Vous êtes sur la page 1sur 7

Sensory Properties of Californian and Imported

Extra Virgin Olive Oils


Claudia Delgado and Jean-Xavier Guinard

Abstract: Production and consumption of extra-virgin olive has been increasing in the United States, particularly in
California. The objective of this study was to compare the sensory characteristics of 22 extra virgin olive oils (EVOO)
from California, Italy, Spain, Chile, and Australia using a generic descriptive analysis. A total of 22 sensory attributes
were identified and defined by the descriptive panel. With the exception of thick and citrus, all sensory attributes
were significantly different among the oils. Canonical Variate Analysis (CVA) showed that California oils differed from
some imported EVOOs, mainly by their absence of defects. A second analysis, of only those attributes included in the
International Olive Council (IOC) official scorecard, provided a less detailed description of the samples and did not allow
for a full characterization of the oils. While the IOC attributes allowed for faster classification in terms of clean versus
defective EVOOs, the more comprehensive descriptive analysis provided both more information and a more refined
classification of the samples. Variety and region of origin were important factors in the classification of both Californian
and imported EVOOs.
Keywords: descriptive analysis, extra virgin olive oil, imported compared with domestic olive oils, International Olive
Council sensory certification

Practical Application: Measuring olive oil sensory quality using the IOC methodpositive attributes of fruitiness,
bitterness, and pungency, and defects including fusty, musty, winey, and rancidallows for the certification of oils as
extra virgin but it provides limited information on the sensory characteristics of the oils. A full descriptive profile, on the
other hand, provides information that can be used by producers in the processing and marketing of their oils, and is a
useful tool in the education of consumers about the wide range of (positive) sensory attributes in EVOO and the various
sensory styles of EVOO.
S: Sensory & Food
Quality

Introduction org/COIAdmin/resources/pdf/MarketMarch08.pdf). Currently,


Olive oil has gained the interest of consumers because of its mul- the United States ranks fourth worldwide in olive oil consump-
tiple culinary applications and nutritional content (high content of tion; however, most of the oil that is consumed is imported (Vossen
oleic acid and balance between monounsaturated and polyunsat- 2007). In terms of olive oil production, the California Olive Oil
urated fatty acids) and its content of minor compounds (polyphe- Council (COOC) states that California has the most developed
nols and tocopherols that have antioxidant properties and volatile olive oil industry among the 50 United States. California has over
compounds that bestow an oil its unique flavor) (Bendini and oth- 200 producers of extra virgin olive oil (EVOO) and accounts for
ers, 2007). These properties might fulfill consumer expectations 99% of the U.S. olive oil production. It is therefore important to
regarding healthier, more nutritious, more flavorful, and more understand how the sensory profiles of California oils compare to
natural food even in countries outside the Mediterranean region those of imported oils.
where most olive oil is consumed (Zampounis, 2006). Since olive The IOC defines EVOO as the commercial grade for olive oil
oil is a relatively new product to most American consumers, it that has been extracted by physical processes alone, without the
is necessary for American producers to acquire a broader under- use of solvents or refining agents; further, EVOO must have an
standing of the sensory properties of olive oils, and how they affect acidity below 0.8% (as oleic acid) and flavor that results purely
sensory quality and in turn consumer preferences. from the fruits metabolism, which yields fruity, bitter, and pun-
According to the Olive Products Market report (March 2008), a gent characteristics, with no taste defects (fusty, musty, winey, or
publication of the Intl. Olive Council (IOC), there was a 92.41% rancid). Physicochemical (free fatty acid [FFA]) content, perox-
increase, of 104700 tons, in the consumption of olive oil in the ide value (PV), UV specific extinction coefficients (K232 , K270 ),
United States between the 1992/93 to 1996/97 period and the and the sensory properties are measured following official IOC
2002/03 to 2006/07 period (http://www.internationaloliveoil. protocols as indicators of the quality of the oil. Both the bitter
taste and the trigeminal sensation of pungency stimulate the taste
receptors and the trigeminal nerve; those sensory properties have
MS 20100355 Submitted 3/31/2010, Accepted 12/14/2010. Authors are with been associated with phenolic compounds present in EVOO (Ben-
Univ. of California, Davis, Dept. of Food Science and Technology, One Shields
Ave., Davis, CA 95616-8598, U.S.A. Direct inquiries to author Guinard (E-mail: dini and others 2007). Volatile compounds stimulate the olfactory
jxguinard@ucdavis.edu). receptors causing the perception of the whole aroma of the virgin
olive oil (Angerosa and others 2004).

C 2011 Institute of Food Technologists


 R

S170 Journal of Food Science r Vol. 76, Nr. 3, 2011 doi: 10.1111/j.1750-3841.2011.02040.x
Further reproduction without permission is prohibited
Sensory properties of extra virgin olive oil . . .

The sensory properties are evaluated by a sensory panel that is Sensory panel. A panel of 18 judges (14 female, 4 male; age
trained using the IOC procedures (COI/T.20/Doc. nr 15/Rev 2) range 20 to 35 y) from the Univ. of California, Davis, was em-
and accredited by the IOC (IOC 2007). This method is a standard- ployed. FIZZ software (Biosystemes, Couternon, France) was used
ized procedure used to classify and establish the commercial grade to construct an automated session. Each panelist completed 10
of olive oil according to current market regulations. The IOC training sessions. In those sessions, panelists worked on the devel-
procedure is well suited to identify the main sensory properties in opment of the language to describe EVOO, concept alignment,
olive oil as well as to establish the commercial grade of the prod- and agreement among terms and sequence of the attributes. The
uct. However, given the complexities in olive oil composition, the first 3 sessions were dedicated to developing the terminology. The
IOC approach may be limited in its ability to evaluate in detail next 4 sessions were devoted to concept alignment. The objective
the sensory profile of EVOOs, as it overlooks many attributes that was to refine the list of attributes while avoiding redundancies
may influence taste preferences and market segmentation. and keeping the most important terms to characterize the sam-
Several factors affect olive oil flavorfor instance, cultivar, ples. In these sessions, panelists were exposed to references that
geographic region, agronomic techniques, climatic conditions, helped define the attributes. Panelists decided to use a contin-
fruit maturity, fruit storage, processing method, and oil storage uous unstructured line scale of 10 points, ranging from low to
(Angerosa and others 2004; Kalua and others 2007). Due to the high intensity. The sequence of the attributes was established by
defects that may be introduced by the presence of microorganisms the panelists, starting with the main aromas, then defects, and fi-
and the oxidation of fatty acids, the most critical of these factors nally taste and mouthfeel attributes. During the last 2 sessions, the
is the storage of the olive fruit and of the oil (Angerosa and others panelists became familiar with the final set of attributes and their
2004; Rotondi and others 2004; Servili and others 2004; Tripoli definitions. To assess panel readiness after the training, a duplicate
and others 2004; Kalua and others 2007). Rancidity is one of the evaluation was conducted of a set of 5 samples randomly selected
major defects in fats and oils, and olive oil is no exception. This from the 22 EVOOs. Univariate (analysis of variance; ANOVA)
defect develops under improper storage conditions, any condition and multivariate statistics (canonical variate analysis; CVA) were
that exposes the oil to oxygen, high temperatures, or light (Ben- used to measure judge reproducibility, discrimination among prod-
dini and others 2009). Other defects in olive oil are characterized ucts, and agreement in rating of the attributes. More training was
by fusty, winey, or musty aromas, which result when the improper provided to those judges whose evaluations showed disagreement
handling of the olive fruit allows the growth of microorganisms, or confusion of the attributes. After these sessions, the panelists
whose metabolic products alter the flavor of the olive oil (Morales rated the 22 EVOO samples. A total of 5 samples were evaluated
and others 2000, 2005). per session. The order of presentation of the samples was com-
While other researchers have used descriptive analysis to charac- pletely randomized and a Williams Latin square design provided
terize EVOOs from the same regions of origin or cultivars (Apari- by the FIZZ software was applied to balance out any potential
cio and others 1992, 1996, 1997; Aparicio and Morales 1995; serving order or carry-over effects. The samples were evaluated in
Morales and others 1995; Angerosa and others 2000; Guerrero triplicate.
and others 2001; Angerosa 2002; Aparicio and Luna 2002; Luna Data analysis. SAS version 9.1 (SAS Inst. Inc., Cary, N.C.,
and others 2006; Ollivier and others 2006; DiBella and others U.S.A.; 2002 to 2008) was used for all statistical analyses. The

S: Sensory & Food


2007), the majority of them have concentrated on sensory char- level of confidence was set at = 0.05. To describe the differ-
acteristics produced by the EVOOs volatile compounds. As far as ences among products and to assess panel performance, univariate

Quality
the authors know, there is no published research on the sensory
properties of California EVOOs.
The main objective of this research was to characterize the Table 1EVOO codification.
sensory properties of a set of California EVOOs and compare
them to a set of imported EVOOs. A generic descriptive analysis ID nr Country of origin Variety
(Lawless and Heymann 1998) was applied for this purpose. A U1 United States (California) Arbequina
second objective was to compare the profiles obtained by using U2 United States (California) Blend
U3 United States (California) Arbequina/Arbosana/Koroneiki
only the attributes measured by the IOC compared with the full
U4 United States (California) Sevillano
set of attributes in the generic descriptive analysis. U5 United States (California) Frantoio
U6 United States (California) Mission/Manzanillo/Sevillano/
Materials and Methods Barouni/ Ascolano
Samples. A total of 22 samples of EVOO from different culti- U7 United States (California) Manzanillo/Mission
U8 United States (California) Blend
vars, geographic regions (California, Italy, Spain, Chile, and Aus- U9 United States (California) Arbequina
tralia), and producers were used for this research. Samples were U10 United States (California) Frantoio/Leccino/
coded using 3 random digit numbers. The samples, 15 mL of each Pendolino/Coratina
oil at room temperature, were served in votive glasses (Libbey 1965 U11 United States (California) Mission
S1 Spain Picual
model, 4 3/4 oz capacity, 2 inch diameter) covered with a trans- S2 Spain Hojiblanca
parent plastic lid (Solo Cup model PL2 2-ounce). The samples C1 Chile Picual
were poured about 30 min before the tasting. Water, previously C2 Chile Arbequina

R
filtered in the Millipore Milli-Q water filtration system, slices I1 Italy Frantoio
of Granny Smith apple, and unsalted crackers were provided as I2 Italy Taggiasca (Late harvest)
I3 Italy Taggiasca
palate cleansers. With the exception of the generic brand oils (GS I4 Italy Picholino
and GI), which were bought in a local supermarket, the oils were A1 Australia Hojiblanca
donated by either producers or distributors in California. Upon GS Spain Generic Brand (Oils from
receipt, the oils were stored in dark and cool conditions. See Table several countries)
GI Italy Generic Brand (Oils from Italy)
1 for a detailed description of the products.

Vol. 76, Nr. 3, 2011 r Journal of Food Science S171


Sensory properties of extra virgin olive oil . . .

(3-way analysis of variance), multiple mean comparisons (Fishers this effect is not critical and may be due to different judges estab-
LSD), and multivariate analysesmultivariate analysis of variance lishing different initial baselines on the rating scale. The replication
(MANOVA) and CVA were used to evaluate panel ratings. CVA effects were not significant (P > 0.05) for the majority of the at-
was selected because it restricts the classification of the sensory tributes evaluated, except for the grassy, citrusy, minty, and bitter
properties of the EVOOs to fewer groups and categories (see attributes. Particularly in the case of bitterness, the ANOVA may
Figure 1A to 1C), and also reveals significant differences among have included carry-over effects. Even though this research took
the products. Each circumference represents the 95% confidence into account the effects of carry-over through the Williams Latin
interval (Chatfield and Collins 1980), with 2 overlapping circles Square design, judges perception of bitterness may have decreased
indicating the products are not significantly different. To compare more slowly (Sinesio and others 2005), or the influence of the
the outcomes of a full descriptive analysis and of the IOCs sen- matrix lipid may have affected the perception of bitterness during
sory certification protocol, the same statistical procedure (CVA) repeated tastings of the same oils (Garca-Mesa and others 2008).
was applied a second time but only to those attributes measured Regarding the grassy attribute, previous research has shown that a
in the IOC method. perception of fresh cut grass or green notes is associated with bit-
terness (Caporale and others 2004). Therefore, carry-over effects
Results and Discussion should be studied in evaluating perceptions of bitterness. Given
the complexity of EVOO and the high number of judges on the
ANOVA and panel performance panel, it does not come as a surprise that the judge product inter-
A total of 22 attributes (see Table 2) were defined by the judges action was significant (P < 0.05) for all the attributes, indicating
to describe the different profiles of the EVOO samples. In general, that there was some disagreement in their ranking of the sam-
the panel performance was satisfactory. Table 3 shows the ANOVA ple oils properties. To evaluate the effect of the judge product
F-ratios and Table 4 the overall mean for each attribute. Overall interaction in the discrimination of the samples, a pseudo
the variance among judges was significant (P < 0.05); however, mix model (F-ratio = MS Product/MS Judge Product) was
S: Sensory & Food
Quality

Figure 1EVOO Canonical Variate Analysis (Full descriptors). Canonical Analysis of Variance (CVA) for the significant attributes (a = 20) across
products (n = 22 EVOO). Products are represented by spheres; please refer to Table 1 for a description of the EVOOs. Vectors represent attributes.
Overlapping spheres indicate no significant differences among products. (A) CV1 compared with CV2; (B) CV1 compared with CV3, and (C) CV2
compared with CV3.

S172 Journal of Food Science r Vol. 76, Nr. 3, 2011


Sensory properties of extra virgin olive oil . . .

calculated for all the attributes. The interaction judge product tributes are not as significant as those of aroma and taste (Angerosa
only affected the perception of a citrus aroma (P > 0.05); for the 2000).
rest of the attributes, the differences among products was still sig-
nificant (P < 0.05). The perception of thick texture was not signif- Canonical analysis of variance for all attributes
icant either before or after the pseudo mix model, a finding which Figure 1A to 1C, presents the CVA analysis for the 22 EVOOs.
confirms that in the case of olive oil, a liquid product, textural at- Canonical variate 1 (CV1) accounts for 29.53% of the variance,

Table 2Vocabulary and references used for training.


Definition Referencea
Grassy Fresh cut grass 1 drop of 1-cis-3-hexenol (Courtesy of Robertet Flavors, Mexico)
in 200mL of Carapelli Extra Light Olive Oil
Green fruit Unripe fruit (Green olive, banana, green Verbal description.
apple, and so on)
Ripe fruit Ripe fruit (olives, apple, banana, berries, Verbal description
and so on) Sweet aroma
Tropical fruit Aroma of Pineapple, Guava, Lychee, 100% Spanish EVOO from Wholefoods.
Mango, and so on
Green tomato Aroma of green tomato/tomato leafs 1492 EVOO
and tomato seeds
Tea Aroma of green tea Bigelow Green Tea (8 bags/5g per 4 bags) in 100 mL of Carapelli
Extra Light Olive Oil
Herbs Fresh herbs (for example, rosemary, Basil:10 g of fresh basil in a coffee filter bag in 150 mL of Carapelli Extra
basil, oregano) Light Olive Oil Rosemary:10 g fresh rosemary in a coffee filter bag) in
150 mL of Carapelli Extra Light Olive Oil
Citrus Lemon aroma Bartolini Lemon Oil EVOO flavored with lemon (Product of Italy)
Floral Flowers aroma 1 g of Miracle by Lancome Solid Fragrance in 100 mL of Carapelli
Extra Light Olive Oil
Nutty Almonds, pecans, walnuts, and so on Pecans: 28 g pecans in a coffee filter bag in 150 mL of Carapelli Extra light
Olive Oil Almond: Sweet Almond Oil International Collection
Butter Aroma of butter Fresh Butter bar Imperial Butter from Unilever
Spicy Spices, for example, pepper/pungent smell Verbal description
Mint Mint/Eucalyptus aroma 1 drop artificial flavor (Courtesy of Robertet Flavors, Mexico) in 100 mL
of Carapelli Extra Light Olive Oil
Musty [musty/humidity/earthy] Verbal description
Fusty Acetic acid fermented aroma Verbal description
Winey Aroma of alcohol, wine, and/or vinegar 10 mL Vinegar from olives in 150 mL of Carapelli Extra Light Olive Oil
30mL Vodka in 150mL of Carapelli Extra Light Olive Oil
Rancid Not fresh oil. Aroma of oxidized oil Canola Oil (2 y old) storage at 80F.

S: Sensory & Food


Bitter Bitter taste Caffeine (0.7g/L) Fisher Scientific
Pungent Trigeminal sensation /chemical irritation Capsaicin (0.08mg/L) Fisher Scientific

Quality
Thick Thinthick Verbal description
Astringent Dryness in the mouth Tannic Acid (1.1g/L) Sigma-Aldrich
a
Fresh ingredients were weighed over a coffee filter bag, sealed, stapled, and dipped into the oil. The standards were prepared at least 3 to 4 d in advance to let the oil absorb the
aromas of the ingredients.

Table 3ANOVA F-ratio for 22 attributes rated by the descriptive analysis panel for 22 EVOO sample attributes.
Replication Product Judge replication Judge product Replication product F-value
Attribute Judge F-ratio F-ratio F-ratio F-ratio F-ratio F-ratio producta
Grassy 82.05 3.34 55.21 1.57 5.11 0.67 NS 10.80
Green fruit 112.90 0.84NS 35.67 0.76 NS 9.08 1.16 NS 3.93
Ripe fruit 47.78 0.57NS 21.93 2.06 8.86 0.94 NS 2.48
Tropical fruit 33.64 0.24NS 79.18 2.14 5.83 0.89 NS 13.57
Green tomato 72.16 1.51NS 39.22 2.10 5.35 0.96 NS 7.33
Tea 115.87 1.68NS 19.88 1.49 8.16 0.69 NS 2.44
Herbs 100.55 0.24NS 73.25 1.65 5.41 1.02 NS 13.53
Citrus 39.77 10.89 3.59 1.42 NS 2.30 1.27 NS 1.56NS
Floral 79.00 1.44NS 31.39 0.96 NS 10.01 0.82 NS 3.14
Nutty 33.48 0.34NS 28.38 1.02 NS 6.97 0.89 NS 4.07
Butter 65.48 0.89NS 18.32 1.08 NS 9.52 1.19 NS 1.93
Spicy 59.71 2.66NS 9.68 1.60 3.05 1.16 NS 3.18
Mint 39.65 4.11 8.83 2.44 2.56 1.35 NS 3.45
Wood 16.94 0.35NS 4.20 1.26 NS 1.82 1.14 NS 2.31
Musty 15.27 1.48NS 23.45 0.74 NS 6.16 1.13 NS 3.81
Fusty 9.08 0.95NS 16.09 1.28 NS 4.79 1.44 3.36
Winey 9.08 0.29NS 29.84 0.86 NS 4.68 0.53 NS 6.38
Rancid 30.64 0.07NS 55.55 0.92 NS 8.52 1.06 NS 6.52
Bitter 24.47 5.82 83.86 2.51 1.91 0.97 NS 44.01
Pungent 28.43 0.31NS 33.14 1.90 1.75 0.87 NS 18.97
Thick 35.53 0.79NS 1.30 NS 3.17 1.38 1.55 NS 0.95NS
Astringent 54.30 0.88NS 9.58 4.53 1.61 0.94 NS 5.95
a
AfterPseudo model F = MSproduct/MS product judge.

Significant at P < 0.05; Significant at P < 0.01; Significant at P < 0.001; NS = non significant.

Vol. 76, Nr. 3, 2011 r Journal of Food Science S173


Sensory properties of extra virgin olive oil . . .

Canonical variate 2 (CV2) accounts for 23.28% of the variance, nificantly different from imported oils A1, GI, and I4. U5 was
and Canonical variate 3 (CV3) accounts for 17.05% of the vari- associated with the nutty attribute and was the only California
ance, so together the 3 canonicals account for 69.86% of the total EVOO with a slightly rancid aroma. This oil was not significantly
variance. The 3 dimensions were significant (Bartlet test P < 0.05). different from oils U3, GI, and A1. U11 was the only oil from
There was a significant simultaneous effect among all the attributes California that differed significantly from the rest of the samples,
and the products (MANOVA, Wilks Lambda, F = 18.24; df 462, as it was characterized mainly by its herb and floral aromas, with
P < 0.05). no bitterness or pungency.
The attribute terms grassy, green fruit, green tomato, tropical Spanish EVOOs S1 and S2 were not significantly different, and
fruit, tea, and floral were the variables most closely related to were characterized by their tropical fruit, herb, mint, floral, bitter,
CV1 (first dimension), while those of nutty, bitter, pungent, fusty, and pungent attributes. Both of the generic brands, GI and GS,
and rancid were more closely related to CV2 (second dimension). presented some defects; however, the musty and winey qualities
Finally, ripe fruit, herbs, butter, mint, astringent, musty, and winey were even more evident in the generic brand from Spain. In addi-
were the attributes most important for CV3 (third dimension). tion to the presence of defects, this oil was characterized by floral
Short vectors showed that certain attributesnamely, spicy and and butter properties. The Italian generic oil proved less defec-
wooddiscriminated less among the oils (see Figure 1A to 1C). tive, although it was noted as having slight rancidity and fustiness.
The first 2 dimensions (CV1 and CV2) show the main differ- Among the 4 Italian EVOOs, 3 showed no significant differences
ences in the EVOO samples (Figure 1B). California EVOOs can (I1, I2, and I3), and were characterized by buttery and nutty fla-
be divided into 2 groups, the first characterized mainly by grassy, vors that were marred somewhat by rancidity. Characterized by
green fruit, green tomato, bitter, and pungent attributes and the properties of slight green fruit, grass, green tomato, and ripe fruit,
second characterized by lack of bitterness and pungency and a with somewhat astringent and tea-like characteristics, I4 was ex-
presence of nutty, tea, and ripe fruit characteristics. U8 and U10 perienced as being significantly different from the other 3 Italian
do not differ significantly; these oils had the highest means for samples.
bitter, pungent, grassy, and green fruit attributes. Because of the significant differences among the imported olive
Chilean and Australian EVOOs displayed similar sensory prop- oils in the design, we did not feel it was justified to treat California-
erties to those found in California EVOOs. All 3 regions are rela- produced and imported oils as 2 blocks or factors in the experi-
tively new olive oil producers. New producers are more likely to mental design and statistical analyses.
apply similar harvesting/process technologies. Despite California As revealed in the CVA results (Figure 1A), the cultivar proved
being one of the newest and fastest growing producers of EVOOs, an important factor in the characterization of the EVOOs. There
the majority of California EVOOs used in this study displayed was a tendency for the oils from the olive varieties Arbequina (oils
positive characteristics, making them a potentially important pro- U9, C2, and U1), Frantoio (U5 and I1), and Taggiasca (I2 and
ducer in the near future. Chilean oils, C1 and C2, were similar; I3) to present similar sensory properties, while the oils made from
there were no significant differences among oils U9, U1 and C2, Hojiblanca (S2 and A1) and Picual (C1 and S1) cultivars did not
which were evaluated as having tea, nutty, green tomato, grassy, show similar sensory profiles. The latter result may have been due
ripe, or somewhat green fruity characteristics with no bitterness or to differences in fruit maturity, as has been shown by previous
S: Sensory & Food

pungency. For its part C1 was similar to U6, both being described experimenters (Morales and others 1995; Stefanoudaki and others
as having green tomato, green fruit, grassy, astringent, bitter, and 2000; Rotondi and others 2004).
Quality

pungent attributes. The Australian oil A1 showed no significant Figure 1B includes additional information regarding the clas-
differences from oils U4 and U3, with sensory properties described sification of the oils. Based on country of origin, there were 3
as nutty and ripe fruit with no bitterness or pungency. The main major groups, one composed mainly of European oils (S2, GS,
difference in U4 was in its intensity of tea flavor, similar to C2 S1, I1, I3, and I2), the second of oils mainly from California,
and U1. U7 and U2 had similar sensory properties, and both were Chile, and Australia, representing the newer olive oil producers in
minty, somewhat bitter, pungent, and spicy. They were not sig- the world (I4 is the only European oil in this group), and U11,

Table 4Overall meansa,b for 22 attributes rated by the descriptive analysis panel for 22 EVOO samples.
Green Ripe Tropical Green
EVOO Grassy fruit fruit fruit Tomato Tea Herbs Floral Nutty Butter Spicy Mint Wood Musty Fusty Winey Rancid Bitter Pungent Astringent
A1 3.49DE 2.79AB 2.74A 2.38C 2.29C 1.35HIJKL 0.89H 1.66BC 1.84DE 0.64GHIJ 1.97BC 0.52CD 0.70FG 0.05FG 0.09DE 0.11FG 0.07F 1.09J 1.99G 1.54GH
C1 3.18EF 2.85AB 1.02KL 0.57HIJ 4.68A 1.28IJKLM 1.06DEFGH 0.97FGH 0.83KL 0.42JKL 2.13B 0.51CD 0.78EFG 0.12EFG 0.12DE 0.17EFG 0.06F 2.96GH 3.12CD 2.18DE
C2 4.23B 2.84AB 1.73EFGH 0.24J 2.39C 2.28BC 1.18DEFGH 0.68IJKL 1.79DEF 0.55HIJK 2.19B 0.46CDEF 0.91EFG 0.12EFG 0.06E 0.09FG 0.06F 1.71I 2.31FG 1.75EFGH
GI 2.90FG 1.44I 1.15IJK 0.99FG 1.57EF 1.01M 0.89H 0.52L 1.46FGH 1.57A 1.24EFG 0.21EF 1.02DEFG 0.64B 0.24CD 0.53CD 0.59D 3.23FG 3.65CD 2.90ABC
GS 2.18HI 1.89GH 1.94DEF 2.56C 1.84DE 1.05LM 1.39CD 1.81B 1.43GH 1.03CD 1.83BCD 0.56BCD 0.80EFG 0.98A 0.93A 1.74A 0.48DE 2.85GH 2.25G 1.99DEFG
I1 1.98I 1.62HI 1.45HIJ 1.32DEF 1.53EFG 1.23JKLM 0.93GH 1.51CD 2.05CDE 1.24BC 1.21EFG 0.44CDEF 0.83EFG 0.39CD 0.35C 0.63C 1.49A 2.68GH 2.35EFG 2.00DEFG
I2 1.96I 0.85K 2.34BC 1.04FG 1.02H 1.37GHIJK 1.10DEFGH 1.63BC 2.09CDE 1.18BC 1.01G 0.46CDEF 1.36BCD 0.50BC 0.66B 0.63C 0.79C 0.89J 0.44J 1.46GH
I3 2.59GH 1.35IJ 1.88DEFG 1.27EF 1.57EF 1.65EFGH 1.53C 1.23FE 3.16A 1.28B 1.38EFG 0.39DEF 0.95DEFG 1.02A 0.79AB 0.88B 1.33B 1.28IJ 0.91IJ 1.74EFGH
I4 3.19EF 2.53BCDE 1.26IJK 1.07FG 2.15CD 1.68EFG 1.31CDE 0.90GHI 1.03JK 0.44IJKL 1.35EFG 0.52CD 0.92EFG 0.17EFG 0.18CDE 0.27EF 0.18F 2.51H 1.82GH 1.81EFGH
S1 1.94I 1.01K 2.89A 4.05B 1.12GH 1.35HIJKL 2.46B 1.64BC 0.83KL 0.37KL 1.18EFG 1.26A 1.70AB 0.13EFG 0.05E 0.12FG 0.06F 4.99D 3.74C 2.78BC
S2 2.16HI 2.50CDE 2.26BCD 5.11A 1.08H 1.08KLM 1.21CDEFGH 1.78BC 0.83KL 0.62HIJ 1.09FG 0.72BC 0.62G 0.21EF 0.15DE 0.33DE 0.18F 6.09C 4.60B 2.87ABC
U1 4.44B 2.36DE 1.66FGH 0.84GH 2.17CD 2.47B 1.27CDEF 0.76HIJKL 1.41GHI 0.92DE 1.89BCD 0.39DEF 0.88EFG 0.25DE 0.14DE 0.11FG 0.13F 1.45IJ 2.27G 2.10DEF
U10 4.98A 2.66ABCD 1.21IJK 0.81GHI 3.37B 1.49FGHIJ 1.09DEFGH 0.86GHIJ 1.07IJK 0.33KL 2.04B 0.51CD 0.95DEFG 0.03G 0.06E 0.08FG 0.05F 7.60A 5.44A 3.29AB
U11 1.47J 1.11JK 1.45HIJ 1.58DE 1.24FGH 1.19JKLM 5.70A 2.49A 0.81KL 0.30L 1.31EFG 1.55A 0.82EFG 0.06FG 0.17DE 0.07FG 0.09F 0.95J 0.43J 1.28H
U2 1.84IJ 0.95K 0.69L 1.27EF 1.18FGH 2.13CD 1.02EFGH 0.59KL 2.19C 1.39AB 1.13EFG 0.33DEF 1.52ABC 0.26DE 0.10DE 0.13EFG 0.08F 5.47CD 4.46B 3.40A
U3 3.77CD 2.02FG 2.66AB 1.65D 1.51EFG 2.80A 1.23CDEFG 1.57BC 1.75EFG 0.89DEF 1.45DEF 0.33DEF 0.91EFG 0.06FG 0.04E 0.07G 0.10F 0.96J 0.73IJ 1.53GH
U4 4.08BC 1.43IJ 1.09JK 0.48HIJ 2.40C 1.87DE 1.00EFGH 0.55KL 2.74B 0.86DEFG 2.08B 0.42DEF 1.82A 0.14EFG 0.09DE 0.06G 0.04F 1.26IJ 1.01IJ 1.60FGH
U5 2.26HI 1.88GH 2.07CDE 1.02FG 1.28FGH 1.76EF 0.91GH 1.25DE 2.10CD 0.66FGH 1.04FG 0.17F 0.85EFG 0.06FG 0.04E 0.12FG 0.40E 1.03J 1.21HI 1.74EFGH
U6 4.06BC 2.31EF 1.54GHI 0.80GHI 2.01CD 2.20BC 1.29CDEF 0.81HIJK 1.28HIJ 0.72EFGH 1.89BCD 0.51CDE 1.22CDE 0.08FG 0.16DE 0.10FG 0.10F 3.77EF 3.02DE 2.48CD
U7 2.85FG 1.59HI 2.34BC 1.59DE 2.06CD 1.57EFGHI 1.54C 1.11FEG 1.33HIJ 0.76EFGH 1.54CDE 0.51CD 0.87EFG 0.11EFG 0.06E 0.16EFG 0.10F 4.33E 3.00DEF 2.52CD
U8 5.13A 2.90A 0.65L 0.33J 3.47B 1.26JKLM 2.52B 0.61JKL 0.66L 0.65GHI 2.89A 0.85B 0.68G 0.09FG 0.08DE 0.10FG 0.03F 6.83B 4.74AB 2.82BC
U9 4.92A 2.72ABC 1.23IJK 0.46JI 3.45B 2.10CD 0.96FGH 0.68IJKL 2.21C 0.90DE 2.01B 0.34DEF 1.14CDEF 0.07FG 0.09DE 0.07G 0.06F 2.48H 2.25G 2.21DE
LSD 0.429 0.324 0.395 0.371 0.421 0.311 0.338 0.264 0.349 0.232 0.438 0.297 0.438 0.165 0.175 0.200 0.155 0.624 0.703 0.552
a In columns, means sharing superscript were not significant different P < 0.05. Please refer to Table 1 for a description of the EVOOs.
b The highest mean for each attribute is shown in a black square.

S174 Journal of Food Science r Vol. 76, Nr. 3, 2011


Sensory properties of extra virgin olive oil . . .

which was the only EVOO with no associations to region of ori- A total of 5 out of the 22 oils were identified as defective; besides
gin. The only significant difference between oils from new world having musty, fusty, winey, and/or rancid qualities, those 5 oils are
producers and those from European ones was the absence of de- all imported. The authors do not mean to imply that imported oils
fects in the new world group. While the reason for this difference are in general defective. In this case there might be other factors
cannot be pinpointed, it does highlight the critical role that stor- influencing the presence of defects in this sample. Defects could
age conditions play when imported oil is shipped into the U.S. be due, for instance, to other conditions: (1) 2 of them are generic
Improper storage can lead to alterations in the sensory profile of brands, which therefore may not have the same standard of quality
the EVOO. as more select oils exported to America; and/or (2) there may have
In the case of region of origin, there were similarities in the been improper conditions in storage while the product was being
sensory profiles of the oils produced from within a regionoils shipped, which has been proven to play a significant role in the
from Spain (S1 and S2), from Italy (I1, I2, and I3), and from development of defects in general and of rancidity in particular
Chile (C1 and C2) shared similar sensory propertiesa finding (Morales and others 2005); and/or (3) shelf conditions in the
that coincides with previous research (Aparicio and others 1997; supermarket or store may have exposed the oils to intense light,
Ollivier and others 2006). In addition to country or region of which facilitates oxidation of the oil; and/or (4) because imported
origin, previous research also shows that variety and fruit ripeness oils take a long time to arrive in America, and with the difficulty
are intrinsic factors that have a great influence in the final sensory of monitoring and standardizing shipping and storage conditions,
profile of the EVOO (Angerosa and others 2004; Rotondi and especially proper temperature, light exposure and time, the oils
others 2004; Servili and others 2004; Tripoli and others 2004; may have arrived already aged, degraded, or defective.
Kalua and others 2007).
Evaluations of CV2 and CV3 (Figure 1C) indicate that the Canonical analysis of variance of the IOC method
oils with the highest bitterness and pungency in the whole set of attributes
EVOO samples were U10, S1, U8, and S2; the 2 California oils Figure 2A and 2B presents the CVA analysis for the 22 EVOOs
were similar to the Spanish only in their bitterness and pungency. as measured by IOC standards, using only 8 attributes. Canon-
It is probable that these samples had a higher phenol content, ical variate 1 (CV1) accounts for 46.73% of the variance, while
since phenols have been associated with pungency and bitterness Canonical variate 2 (CV2) accounts for 23.68%, for a total of
(Bendini and others 2007). 70.41% There was a significant difference among the products
There were no significant differences between the generic across all the attributes (MANOVA, Wilks Lambda, F = 18.24;
brands (GI and GS) in terms of dimensions CV2 and CV3. Total df 462, P < 0.05). In this case only the first 2 dimensions were
of 3 of the Italian oils likewise showed no significant differences significant (Bartlet test, P < 0.05). The differences regarding the
in these dimensions. Both the Italian oils and the generic brands samples were mainly explained by the first CV, with the most
were defective, despite some nutty and buttery aromas. important attributes in this dimension being bitter, pungent, ripe

S: Sensory & Food


Quality

Figure 2Canonical analysis of variance (CVA) for the attributes measured in the IOC method across products. (A) Dimension 1 compared with
Dimension 2 (CV1 compared with CV2). Products (n = 22 EVOO) are represented by spheres. Please refer to Table 1 for a description of the EVOOs.
Vectors represent attributes (a = 8). Overlapping spheres indicate no significant differences among products. (B) Biplot of attributes and products
(represented by solid circles) and classification of the products.

Vol. 76, Nr. 3, 2011 r Journal of Food Science S175


Sensory properties of extra virgin olive oil . . .

fruit, and fusty qualities, while those for CV2 were green fruit, tion of flavors and Paul Vossen and Dr. Hildegarde Heymann for
rancid, musty, and winey characteristics. their helpful suggestions and assistance with SAS codes. The panel
Generic brands GS and GI, together with I1, I2, and I3, were of experts consisted of Ramon Aparicio, Nancy Ash, Milagros
identified as defective oils, which coincided with the previous Castro, Sarah Chironi, Thomas Curry, Elena Franceschi,Veronica
CVA for the whole set of attributes; however, these oils have Gaynor, Richard Gawel, Fran Gage, Luis Guerrero, Bruce Golino,
other attributes in common with other EVOOs that are not as fully Louie Gonzalez, John Hadley, Arden Kremer, Nancy Lilly, Julie
explained by the IOC analysis as by the one using the whole set of Menge, Marvin Martin, Frank Menacho, Jeffers Richardson, Deb-
attributes. Part of the lack of precision in the IOC analysis of these orah Rogers, Sandy Sonnenfelt, Paul Vossen, and Dean Wilkinson.
oils is due to a second group of oils, related to each other, that make
sub-categorization of those oils, which was possible in the previous References
CVA analysis, more difficult. Oils U5, U3, U11, U4, C2, and U1 Angerosa F. 2000. Sensory quality of olive oils. In: Harwood J. & Aparicio R., editors. Handbook
of olive oil. analysis and properties. Gaithersburg, Md.: Aspen Publication Inc. p. 35592.
were not significantly different; however, it is not possible using the Angerosa F. 2002. Influence of volatile compounds on virgin olive oil quality evaluated by
IOC methodology to identify the sensory characteristics in those analytical approaches and sensor panels. Eur J Lipid Sci Technol 104(910):63960.
Angerosa F, Mostallino R, Basti C, Vito R. 2000. Virgin olive oil odour notes: their relationships
oils, as the only attribute they had in common was ripe fruit. In with volatile compounds from the lipoxygenase pathway and secoiridoid compounds. Food
addition, oils I4, C2, U9, C1, and U6 were mainly identified as Chem 68:2837.
Angerosa F, Servili M, Selvaggini R, Taticchi A, Esposto S, Montedoro G. 2004. Volatile com-
possessing characteristics of green fruit, while U7, S1, U2, S2, U8, pounds in virgin olive oil: occurrence and their relationship with the quality. J Chromatogra
and U10 were mainly bitter and pungent. Compared to the earlier A 1054(12):1731.
Aparicio R, Morales MT. 1995. Sensory wheels: a statistical technique for comparing QDA
analysis, then, this approach was not helpful even in identifying the panels. J Sci Food Agric 67:24757.
similarities in green fruit characteristics among the last samples. Aparicio R, Luna G. 2002. Characterisation of monovarietal virgin olive oils. Eur J Lipid Sci
Using the attributes mentioned in the IOC method (IOC 2007) Technol 104(910):61427.
Aparicio R, Gutierrez F, Rodriguez Morales J. 1992. Relationship between flavor descriptors
allows a classification of the EVOO into 3 main categories (see and overall grading of analytical panels for virgin olive oil. J Sci Food Agric 58:55562.
Figure 2B), oils that are bitter and pungent; fruity; or defective. Aparicio R, Morales MT, Alonso MV. 1996. Relationship between volatile compounds and
sensory attributes of olive oils by the sensory wheel. JAOCS 73(10):125364.
Even though it is possible to recognize which products present de- Aparicio R, Morales MT, Alonso V. 1997. Authentication of European virgin olive oils by
fective characteristics, the IOC categories are not flexible enough their chemical compounds, sensory attributes, and consumers attitudes. J Sci Food Agric
45:107683.
to adequately classify nondefective oils, which may lead to misclas- Bendini A, Cerretani L, Carrasco-Pancorbo A, Gomez-Caravaca AM, Segura-Carretero A,
sification of an EVOO. The IOC method is in fact very useful for Fernandez-Gutierrez A, Lercker G. 2007. Phenolic molecules in virgin olive oils: a survey
of their sensory properties, health effects, antioxidant activity and analytical methods. An
preliminary classification of oils, but if the objective is to character- overview of the last decade. Molecules 12(8):1679719.
ize EVOOs, a descriptive analysis offers a more precise and through Bendini A, Cerretani L, Salvador MD, Fregapane G, Lercker G. 2009. Stability of the sensory
quality of virgin olive oil during storage: an overview. Ital J Food Sci. 21:389406.
characterization of the EVOOs sensory properties. This opinion Caporale G, Policastro S, Monteleone E. 2004. Bitterness enhancement induced by cut grass
is shared by Aparicio and Morales (1997), who have referred to odorant (cis-3-hexen-1-ol) in a model olive oil. Food Qual Prefer 15(3):21927.
Chatfield C, Collins AJ. 1980. Introduction to multivariate analysis. London: Chapman & Hall.
the limitations of the IOC method and proposed the addition of 248 p.
green banana, butter, tomato, floral, and astringent to the descrip- DiBella G, Maisano R, LaPera L, LoTurco V, Salvo F, Dugo G. 2007. Statistical characterization
of sicilian olive oils from the peloritana and maghrebian zones according to the fatty acid
tors used. The present study confirms that other attributes should profile. J Agric Food Chem 55(16):656874.
be included in the evaluation of the sensory qualities of EVOOs.
S: Sensory & Food

Garca-Mesa JA, Pereira-Caro G, Fernandez-Hernandez A, Garca-Ortz Civantos C, Mateos


R. 2008. Influence of lipid matrix in the bitterness perception of virgin olive oil. Food Qual
Prefer 19(4):42130.
Quality

Conclusions Guerrero L, Romero A, Tous J. 2001. Importance of generalised procrustes analysis in sensory
characterisation of virgin olive oil. Food Qual Prefer 12(8):51520.
In terms of sensory properties, the majority of the California oils [IOC] Intl. Olive Council. 2007. Sensory Analysis of Olive Oil. Method for the organoleptic
assesment of virgin olive oil. COI/T.20Doc.No15/Rev.2 September 2007.
in the experimental set were defined by the green-bitter, nutty, tea, Kalua CM, Allen MS, Bedgood JDR, Bishop AG, Prenzler PD, Robards K. 2007. Olive
ripe fruit, and pungent characteristics. Spanish oils in the set were oil volatile compounds, flavour development and quality: a critical review. Food Chem
mainly defined by tropical, herbs, and floral attributes. Chilean and 100(1):27386.
Lawless HT, Heymann H. 1998. Sensory evaluation of food. N.Y.: Springer.
Australian oils shared some attributes with the California profile. Luna G, Morales MT, Aparicio R. 2006. Characterisation of 39 varietal virgin olive oils by their
Some of the main attributes for the Italian oils in the set were nutty, volatile compositions. Food Chem 98(2):24352.
Morales MT, Alonso MV, Rios JJ, Aparicio R. 1995. Virgin Olive oil aroma: relationship
butter, and the defect rancid. It is important to stress; however, that between volatile compounds and sensory attributes by chemometrics. J Agric Food Chem
the findings reported in this study only apply to the sample set that 43:292531.
Morales MT, Luna G, Aparicio R. 2000. Sensory and chemical evaluation of winey-vinegary
was considered and that they may not be generalized to all oils from defect in virgin olive oils. Eur Food Res Technol 211(3):2228.
the countries represented in our design. Variety, region of origin, Morales MT, Luna G, Aparicio R. 2005. Comparative study of virgin olive oil sensory defects.
Food Chem 91(2):293301.
and fruit ripeness are intrinsic factors that have a great influence Ollivier D, Artaud J, Pinatel C, Durbec J-P, Guerere M. 2006. Differentiation of French virgin
in the EVOO sensory profile. In the case of some imported oils olive oil RDOs by sensory characteristics, fatty acid and triacylglycerol compositions and
chemometrics. Food Chem 97(3):38293.
improper storage and handling could have led to the development Rotondi A, Bendini A, Cerretani L, Mari M, Lercker G, Toschi TG. 2004. Effect of olive
of defects, particularly rancidity. We found that descriptive analysis ripening degree on the oxidative stability and organoleptic properties of Cv. Nostrana di
Brisighella extra virgin olive oil. J Agric Food Chem 52(11):364954.
was a more effective method to provide a more detailed classifica- Servili M, Selvaggini R, Esposto S, Taticchi A, Montedoro G, Morozzi G. 2004. Health and
tion of the EVOOs than the limited profile used by the IOC. We sensory properties of virgin olive oil hydrophilic phenols: agronomic and technological aspects
of production that affect their occurrence in the oil. J Chromatogra A 1054(12):11327.
are investigating the implication of the differences in the sensory Sinesio F, Moneta E, Esti M. 2005. The dynamic sensory evaluation of bitterness and pungency
profiles of the California-produced and imported oils on consumer in virgin olive oil. Food Qual Prefer 16(6):55764.
acceptance and EVOO market dynamics in the United States. Stefanoudaki E, Kotsifaki F, Koutsaftakis A. 2000. Sensory and chemical profiles of three Euro-
pean olive varieties (Olea europea L.); an approach for the characterisation and authentication
of the extracted oils. J Sci Food Agric 80:3819.
Tripoli E, Giammanco M, Tabacchi G, Di Majo D, Giammanco S, La Guardia M, 2004. The
Acknowledgments phenolic compounds of olive oil: structure, biological activity and beneficial effects on human
The authors thank Corto Olive, The Olive Press, Hojiblanca health. Nutr Res Rev 18(01):98112.
Vossen PM. 2007. Current opportunities in the California olive oil industry. Plant and soil
Enrique Escudero, Golden Hill Olive Oil Co., Jovia Groves, Dos Conference of the California Chapter of the American Society of Agronomy, Radisson
Colinas, McEvoy Ranch, Veronica Foods, Calolea, Sciabicas, Hotel, Sacramento, Calif.
Zampounis V, 2006. Olive oil in the world market. In Boskou D. Olive oil chemistry and
Corti Brothers, and the UC Davis Olive Center for their do- technology, pp. 2139. AOCS Press.
nations of oils, Robertet Mexico (Flavor Division) for their dona-

S176 Journal of Food Science r Vol. 76, Nr. 3, 2011

Vous aimerez peut-être aussi