Eur Food Res Technol (2018) 244:245–258

https://doi.org/10.1007/s00217-017-2951-x

ORIGINAL PAPER

Influence of the volatile substances released by oak barrels

into a Cabernet Sauvignon red wine and a discolored Macabeo
white wine on sensory appreciation by a trained panel
María Navarro1 · Nikolaos Kontoudakis1 · Sergio Gómez‑Alonso2,4 ·
Esteban García‑Romero3 · Joan Miquel Canals1 · Isidro Hermosín‑Gutíerrez2 ·
Fernando Zamora1

Received: 10 March 2017 / Revised: 25 June 2017 / Accepted: 8 July 2017 / Published online: 22 July 2017
© Springer-Verlag GmbH Germany 2017

Abstract The analytical and sensory analysis of wines species. In a triangle test, tasters could distinguish between
aged in barrels of French and American oak with three dif- wines aged in new American and French oak barrels when
ferent levels of toasting was performed to determine the the toasting level was light or medium but not when it was
relationship between the volatile substances released by heavy. They were also able to distinguish between wines
oak wood and sensory appreciation by a trained panel. For aged in new and 1-year-used barrels. It can be concluded
that purpose, a discolored white wine of Macabeo and a that the botanical origin of the oak, the toasting level of the
red wine of Cabernet Sauvignon from 2012 vintage were staves and the number of times that the barrels have been
aged for 12 months in new barrels. Similar wines from the used previously have a real impact on the volatile composi-
following vintage were aged in the same barrels for know- tion of the wine and in its sensory impact.
ing how the use of the barrels affects their capacity to
release volatile substances and its organoleptic impact. A Keywords Wine · Oak aging · Oak botanical origin ·
significant correlation was found between the appreciation Toasting level · Barrel use · Aroma perception
of coconut and smoked/toasted notes by the panel and the
theoretical sensory impact of β-methyl-γ-octalactone and
volatile phenols, respectively, in two different wines aged Introduction
in new oak barrels. Vanillin correlated significantly in only
one of the wines tasted. The panelists generally preferred Aging in oak barrels is a common practice used to improve
wines aged in medium-toasted new barrels for both oak the quality of wine because of its beneficial effects on fla-
vor, aromatic complexity, color stabilization and astrin-
Electronic supplementary material The online version of this gency modulation [1, 2]. Different phenomena take place
article (doi:10.1007/s00217-017-2951-x) contains supplementary inside the barrel during aging. First, oak wood releases
material, which is available to authorized users. several volatile substances into the wine which enrich the
aroma, making it more intense and complex [3, 4]. Second,
* Fernando Zamora
fernando.zamora@urv.cat oak wood also releases phenolic compounds, especially
ellagitannins, which contribute to texture sensations such
1
Departament de Bioquímica i Biotecnologia, Facultat as astringency and mouthfeel [1, 5]. Third, the slow and
d’Enologia de Tarragona, Universitat Rovira i Virgili, C/
continuous diffusion of oxygen through the wood pores, the
Marcel.li Domingo s/n, 43007 Tarragona, Spain
2
interstices between staves and the bunghole causes differ-
Universidad de Castilla-La Mancha, Instituto Regional de
ent reactions between anthocyanins and proanthocyanidins
Investigación Científica Aplicada, Campus Universitario s/n,
13071 Ciudad Real, Spain which stabilize the wine color and smooth wine astrin-
3 gency [6, 7]. Finally, the repose of the wine inside the bar-
Instituto de la Vid y el Vino de Castilla-La Mancha,
Ctra. Toledo‑Albacete s/n, 13700 Tomelloso, Ciudad Real, rels allows the natural sedimentation of unstable colloidal
Spain matter, leaving the wine limpid and stable [8].
4
Fundación Parque Científico y Tecnológico de Castilla The volatile compounds that oak barrels transfer to the
La-Mancha, Paseo de la Innovación 1, 02006 Albacete, Spain wine may be naturally present in the original oak wood or

13
Vol.:(0123456789)
246 Eur Food Res Technol (2018) 244:245–258

may derive from other wood compounds during the bar- by the wood of both oaks and also their organoleptic
rel manufacture [9]. Many volatile compounds have been impact [13, 20]. These studies have shown that Q. alba
described belonging to very different chemical families contributes mainly with coconut flavor because it releases
[10], but few of them are significant with regard to their high amounts of β-methyl-γ-octalactone, especially from
impact on the sensory characteristics of wines. From its isomer cis (more than 90%) [11, 29], whereas Q. pet-
the point of view of their sensory impact, the main vola- raea provides a more complex aroma because it releases
tile substances released by oak wood are furans, phenolic lower amounts of β-methyl-γ-octalactone with a lower
aldehydes and ketones, volatile phenols and β-methyl-γ- proportion of the cis isomer, which produces a better bal-
octalactones [11, 12]. ance with the other flavors (vanillin, eugenol,…) [2, 10,
The family of furans, which includes mainly furfural, 20, 23].
methyfurfural, hydroxymethylfurfural and furfurilic alco- The toasting process necessary for the construction of
hol, is produced from wood polysaccharides by means the barrels is also reported to have a major influence on
of the Maillard reaction during the toasting of the staves the substances that oak wood releases into the wine [30].
needed for barrel assembly [13, 14]. Furans contribute to Toasting the inner surface of the barrels severely modifies
smoked and toasted nut notes [15, 16]. The family of phe- the chemical composition of the wood, which in turn will
nolic aldehydes and ketones, which mainly includes vanil- influence the composition of the wine during aging. Con-
lin, syringaldehyde, acetovanillone and propiovanillone, sequently, winemakers consider the level of toasting to be
contributes to the characteristic vanilla aroma of aged a key point when they choose the barrels for wine aging.
wines, with vanillin being the major contributor. Phenolic In short, furanic compounds, vanilla and volatile phenols
aldehydes and ketones result from the thermal degradation increase with higher toasting levels, although the vanilla
of lignin during barrel toasting. The family of volatile phe- decreases when the level of toasting is very high [31,
nols includes ethylphenol, guaiacol, methylguaiacol, eth- 32]. In contrast, the opposite happens with β-methyl-γ-
ylguaiacol, vinylguaiacol, eugenol and trans-isoeugenol. octalactones and ellagitannins [10, 22, 23]. For that reason,
Eugenol contributes to an interestingly spicy note of clove, cooperages usually commercialize at least three levels of
while all the other volatile phenols contribute to smoked/ toasting—light, medium and heavy—so as to offer differ-
toasted notes. These volatile phenols are also originated ent sensory profiles. However, medium toasted is the level
from lignin during barrel toasting. Finally, β-methyl-γ- most requested by winemakers because it provides a good
octalactones (also known as whiskey lactones), which balance between the different sensory attributes that oak
are present in the form of two isomers (cis and trans), are wood can release into wine.
responsible for the coconut flavor. It should be noted that The successive use of the barrels leads to a depletion of
the cis isomer has a much lower perception threshold than these substances. The capacity of the barrels to enrich the
the trans isomer, making its contribution to coconut per- wine with aromas and ellagitannins consequently decreases
ception much more important [17, 18]. with use [13]. The number of times a barrel has previously
The release of volatile compounds from oak barrels into been used is, therefore, another key point considered by
the wine depends on several factors including the botani- winemakers.
cal and geographic origin of the wood [13, 19, 20], season- The interest of wineries and cooperages to discover how
ing [10, 21], the degree of oak toasting [9, 15, 22, 23] and a barrel’s botanical origin, toasting level and age affects
the number of times the barrel was previously used [24]. the composition and quality of the wine has led to numer-
In addition, the quantity of volatile compounds released is ous studies being carried out. However, most of these have
influenced by the aging time of the wine in the barrels [14]. focused on chemical composition [1, 5, 10, 32–36], with
Although the genus Quercus is composed of more than only a few looking into the sensory aspects [2, 20]. It is
150 species of oak, only three are currently used in coop- important to remember that knowledge of a wine’s vola-
erage [25], specifically the American white oak (Quercus tile composition alone, without sensory evaluation, is not
alba) and two European species, the sessile oak (Quercus enough when it comes to predicting the flavor of the whole
petraea) and the pedunculated oak (Quercus robur), usu- system as perceived by a trained sensory panel. Indeed, the
ally known as French oaks. It is generally accepted that interactions between odorants and between the odorant and
Q. petraea and Q. alba provide greater aromatic richness different elements of the wine’s non-volatile matrix can
than Q. robur, whereas Q. robur releases more ellagitan- affect odorant volatility, flavor release and the overall per-
nins [26, 27]. Since winemakers today tend to be seeking ceived flavor, intensity and quality [37].
wines with aromatic complexity rather than high struc- In light of the above considerations, the aim of this study
ture, Q. petraea and Q. alba are the main oaks used for is to determine the influence of botanical origins and toast-
wine aging [28]. Much scientific research has, therefore, ing levels on the volatile composition and sensory appre-
been carried out to study the volatile compounds released ciation of wines aged in new and used oak barrels and the

13
Eur Food Res Technol (2018) 244:245–258 247

relationship between the volatile substances released by Table 1  Analytical characteristics of wines at the beginning of the
oak wood and sensory appreciation by a trained panel. experiment
Parameter Discolored Red wines
white wines
Vintage Vintage
Materials and methods
2012 2013 2012 2013

Chemical Ethanol content (% v/v) 13.3 13.5 13.5 13.4

Titratable acidity (g of tartaric acid/L) 6.2 6.5 5.4 5.7
Enological charcoal and bentonite were purchased from Volatile acidity (g of acetic acid/L) 0.22 0.18 0.42 0.38
Agrovin (Alcázar de San Juan, Ciudad Real, Spain). pH 3.14 3.09 3.74 3.62
Dichloromethane and methanol (LiChrosolv quality) Total phenolic index (TPI) 3.5 2.0 56.2 53.3
were purchased from Merck (Darmstadt, Germany), pen- Free ­SO2 (mg/l) 24 27 23 25
tane from Fluka (Buchs, Switzerland), and ethyl acetate, Total ­SO2 (mg/l) 85 78 60 52
absolute ethanol and sodium hydroxide from Panreac (Bar-
celona, Spain). Pure water was obtained from a Milli-Q
purification system (Millipore, USA). The chemical stand- Experimental design
ards used were purchased from Aldrich and PolyScience
(Niles, USA). The internal standard solution contained The wines of the 2012 vintage were placed in the new bar-
4-methyl-2-pentanol at a concentration of 300 μg/g of rels in April 2013 and aged for 12 months. Once the barrels
dichloromethane. were emptied and cleaned, they were immediately filled
with the same wines from the 2013 vintage, which were
also aged for 12 months. Samples from all barrels were
Barrels taken and bottled at 12 months for volatile compound and
sensory analysis.
A total of 24 new barrels (225 L) were purchased from the
Boteria Torner Cooperage (Sant Cugat Sesgarrigues, Bar- General analysis
celona, Spain). The cooperage selected wood staves with
the criterion that they were as homogeneous as possible Analytical methods recommended by the International
within each species. Twelve barrels were made with Amer- Organization of Vine and Wine [38] were used to deter-
ican white oak from Missouri (Quercus alba) and the other mine the ethanol content, titratable acidity, volatile acid-
12 with French oak from Allier (Quercus petraea). Three ity, free ­SO2 and total ­SO2. The total phenolic index (TPI)
toasting levels (light, medium and heavy) were used for was determined by measuring absorbance at 280 nm and
each type of oak, and a similar toasting level was applied expressed as absorbance units [8].
to the barrel heads. Each experimental condition was per-
formed in duplicate, using 2 barrels. Volatile compound analysis

Volatile compounds released from the oak wood were

Wines and barrel aging analyzed by GC/MS using a gas chromatograph TraceGC
Ultra coupled to a mass spectrometer DSQ II with electron
The study was carried out with two Cabernet Sauvignon red impact ionization source and quadrupole analyzer equipped
wines and two discolored Macabeo white wines obtained with an autosampler TriPlus, all from ThermoQuest. The
from the experimental cellar of the Faculty of Enology conditions of the detector were as follows: electron mul-
of the Rovira i Virgili University in Constantí (AOC Tar- tiplier voltage, 1592 V; impact energy, 70 eV; ion source
ragona, Spain) from the 2012 and 2013 vintages. The red temperature 250 °C; and mass scanning range, 39–400
wines (RW) were used without any stabilization treat- amu. A BP21 column (SGE) 50 m–0.32 mm internal diam-
ment, whereas the white wines were drastically decolored eter and 0.25 mm thick of free fatty acid phase (FFAP)
(DWW) with enological charcoal (2 g/L) and bentonite (polyethylene glycol treated with nitroterephthalic acid)
(1 g/L) to minimize the presence of aromas and phenolic was used.
compounds. The aim was to obtain a very neutral medium Volatile compounds were extracted using the method
to facilitate chemical and sensory analysis. developed by Ibarz et al. [39]. 25 mL of wine was passed
The analytical characteristics of wines at the beginning through columns filled with 0.2 g of LiChrolut EN
are shown in Table 1. (40–120 µm, Merck) using 4-nonanol as internal standard.

13
248 Eur Food Res Technol (2018) 244:245–258

The columns were then washed with 25 mL of water to 1 to 5 (1 = ‘slight intensity’, 5 = ‘maximum intensity’):
remove sugars, acids and other polar substances. The frac- coconut, vanilla and smoked/toasted. The intensity level of
tion of volatile compounds was eluted with 15 mL of pen- each descriptor was then expressed as the mean value of all
tane: dichloromethane (2:1 v/v). Extracts were concentrated the judges. A sensory training session was held beforehand
by distillation in a Vigreux column and then under nitrogen so that the panelists could agree on the criteria for each
stream to 100 µL and then kept at −20 °C until analysis. of the different sensory attributes. Samples were served
Separated compounds were identified by their mass randomly to avoid the influence of the tasting order. Dark
spectra and chromatographic retention times, using com- glasses were used in the case of the samples correspond-
mercial products as a standard. Quantification was per- ing to the DWW to avoid the influence of the color. Normal
formed by analyzing the characteristic m/z fragment for glasses were used in the case of the RW samples because
each compound using the internal standard method. Results the color was similar in all of them. Since in the wineries
for non-available compounds were expressed in concentra- the selection of the barrels is usually under the criterion of
tion units (μg/L or mg/L) as internal standard equivalents the winemakers, tasters were also asked which of the sam-
obtained by normalizing the compound peak to that of the ples they preferred. The aim of this question was not to
internal standard and multiplying by the concentration of determine the preferences of the consumers but for know-
the internal standard. Only volatile compounds released ing which type of barrels would be selected by the trained
by oak wood (furanic compounds, phenolic aldehydes and professionals who will be responsible of this decision.
ketones, β-methyl-γ-octalactones and volatile phenols) are Two trials of sensory triangle tests were also conducted
shown because the aim of this study focuses on the impact (third and fourth sessions) in accordance with ISO 4120
of oak aging. [42]. The third session consisted of six triangle tests to
compare the wines aged in new French oak barrels with
Sensory analysis those aged in new American oak barrels for each one of
the toasting levels. The fourth session also consisted of
All sensory analyses were performed in the tasting room six triangle tests, but in this case the aim was to compare
of the Faculty of Enology of Tarragona (University Rovira wines aged in new and 1-year-used barrels. This was only
i Virgili), which was designed in accordance with UNE performed with the DWW in order to minimize the effect
87004.197 [40]. Tasting was carried out using ISO official of the vintage. This experiment was carried out over two
tasting glasses [41]. Each sample consisted of 30 mL of consecutive years because the aim was to determine the
wine presented at room temperature (20 °C) covered with effect of the depletion caused using the barrels. Since the
clear plastic petri dishes to minimize the escape of volatile difference between the red wines from the 2012 and 2013
components and randomly coded with three-digit numbers. vintages was so obvious, a triangle test was of no use. In
All the samples were tasted by 16 students on the final contrast, the discoloration of the white wines with charcoal
year of the Bachelor in Enology at the University Rovira and bentonite minimized the differences in color and aroma
i Virgili (four-year degree). This panel was made up of 10 between the two vintages in such a way that the differences
males and 6 females aged between 21 and 25, who had were mainly due to the aromas released by oak aging.
been training together for 3 years while studying sensory All these triangle tests were performed to determine
analysis as part of the enology degree. whether tasters were able to recognize the wine that was
Six months after bottling, sensory analyses of the wines different. The second objective was to determine which
aged in the different barrels were carried out in four ses- wine was preferred by the panelists who correctly identi-
sions to prevent fatigue in the panelists. The first and the fied the different wines. The purpose of these triangular
second sessions were used for descriptive sensory analysis, tests was to know if oenologist (in that case students), who
whereas the third and the fourth sessions were used for tri- will must take the decision of buying oak barrels, were able
angular tests. To simplify the tasting, the wines from two to distinguish between wines aged in different types of oak
similar (duplicate) barrels of each experimental group were barrels and after know which one they will select.
mixed.
During the first session, two trials were carried out to Calculation of the theoretical sensory impact of aromas
determine the influence of the toasting level of both French present in wines
and American oaks on the discolored white wines. Conse-
quently, 6 wines were tasted in two trial of 3 wines each The estimation of the theoretical sensory impact of the
one. Similar tasting was performed with red wines in the main aromas in the different samples was calculated by
second session. dividing the individual concentration of each aroma by
For each sample, tasters were required to evaluate the its corresponding threshold. These odor activity values
intensity of three olfactory sensory attributes on a scale of [43] were obtained using the average of the thresholds

13
Eur Food Res Technol (2018) 244:245–258 249

previously reported by Spillman et al. [16] for the different furanic compounds, vanillin, trans and cis β-methyl-γ-
volatile compounds in different matrices. Since we have not octalactones, the cis/trans ratio and the global volatile
determined the specific thresholds of the different volatile phenols so as to simplify the great amount of informa-
compounds in the matrix used (DDW and RW), we decided tion. The full data are shown in Supplementary Tables S1
to use the term ‘theoretical sensory impact of aroma’ (furanic compounds), S2 (phenolic aldehydes and ketone
(TSIA) instead of odor value. compounds), S3 (β-methyl-γ-octalactones) and S4 (volatile
Specifically, the thresholds considered were for coconut: phenols).
cis β-methyl-γ-octalactones: 83 µg/L; trans β-methyl-γ- Generally speaking, both matrices (DWW and RW)
octalactones: 390 µg/L; for vanilla: vanillin: 65 µg/L; and showed a similar trend. In brief, American oak released
for smoked/toasted: guayacol and the other volatile phe- significantly higher amounts of β-methyl-γ-octalactones,
nols: 20 µg/L. Furanic compounds were not considered especially of the cis isomer, than the French oak at all toast-
because their concentration in the samples was much lower ing levels and in new and 1-year-used barrels. In contrast,
than their sensory thresholds. The TSIA of each sensory no clear differences were found between the two oak spe-
attribute was considered as the summatory of the total of cies for furanic compounds, vanillin and volatile phenols.
all the quotients of the individual compound concentrations The toasting level clearly affected all volatile substances
divided by their corresponding threshold (∑Ai/Ti). It should in both species in a similar way. β-methyl-γ-octalactones
be noted that these thresholds were determined in different decreased when the toasting level was higher, whereas
matrices to those of this study, so they should be consid- furanic compounds and volatile phenols did the opposite.
ered with caution. We decided to use these values in the Vanillin also increased between light and medium toasting
belief that they would at least be good indicators. but decreased in the heavy toasting. Finally, the levels of all
volatile substances were significantly lower in wines aged
Statistics in 1-year-used barrels than in new barrels in all the experi-
mental conditions.
All analytical data are expressed as the arithmetic
mean ± standard deviation of the samples from two bar- Evaluation of panel performance and of the
rels. The multifactor analysis of variance (ANOVA) was homogeneity of criteria between the different panelists
carried out using XLSTAT software in order to compare
the wines in function of the oak origin, toasting level and Figure 1 shows the results of the multivariate analysis
barrel use. All sensory data are expressed as the arithmetic following the Tucker method that was applied to verify
mean ± standard deviation of the scores of the 16 tasters. panel performance and assess the homogeneity of criteria
The performance and homogeneity of criteria between between the different panelists with regard to the differ-
the different panelists with respect to different sensorial ent sensorial attributes. This analysis produced a total of
attributes were analyzed by Tucker’s method [44, 45] using six Tucker correlation plots. The first three show how the
PanelCheck software [45, 46]. The Tucker correlation plot 16 panelist are aligned according to the sensory attribute:
is essentially a principal component analysis (PCA) on an coconut (Tucker-1), vanilla (Tucker-2) and smoked/toasted
unfolded data matrix, and is very useful for visualizing (Tucker-3) for the DWW. The other three are for the same
similarities and dissimilarities between the tasters’ criteria sensory attributes but for the RW. By screening through
for the different sensory attributes. the plots, it can be seen that the overall performance by the
Pearson’s correlation of data and principal component 16 tasters can be considered reasonably good for the three
analysis was carried out using IBM SPSS Statistics 19 sensory attributes in both matrices, since in all cases the
software (International Business Machines S.A., Madrid, results were statistically significant (p < 0.05). The differ-
Spain). The level of significance of the sensory triangle ent Tucker correlation plots show that the discolored white
tests was determined following the Jackson’s method [47]. wine presents a better arrangement with fewer tasters stray-
ing from the common alignment than in the case of the red
wine. The plots also indicate that tasters are better aligned
Results with regard to the coconut and smoked/toasted notes than
with the vanilla.
Volatile compounds released into the wine
by the different barrels Sensory evaluation of the main oak aromas
and preference selection
Table 2 shows a synthesis of the concentration of the main
volatile substances released by oak wood from the differ- Figure 2 shows the panelists’ sensory evaluation of the
ent barrels into the different wines. It only shows the global main oak aromas of the wines aged in new barrels. No

13
 Table 2  Main volatile compounds of oak wood from the different samples
250

Matrix Compounds Oak origin Barrel age Light Toast Medium Toast Heavy Toast

13
Discolered Total Furanic com- French New Barrel 1231.7 ± 603.7 a B α 3717.2 ± 767.7 b B α 3232.6 ± 370.6 b B α
white wine pounds (µg/L) 1 year barrel 166.1 ± 57.2 a A α 421.4 ± 131.6 b A α 1095.7 ± 157.4 c A α
American New Barrel 716.1 ± 196.6 a B α 4554.1 ± 1298.7 b B α 5688.8 ± 1642.3 b B β
1 year barrel 121.6± 15.0 a A α 435.9 ± 77.2 b A α 1713.9 ± 421.1 c A β
Vanillin (µg/L) French New Barrel 547.1 185.7 a B α 621.2 ± 16.7 a B α 497.0 ± 148.9 a B α
1 year barrel ± 157.5 ± 54.5 a A α 239.1 ± 76.4 ab A α 281.1 ± 27.3 b A α
American New Barrel 390.1 ± 90.4 a B α 1152.6 ± 100.0 b B β 568.2 ± 128.2 a B α
1 year barrel 147.1 ± 17.4 a A α 361.6 ± 34.0 c A β 288.2 ± 25.5 b A α
trans-β -methyl-γ- French New Barrel 1 year 183 ± 58 b B α 155 ± 10 b B α 58 ± 20 a B α
octalactone (µg/L) barrel 38 ± 8 b A α 38 ± 4 b A α 24 ± 5 a A α
American New Barrel 1 year 366.3 ± 56.9 b B β 282.9 ± 60.5 b B β 79.0 ± 12.3 a B α
barrel 74.3 ± 7.6 b A β 60.6 ± 7.6 b A β 29.3 ± 3.7 a A α
cis-β -methyl-γ- French New Barrel 1 year 349 ± 231 ab B α 267 ± 53 b B α 103 ± 25 a B α
octalactone (µg/L) barrel 60 ± 32 ab A α 56 ± 10 b A α 32 ± 8 a A α
American New Barrel 1 year 2507.5 ± 96.9 b B β 2597.4 ± 754.7 b B β 712.6 ± 107.6 a B β
barrel 472.0 ± 34.0 b A β 495.4 ± 82.4 b A β 224.2 ± 27.9 a A β
cis/trans ratio French New Barrel 1 year 1.8 ± 0.7 a A α 1.7 ± 0.2 a A α 2.0 ± 1.1 a A α
barrel 1.5 ± 0.6 a A α 1.5 ± 0.4 a A α 1.4 ± 0.6 a A α
American New Barrel 1 year 6.9 ± 1.3 a A β 9.0 ± 1.0 a A β 9.0 ± 1.7 a A β
barrel 6.4 ± 1.1 a A β 8.1 ± 0.5 b A β 7.7 ± 0.3 ab A β
Volatile Phenols French New Barrel 1 year 125.3 ± 34.0 a B α 165.3 ± 39.8 a B α 248.4 ± 23.4 a B α
(µg/L) barrel 25.7 ± 7.5 a A α 35.8 ± 4.3 a A α 61.6 ± 4.9 a A α
American New Barrel 1 year 173.8 ± 43.7 a B α 246.5 ± 46.5 b B β 313.3 ± 61.8 a B α
barrel 81.5 ± 13.7 a A β 99.7 ± 16.7 a A β 117.2 ± 27.7 a A β
Red wine Total Furanic com- French New Barrel 1 year 253.0 ± 14.6 a B α 1273.4 ± 150.0 a B α 1480.4 ± 307.5 a B α
pounds (µg/L) barrel 48.3 ± 7.4 a A α 138.4 ± 16.4 a A α 381.3 ± 98.6 a A α
American New Barrel 1 year 205.1 ± 36.1 a B α 493.1 ± 159.9 b B β 1724.9 ± 100.7 a B α
barrel 47.9 ± 5.4 a A β 80.8 ± 13.2 a A β 457.0 ± 49.3 a A β
Vanillin (µg/L) French New Barrel 1 year 437.7 ± 27.1 a B β 490.6 ± 46.7 a B α 330.2 ± 120.1 a A α
barrel 122.0 ± 7.1 a A α 206.4 ± 12.5 b A α 197.9 ± 16.9 b A α
American New Barrel 1 year 324.5 ± 70.6 a B α 502.6 ± 102.8 b B α 361.4 ± 1.9 a B α
barrel 120.3 ± 13.6 a A α 208.2 ± 10.5 b A α 227.0 ± 17.7 b A α
trans-β -methyl-γ- French New Barrel 1 year 157 ± 19 c B α 82 ± 51 b A α 27 ± 3 a B α
octalactone (µg/L) barrel 34 ± 3 a A α 25 ± 14 a A α 12 ± 2 a A α
American New Barrel 1 year 365.1 ± 79.6 c B β 195.7 ± 62.3 b B α 79.2 ± 8.9 a B β
barrel 75.4 ± 15.2 b A β 61.7 ± 22.5 b A β 31.0 ± 2.6 a A β
Eur Food Res Technol (2018) 244:245–258
 Eur Food Res Technol (2018) 244:245–258 251

significant differences were found in any of the sensory

All data are expressed as the average of 2 barrels ± standard deviation. Different letters indicate statistical differences (p < 0.05). Roman lower case letters are used to compare toasting levels.
α
α
α
α
α
α
α
α

β
β
β
β
attributes in the wines aged in French oak. In contrast, sig-
nificant differences were found in the coconut and smoked/

A
A
A
A
A
A

B
B
B

B
toasted notes of the wines aged in American oak. How-

b
a
a

b
b

a
ever, despite the lack of significance in some of these data,

a
a
a
a
a
a number of interesting trends can be pointed out. Briefly,
coconut was scored higher in wines aged in American oak

117.4 ± 23.7

130.0 ± 10.3
234.9 ± 11.0

201.8 ± 67.0

257.0 ± 26.0
760.6 ± 42.7
Heavy Toast

than in French oak in both matrices, and the intensity of this

136 ± 42

10 ± 1.6
8 ± 0.5
sensory attribute decreased in line with the toasting level
3 ± 0.3
33 ± 7

5 ± 1.0

in a similar way as observed with the levels of β-methyl-

γ-octalactones. Vanilla tended to increase between light
and medium-toasted barrels in both oak species and in both
α
α
α
α
α
α
α
α

β
β
β
β

matrices, thus showing a similar trend to that determined

A

A
A

A
A

A
A

B
B

with the levels of vanilla. Finally, smoked/toasted notes

Roman capital letters are used to compare new and 1-year-used barrels. Greek letters are used to compare the botanical origin of the oak barrels
ab

ab
ab
b

increased with toasting in both oak species and both matri-

a

b
b

a
a
a

ces in a similar way as for volatile phenols.

The panelists were also asked to indicate their prefer-
1837.5 ± 160.2
Medium Toast

459.3 ± 11.0

± 51.1 ± 81.9

ences during the descriptive sensory analysis of the wines

140.3 ± 12.1

191.1 240.9
55.7 ± 6.6

aged in the various different new barrels. Figure 3 shows

10 ± 4.0
3 ± 1.6

8 ± 3.0
297 ± 1
59 ± 4

4 ± 2.8

the results obtained. Briefly, in the case of French oak

the panelists clearly preferred the wine aged in medium-
toasted barrels regardless of the matrix (55% in DWW
and 60% in RW). However, a different order of preference
α
α
α
α
α
α

β
β
β
β
β
β

was obtained for the other two levels of toasting depend-

A

A
A
A
A
A
A

B
B

ing on the matrix. In the case of DWW, 35% of the tasters

ab

ab
ab

preferred the wine aged in light-toasted barrels and only

a

b
b

a
a
a
a
c

10% preferred the wine aged in the heavy toasted. In con-

trast, in the case of RW 30% of the tasters preferred the
wine aged in heavy-toasted barrels and only 10% selected
147.8 ± 11.7
2887.3 ± 51.2
466.0 ± 19.8

122.5 ± 10.0

172.4 ±11.1
98.3 ± 8.7
Light Toast

the wine aged in light-toasted oak.

275 ± 179
51 ± 25

2 ± 0.9

6 ± 1.5
2 ± 1.4

8 ± 1.9

In the case of American oak, important differences

were found in the tasters’ preferences depending on the
matrix. Specifically, in the case of DWW 85% of the
panelists preferred the wine aged in medium-toasted bar-
New Barrel 1 year

New Barrel 1 year

New Barrel 1 year

New Barrel 1 year

New Barrel 1 year

New Barrel 1 year

rels, while the remaining 15% selected the wine aged in

light-toasted oak and none preferred the wine aged in
Barrel age

barrel

barrel
barrel

barrel
barrel

barrel

heavy-toasted barrels. In contrast, in the case of RW no

clear preferences were obtained between wines aged in
medium- and light-toasted barrels (45 and 40% respec-
tively), and 15% of the tasters preferred the wine aged in
Oak origin

American
American
American

heavy-toasted barrels.
French
French
French

Theoretical sensory impact of the main aromas

octalactone (µg/L)

Figures 4 and 5 show the theoretical sensory impact of the

Volatile Phenols
cis-β -methyl-γ-

cis/trans ratio

main aromas (TSIA) released by the different barrels in

Compounds

the DWW and RW, respectively. The three main olfactory

(µg/L)

sensory attributes associated with oak aging were selected:

Table 2  (continued)

coconut, vanilla and smoked/toasted notes. The total TSIA

was clearly higher in wines aged in new American oak bar-
rels than in new French oak barrels. This higher total TSIA
was mainly due to the coconut attribute resulting from the
Matrix

higher amounts of cis-β-methyl-γ-octalactones released

13
252 Eur Food Res Technol (2018) 244:245–258

Discolored white wine

Tucker 1 - Correlaon loadings: Coconut Tucker 2 - Correlaon loadings: Vanilla Tucker 3 - Correlaon loadings: Smoked/Toasted
1,0 1,0 1,0

4 6
0,5 9 0,5 0,5 6
16 8 11
9 3
15
PC2 (17,9%)

PC2 (17,9%)

PC2 (17,9%)
1 13
14 10
0,0
8
14 p < 0.001 0,0
12 p < 0.05 2
15 0,0 p < 0.001 10
7
4 1
8
7 2 5
1 11 13
12
5 7
-0,5 10 13 12 -0,5 -0,5 414
2 16 16
3
515 6 9

-1,0 -1,0 -1,0

-1,0 -0,5 0,0 0,5 1,0 -1,0 -0,5 0,0 0,5 1,0 -1,0 -0,5 0,0 0,5 1,0
PC1 (48,7%) PC1 (48,7%) PC1 (48,7%)

Red wine
Tucker 4 - Correlaon loadings: Coconut Tucker 5 - Correlaon loadings: Vanilla Tucker 6 - Correlaon loadings: Smoked/Toasted
1,0 1,0 1,0
15
16 5
7 12
8 4 11 13 7 9
0,5 5 14 0,5 0,5 10
6 1
12 2
p < 0.05 p < 0.001 8
PC2 (24,7%)

PC2 (24,7%)

PC2 (24,7%)
6
2
3
0,0 1
p < 0.001 0,0 2 0,0
11
4
15
10 7 11
13 16 3
3 84 1 5
9 12
-0,5 -0,5 -0,5
14
10 9 16
6
15 14
13
-1,0 -1,0 -1,0
-1,0 -0,5 0,0 0,5 1,0 -1,0 -0,5 0,0 0,5 1,0 -1,0 -0,5 0,0 0,5 1,0
PC1 (34,9%) PC1 (34,9%) PC1 (34,9%)

Fig. 1  Assessment of the panelists’ homogeneity of criteria with respect to the different sensorial attributes

by this specie. This trend was observed in both matrices, Statistical correlation between sensory appreciation
DWW and RW. The wines aged in 1-year-used American of the different descriptors and the corresponding TSIA
barrels also showed a higher total TSIA, but the differences value
were much smaller.
The toasting level also had an important effect on the The similarities observed between chemical analysis, TSIA
total TSIA, but the trend was different in the French and and sensory evaluation by the panel suggested a Pearson’s
American oaks. In French oak, the total TSIA increased correlation analysis between the sensory evaluation and
with the toasting level because the global increase in the TSIA of each attribute. The results obtained (Table 3)
smoked/toasted and vanilla was higher than the decrease in confirm the existence of a significant correlation (p < 0.05)
coconut. In contrast, the decrease in coconut notes in the with high Pearson’s correlation coefficients between coco-
case of American oak was more drastic than the increase nut perception and coconut TSIA in both matrices (0.964
in smoked/toasted and vanilla. This balance means that in DWW and 0.903 in RW). The other Pearson’s correla-
the total TSIA tended to decrease when the toasting level tion coefficients between coconut perception and the other
was higher in the case of the RW. In the case of the DWW, TSIA attributes were not significant. Similar results were
the total TSIA increased between the light-toasting and obtained for smoked/toasted perception, which showed sig-
medium-toasting levels but clearly decreased in the heavy- nificant correlation, and high Pearson’s correlation coeffi-
toasted level. cients with volatile phenol TSIA (0.945 in DWW and 0.959
Finally, the total TSIA of the new barrels was clearly in RW), but no significant correlations with other TSIA
higher than for the 1-year-used barrels in both wines and at attributes were found. Vanilla sensory perception also
all toasting levels. showed a significant correlation and high Pearson’s cor-
relation coefficient with vanilla TSIA (0.915) in RW, but

13
Eur Food Res Technol (2018) 244:245–258 253

Fig. 2  Sensory evaluation of Discolored white wine

the main oak aromas. All data a
are expressed as the average 5 A French Oak B American Oak b
of 16 tasters. Different letters a b ab a
indicate statistical differences 4 a
a a
(p < 0.05) a a
a

Score
3 a a
a a a
2 a
1

0
Red wine
5 C French Oak D American Oak
a b
a a b
4 a ab
a
a ab a a
Score

3 a a a a
a a
2

0
Smoked Smoked
Coconut Vanilla Coconut Vanilla
Toasted Toasted
Light toasted Medium toasted Heavy toasted

Fig. 3  Preferences during the Discolored white wine

sensory analysis of the wines
aged in the different new bar- A French Oak B American Oak
rels. All data are expressed as HT LT
the average and percentage of 10% 15%
16 tasters. LT light toasting, LT
MT medium toasting, HT heavy 35%
toasting
MT
MT
85%
55%

Red wine
C French Oak D American Oak
LT
10% HT
HT 15%
30% LT
40%

MT MT
60% 45%

13
254 Eur Food Res Technol (2018) 244:245–258

Fig. 4  Theoretical sensory

impact of the main aromas in a
discolored white wine. All data
are expressed as the average of
two barrels. N new barrels, 1Y
one-year-used barrels, L light
toasting, M medium toasting, H
heavy toasting. The theoretical
sensory impacts of the coconut,
vanilla and smoked/toasted
aromas (ΣAi/Ti) were obtained
by applying the summatory
of all the quotients of the cor-
responding individual volatile
compounds concentrations by
their corresponding threshold

Fig. 5  Theoretical sensory

impact of the main aromas in a
red wine. All data are expressed
as the average of two barrels.
N new barrels, 1Y one-year-
used barrels, L light toasting,
M medium toasting, H heavy
toasting. The theoretical sensory
impacts of the coconut, vanilla
and smoked/toasted aromas
(ΣAi/Ti) were obtained by apply-
ing the summatory of all the
quotients of the corresponding
individual volatile compounds
concentrations by their cor-
responding threshold

was not significant in DWW. No other significant correla- and American oak at the different toasting levels. The pan-
tion was found between vanilla perception and other TSIA elists were able to significantly distinguish between wines
attributes. of both matrices, DWW and RW, aged in new French and
new American oak barrels in the case of light and medium
Sensory comparisons using triangle tests toasting levels but not in the case of the heavy toasting
level. Specifically, the number of tasters out of a total of
Table 4 shows the results of sensory comparisons by tri- sixteen that were capable of distinguishing between DWW
angle test between wines aged in new barrels of French aged in new French and American oak barrels was 12 in the

13
Eur Food Res Technol (2018) 244:245–258 255

Table 3  Statistical correlation Matrix Sensory evaluation TSIA

between sensory appreciation
of the different descriptors and Coconut Vanilla Smoked/Toasted
the corresponding odor activity
value PCC p PCC p PCC p

Discolored Coconut 0.964 0.002 0.362 0.481 -0.591 0.217

white wine Vanilla 0.324 0.531 0.686 0.132 -0.294 0.572
Smoked/Toasted −0.634 0.177 −0.633 0.177 0.945 0.004
Red wine Coconut 0.903 0.014 −0.097 0.854 -0.415 0.413
Vanilla −0.194 0.713 0.915 0.011 -0.360 0.483
Smoked/Toasted −0.627 0.183 −0.699 0.699 0.959 0.003

All values correspond to Pearson’s correlation coefficients. Values in bold indicate the existence a statisti-
cally significant correlation (p < 0.05)
TSIA theoretical sensory impact of aroma

Table 4  Sensory comparison by triangle test between wines aged in In the case of RW, the preferences were not similar.
new barrels of French and American oak at the different toasting lev- Only 4 of the 12 tasters who correctly identified the dif-
els
ferent samples preferred wines aged in French oak and 8
Matrix Triangle test Positive p Preferences preferred wines aged in American oak. However, this pref-
identifica- erence changed completely when the barrels were medium
tions French American
toasted, since 8 tasters preferred the wine aged in French
Discolored Light toasted 12/16 0.001 8 4 oak and only 4 selected the wine aged in American oak.
white Medium 10/16 0.05 5 5 Finally, when the barrels were heavy toasted, although the
wine toasted panel did not give statistically significant results, the tasters
Heavy toasted 7/16 NS 3 4 who correctly identified the different sample clearly pre-
Red wine Light toasted 12/16 0.001 4 8 ferred the wine aged in American oak (7 out of a total of 8).
Medium 12/16 0.001 8 4 Table 5 shows the results of the sensory comparison by
toasted
triangle test between DWW aged in new and 1-year-used
Heavy toasted 8/16 NS 1 7
barrels of French and American oak at the different toast-
ing levels. Panelists were able to distinguish between wines
aged in new barrels and wines aged in 1-year-used barrels
case of light toasted (p = 0.001), 10 in the case of medium in all cases, in different oak species and at different toasting
toasted (p = 0.05) and only 7 in the case of heavy toasted levels.
(p > 0.05). Similar results were obtained in the case of RW, To be specific, 13 (p = 0.01) out of a total of sixteen
since 12 tasters were able to differentiate between wines tasters were capable of distinguishing between new and
aged in new French and American oak barrels (p = 0.001) 1-year-used French oak barrels in the case of light toasted,
in the case of light- and medium-toasted barrels, but only 8 14 (p = 0.001) in the case of medium toasted and 16
(p > 0.05) when the barrels were heavy toasted. (p < 0.001) in the case of heavy toasted. Similar results
Table 4 also shows the preferences of the tasters who were found in the case of American oak: 16 (p < 0.001)
correctly identified the different samples in the differ- for light toasted, 13 (p = 0.01) for medium toasted and 16
ent triangle test comparisons. In the case of DWW aged (p < 0.001) for heavy toasted.
in light-toasted new barrels, the wine aged in French oak Table 5 also shows the preferences of the tasters who cor-
was clearly preferred over the wine aged in American oak rectly identified the different samples in the various triangle
(8 and 4, respectively). However, the preferences tended to test comparisons. In the case of French oak, tasters clearly
become more balanced between both oak species when the preferred the wines aged in new barrels when the toasting
toasting level increased. Specifically, 5 panelists selected level was light (8 out of 13) and medium (12 out of 14). How-
the wine aged in French oak and 5 the wine aged in Amer- ever, this changed completely when the barrels were heavy
ican oak when the barrels were medium toasted, and, toasted, since 13 out of 16 preferred the wine aged in the
although the panel was not able to significantly distinguish 1-year-used barrel. As for the American oak, tasters always
when the barrels were heavy toasted, 3 of the 7 tasters who preferred the wine aged in the new barrels: 13 out of 16 in
correctly identified the different sample preferred the wine the case of light toasted, 10 out of 13 in the case of medium
aged in French oak and 4 the wine aged in American oak. toasted and 10 out of 16 in the case of heavy toasted.

13
256 Eur Food Res Technol (2018) 244:245–258

Table 5  Sensory comparison by triangle test between decolored the panelists were more aware of and had more homoge-
white wines aged in new barrels and 1-year-used barrels of French neous criteria for the sensory attributes of coconut and
and American oak at the different toasting levels
smoked/toasted notes than for vanilla. In fact the dispersion
Oak Triangle test Positive ρ Preferences of panelists in the Tucker plot for vanilla in the RW matrix
identifica- is very clear. In any case, the variance analysis was statisti-
tions New 1 Year
cally significant (p < 0.05) in all cases, which confirms that
French Light toasted 13/16 0.01 8 5 the panel is suitably trained for the sensory evaluation of
Medium toasted 14/16 0.001 12 2 these attributes.
Heavy toasted 16/16 <0.001 3 13 The trained panel’s sensory evaluation of the main aro-
American Light toasted 16/16 <0.001 13 3 mas released by oak wood in the different wines generally
Medium toasted 13/16 0.01 10 3 followed a similar behavior to the calculated TSIA. Thus,
Heavy toasted 16/16 <0.001 10 6 the correlation analysis between both data was performed
with very satisfactory results in the case of coconut and
smoked/toasted notes, since in all cases the Pearson’s cor-
Discussion relation coefficients were high and significant. However, in
the case of vanillin the results were rather surprising. The
The aim of the present study was to determine the influence Tucker plots for vanilla indicated that the panel performed
of botanical origins and toasting levels on the volatile com- better with the DWW than the RW matrix. However, the
position and sensory appreciation of wines aged in new and Pearson’s correlation coefficient between the sensory score
1-year-used oak barrels and also to study the relationship of the panel and the TSIA was only significant in the case
between the volatile substances released by oak wood and of the red wine. In the case of DWW, the Pearson’s cor-
sensory appreciation by a trained panel. relation coefficient for vanilla was relatively high (0.686)
In general terms, the results obtained for the main vola- but did not reach a significant level (p = 0.132). Another
tile aroma levels released by new and 1-year-used French interesting fact arising from these correlation analyses is
and American oak barrels with different toasting levels that vanilla and coconut showed relatively high negative
into the wines analyzed match previously reported data. Pearson’s correlation coefficients (lower than −0.6 in all
Thus, our data agree with those of other authors who have cases) with the sensory perception of smoked/toasted notes,
previously described that American oak releases higher which suggests a possible antagonism between “sweetish”
β-methyl-γ-octalactone amount, especially of the cis iso- and “smoky” aromas.
mer, than French oak [2, 10, 11, 20, 23, 29]. In contrast, Regarding preferences, the panelists almost always pre-
Spillman et al. [36] have reported lower levels of β-methyl- ferred the medium-toasted barrels independently of the
γ-octalactone in wines aged in American oak than in matrix (DWW or RW) or the oak species, probably because
French oak, but to our knowledge there are no other works this toasting level gives the best balance of the three main
in that sense. In turn, the literature describes that toasting aromas. However, the light toasting level was also preferred
increases the levels of furanic compounds, vanillin and by a substantial proportion of the tasters in the case of the
volatile phenols [2, 10, 22, 31, 32] and decreases the levels DWW aged in French oak and the RW aged in American
of β-methyl-γ-octalactone [10, 22, 23]. Our results match oak. The heavy-toasted level was also relatively preferred
perfectly with that. Finally, the scarce information on the in the case of the red wine aged in French oak. These data
influence of the use of barrels on the release of volatile sub- explain why medium-toasted barrels are the most fre-
stances [13, 33] coincides with our data, confirming that quently used for wine aging in wineries.
after one year of use the capacity of the barrels to aroma- The triangle test comparisons between wines aged in
tize the wine diminishes a lot. new French and American oak barrels at different levels
The panel’s performance in the sensory analysis of the of toasting produce some very interesting results. Tast-
three main aromas released by oak wood can be considered ers were able to distinguish significantly between French
quite satisfactory in all cases. However, their performance and American oak when the barrels were light or medium
was better with regard to the DWW matrix than the RW toasted. In contrast, when the barrels were heavy toasted,
matrix, probably because the red wine was rather more the panel could not significantly identify the differences.
complex than the decolored white wine. Indeed, the total This suggests that the excess of toasted notes masks the
amount of variance explained by PC1 and PC2 is 66.6% other aromas, making it impossible to distinguish between
(PC1: 48.7% and PC2: 17.9%) in the case of the DWW, oak species.
whereas in the case of the RW they explained only 59.6% The panel could also significantly distinguish between
(PC1: 34.9% and PC2: 24.7%). It was also observed that DWW wines aged in new and 1-year-used barrels of both

13
Eur Food Res Technol (2018) 244:245–258 257

oak species at all toasting levels. This was to be expected, Compliance with Ethics requirements This article does not con-
since the concentrations and TSIAs of the different aro- tain any studies with human or animal subjects.
mas were significantly higher in the new barrels than in the
1-year-used barrels. However, the panelists’ preferences
produced more interesting data. In the case of American
References
oak barrels, the tasters mostly preferred the DDW aged in
new barrels for the three toasting levels. However, in the 1. Fernández de Simón B, Martínez J, Sanz M, Cadahía E, Ester-
case of French oak, the tasters preferred the DDW aged uelas E, Muñoz AM (2014) Volatile compounds and sensorial
in new barrels only when the toasting level was light or characterization of red wine aged in cherry, chestnut, false aca-
medium but not when it was heavy. This suggests that the cia, ash and oak wood barrels. Food Chem 147:346–356
2. González-Centeno MR, Chira K, Teissedre PL (2016) Ellagitan-
excess of smoked/toasted notes caused by heavy toasting is nin content, volatile composition and sensory profile of wines
considered a negative characteristic by tasters, since they from different countries matured in oak barrels subjected to dif-
prefer wines aged in the used barrel in which these aromas ferent toasting methods. Food Chem 210:500–511
have a noticeably lower impact. 3. Návojská J, Brandes W, Nauer S, Eder R, Frančáková H (2012)
Influence of different oak chips on aroma compounds in wine. J
Microbiol Biotechnol Food Sci 1:957–971
4. Pérez-Juan PM, Luque de Castro MD (2014) Use of Oak wood
to Enrich wine with volatile compounds. In: Preedy VR(ed)
Conclusions Processing and impact on active components in food. Academic
Press: London, pp 471–48
It has been possible to establish a significant correlation 5. Jourdes M, Michel J, Saucier C, Quideau S, Teissedre PL
between a trained panel’s sensory appreciations of coco- (2011) Identification, amounts, and kinetics of extraction of
C-glucosidic ellagitannins during wine aging in oak barrels
nut and smoked/toasted notes and the theoretical sensory or in stainless steel tanks with oak chips. Anal Bioanal Chem
impact of β-methyl-γ-octalactone and volatile phenols, 401:1531–1539
respectively, in two different wines aged in new oak bar- 6. Singleton VL (2000) A survey of wine aging reactions, espe-
rels. The correlation between the sensory appreciation of cially with oxygen. In: Proceedings of the ASEV 50th Anniver-
sary Annual Meeting. pp.323-336. ASEV, Seattle
vanilla and the theoretical sensory impact of vanillin was 7. Del Alamo-Sanza M, Nevares I (2014) Recent advances in the
significant in the case of the red wine but not in the dis- evaluation of the oxygen transfer rate in oak barrels. J Agric
colored white wine. In general, the panelists clearly pre- Food Chem 62:8892–8899
ferred wines aged in medium-toasted new barrels for both 8. Ribéreau-Gayon P, Glories Y, Maujean A, Dubordieu D (2006)
Phenolic compounds. In: Ribéreau-Gayon P, Glories Y, Mau-
oak species. However, a substantial number of tasters jean A, Dubordieu D (eds) The chemistry of wine stabilization
selected discolored white wine aged in light-toasted French and treatments. Handbook of Enology, vol 2, 2nd edn. Wiley,
oak barrels and the red wine aged in light-toasted Ameri- Chichester
can oak barrels. The wines aged in heavy-toasted new 9. Chira K, Teissedre PL (2013) Relation between volatile compo-
sition, ellagitannin content and sensory perception of oak wood
barrels tended to be the least appreciated, although a non- chips representing different toasting processes. Eur Food Res
negligible proportion of the panelists selected this toasting Technol 236:735–746
level in the case of French oak. In a triangle test, they were 10. Cadahía E, Fernández de Simón B, Jalocha J (2003) Vola-
able to distinguish between wines aged in new American tile compounds in Spanish, French, and American oak woods
after natural seasoning and toasting. J Agric Food Chem
and French oak barrels when the toasting level was light or 51:5923–5932
medium but not when it was heavy. The panelists generally 11. Garde-Cerdán T, Torrea-Goñi D, Ancín-Azpilicueta C (2004)
selected wines aged in French oak, but not in all the cases. Accumulation of volatile compounds during ageing of two red
They were also able to distinguish between wines aged in wines with different composition. J Food Eng 65:349–356
12. Prida A, Chatonnet P (2010) Impact of oak-derived compounds
new and 1-year-used barrels and preferred wines aged in on the olfactory perception of barrel-aged wines. Am J Enol
new barrels, the only exception being the discolored white Vitic 61:408–413
wine aged in heavy-toasted French oak barrels. 13. Garde-Cerdán T, Rodriguez S, Ancín-Azpilicueta C (2002) Vol-
atile composition of aged wine in used barrels of French oak and
Acknowledgements We would like to thank CICYT (AGL2014- American oak. Food Res Int 35:603–610
56594-C2-1-R and AGL2014- 56594-C2-2-R) for its financial sup- 14. Garde-Cerdán T, Ancín-Azpilicueta C (2006) Effect of oak bar-
port. We are also grateful to the “Boteria Torner” cooperage for pro- rel type on the volatile composition of wine: storage time optimi-
viding the barrels. zation. Food Sci Technol 39:199–205
15. Chatonnet P, Cutzach I, Pons M, Dubourdieu D (1999) Monitor-
Compliance with ethical standards ing toasting intensity of barrels by chromatographic analysis of
volatile compounds from toasted oak wood. J Agric Food Chem
47:4310–4318
Conflict of interest The authors declare that they have no competing 16. Spillman PJ, Sefton MA, Gawel R (2004) The effect of oak wood
interest. source, location of seasoning and coopering on the composition

13
258 Eur Food Res Technol (2018) 244:245–258

of volatile compounds in oak-matured wines. Aust J Grape Wine 31. Alañón ME, Díaz-Maroto MC, Pérez-Coello MS (2012) Analy-
Res 10:216–226 sis of volatile composition of toasted and non-toasted commer-
17. Günther C, Mosandl A (1986) Stereoisomere Aromastoffe, XII. cial chips by GC-MS after an accelerated solvent extraction
3-Methyl-4-octanolid “Quercuslacton, Whiskylacton” Struk- method. Int J Food Sci Technol 47:816–826
tur und Eigenschaften der Stereoisomeren. Liebigs Ann Chem 32. Collins TS, Miles JL, Boulton RB, Ebeler SE (2015) Targeted
1986:2112–2122 volatile composition of oak wood samples taken during toasting
18. Abbott N, Puech JL, Bayonove C, Baumes R (1995) Determina- at a commercial cooperage. Tetrahedron 71:2971–2982
tion of the aroma threshold of the cis and trans racemic forms of 33. Vivas N, Saint-Cricq de Gaulejac N (1998) Influence de la durée
β-methyl-γ-octalactone by gas chromatography-sniffing analysis. d´utilisation des barriques sur leurs apports aux vins. Actes du
Am J Enol Vitic 46:292–294 Colloque Sciences et Techniques de la Tonnellerie. Vigne et Vin
19. Chatonnet P, Dubourdieu D (1998) Comparative study of the publications Internationales, Bordeaux, pp 65–74
characteristics of American white oak (Quercus alba) and Euro- 34. García R, Soares B, Barrocas-Dias C, Costa-Freitas AM, Cabrita
pean oak (Quercus petraea and Q. robur) for production of bar- MJ (2012) Phenolic and furanic compounds of Portuguese chest-
rels used in barrel aging of wines. Am J Enol Vitic 49:79–85 nut and French, American and Portuguese oak wood chips. Eur
20. Chira K, Teissedre PL (2015) Chemical and sensory evaluation Food Res Technol 235:457–467
of wine matured in oak barrel: effect of oak species involved and 35. Navarro M, Kontoudakis N, Gómez-Alonso S, García-Romero
toasting process. Eur Food Res Technol 240:533–547 E, Canals JM, Hermosín-Gutíerrez I, Zamora F (2016) Influence
21. Masson E, Baumes R, Moutounet M, Puech JL (2000) The effect of the botanical origin and toasting level on the ellagitannin con-
of kiln-drying on the levels of ellagitannins and volatile com- tent of wines aged in new and used oak barrels. Food Res Int
pounds of European oak (Quercus petraea Liebl.) stave wood. 87:197–203
Am J Enol Vitic 51:201–214 36. Spillman PJ, Sefton MA, Gawel R (2004) The effect of oak wood
22. Fernández de Simón B, Cadahía E, Muiño I, Del Álamo M, source, location of seasoning and coopering on the composition
Nevares I (2010) Volatile composition of toasted oak chips of volatile compounds in oak-matured wines. Aust J Grape Wine
and staves and of red wine aged with them. Am J Enol Vitic Res 10:216–226
61:157–165 37. Saénz-Navajas MP, Campo E, Cullere L, Fernández-Zurbano
23. Chira K, Teissedre PL (2013) Extraction of oak volatiles and P, Valentin D, Ferreira V (2010) Effects of the nonvolatile
ellagitannins compounds and sensory profile of wine aged with matrix on the aroma perception of wine. J Agric Food Chem
French winewoods subjected to different toasting methods: 58:5574–5585
behaviour during storage. Food Chem 140:168–177 38. O.I.V. Organisation Internationale de la Vigne et du Vin (2014)
24. Towey JP, Waterhouse AL (1996) The extraction of volatile com- Methods of analysis of wines and must. http://www.oiv.int/oiv/
pounds from French and American oak barrels in Chardonnay info/enmethodesinternationalesvin
during three successive vintages. Am J Enol Vitic 47:163–172 39. Ibarz M, Ferreira V, Hernández-Orte P, Loscos N, Cacho J
25. Fernández de Simón B, Cadahía E, Sanz M, Poveda P, Pérez- (2006) Optimization and evaluation of a procedure for the gas
Magariño S, Ortega-Heras M, González-Huerta C (2008) Vola- chromatographic-mass spectrometric analysis of the aromas gen-
tile compounds and sensorial characterization of wines from erated by fast acid hydrolysis of flavors precursors extracted from
four Spanish denominations of origin aged in Spanish rebollo grapes. J Chromatogr A 1116:217–229
(Quercus pyrenaica Wild.) oak wood barrels. J Agric Food 40. AENOR (2010) Manual de normas UNE. Análisis sensorial
Chem 56:9046–9055 AENOR, Madrid
26. Masson G, Puech JL, Moutounet M (1994) Localization of the 41. ISO-3591 (1997) Sensory analysis - Apparatus - wine tasting
ellagitannins in the tissues of Quercus robur and Quercus pet- glass. International Organization for Standardization, Geneva
raea woods. Phytochemistry 37:1245–1249 42. ISO-4120 (2004) Sensory analysis—methodology—triangle test.
27. Díaz-Maroto MC, Guchu E, Castro-Vázquez L, de Torres C, International Organization for Standardization, Geneva
Pérez-Coello MS (2008) Aroma active compounds of Ameri- 43. Escudero A, Gogorza B, Melús MA, Ortín N, Cacho J, Ferreira
can, French, Hungarian and Russian oak woods studied by Gas V (2004) Characterization of the Aroma of a wine from Mac-
Chromatography-Mass Spectrometry and Gas Chromatography- cabeo. Key role played by compounds with Low Odor activity
Olfactometry. Flavour Frag J 23:93–98 values. J Agric Food Chem 52:3516–3524
28. Vivas N (2002) Travail et traitement du bois en tonnellerie. 44. Tucker LR (1996) Some mathematical notes on three-mode fac-
Manuel de tonnellerie à l’usage des utilisateurs de futaille, pp. tor analysis. Psychometrika 31:279–311
169–179. Féret, Bordeaux 45. Tomic O, Luciano G, Nilsen A, Hyldig G, Lorensen K, Næs T
29. Schumacher R, Alañón ME, Castro-Vázquez L, Pérez-Coello (2010) Analysing sensory panel performance in a proficiency
MS, Díaz-Maroto C (2013) Evaluation of oak chips treatment on test using the PanelCheck software. Eur Food Res Technol
volatile composition and sensory characteristics of Merlot wine. 230:497–511
J Food Qual 36:1–9 46. PanelCheck software (2006) Nofima Mat, Ås, Norway. http://
30. Jordão AM, Ricardo Da Silva JM, Laureano O, Adams A, www.panelcheck.com
Demyttenaere J, Verhé R, De Kimpe N (2006) Volatile com- 47. Jackson RS (2002) Quantitative (technical) wine assessment.
position analysis by solid-phase microextraction applied to oak In: Taylor SL (ed) Wine tasting. A profession. Academic Press,
wood used in cooperage (Quercus pyrenaica and Quercus pet- London
raea): effect of botanical species and toasting process. J Wood
Sci 52:514–521

13