Você está na página 1de 8

LWT - Food Science and Technology 43 (2010) 393400

Contents lists available at ScienceDirect

LWT - Food Science and Technology


journal homepage: www.elsevier.com/locate/lwt

Application of a sorting procedure to greenhouse-grown cucumbers and tomatoes


Kevin C. Deegan a, b, *, Laura Koivisto a, Juha Nakkila c, Lea Hyvonen a, Hely Tuorila a
a

Department of Food Technology, University of Helsinki, Finland


Department of Applied Chemistry and Microbiology, University of Helsinki, Finland
c
, Finland
MTT Agrifood Research Finland, Plant Production Research, Piikkio
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 6 March 2009
Received in revised form
20 August 2009
Accepted 21 August 2009

Sorting method, a simple procedure to evaluate sensory differences in cucumbers and tomatoes,
requiring minimal training and labour, was applied to cucumbers and tomatoes grown under natural and
articial light in Finland and natural light in Spain, and evaluated after set storage times of up to 22d.
Descriptive sensory proling was also performed on the same samples. Sorting data were analysed using
multidimensional scaling (MDS), while principal component analysis (PCA) was used to interpret
proling data. For both cucumbers and tomatoes, similar trends were observed by use of either sorting or
descriptive sensory analysis. Differences between sources and storage times were found for all samples at
all sampling times. Natural light-grown cucumbers were rated higher for odour and taste intensity than
those grown in articial light. Both consistency and intensity of red colour contributed signicantly to
differences between Finnish and Spanish tomatoes. Sorting was less time consuming and labour
demanding, yet it resulted in the positioning of samples equivalent to the more complex proling
method. Thus, sorting appears to be a viable alternative to proling when sensory similarities and
differences, rather than accurate sensory proles, are required.
2009 Elsevier Ltd. All rights reserved.

Keywords:
Sorting
Sensory
Cucumber
Tomato

1. Introduction
The ability of humans to organise their environment and to
categorise based on similarities and dissimilarities in perceptive
features between objects allows the utilisation of sorting in sensory
studies (Faye, Bremaud, Durand Daubin, Courcoux, Giboreau, &
Nicod, 2004). Sorting has been used as a procedure in a range of
sensory studies in order to ascertain the qualitative variation
between products. Foodstuffs including beers (Chollet & Valentin,
2001), yoghurts (Saint Eve, Paci-Kora, & Martin, 2004), wines
(Gawel, Iland, & Francis, 2001) and cheese (Lawless, Sheng, &
Knoops, 1995) amongst others, have been evaluated. Sorting has
several advantages relative to quantitative descriptive analysis as
described by Cartier et al. (2006). It requires little training relative
to descriptive analysis, it doesnt require the use of a quantitative
rating system or forced-agreement between panellists and, as it is
not required to provide panellists with a list of attributes, it avoids
biasing panellists, as well as being less fatiguing (Cartier et al.,
2006; Faye et al., 2004; Lawless et al., 1995; Saint Eve et al., 2004).

* Corresponding author. Department of Food Technology, University of Helsinki,


P.O. Box 66, FI-00014 Helsinki, Finland.
E-mail address: kevin.deegan@helsinki. (K.C. Deegan).
0023-6438/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.lwt.2009.08.014

Given the apparent advantages, it could be speculated that the


results of sorting procedures would not be comparable with the
results obtained from a conventionally trained sensory panel performing more complex quantitative descriptive analysis techniques. However, Cartier et al. (2006) and Faye et al. (2004) have
shown that the plots based on the results obtained from both
techniques led to the same conclusions from the products tested.
The basic sorting task requires the assessors to examine samples
and group them based on their similarities (Lawless et al., 1995).
Multidimensional scaling (MDS) is routinely used in the analysis
of data provided by sorting (Abdi, Valentin, Chollet, & Chrea, 2007;
Lawless et al., 1995). The procedure produces a spatial or pictorial
representation of the similarities and dissimilarities between
samples, where similar samples are displayed close to each other,
while different samples will appear as points which are far apart
(Bertino & Lawless, 1993; King, Cliff, & Hall, 1997). MDS can be
utilised to provide clues of the perceptual features of the products
tested by interpretation of the dimensional structure or patterns of
clustering in the plots generated (Lawless, 1993; Popper & Heymann, 1996). MDS has been widely used in sensory studies for
a variety of food and beverage products including drinking water
(Falahee & MacRae, 1997), soft drinks (Chauhan & Harper, 1986), rye
bread (Hellemann, Tuorila, Salovaara, & Tarkkonen, 1987) and
extra-virgin olive oil (Aparicio, Calvente, & Morales, 1996). In

394

K.C. Deegan et al. / LWT - Food Science and Technology 43 (2010) 393400

combination with sorting, it has also been used in the case of cheese
by Lawless et al. (1995) in which the authors outlined the quality
criteria which are desired in perceptual mapping, namely; that the
resultant solutions/plots should be accurate and reliable, that maps
should relate to descriptive attributes or preferences, that plots
should have a few dimensions and be interpretable, that the data
collection should be fast and cost-effective and, nally, that maps
should suggest old, or conrm new hypothesis.
Finlands location in the north-east corner of Europe means that
natural light levels during the winter months are very low, resulting in the necessity for supplemental light in the growth of both
cucumbers and tomatoes (Hovi, Nakkila, & Tahvonen, 2004) as light
is essential for owering, pollination and growth of fruit (Rodriguez
& Lambeth, 1975). The benecial and negative effects of such
supplemental lighting will be investigated as part of this study.
Storage of tomatoes and cucumbers is necessary in retail or
domestic situations and the conditions of this storage can affect
sensory attributes of the fruit, for example, storing below 13  C is
known to cause signicant chilling injury to tomatoes leading to
reduced avour, softening, surface pitting, shrivelling and uneven
ripening (Maul et al., 2000) and surface pitting and increased
susceptibility to decay in cucumbers (Cantwell & Kasmire, 2002).
In the present study, the suitability of sorting was investigated as
a procedure to distinguish between tomatoes and cucumbers grown
under different light conditions and thereafter stored for different
time periods. First, sorting by visual inspection resembles the way
consumers choose their fruit and vegetables at the grocery store.
Sorting by appearance was therefore separated from sorting by other
sensory qualities (odour, avour, and texture). Second, sorting was
considered appropriate given that fruit and vegetables are typically
graded into quality categories based on their sensory quality again,
mainly visually. On the other hand, sorting was expected to be
particularly demanding because of biological variability and subtle
differences between the samples. Sensory proling was used on the
same samples to assist in the interpretation of the sorting results.
2. Materials and methods
2.1. Panellists
Fifteen subjects were recruited from the staff of the University of
Helsinki, according to their motivation and availability. Training
sessions for panellists consisted of basic information about
cucumbers and tomatoes and information about the sorting and
proling methods. Terms used in sensory proling were suggested
by panellists and were based on similar reasoning used by
consumers in a retail situation. Both the cucumber panel and the
tomato panel consisted of 10 panellists for proling and 15 panellists for sorting and some of the panellists were common to both.
2.2. Samples
Tomatoes (Lycopersicon esculentum L. cv. Encore) and cucumbers
(Cucumis sativus L. cv. Cumuli) were grown under natural and articial light and stored at 12  C in MTT Agrifood Research Finland.
Reference tomatoes and cucumbers were obtained from a farm in
Finland (of the same varieties, grown in greenhouses under natural
light) and from a Spanish supplier (same size classes, grown under
natural light) and were stored in MTT Agrifood Research under the
same storage conditions. In MTT, tomato and cucumber plants were
fertilised using a commercial complete nutrition solution (TailorMade Superex, Kekkila, Finland) according to a specic programme
using a drip irrigation system. The irrigation system was controlled by
a timer. In tomato cultivation, the electrical conductivity (EC) of the
feeding solution for the rst 2 weeks ranged from 3.5 to 3.0 mS cm 1

and subsequently gradually lowered to 1.3 mS cm 1. In cucumber


cultivation, the electrical conductivity (EC) of the feeding solution for
the rst 4 weeks ranged from 3.0 to 3.5 mS cm 1 and subsequently
gradually lowered to 2.4 mS cm 1. The Finnish farms tomato and
cucumber plants were fertilised using a commercial complete
nutrition solution according to a specic programme using a drip
irrigation system. Cucumbers from MTT were grown with a high-wire
method, while Finnish farm cucumbers were grown with an umbrella
method. Tomato samples were received in September (from MTT and
the Finnish farm) and January (from MTT and the Spanish supplier)
while cucumber samples were received in August (from MTT and the
Finnish farm) and November (from MTT and the Spanish source).
Henceforth, samples will be named according to their source (i.e. M, F
or S, for MTT, Finnish farm or Spanish source, respectively). M and F
tomatoes were harvested at the red ripening stage the point at which
there was no green colour remaining (USDA, 1997). Cucumbers were
picked when the minimum weight of 250 g was exceeded.

2.3. Sorting
Samples were stored at room temperature (20  C) from arrival
at the laboratory until sorting (1d). Depending on the session,
each assessor was presented with a tray containing 16 cucumbers
or 16 tomatoes (each presented with a white sticker bearing
a three digit code), in a random order. The origin and codes of
the 16 cucumbers/tomatoes are shown in Table 1. For M tomatoes, there were four replicates (denoted by a, b, c and d) for each
storage time (1, 12 or 22d). Due to supply constraints, F tomatoes
in September had two replicates for 1d storage, and 22d S
samples in January had two replicates. Similarly, for M cucumbers, there were four replicates (denoted by a, b, c or d) for each
storage time (1, 8 or 11d). Again, due to supply constraints,
replicates were limited to 1d for August F cucumbers and 8d for
November S cucumbers. Harvesting times were staggered over 1,
12 and 22d, for tomatoes, and 1, 8 and 11d, for cucumbers, prior
to transportation to the laboratory. Firstly, the panellists were
asked to sort the 16 cucumbers or 16 tomatoes based on similarities in the combination of taste, odour and texture. This was
carried out under green light (for cucumbers) and red light (for
tomatoes) to eliminate bias based on appearance. Each grouping
formed by sorting was required to contain a minimum of 1

Table 1
Codes and origin of cucumber and tomato samples.
Code

M1a
M1b
M1c
M1d
M8a/M12a
M8b/M12b
M8c/M12c
M8d/M12d
M11a/M22a
M11b/M22b
M11c/M22c
M11d/M22d
F1a/S1
F1b
F8/F12/S8/S12
F22/S22a
S22b
a

Growth linea

Storage time (d)


Cucumber

Tomato

MTT

Farm/Spain

MTT

Farm/Spain

1
1
1
1
8
8
8
8
11
11
11
11

1
1
8
22
22

1
1
1
1
12
12
12
12
22
22
22
22

1
1
12
22
22

1
2
3
4
1
2
3
4
1
2
3
4
1
1
1
1
1

Growth line represents the location of the samples growth in the glasshouse.

K.C. Deegan et al. / LWT - Food Science and Technology 43 (2010) 393400

2.4. Proling
Samples were stored at room temperature (20  C) from arrival at
the laboratory until proling (3d). Training sessions were carried out
in the weeks preceding the arrival of the rst samples, where assessors developed a list of proling attributes based on texture, taste,
odour and appearance attributes, which are shown in Table 2. The
presentation of samples was randomised. Samples were presented to
assessors in white plastic tubs with transparent plastic covers. White
stickers with three digit codes were displayed clearly on the transparent covers. Cucumber samples were presented as a cut cucumber
segment (approx. 3 cm) while tomato samples were presented as
a half tomato. Evaluation was carried out using a 10 cm unstructured
line scale with the anchor points 0 no intensity and 100 very high
intensity. Following the examination of taste and odour attributes,
separate trays containing full cucumbers or tomatoes were presented
to panellists to allow proling of appearance attributes. The proling
procedure was repeated 4 h later with the same samples.

Table 2
Sensory proling: attributes and their descriptions.

2.5. Statistical analysis


In each sorting session, the number of times each sample was
grouped with another sample or samples was added and averaged
between the morning and afternoon sessions to create a similarity
matrix, which is an essential requirement of MDS (Tang & Heymann, 2002). A similarity matrix was generated in each of the two
sorting sessions and the matrices were averaged to give the input
required for MDS. The matrix was subjected to an Alternating Least
Square sCALing (ALSCAL) procedure, which is an MDS technique
used commonly in sensory evaluation (Abdi et al., 2007). The
procedure was carried out using an Euclidean distance scaling
model. The procedure was carried out using SPSS version 13.0 (SPSS
Inc, Chicago, IL, USA). Principal component analysis (PCA) was
performed on the means of the sensory proling data from the
proling task using The Unscrambler version 9.7 (Camo Software
AS, Oslo, Norway). The scores and loadings of the data analysed by
PCA were displayed as bi-plots.

2
F1b
M22b
M1b
M22a

Dimension 2

sample and the maximum number of groups formed was not


restricted. Panellists were encouraged to sort the samples at their
own pace and tap water was served to allow rinsing between
samples. Following completion of the sorting task the assessors
were required to record the groupings by writing the codes of the
samples on a log sheet. Next, panellists were required to sort the
samples based on appearance (under normal laboratory lighting)
using the same procedure as described above. The sorting
procedure was then replicated 4 h later with the same samples.

395

M1c
F22

M1a

0
M1d

F12

F1a

M12c
M12b

Tomato
Taste
Texture

Attribute
Sour
Sweet

Intensity of sour taste


Intensity of sweet taste

Hand hardness

The hardness when squeezed in the


hand
Hardness during mastication
Toughness during mastication

Mouth hardness
Toughness
Appearance

Red colour
Colour consistency
Surface aws

Cucumber
Odour
Taste

Texture

Intensity of red colour on the


surface
Consistency of the surface colour
Extent of aws on the surface

M12d
M22c

-1

M12a

-2

-1

Dimension 1

Odour

Intensity of cucumber odour

Taste intensity
Bitterness
Sweetness

Intensity of taste when ingested


Intensity of bitter taste
Intensity of sweet taste

Dryness

Extent of dryness during


mastication
Extent of crispiness during
mastication
The hardness when squeezed in the
hand

Hardness
Darkness
Colour consistency
Surface shrivelling
Surface roughness
Curving
Narrowing

Grooving

How dark the surface is


Consistency of the surface colour
Extent of shrivelling of the skin on
the surface
Extent of shrivelling on the surface
Extent of curved shape
Extent of narrowing at the ends of
the
cucumber
Extent of grooves in the surface

M22d

M12c
M22d

M12d
F12

Crispiness

Appearance

Description

Dimension 2

Grouping

M12a

F22
M22b

F1a

0
M1d

M1a

M22a

M1c
M12b

-1

M22c
F1b

M1b

-2

-1

Dimension 1
Fig. 1. Two-dimensional MDS solution from sorting of the September samples of
tomatoes, based on (a) appearance and (b) odour, taste and texture. Letters represent
samples; M1ad; M12ad and M22ad represent MTT at 1, 12 and 22d storage
respectively, while F1 a, b; F12 and F22, represent Finnish farm tomatoes at 1, 12 and
22d storage time respectively.

396

K.C. Deegan et al. / LWT - Food Science and Technology 43 (2010) 393400

3. Results
3.1. General

M12b

M22b

M22c
M12c

Dimension 2

The matrices constructed from sorting data were subjected to


MDS resulting in two-dimensional plots (Figs. 1, 3, 5 and 7) while
proling data were used to construct the PCA bi-plots (shown in
Figs. 2, 4, 6 and 8). In general, MDS plots clearly presented the data
and showed differences/similarities between samples as distances.
Thus, they allow quick and efcient interpretation of sample
groupings. PCA bi-plots further aid in the explanations of such
similarities and differences.

M22a

S22b
S1
S22a

M12a

M1d
M12d

M1b

S12

M22d

3.2. Tomatoes

-1

1,3

Princi pal com ponent 2 (27% )

F22

F1
0,3

-1

-0,5

-0,2

M22

Toughness
0,5
1
Red colour Mouth hardness

M12

Sw eet
Surface flaw s

-2

-1

Dimension 1

M1c

M1a

M1d

M1b

M12a

S1
M12b
S22a
M12c

0
M22b
S12
M22d

S22b

-1

M22a
M12d

-2

-1

Dimension 1

M22c

Fig. 3. Two-dimensional MDS solution from sorting of the January samples of tomatoes, based on (a) appearance and (b) odour, taste and texture. Letters represent
samples; M1ad; M12ad and M22ad represent MTT at 1, 12 and 22d storage
respectively, while S1; S12 and S22a, b, represent Spanish tomatoes at 1, 12 and 22d
storage time respectively.

Sour
0,8

-1,5

M1c

Dimension 2

Sorting of the September samples of tomatoes based on


appearance (Fig. 1a), showed a separation, based mainly on the
source, with tomatoes at 1, 12 and 22d storage grown in natural
light (F1a, F12 and F22) clustered relatively tightly on the positive
side of the Dimension 1 axis. Conversely, M tomatoes were shown
to be more sparsely clustered and generally increase from negative
to positive values of Dimension 1. In terms of odour, taste and
texture (Fig. 1b), M samples were, generally, grouped together at
each storage time, suggesting a separation based on storage time.
Natural light-grown F samples were closely correlated with
equivalent storage time M samples (especially at 12 and 22d),
suggesting that odour, taste and texture attributes were less
discriminating reasons than storage time for perceived differences
during sorting of tomatoes. Both M and F samples which were
stored for 22d were grouped with positive values for Dimension 1,
indicating similarities between M and F tomatoes based on sorting
by odour, texture and taste.
The PCA score plot corresponding to the sensory proling of
tomatoes in September is shown in Fig. 2. PC1 (67% of the total
variance) was shown to separate samples based on their storage
time while the M and F tomatoes were separated extensively by
PC2 (27% of the total variance). The appearance attributes colour
consistency and surface aws were more closely correlated with
M tomatoes, indicating that the differences between sources of
tomatoes encountered during sorting based on appearance (Fig. 1a)
are most likely due to these attributes. In terms of odour, taste and

M1a

-0,7

1,5

Hand hardness

M1

Colour
consistency

-1,2

Principal component 1 (67%)


Fig. 2. Score plot obtained from principal component analysis of data from descriptive
sensory proling of September tomatoes (M represents tomatoes grown and stored in
MTT, F represents Finnish farm tomatoes, both followed by the storage time in days).

texture attributes, M1 and F1 tomatoes were correlated closely


with both hand hardness and mouth hardness (both with positive values for PC1) while M22 and F22 tomatoes had low negative
values for PC1 indicating a large difference based on sensory
proling and possibly explaining the large differences seen
between 1d and 22d storage time tomatoes in Fig. 1b. Sweet
attribute was more closely correlated with M tomatoes, while sour
was correlated with F samples, however, these attributes seemed to
be somewhat overlooked, when sorted by odour, taste and texture,
as Fig. 1b showed similarities between M and F tomatoes at the
same storage time.
The MDS plots for the January set of tomatoes are shown in
Fig. 3(a, b). Sorting based on appearance (Fig. 3a) displayed a large
separation of tomatoes from M and S, with a much greater separation due to source as compared to Fig. 1, with S samples occupying the positive side of Dimension 1. This separation might be the
effect of a different effect of natural light on the growth of S
tomatoes relative to F tomatoes (grown in natural light) and, as S
samples at different storage times were clustered more closely

K.C. Deegan et al. / LWT - Food Science and Technology 43 (2010) 393400

1,3

397

2
M1a

0,8

M1c M1d

Surface flaw s

0,3

S1
-1,5

-1,0

Colour
Sour consistency
Red colour

Toughness
-0,5

S22

-0,2

0,0

0,5

M12

1,0

1,5

Dimension 2

Principal com ponent 2 (23%)

M22

M1

S12

M8b

F8 F1a

M1b
M11a
F1b

0
F11

M8c
M8a

-1

M8d

M11d

M11b

Mouth hardness
-0,7

M11c

Hand hardness

-2

Sw eet

-2

-1,2

Principal component 1 (58%)


Fig. 4. Score plot obtained from principal component analysis of data from descriptive
sensory proling of tomatoes in January (M represents tomatoes grown and stored in
MTT, S represents Spanish tomatoes, both followed by the storage time in days).

-1

3.3. Cucumbers
Fig. 5 shows the two-dimensional MDS solution from sorting of
the August samples of cucumbers from F and M. The plot from
sorting based on appearance attributes (Fig. 5a) showed a clear
separation of the M and F cucumbers, with F samples on the
negative side of Dimension 1 (with the exception of M11a) and M
samples on the positive side. F samples at 1, 8 and 11d storage were
closely correlated. M cucumbers showed a general trend of
decreasing in value of Dimension 2 with increased storage time,
which would indicate a change in appearance characteristics with
storage time. In terms of odour, taste and texture (Fig. 5b) the M
cucumbers samples stored for 1d were separated from samples
stored at 8 and 11d (which were in turn, more closely correlated).

Dimension 2

M1aM1b

together relative to F tomatoes, that storage time had a relatively


low effect on appearance in the S tomatoes. M samples were more
sparsely separated by storage time on Dimension 2, with a general
trend of decreasing values of Dimension 2 with increased storage
time. The odour, taste and texture plot (Fig. 3b) showed a similar
trend to the plot based on appearance, with a clear separation of
S tomatoes from M tomatoes, while M tomatoes were separated by
Dimension 2 based primarily on their storage times.
Similar trends found in the MDS plots from sorting of the
January tomatoes can be seen in the PCA plot resulting from
proling of the same samples (Fig. 4). M and S samples were
extensively separated from each other, primarily by PC1 (which
accounted for 58% of the total variance) indicating that the source
was the primary reason for perceived differences. In terms of
appearance, the attributes red colour and colour consistency
were rated higher for M tomatoes relative to S tomatoes. This may
explain the large differences in sorting based on appearance seen in
Fig. 3a. Also, there was little change in both these attributes with
storage time for S tomatoes, which is suggested by the close clustering of S tomatoes on the right of the sorting by appearance plot.
Examining odour, taste and texture attributes, S tomatoes were
rated less sour than M tomatoes, but closely correlated with both
mouth and hand hardness on the PCA plot on the negative side of
PC1 (Fig. 4) which could represent the large differences perceived
between M and S tomatoes during sorting (Fig. 3b).

Dimension 1

M8a

F8 F1

M11b

M8c

M11d

0
M1d
M11c

M11a

F11

-1

M1c

M8b

M8d

-2
-2

-1

Dimension 1

Fig. 5. Two-dimensional MDS solution from sorting of the August samples of


cucumbers, based on (a) appearance and (b) odour, taste and texture. Letters represent
samples; M1ad; M8ad and M11ad represent MTT at 1, 8 and 11d storage respectively, while F1; F8 and F11a,b, represent Finnish farm cucumbers at 1, 8 and 11d
storage time respectively.

The PCA plot from descriptive sensory proling of cucumbers


in August (Fig. 6) shows the M samples occupying the negative
half of the rst principal component (61% of total variance) while
F samples were in the positive side. The appearance attributes
darkness and grooving were more closely correlated with F
cucumbers, while the attribute curving was closer to M
cucumbers. The large separation of F and M cucumbers on the
appearance sorting plot (Fig. 5a) could possibly be explained by
the differences observed in these attributes. Also, M cucumbers
were less closely correlated with the appearance attributes
colour consistency and more correlated with narrowing and
surface roughness with storage time, indicating a change in
these attributes with storage time, as was seen in Fig. 5a. In terms
of odour, taste and texture, taste intensity was rated higher,
while both dryness and hardness were much lower in F
cucumbers. Sweetness was similarly correlated for 1d storage of
both F and M cucumbers and decreased with storage time for
both, while bitterness and crispiness increased, which could
explain the general shift from positive to negative values for

398

K.C. Deegan et al. / LWT - Food Science and Technology 43 (2010) 393400

Surface shrivelling

F11

0,8
Crispiness
Bitterness

M8c
S11
S8b

Odour
0,3
Narrow ing
Darkness

M8

M8b

M8d

M1b

M11

-1,0

S1

M1a

Grooving
Taste intensity

-0,5

0,0
-0,2
Hardness

0,5
Colour consistency

1,0

F1

Dimension 2

Principal component 2 (24%)

Surface roughness

Sweetness

M1d

M8a
S8a

-1

M11b
M11a

M11c
M11d

Dryness
-0,7
Curving

M1c

-2

M1
-1,2

-2

Principal component 1 (61%)


Fig. 6. Score plot obtained from principal component analysis of data from descriptive
sensory proling of cucumbers in August (M represents tomatoes grown and stored in
MTT, F represents Finnish farm cucumbers, both followed by the storage time in days).

-1

M11c

4. Discussion
4.1. Tomato quality and differences
Tomatoes grown under a combination of natural and articial
light and stored in MTT (M) were greatly separated from tomatoes
grown under natural light from the Finnish farm (F) in terms of
appearance. This separation was displayed by the sorting method
and conrmed by the proling method. Colour consistency and the
intensity of red colour, which had a large effect on the separation
of source, had higher ratings in M relative to F tomatoes, but
changed signicantly with storage time in the M samples. In

M11d
S8b

M11b

S1
M11a

M8a

Dimension 2

Dimension 2 in the plot based on sorting by odour, taste and


texture (Fig. 5b).
The MDS solutions from sorting the November cucumbers are
shown in Fig. 7(a, b). The plot based on sorting by appearance only
(Fig. 7a) showed grouping of M cucumbers at similar storage times.
M1 cucumbers (with the exception of M1c) were grouped with
negative Dimension 1 values and positive Dimension 2 values, M8
were grouped with positive values for both Dimensions, while the
11d storage M cucumbers had negative values for Dimension 2.
Storage time in M cucumber thus progressed in a roughly clockwise
manner. This suggests that appearance attributes between replicates were similar, but that an appearance-related change took
place as storage time progressed. A similar trend is presented in the
MDS plot from the sorting by odour, taste and texture. S cucumbers
stored for 1d (S1) were separated from the M1 cucumbers in both
plots, as were S11 cucumbers. Descriptive sensory proling of
cucumbers in November resulted in the PCA score plot shown in
Fig. 8. Cucumbers from the two sources (M and S) were extensively
separated by PC1 (which represented almost half, 46%, of the total
variance). The appearance attribute surface roughness increased
with storage time, while darkness, colour consistency and
grooving decreased in proling ratings which would most likely
explain the separation of samples based on storage time seen in
Fig. 7a. The attributes odour and taste intensity were more closely
correlated with S cucumbers, while sweetness was similarly
correlated for M1 and S1, which occupied opposing sides of
Dimension 1 in Fig. 7b.

Dimension 1

M8c

M1c

0
M1b

M8b
S8a
M8d

-1

S11
M1a
M1d

-2
-2

-1

Dimension 1
Fig. 7. Two-dimensional MDS solution from sorting of the November samples of
cucumbers, based on (a) appearance and (b) odour, taste and texture. Letters represent
samples; M1ad; M8ad and M11ad represent MTT at 18 and 11d storage respectively, while S1; S8 and S11a,b, represent Spanish cucumbers at 1, 8 and 11d storage
time respectively.

a study by Auerswald, Peters, Bruckner, Krumbein, and Kuchenbuch (1999), where tomatoes (L. esculentum Mill.) were stored at
20  C for up to 7d, a signicant increase in both red colour and the
intensity of the red colour was observed with increased storage
time. Odour, texture and taste attributes were less discriminating
between M and F, however the sweet attribute was rated higher
for M tomatoes. Overall tomato avour is generally believed to
depend on reducing sugars and volatile organic compounds,
sweetness is related to sugar and acid content while overall
consumer preference can depend on sweetness and aroma
(Causse, Buret, Robini, & Verschave, 2003). The change observed in
odour, taste and texture with storage time is to be expected, as
tomatoes are climacteric fruit which continue to ripen after
harvest (Auerswald, Peters, et al., 1999; Auerswald, Schwarz, Kornelson, Krumbein, & Bruckner, 1999). The relative similarities
between M and F tomatoes, when sorting was based on odour,
texture and taste, indicate that there is little difference between
the supplemented light/natural light combination of the growth
conditions in M and the natural light during growth in F tomatoes.

K.C. Deegan et al. / LWT - Food Science and Technology 43 (2010) 393400

S1

Grooving

Principal component 2 (37%)

Sweetness

399

separation of 1d storage samples of S cucumbers and M cucumbers


and 11d samples of S and M in the sorting plots based on odour,
taste and texture could be, in part, explained by the higher ratings
for odour and taste intensity in S cucumbers and shows an
advantage of growth in natural light.

Surface shrivelling
Taste intensity

0,8

Darkness Colour consistency


Odour

4.3. Limitations of the study

0,3

Dryness

M1
-1

-0,5
Crispiness

M8

0,5

1
Bitterness

-0,2

S11

Hardness
Narrow ing

M11

Surface roughness

Curving

-0,7

-1,2

Principal component 1 (46%)


Fig. 8. Score plot obtained from principal component analysis of data from descriptive
sensory proling of cucumbers in November (M represents cucumbers grown and
stored in MTT, S represents Spanish cucumbers, both followed by the storage time in
days).

A large separation occurred between M and S tomatoes (Spanish


tomatoes grown under natural light) in terms of appearance and by
odour/taste and texture, clearly indicating the advantages of
growth in natural light. Red colour was higher in M tomatoes
relative to S tomatoes. The red colour of tomatoes is due to the
carotenoid lycopene and both light and temperature during fruit
growth affect its production (Barringer, 2004). Lycopene production is retarded by excessive temperature (i.e. greater than 30  C,
which inhibits ethylene production; Cantwell & Kasmire, 2002),
which may explain why S tomatoes were less red than their M
counterparts.
The use of a higher storage temperature may be warranted in
future studies, as the storage temperature below 20  C has been
shown to have a detrimental effect on tomato (L. esculentum Mill.)
quality by Maul et al. (2000). These authors attributed stronger
tomato aroma and avour after storage at 20  C to faster ripening
rates and metabolic activity at this temperature. The use of a higher
storage temperature than used in this study would also allow
a comparison to be made to the conditions of storage in the home
and retail outlets, where tomatoes would be stored at room
temperature (Krumbein, Peters, & Bruckner, 2004).
4.2. Cucumber quality and differences
Cucumbers from M and F were extensively separated based on
source and storage time with regards appearance attributes by the
sorting task. F cucumbers were rated darker, which may be due to
the greater intensity of light when grown under natural light as
opposed to supplemented light, as in the case of the M cucumbers.
The green colour of cucumbers is due to the chlorophyll pigment
which in turn is dependent on light intensity, i.e. the formation of
chlorophyll increases with the intensity of light (Lin & Jolliffe, 1996).
Odour and taste intensity, as well as the texture attributes, crispness and hardness, also contributed to differences between M
cucumbers and F cucumbers. This would seem to indicate the
apparent advantage of use of full natural light during growth
(F cucumbers) as opposed to natural light supplemented with
articial light (M cucumbers). Odour and taste intensity were
rated higher in S cucumbers relative to M cucumbers. The large

The cultivars of tomatoes and cucumbers used in this study may


play a big part in chemical and sensory results of this study. As the
cultivar of S tomatoes and cucumbers was unknown, the role of
cultivars cannot be currently speculated on. Also, EC values of the
nutrient solutions used in both F and S are not available. EC values
have an impact on the intensity of sensory characteristics of
vegetables and on sugar and acid levels (Auerswald, Schwarz et al.,
1999). These limitations need to be considered in the interpretation
of results.
The use of the same assessors for both sorting and proling
tasks in the same study may have somewhat contributed to error in
the investigation of suitability of both tasks. In the sorting task,
panellists were asked to record descriptive words for each sorted
group. These words were not agreed on beforehand and were
somewhat individual and ambiguous. Examples of such words
include solid, long, dark, dull, thick, sweetness, grassy taste
and strong smelling for cucumbers, and, red colour, darkness,
small, soft, strong taste, sweet and tasteless for tomatoes. As
assessors received training in the use of sensory proling and
agreed on the vocabulary used, this knowledge could provide the
assessors with supplementary know-how unneeded or even
detrimental when performing the sorting task. However, the
vocabulary and the basic knowledge regarding tomatoes and
cucumbers given in training would not have given assessors any
advantage above an informed consumer in a retail situation.
Attention must also be paid to the relative differences seen
between replicates in the MDS plots (Figs. 1, 3, 5 and 7). For
instance, in Fig. 3b, replicates of storage time points are clustered
closely together (with the exception of M12d) while in Fig. 1a, M1
replicates are quite separated. Visual inspection of both tomatoes
and cucumbers on arrival to the sensory laboratory showed, in
some instances, considerable differences between replicates for the
same storage time. These were deemed unavoidable due to the
nature of biological samples, which can invariably be different.
4.4. Sorting
This study has demonstrated the application of a simple sorting
procedure as an alternative to classical sensory proling or as an
additional analysis prior to normal proling. Information extrapolated from MDS plots produced from sorting data can be compared
to the information resulting from sensory proling of the same
samples to conrm details already found or to add a new insight
into classical proling. Sorting is easier to organise and carry out
than proling. As the assessors are not trained extensively, is
quicker and easier to comprehend. Lawless and Glatter (1990)
found that when sorting was applied to odour perception, the plots
obtained from MDS were relatively unaffected by learning or
linguistic inuences, which holds true in this study, as the extra
training and vocabulary development used in the proling procedure still highlighted the same variation between samples on
analysis relative to sorting. Therefore, the use of sorting alone may
be considered as an investigative or preliminary method of
assessment. Yet proling may have advantages above sorting, as it
can be used to describe the attributes which cause differences
between samples (Matuszewska, Barylko-Pikielna, Tarkkonen,
Hellemann, & Tuorila, 1992). Sorting procedure and

400

K.C. Deegan et al. / LWT - Food Science and Technology 43 (2010) 393400

multidimensional scaling describe only the internal relationship


within a particular group of samples without yielding quantitative
information about specic attributes. As a result, sorting alone
yields relatively low information.
The present study suggests that sorting is not only useful for
demonstration but also as one way of categorisation of stimuli
(e.g., Faye et al., 2004). On the contrary, it provides an indicative
way of evaluating everyday produce. With panellists having
a relatively free hand in their implementation of sorting, they
ended up producing meaningful mapping of the samples. In spite of
the complexity of the task when compared with earlier sorting
exercises, the sorting procedure easily showed differences between
both sources of growth and storage time for both cucumbers and
tomatoes, differences which were, in general, conrmed by sensory
proles. The present data should encourage consideration of sorting as a possible alternative in sensory research in which differentiation of a large number of samples is an issue.
Acknowledgements
The kind assistance of Mrs. Kaisu Taskila in organising and
performing the sorting and sensory proling tasks is greatly
appreciated. This study was funded as part of a VILA project funded
by the Finnish Ministry of Agriculture and Forestry.
References
Abdi, H., Valentin, D., Chollet, S., & Chrea, C. (2007). Analyzing assessors and
products in sorting tasks: DISTATIS, theory and applications. Food Quality and
Preference, 18, 627640.
Aparicio, R., Calvente, J. J., & Morales, M. T. (1996). Sensory authentication of
European extra-virgin olive oil varieties by mathematical procedures. Journal of
the Science of Food and Agriculture, 72, 435447.
Auerswald, H., Peters, P., Bruckner, B., Krumbein, A., & Kuchenbuch, R. (1999).
Sensory analysis and instrumental measurements of short-term stored tomatoes (Lycopersicon esculentum Mill.). Postharvest Biology and Technology, 15,
323334.
Auerswald, H., Schwarz, D., Kornelson, C., Krumbein, A., & Bruckner, B. (1999).
Sensory analysis, sugar and acid content of tomato at different EC values of the
nutrient solution. Scientia Horticulturae, 82, 227242.
Barringer, S. A. (2004). Production, freezing, and storage of tomato sauces and slices.
In Y. H. Hui, P. Cornillon, I. Guerrero Legaretta, M. H. Lim, K. D. Murrell, & W.K. Nip (Eds.), Handbook of frozen foods (pp. 415432). London, UK: CRC Press.
Bertino, M., & Lawless, H. T. (1993). Understanding mouthfeel attributes: a multidimensional scaling approach. Journal of Sensory Studies, 8, 101114.
Cantwell, M. I., & Kasmire, R. F. (2002). Postharvest handling systems: fruit vegetables. In A. A. Kader (Ed.), Postharvest technology of horticultural crops (pp. 407
421). CA, USA: University of California, Agriculture and Natural Resources.
Cartier, R., Rytz, A., Lecomte, A., Poblete, F., Krystlik, J., Belin, E., et al. (2006). Sorting
procedure as an alternative to quantitative descriptive analysis to obtain
a product sensory map. Food Quality and Preference, 17, 562571.

Causse, M., Buret, M., Robini, K., & Verschave, P. (2003). Inheritance of nutritional
and sensory quality traits in fresh market tomato and relation to consumer
preferences. Journal of Food Science, 68, 23422350.
Chauhan, J., & Harper, R. (1986). Descriptive proling versus direct similarity
assessments of soft drinks. Journal of Food Technology, 21, 175187.
Chollet, S., & Valentin, D. (2001). Impact of training on beer avour perception and
description: are trained and untrained panellists really different? Journal of
Sensory Studies, 16, 601618.
Falahee, M., & MacRae, A. W. (1997). Perceptual variation among drinking waters:
the reliability of sorting and ranking data from multidimensional scaling. Food
Quality and Preference, 8, 389394.
Faye, P., Bremaud, D., Durand Daubin, M., Courcoux, P., Giboreau, A., & Nicod, H.
(2004). Perceptive free sorting and verbalization tasks with nave subjects:
an alternative to descriptive mappings. Food Quality and Preference, 15,
781791.
Gawel, R., Iland, P. G., & Francis, I. L. (2001). Characterizing the astringency of red
wine: a case study. Food Quality and Preference, 12, 8394.
Hellemann, U., Tuorila, H., Salovaara, H., & Tarkkonen, L. (1987). Sensory proling
and multidimensional scaling of selected Finnish rye breads. International
Journal of Food Science and Technology, 22, 693700.
Hovi, T., Nakkila, J., & Tahvonen, R. (2004). Interlighting improves year-round
cucumber production. Scientia Horticulturae, 102, 283294.
King, M. C., Cliff, M. A., & Hall, J. W. (1997). Comparison of projective mapping and
sorting data collection and multivariate methodologies for identication of
similarity-of-use of snack bars. Journal of Sensory Studies, 13, 347358.
Krumbein, A., Peters, P., & Bruckner, B. (2004). Flavour compounds and a quantitative descriptive analysis of tomatoes (Lycopersicon esculentum Mill.) of
different cultivars in short-term storage. Postharvest Biology and Technology,
32, 1528.
Lawless, H. T. (1993). Characterization of odor quality through sorting and multidimensional scaling. In C.-T. Ho, & C. H. Manley (Eds.), Flavour measurement (pp.
159183). New York, NY: Marcel Dekker Inc.
Lawless, H. T., & Glatter, S. (1990). Consistency of multidimensional scaling models
derived from odor sorting. Journal of Sensory Studies, 5, 217230.
Lawless, H. T., Sheng, N., & Knoops, S. S. C. P. (1995). Multidimensional scaling of
sorting data applied to cheese perception. Food Quality and Preference, 6, 9198.
Lin, W. C., & Jolliffe, P. A. (1996). Light intensity and spectral quality affect fruit
growth and shelf life of greenhouse-grown long English cucumber. Journal of
the American Society for Horticultural Science, 121, 11681173.
Matuszewska, I., Barylko-Pikielna, N., Tarkkonen, L., Hellemann, U., & Tuorila, H.
(1992). Similarity proling versus proling of spreads: do we need both? Food
Quality and Preference, 3, 4750.
Maul, F., Sargent, S. A., Sims, C. A., Baldwin, E. A., Balaban, M. O., & Huber, D. J.
(2000). Tomato avour and aroma quality as affected by storage temperature.
Journal of Food Science, 65, 12281237.
Popper, R., & Heymann, H. (1996). Analyzing differences among products and
panellists by multidimensional scaling. In T. Naes, & E. Risvik (Eds.), Multivariate analysis of data in sensory science (pp. 159184). Amsterdam: Elsevier
Science B.V.
Rodriguez, B. P., & Lambeth, V. N. (1975). Articial lighting and spacing as photosynthetic and yield factors in winter greenhouse tomato culture. Journal of the
American Society for Horticultural Science, 100, 694697.
Saint Eve, A., Paci-Kora, E., & Martin, N. (2004). Impact of the olfactory quality and
chemical complexity of the avouring agent on the texture of low fat stirred
yogurts assessed by three different sensory methodologies. Food Quality and
Preference, 15, 655668.
Tang, C., & Heymann, H. (2002). Multidimensional sorting, similarity scaling and
free-choice proling of grape jellies. Journal of Sensory Studies, 17, 493509.
USDA. (1997). United States standards for grades of fresh tomatoes.

Você também pode gostar