 50 ppm PMS to prevent oxidation and 500 mg L−1 bentonite for protein fining were made to finished wines before bottling with Stelvin aluminum closures under CO2 gas. A 1 mL aliquot of wine from each sample was diluted and plated to ensure that only yeast colonies were present at the end of fermentation. Wines were cold-settled and stored at 4 °C for 2 weeks. Biological triplicates were pooled to allow for the volumes required for
Table 1. Sauvignon Blanc Sensory Reference Standards Used for Panel Training reference
standard
fruity
sweet sweaty passionfruit tropical
stone fruit
apple green
fresh asparagus canned asparagus capsicum grassy
components (375 mL final volume) 2000 ng L−1 3-mercaptohexyl acetatea 71.43 mL L−1 Golden Circle Mango juice, 71.43 mL L−1 Golden Circle Golden Pash juice, 357.14 mL L−1 Just Juice Mandarinb 500 mL L−1 juice from Watties canned apricots and peaches in clear juiceb 187 g L−1 “Sciros” Pacific Rose apple peeled and soaked in diluted base wine for 30 minb 500 mL L−1 steamed asparagus water from one bunchb 26.7 mL L−1 juice from Watties canned asparagusb 1000 ng L−1 2-methoxy-3-isobutylpyrazinea 28.8 μg L−1 (Z)-3-hexenola
a
Added to diluted base wine (50% Corbans White Label Sauvignon blanc 2009 and 50% water). bAdded equal parts to base wine (Corbans White Label Sauvignon blanc 2009). were mixed together at different ratios to produce five standards for training: 100%:0% fruity/green, 75%:25% fruity/green, 50%:50% fruity/green, 25%:75% fruity/green, and 0%:100% fruity/green (Table 1). Standards (50 mL) were presented blind in transparent XL5 glasses with random three-digit codes. Each panelist determined descriptors for the five standards in a written questionnaire (Table 2). The third standard (50%:50% fruity/green) was formally deemed by the panel B
DOI: 10.1021/acs.jafc.7b03229 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry
Data Analysis. Analysis of variance (ANOVA) coupled with Tukey’s HSD was computed to determine significant differences between samples for each aroma compound at p < 0.05. Significance letters were generated using online software by Assaad et al.23 In order to understand the relationships among temperature, juice, and yeast strain in terms of wine aroma chemistry, permutational multivariate analysis of variance (PERMANOVA) was conducted across all wines using R software (https://www.r-project.org/). Model construction and visualization via principal component analysis (PCA) was conducted using the R-package vegan24 using a modification of the R script indicated in Parish et al.25 Student’s t-test was used to determine significant differences between 12.5 and 25 °C pairs for sensory analysis. Partial least-squares regression (PLS) was used to correlate aroma data with fruitiness values using the R package plsdepot.26
Table 2. Descriptors for Five Sauvignon Blanc Sensory Reference Standards Generated by Panelists (n = 11) reference 100%:0% fruity/ green 75%:25% fruity/ green 50%:50% fruity/ green 25%:75% fruity/ green 0:1 fruity/ green
descriptors apricot, fruity, kiwifruit, mango, passionfruit, red capsicum, slightly grassy, socks, sweaty, tropical apple, asparagus, capsicum, fruity, grapefruit, grassy, mango, passionfruit, peach, red guava, slightly green, sweaty, tinned peas, tropical asparagus, capsicum, cucumber, fruity, grapefruit, grassy, green, mango, passionfruit, peach, pickles, “Sauvignon blanc-like”, tropical asparagus, canned pea, grapefruit, grassy, green, green capsicum, passionfruit, slightly fruity, vegetal
■
asparagus, canned pea, grassy, green, slightly fruity
RESULTS AND DISCUSSION Fermentation Progress and WineScan Quality Parameters. As anticipated, the single largest effect of low temperature was to decrease the speed of fermentation. Fermentations at 12.5 °C required at least 35 days to finish, with some fermentations becoming sluggish and still releasing CO2 after 70 days. Fermentations at 25 °C required 14−31 days to stop releasing CO 2 (Table 3). For all yeast strains except EC1118, fermentations at 12.5 °C had higher residual sugar and subsequently lower ethanol than those fermented at 25 °C (Table 3). The presence of residual sugar in wines fermented by L-1528, M2, and X5 at both temperatures suggests that these fermentations became stuck or sluggish. Causes of problem fermentations can be difficult to identify. Clearly, EC1118 did not have the same problems as the other yeast strains, suggesting that there could be differences in YAN requirements, fructose utilization, or tolerance to stress, ethanol, SO2, or other inhibitors.27 Lower temperatures only served to enhance these stresses, as evident for X5. Because of these issues, fermentation completion was included as an essential variable when comparing aroma chemistry between samples.
(blind test with repetition) to represent the classic Marlborough Sauvignon blanc style. The panel was asked to order the standards on a ranked scale from most fruity to least fruity based on their aroma. Training ceased when panelists were able to place the standards in the correct order of fruitiness three consecutive times with a Friedman (F) statistic greater than the critical value for α = 0.05. For sample analysis, transparent XL5 glasses were filled with 50 mL wine and covered with a plastic Petri dish. Twenty-four samples arranged in pairs (12 × 2) were arranged at separate randomized stations and labeled with random three-digit codes. Two known reference standards (75%:25% fruity/green vs 50%:50% fruity/green and 100%:0% fruity/green vs 0%:100% fruity/green) and two pairs with identical test samples (wines) served as internal controls. The remaining eight comparisons assessed the effect of low temperature fermentation by pairing wines produced using the same yeast strain and juice at 12.5 °C vs 25 °C. Panelists assessed the wines using an unstructured 150 mm scale where they sniffed and rated pairs of coded samples in terms of their intensity of “fruitiness” anchored at 0 (fruitiness absent; greenness extreme) and 150 (fruitiness extreme; greenness absent). Scores from three panelists who were unable to correctly rate the internal controls based on fruitiness level were excluded from the analysis.
Table 3. Mean Quality Parameters and Pooled Standard Error of the Mean (SEM) in Wines Fermented at 12.5 and 25 °C (n = 3 for Fermentation Parameters and n = 4 for WineScan Parameters)a
final weight loss (g) approximate time taken until daily weight loss remained unchanged (d) ethanol (% v/v) residual sugar (g L−1) pH total acidity (g L−1) volatile acidity (g L−1) malic acid (g L−1) citric acid (mg L−1) lactic acid (g L−1)
a
EC1118
EC1118
L-1528
L-1528
M2
M2
X5
X5
pooled
°C
BLANW
M1016
BLANW
M1016
BLANW
M1016
BLANW
M1016
SEM
12.5 25 12.5 25 12.5 25 12.5 25 12.5 25 12.5 25 12.5 25 12.5 25 12.5 25 12.5 25
65.6 b 65.0 bc 35 14 11.67 e 11.57 f 2.60 n 2.63 n 3.00 fg 2.99 gh 7.67 e 7.80 cd 0.28 j 0.30 j 4.33 a 4.26 ac 0.30 ab 0.31 a 70 29 9.82 n 11.35 g 45.81 a 21.22 e 3.23 de 3.28 c 6.44 f 6.20 g 0.68 a 0.55 c 3.01 de 2.76 f 0.16 k 0.22 j 0.17 ab 0.18 a
57.5 d 61.8 bc 68 31 10.33 m 10.84 j 22.93 c 12.61 i 2.96 j 2.96 ij 7.92 b 7.96 ab 0.47 de 0.34 hi 4.28 ab 4.29 ab 0.22 hj 0.27 d
