Preference mapping of apple varieties in Europe

Food Quality and Preference 32 (2014) 317–329

Contents lists available at ScienceDirect

Food Quality and Preference journal homepage: www.elsevier.com/locate/foodqual

Preference mapping of apple varieties in Europe J. Bonany a,⇑, C. Brugger b, A. Buehler b, J. Carbó a, S. Codarin c, F. Donati d, G. Echeverria e, S. Egger b, W. Guerra f, C. Hilaire c, I. Höller f, I. Iglesias e, K. Jesionkowska g, D. Konopacka g, D. Kruczyn´ska g, A. Martinelli i, C. Petiot j, S. Sansavini d, R. Stehr k, F. Schoorl l a

IRTA-Mas Badia, E-17134 La Tallada, Spain Agroscope Changins-Wädenswil Research Station ACW, Schloss 1, P.O. Box, 8820 Wädenswil, Switzerland CTIFL-Lanxade, 28 Route des Nebouts, 24130 Prigonrieux, France d University of Bologna, Viale Fanin, 44, 40127 Bologna, Italy e IRTA-Lleida, Av. Alcalde Rovira i Roure, 191, 25198 Lleida, Spain f Research Centre for Agriculture and Forestry Laimburg, Laimburg 6, I-39040 Posta Ora (BZ), Italy g Research Institute of Horticulture, Konstytucji 3 Maja 1/3, 96-100 Skierniewice, Poland i Consorzio Italiano Vivaisti, Stat. Romea n. 116, 44020 S. Giuseppe di Comacchio (Ferrara), Italy j NOVADI, 23 rue Jean Baldassini F-69364 Lyon Cedex 07, France k Obstbau Versuchs-und Beratungszentrum Jork, Moorende 53, 21635 Jork, Germany l WUR-PPO, Lingewal 1, 6668 LA Randwijk, Netherlands b c

a r t i c l e

i n f o

Article history: Received 13 September 2012 Received in revised form 25 September 2013 Accepted 27 September 2013 Available online 7 October 2013 Keywords: Apple Varieties Consumer acceptance Preference mapping Consumer segmentation European countries

a b s t r a c t A consumer test carried out in 7 different European countries compared 3 standard apple varieties to 8 new ones. A total of 4290 consumers took part in the test. Data from this test was used to develop a preference map for apple. The preference map was constructed with 3 main dimensions (1 – sweetness, fruitiness, flowery attributes, 2 – acidity, firmness, 3 – juiciness and crispness). Consumers were segmented in 6 clusters according to their preferences. The 6 clusters were grouped into two main mega clusters A (68% of consumers) and B (32% of consumers). Megacluster A (Clusters 1, 2, 5 and 6) was characterized by preferring sweet apples. Clusters 2 and 5 (41% of consumers) liked sweet apples independently of their acidity and firmness and moderate positive values on dimension of juiciness and crispness. Cluster 1 (21% of consumers) had an optimal point in positive values of the sweetness dimension, moderate negative value for acidity and firmness and moderate positive value for juiciness and crispness. Cluster 6 (6% of consumers) besides preferring sweet varieties disliked acid-firm varieties. As to regard to megacluster B (Clusters 3 and 4) (32% of consumers), they preferred varieties that were acidic-firm and juiciy and crisp with values in the mid range of the sweetness dimension. In spite of the difficulties in translating preference dimensions into standard practical values for fruit quality and the fact of being a punctual measurement of consumer behaviour, this preference map could be of practical use of different actors on the fruit value chain like marketers and breeders. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Many of the European countries do not achieve minimal recommended intake of fruit and vegetables proposed by FAO/WHO (Robertson et al., 2004). Increase fruit consumption is therefore a public health objective (WHO, 2003) that has been translated into several campaigns of promotion of fruit consumption (Department of Health. The NHS Plan – A plan for investment, 2000 and Center for Disease Control, 2002 cited in Gilmer, 2005; Lock, Pomerleau, Causer, Altmann, & McKee 2005; Subar et al., 1995) but also research projects promoted from the European Union to diminish ⇑ Corresponding author. Tel.: +34 972780275; fax: +34 9727801517. E-mail address: [email protected] (J. Bonany). 0950-3293/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.foodqual.2013.09.010

barriers to fruit consumption that could hamper the achievement of minimal fruit intake. A possible barrier for increased fruit consumption is insufficient fruit quality (Briz et al., 2008; Cohen, Stoddard, Sarouhkhanians, & Sorensen, 1998; Yeh et al., 2008). On apple, in the recent years, a new generation of apple cultivars with improved fruit quality are now making their way into the markets after their release (‘PINK LADYÒ Cripps Pink cov’, ‘KANZIÒ Nicoter cov’, ‘ARIANE cov’, among others). Many of these new apple varieties have a improved texture, higher soluble solids and higher total titratable acidity than currently cultivated varieties like ‘GOLDEN DELICIOUS’ or ‘JONAGOLD’, the two most cultivated varieties in Europe. On the other hand, to increase fruit consumption it could be helpful to know what the preferences of the consumers are and

318

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

how these consumers are segmented with regard to those preferences. Targeting consumer preferences can lead to more satisfied consumers which in turn can result in increased fruit consumption. Consumer preference mapping is the technique most widely used in the food and beverage industries to evaluate the preferences of the consumers and the segmentation of these consumers on homogeneous groups. Additionally, this methodology provides information on what are the food attributes that explain consumer preference. On fruit this technique has been used on apples (AllanWotjas, Sanford, McRae, & Carby, 2003; Daillant-Spinnler, MacFie, Beyts, & Hedderley, 1996; Jaeger, Andani, Wakeling, & MacFie, 1998; Villatoro, López, Echeverria, Graell, & Lara, 2009) , pears (Harker, Gunson, & Jaeger, 2003; Jaeger, Lund, Lau, & Harker, 2003a) and kiwifruit (Jaeger, Rossiter, Wismer, & Harker, 2003b). Consumer tests comparing consumer acceptance of different apple varieties have been carried out in the past in the United Kingdom, Denmark, Holland, Germany, Switzerland, Poland, France, and other European countries (Barendsee, 1993; DaillantSpinnler et al., 1996; Ellinger, 1987; Höhn & Güggenbühl, 1999; Jaeger et al., 1998; Kellerhals et al., 1999; Konopacka, Jesionkowska, Rutkowski, Płocharski, & Tomala, 2006; Mante, 1973; Tomala, Barylko-Pikielna, Jankowski, Jeziorek, & Wasiak-Zys, 2009; van de Abeele & Reijnders, 1980). In spite of numerous consumer tests, many of them, be either consumer tests comparing acceptance of varieties or using the methodology of consumer preference mapping, do not include the major apple varieties consumed in Europe and logically did not cover the newly released cultivars. Moreover, none of the tests involve consumers of a wide selection of European cities impeding a valid comparison among different areas in Europe. A consumer test of 11 apple varieties carried out in different European countries was used as methodological tool to test the hypothesis that new cultivars with better quality attributes would increase fruit consumption. In a previous paper, Bonany et al., (2013), results on consumer eating quality acceptance as related to variety and demographic factors were reported. In this paper, focus is on the results of applying the preference mapping methodology on the same data set.

2. Materials and methods 2.1. Varieties included in the consumer test, origin of fruit samples and management to simulate fruit chain Varieties included in the apple consumer test are listed in Table 1. The varieties selected for the apple consumer test were selected to represent on one side the most cultivated produced apple varieties in Europe (GOLDEN DELICIOUS’ and JONAGOLD) and some newly introduced varieties (FUJI and PINK LADYÒ Cripps Pink cov) or in process of introduction (KANZIÒ Nicoter cov,

Table 2 Temperature, O2 and CO2 concentration and Relative Humidity of cold storage rooms.

a

Variety

Temperature (°C)

O2 (%)

CO2 (%)

Relative humidity (%)

GOLDEN DELICIOUS JONAGOLD FUJI PINK LADYÒ Cripps Pink cov ARIANE cov RUBENSÒ CIVni cov KANZIÒ Nicoter cov JUNAMIÒ Milwa cov WELLANTÒ Cpro-47 LIGOL GOLDCHIEFÒ Gold Pink cov

1.3 na 1.3 2.1 1–1.5 1 1.3 1 1.5–2 1.7 1.5

1.0 na 1.5 4.6 1.5–2 1.2 1.5 1.2 1.3a 2.0 2

3.0 na 1.3 0.9 <1.3 1.5 1.3 2.5 3 1.8–2.0 3

>95 na 91–93 92 90–93 na 90–93 na na Up to 90 90–95

Some periods temporary higher concentration.

JUNAMIÒ Milwa cov, WELLANTÒ, LIGOL, ARIANE cov’, LIGOL, GOLDCHIEFÒ Goldpink cov or RUBENSÒ CIVni cov). On the other side, they were also selected because it was anticipated from knowledge of the varieties that they would provide a good representation of the sensory space (flavour and texture mainly). Fruits of these varieties were harvested from a single representative commercial orchard with standard management practices for each variety. Harvest dates and location and country of origin for each variety can also be found in Table 1. After harvest, fruits were sorted and stored under the appropriate conditions for each variety for long term storage. Locations for storage and detailed conditions are described in Table 2. All fruit samples were removed from cold storage on 15 January 2007 and transported to consumer test locations by means of refrigerated vehicle where they were held between 3 °C and 4 °C until 48 h prior utilization in consumer test. The last 48 h before the test was carried out, fruit samples were maintained at room temperature. 2.2. Fruit quality measurements Fruit quality (Soluble Solids Content, SSC, °Brix; Total Titratable Acidity, TTA, g/L equivalent malic acid; Firmness, F, kg) on 25 fruits for all cultivars included in the test, was measured in different points in time throughout the simulation of the fruit chain: at harvest, at the end of the cold storage period, after the transportation to the location of the consumer test and just prior the moment of consumer test. Only this later measurement was used in the statistical analysis. The rest were used as quality control of the evolution of the different parameters. Starch Pattern Index was also measured at harvest time. 2.3. Trained panel sensory evaluation of fruit samples Additionally to the fruit quality analysis, all varieties were subjected to a sensory evaluation by the AGROSCOPE trained panel in

Table 1 Varieties used for the apple consumer test.

a

Variety

Country were the variety was bred

Location of orchard

Harvest date

‘Golden Delicious’ ‘Jonagold’ ‘Fuji’ ‘Pink LadyÒ’ Cripps Pinkcov ‘Arianecov’ ‘RubensÒ’ CIVnicov ‘KanziÒ’ Nicotercov ‘JunamiÒ’ Milwacov ‘WellantÒ’ CPRO-47 ‘LigolÒ’ ‘GoldchiefÒ’ ‘Gold Pinkcov’

United States United States Japan Australia France Italy Belgium Switzerland The Netherlands Poland Italy

Merano (Italy) Wijk bij Duurstede (The Netherlands) Tramin (Italy) Nîmes (France) Saint Laurent des Vignes (France) Randwijk (The Netherlands) Schenna (Italy) Randwijk (The Netherlands) Randwijk (The Netherlands) Zalesie (Poland) Coredo (Italy)

30/9/2006 11/10/2006a 16/10/2006 2/11/2006 8/9/2006 27/09/2006 6/10/2006 5/10/2006 3/10/2006 25/09/2006 16/10/2006

2nd Pick.

319

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

Wädenswil (Switzerland). The trained panel consisted of 12 people from the staff of AGROSCOPE who were selected after checking for homogeneity with the other panellists and repeatability of sample scoring. All panellists had previous experience and were previously trained and involved in sensory testing of apples. The performance of the panel was checked using PanelCheck software (2006). The sensorial evaluation was replicated twice and in each replicate two sessions were held in two consecutive days (replicate 1: Jan 25 and 30, replicate 2: Feb 1 and 6, 2007). In each session, 5 or 6 apple varieties samples were presented using three digit codes and according to incomplete block design (latin square). Samples of fruits were served cut in halves. The panellists were advised to bite in from the side, having the fruit flesh side on the top and the skin side below when placing the apple sample in the mouth. A total of 18 attributes were evaluated using a non-structured 10 cm line scale verbally anchored at both ends using the FIZZ Software (Biosystèmes, F-Courternon) (Table 3). 2.4. Consumer tests Consumer tests were carried out in different sites located in France, Netherlands, Germany, Poland, Switzerland, Italy and Spain. The test was organized in 39 sessions involving 110 consumers each with a total of 4290 consumers participating. All sessions took place within a relatively short period of time to minimize the changes in fruit quality attributes ranging from 24/01/2007 the earliest to 10/02/2007 the latest. Site, date, partner carrying out the test, number of consumers and the type of location can be found in Table 4. Target population was people between 15 and 70 years old located in major representative European cities regardless of their apple consumption habit for the objective of increase fruit consumption is for all population. All varieties were tested in each session. Fruit samples were prepared so two fruits wedges from the transition from the well exposed part of the fruit to the non well exposed part were be cut from each fruit. Each fruit wedge piece, unpeeled, was placed in a previously coded labelled disposable Styrofoam cup and placed into a tray with 6 varieties. Fruit wedges were maintained only for a short period of time to avoid browning. Experimental design was incomplete blocks derived from Mutually Orthogonal Latin Squares (MOLS) with 110 consumers in each session tasting 6

varieties out of the 11 varieties to minimize the order and carry over effects (Callier & Schlich, 1997; Wakeling & MacFie, 1995). People were invited, according to the target population, to participate in the test and if they accepted they were given a tray with 6 varieties according to pre-established statistical design. After a brief explanation about the purpose of the European research project ISAFRUIT, under which the tests were carried out, the interviewer did shortly explain how to perform the test and give an indication of the total amount of time estimated to carry it out. Invitation to take part in the test was done so proportions representing the desired combination of ages and genders were achieved. Interviewers worked to prevent any cross consumer influence. Each consumer was given a questionnaire in paper format in which the samples were identified in the same order as in the plastic tray. The questionnaire contained 4 different parts. Part A recorded information about age class, gender and classification for consumer habit for apples. Part B included a 9 scale hedonic test of the 6 fruit samples for eating quality acceptance. Part C consisted of a question addressed to estimate the potential increase of apple consumption potentially brought about by the consumption of that specific variety. Finally, Part D included a 9 scale hedonic test for fruit visual acceptance for all 11 varieties included in test. Consumers were asked to taste each variety individually and answer the questions in Part B and C separately for each variety. After parts A, B and C of the test were completed, they were prompted to move to separate but contiguous part where they were shown a display with fruits of the 11 varieties under test and ask to answer Part D of the questionnaire. In this paper, only data from parts A and B will be analyzed. 2.5. Statistical analysis Preference and perceptual data was analyzed using External Preference Mapping approach as described by Schlich and McEwan, (1992). The first part of this procedure makes uses of Principal Component Analysis applied to a matrix of values with attributes evaluated by the expert panel (sensory) and instrumental variables for each one of the 11 varieties as variables and the varieties as observations. The output of this procedure is what is called a sensory map or biplot where projections of the

Table 3 Attributes evaluated by the trained panel in AGROSCOPE (Wadesnwil, Switzerland). Definition

Beginning scale

End of scale

Smell Overall odour intensity (O/OS) Grassy (O/GR) Fruity (O/FR) Flowery (O/FL)

Overall odour intensity Odour intensity of fresh cut grass Intensity of fruity odours Intensity of flowery odours

Not Not Not Not

Very Very Very Very

Texture Juiciness (T/JU) Firmness (T/FN) Mealiness (T/ME) Crispness (T/CR) Chewiness (T/CH) Toughness of the skin (T/TP) Fineness (T/FI)

Amount of liquid released during chewing Force required to bite the apple Mealy mouth-feeling while chewing the apple Force for first bite and its noise intensity Time and number of mastication steps required to the prepare the apple piece before swallowing Force needed to penetrate the skin Smoothness of the fruit flesh

Not applicable Not applicable Not applicable Not applicable Short Not applicable Fine

Very high Very high Very high Very high Long Very tough Rough

Flavour intensity of fresh cut grass Intensity of flowery flavour Intensity of fruity flavour Intensity of apple flavour

Not Not Not Not Not Not Not

Very Very Very Very Very Very Very

Flavour Overall flavour intensity (F/OF) Sweetness (F/SW) Sourness (F/SO) Grassiness (F/GR) Floweriness (F/FL) Fruitiness (F/FR) Apple flavour (F/AF)

applicable applicable applicable applicable

applicable applicable applicable applicable applicable applicable applicable

high high high high

high high high high high high high

320

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

Table 4 Location, sessions, date, partners, number of consumers and type of location of the apple varieties consumer test developed. Location

Session

Date

Partner

Number consumers

Type of location

Espai Gironés, Girona, Spain Valencia, Spain Barcelona, Spain Barcelona, Spain Zaragoza, Spain Madrid, Spain Madrid, Spain San Sebastian, Spain Warsaw, Poland Warsaw, Poland Warsaw, Poland Warsaw, Poland Quickborn–Hamburg, Germany Stade–Hamburg, Germany Quickborn–Hamburg, Germany Buxtehude–Hamburg, Germany Rome, Italy Rome, Italy Rome, Italy Rome, Italy San Lazzaro–Bologna, Italy San Lazzaro–Bologna, Italy Centro Lame, Bologna, Italy Centro Lame, Bologna, Italy Winkelcentrum Overvecht, Utrecht, Netherlands Winkelcentrum Overvecht, Utrecht, Netherlands Winkelcentrum Schalkwijk,Haarlem, Netherlands Winkelcentrum Schalkwijk,Haarlem, Netherlands La Ville du Bois–Paris, France La Ville du Bois–Paris, France La Ville du Bois–Paris, France French speaking area, Switzerland French speaking area, Switzerland German speaking area, Switzerland German speaking area, Switzerland German speaking area, Switzerland Castelssarrasin, France Bordeaux–Sainte Eulalie, France Bordeaux–Sainte Eulalie, France

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 6 7 8 1 2 3 4 5 6 7 8

24/1/2007 2/2/2007 7/2/2007 9/2/2007 27/1/2007 2/2/2007 3/2/2007 10/2/2007 24/1/2007 26/1/2007 31/1/2007 2/2/2007 24/1/2007 25/1/2007 1/2/2007 30/1/2007 24/1/2007 26/1/2007 31/1/2007 2/2/2007 24/1/2007 31/1/2007 7/2/2007 9/2/2007 24/1/2007 25/1/2007 31/1/2007 1/02/2007 25/01/2007 26/01/2007 31/01/2007 26/01/2007 7/02/2007 26/01/2007 2/02/2007 9/02/2007 31/01/2007 9/02/2007 10/02/2007

IRTA IRTA IRTA IRTA IRTA IRTA IRTA IRTA RIPF RIPF RIPF RIPF JORK JORK JORK JORK LAIMBURG LAIMBURG LAIMBURG LAIMBURG UNIBO, CIV UNIBO, CIV UNIBO, CIV UNIBO, CIV WUR-PPO WUR-PPO WUR-PPO WUR-PPO CTIFL CTIFL CTIFL ACW ACW ACW ACW ACW CTIFL CTIFL CTIFL

110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110 110

Commercial mall Commercial mall Commercial mall Fruit shop Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Phone banking company Restaurant/school Phone banking company School, meeting room Consumer test facility Consumer test facility Consumer test facility Consumer test facility Commercial mall Commercial Mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Commercial mall Supermarket Supermarket Commercial mall Supermarket Commercial mall Commercial mall Commercial mall Commercial mall

observations (varieties) and variables (sensory attributes and instrumental quality variables) are plotted in. XLSTAT-Pro module (XLSTAT, New York, NY, USA) was used for this procedure. Afterwards, a second step consisting of grouping consumers accordingly to their preferences was undertaken. More specifically, given that the data set was incomplete (every consumer tasted only 6 out of the 11 varieties), a specific technique was used: CLIP (Clustering of Incomplete Preferences) developed by Callier and Schlich, (1997). With this technique, a similarity matrix is produced. In this matrix, each element i, j of the matrix represents an index of similarity between consumer i and consumer j, the higher the index, the higher the similarity of the ratings of the varieties that consumer i and consumer j have tasted in common. This similarity matrix was obtained using a specifically written SAS program ( SAS Institute Inc., Cary, NC, USA) following the method described in Callier and Schlich, (1997). The similarity matrix is then used as the input data for the Weighed Multidimensional Scaling procedure using PROC MDS (SAS Institute Inc., Cary, NC, USA tware). The result is a two dimension map of consumers with similar preferences. Based on the position in this map, clusters of consumers with similar varietal preferences are created using PROC FASTCLUS (SAS Institute Inc., Cary, NC, USA). Finally using the average acceptance values for each cluster, 4 different models are tested to predict the acceptance of each consumer group based on the attributes of the varieties: (1) vector model, (2) circular ideal point model, (3) elliptical ideal point model and (4) quadratic surface model. This part was carried out using PREFMAP method in XLSTAT-MX module (XLSTAT, New York, NY, USA).

3. Results 3.1. Descriptive and correlation analysis Overall mean, standard deviation, minimum and maximum of all sensory and instrumental quality variables and means for each variety are listed in Table 5. In Table 6, the correlation matrix among all variables can be observed. Quite an important number of significant correlations were found among variables. Among instrumental variables TTA1 and F were significantly correlated (r = 0.757). The relationship between instrumental quality variables (TTA, F, SSC) and their corresponding sensory variables (F/SO, F/FN, F/SW) was also significant. TTA and F/SO correlation was the highest with an r value of 0.967. The relationship between F and T/FN was 0.767, being the lowest between SSC and F/SW with an r value of 0.710. Additionally, SSC was positively correlated with F/AF, F/FR, F/OF and negatively with F/GR. TTA was found positively correlated with sensory texture variables T/FN and T/FI. Finally, F, was positively correlated with T/CH, T/CR and T/FI. In general, the sensory variables show some degree of correlation with other varieties of the same group (Flavour, Texture or

1 SSC: Soluble Solid Contents %, TTA: Total Titrable Acidity, g/L, F: Firmness, kg, F/ AF: Flavour-Apple Flavour, F/FL: Flavour-Flowery, F/FR: Flavour-fruitness, F/GR: Flavour: grassiness, F/OF: Flavour-Overall Flavour, F/SO: Flavour-Sourness, F/SW: Flavour-Sweetness, O/FL: Odour-Flowery, O/FR:Odour-Fruity, O/GR: Odour-Grassiness, O/OS: Odour-Overall Smell, T/CH: Texture-Chewiness, T/CR: Texture-Crispness, T/FN: Texture-Firmness, T/FI: Texture-Fineness, T/JU: Texture-Juiciness, T/ME: Texture-Mealiness, T/TP: Texture-Toughness Peel.

Table 5 Mean for each apple cultivar of instrumental quality variables and sensory scores from trained panel used in the principal component analysis. (SSC: Soluble Solid Contents%, TTA: Total Titrable Acidity, g/L, F: Firmness, N, F/AF: FlavourApple Flavour, F/FL: Flavour-Flowery, F/FR: Flavour-fruitness, F/GR: Flavour: grassiness, F/OF: Flavour-Overall Flavour, F/SO: Flavour-Sourness, F/SW: Flavour-Sweetness, O/FL: Odour-Flowery, O/FR:Odour-Fruity, O/GR: OdourGrassiness, O/OS: Odour-Overall Smell, T/CH: Texture-Chewiness, T/CR: Texture-Crispness, T/FN: Texture-Firmness, T/FI: Texture-Fineness, T/JU: Texture-Juiciness, T/ME: Texture-Mealiness, T/TP: Texture-Toughness Peel). TTA (g/L)

F (N)

F/AF (0–10)

F/FL (0–10)

F/FR (0–10)

F/GR (0–10)

F/OF (0–10)

F/SO (0–10)

F/SW (0–10)

O/FL (0–10)

O/FR (0–10)

O/GR (0–10)

O/OS (0–10)

T/CH (0–10)

T/CR (0–10)

T/FN (0–10)

T/FI (0–10)

T/JU (0–10)

T/ME (0–10)

T/TP (0–10)

16.4 14.3 11.6 13.7

8.5 2.8 4.5 4.6

82.3 53.3 54.9 54.9

4.7 3.4 2.9 4.2

1.1 1.7 0.8 0.9

4.4 3.5 2.5 4.0

1.0 0.7 2.0 0.7

6.0 4.6 4.2 4.8

6.6 1.4 3.3 3.7

4.5 5.8 3.4 4.8

1.0 1.2 0.8 1.0

3.2 4.0 3.1 2.9

1.4 1.2 1.8 1.7

4.1 5.4 4.8 4.8

5.9 4.0 2.9 3.2

6.8 6.9 5.1 5.7

4.8 3.2 2.7 3.1

7.2 4.9 2.9 4.1

5.2 7.6 6.5 6.2

1.0 0.8 1.5 0.8

4.5 3.9 3.9 4.9

14.6 14.7

5.2 4.1

61.7 5.0 48.0 4.1

1.6 0.9

5.2 4.2

0.5 0.4

5.7 5.0

3.9 2.1

5.7 5.3

0.9 1.3

3.3 4.5

1.2 1.0

4.2 5.4

2.9 1.9

5.9 3.8

3.0 3.1

5.1 2.4

6.3 4.3

1.0 3.3

2.6 2.9

16.4

5.0

59.8 5.3

1.3

5.7

0.4

5.9

3.4

6.6

0.8

4.4

0.7

5.4

3.8

6.0

3.3

4.3

6.0

1.5

3.6

14.5

5.8

67.6 5.0

1.0

4.5

1.0

5.9

4.3

5.5

0.5

2.8

2.0

4.8

4.1

6.9

3.0

5.7

6.5

1.1

4.1

12.3 12.7

5.0 6.8

62.7 3.3 61.7 4.0

0.6 1.2

3.1 3.5

1.4 1.4

4.4 5.6

4.1 5.5

3.3 4.5

0.7 1.0

3.0 3.2

1.4 2.9

4.4 5.3

4.4 3.4

7.2 7.7

4.0 3.4

6.1 5.8

6.4 7.7

0.8 0.3

3.7 3.9

14.4

5.2

65.7 4.7

1.9

5.2

0.4

6.1

3.4

6.0

1.6

5.0

0.9

5.8

4.0

5.8

3.4

5.1

5.7

1.8

4.1

14.1 1.5 11.6 16.4

5.2 1.5 2.8 8.5

61.7 8.8 49.0 82.3

1.2 0.4 0.6 1.9

4.2 1.0 2.5 5.7

0.9 0.5 0.4 2.0

5.3 0.7 4.2 6.1

3.8 1.4 1.4 6.6

5.0 1.0 3.3 6.6

1.0 0.3 0.5 1.6

3.6 0.8 2.8 5.0

1.5 0.6 0.7 2.9

4.9 0.5 4.1 5.8

3.7 1.0 1.9 5.9

6.2 1.1 3.8 7.7

3.3 0.6 2.7 4.8

4.9 1.4 2.4 7.2

6.2 1.0 4.3 7.7

1.3 0.8 0.3 3.3

3.8 0.7 2.6 4.9

O/FR

O/GR

O/OS

T/CH

T/CR

T/FN

T/FI

T/JU

4.2 0.8 2.9 5.3

Table 6 Correlation matrix among variables (sensory and instrumental quality variables) included in principal component analysis. Variables SSC TTA F F/AF F/FL F/FR F/GR F/OF F/SO F/SW O/FL O/FR O/GR O/OS T/CH T/CR T/FN T/FI T/JU T/ME T/TP

SSC 1 0.286 0.437 0.781 0.364 0.813 0.733 0.711 0.117 0.710 0.154 0.386 0.564 0.010 0.344 0.079 0.352 0.240 0.468 0.236 0.036

TTA 0.286 1 0.757 0.420 0.147 0.198 0.214 0.588 0.967 0.208 0.179 0.323 0.352 0.473 0.595 0.426 0.646 0.688 0.151 0.280 0.269

F

F/AF 0.437 0.757 1 0.367 0.201 0.251 0.043 0.557 0.710 0.008 0.065 0.190 0.030 0.426 0.920 0.628 0.767 0.896 0.017 0.436 0.370

0.781 0.420 0.367 1 0.341 0.939 0.658 0.914 0.311 0.723 0.013 0.207 0.272 0.017 0.164 0.030 0.098 0.297 0.313 0.087 0.084

F/FL 0.364 0.147 0.201 0.341 1 0.510 0.562 0.481 0.246 0.711 0.647 0.577 0.327 0.450 0.065 0.119 0.098 0.146 0.215 0.089 0.101

F/FR 0.813 0.198 0.251 0.939 0.510 1 0.811 0.839 0.068 0.840 0.198 0.465 0.542 0.127 0.076 0.119 0.052 0.160 0.378 0.222 0.212

F/GR 0.733 0.214 0.043 0.658 0.562 0.811 1 0.514 0.346 0.875 0.479 0.620 0.648 0.345 0.123 0.279 0.022 0.065 0.379 0.358 0.241

F/OF 0.711 0.588 0.557 0.914 0.481 0.839 0.514 1 0.448 0.646 0.159 0.281 0.123 0.090 0.320 0.201 0.234 0.459 0.202 0.008 0.009

F/SO 0.117 0.967 0.710 0.311 0.246 0.068 0.346 0.448 1 0.349 0.311 0.508 0.473 0.584 0.588 0.529 0.614 0.718 0.004 0.457 0.321

F/SW 0.710 0.208 0.008 0.723 0.711 0.840 0.875 0.646 0.349 1 0.349 0.621 0.512 0.500 0.119 0.149 0.241 0.075 0.117 0.243 0.215

O/FL 0.154 0.179 0.065 0.013 0.647 0.198 0.479 0.159 0.311 0.349 1 0.757 0.349 0.595 0.124 0.281 0.077 0.145 0.231 0.352 0.018

0.386 0.323 0.190 0.207 0.577 0.465 0.620 0.281 0.508 0.621 0.757 1 0.664 0.748 0.211 0.430 0.072 0.327 0.371 0.623 0.295

0.564 0.352 0.030 0.272 0.327 0.542 0.648 0.123 0.473 0.512 0.349 0.664 1 0.159 0.017 0.483 0.091 0.219 0.568 0.528 0.277

0.010 0.473 0.426 0.017 0.450 0.127 0.345 0.090 0.584 0.500 0.595 0.748 0.159 1 0.364 0.248 0.407 0.447 0.077 0.384 0.029

0.344 0.595 0.920 0.164 0.065 0.076 0.123 0.320 0.588 0.119 0.124 0.211 0.017 0.364 1 0.684 0.824 0.857 0.109 0.504 0.554

0.079 0.426 0.628 0.030 0.119 0.119 0.279 0.201 0.529 0.149 0.281 0.430 0.483 0.248 0.684 1 0.459 0.851 0.710 0.881 0.364

0.352 0.646 0.767 0.098 0.098 0.052 0.022 0.234 0.614 0.241 0.077 0.072 0.091 0.407 0.824 0.459 1 0.749 0.211 0.241 0.307

0.240 0.688 0.896 0.297 0.146 0.160 0.065 0.459 0.718 0.075 0.145 0.327 0.219 0.447 0.857 0.851 0.749 1 0.270 0.669 0.324

0.468 0.151 0.017 0.313 0.215 0.378 0.379 0.202 0.004 0.117 0.231 0.371 0.568 0.077 0.109 0.710 0.211 0.270 1 0.808 0.212

T/ME 0.236 0.280 0.436 0.087 0.089 0.222 0.358 0.008 0.457 0.243 0.352 0.623 0.528 0.384 0.504 0.881 0.241 0.669 0.808 1 0.413

T/TP 0.036 0.269 0.370 0.084 0.101 0.212 0.241 0.009 0.321 0.215 0.018 0.295 0.277 0.029 0.554 0.364 0.307 0.324 0.212 0.413 1

321

Bold values are statistically sifnificant from 0 with significance level alfa = 0,05. (SSC: Soluble Solid Contents%, TTA: Total Titrable Acidity, g/L, F: Firmness, N, F/AF: Flavour-Apple Flavour, F/FL: Flavour-Flowery, F/FR: Flavour-fruitness, F/GR: Flavour: grassiness, F/OF: Flavour-Overall Flavour, F/SO: FlavourSourness, F/SW: Flavour-Sweetness, O/FL: Odour-Flowery, O/FR:Odour-Fruity, O/GR: Odour-Grassiness, O/OS: Odour-Overall Smell, T/CH: Texture-Chewiness, T/CR: Texture-Crispness, T/FN: Texture-Firmness, T/FI: TextureFineness, T/JU: Texture-Juiciness, T/ME: Texture-Mealiness, T/TP: Texture-Toughness Peel).

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

ARIANE cov FUJI JONAGOLD WELLANTÒ CPRO47 RUBENSÒ CIVni cov GOLDEN DELICIOUS GOLDCHIEFÒ Goldpink cov JUNAMIÒ Milwa cov LIGOL KANZIÒ Nicoter cov PINK LADYÒ Cripps Pink cov Mean Standard Deviation Minimum Maximum

SSC (°Brix)

322

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

Odour groups). Within the Flavour variables group, F/AF and F/FR, F/AF and F/OF, F/AF and F/SW, F/FL and F/SW, F/FR and F/OF, F/FR and F/SW, F/OF and F/SW were positively correlated whereas F/GR and F/AF, F/GR and F/FR, F/GR and F/SW were negatively correlated. When considering the group Odour sensory variables, the following positive correlations were found: O/FL and O/FR, O/FR and O/OS while O/FR was negatively correlated with O/GR. When Texture sensory variables are considered, the following couple of variables were statistically significantly positively related: T/CH and T/CR, T/CH and T/FN, T/CH and T/FI, T/CR and T/FI, T/CR and T/JU, T/FN and T/FI. On the contrary the following pairs T/CR and T/ME, T/FI and T/ME, T/JU and T/ME were negatively correlated. Some sensory variables showed also a correlation outside of their own group. Thus, between variables of flavour and odour were found the following positive correlations: F/FL and O/FL, F/SW and O/FR, F/GR and O/GR whereas a negative correlation between F/GR and O/FR was found. Among variables of Flavour and Texture very few correlations were found. Only F/SO was positively correlated with T/FN and T/FI. Likewise between odour sensory variables and texture sensory variables only a positive correlation was observed between O/FR and T/ME. 3.2. Sensory map From the results of Principal Component Analysis it can be observed that all instrumental and sensory variables measured on the 11 varieties can be summarized in 3 main orthogonal components accounting for 76.55% of the total variance. In Fig. 1, a scree plot can be observed and relative importance of the 3 main components can be assessed. In order to simplify the interpretation of the different dimensions of the principal component analysis, the VARIMAX transformation (Kaiser, 1958) was applied. A new distribution of the total variance explained by each component while maintaining the accumulated variance for the first 3 components (D1, D2, D3) was obtained (Table 7). Although the variance accumulated by the first 3 components remains to 76.55% of the total, the distribution among the 3 components is more homogeneous. Observing Table 8, it can be interpreted that the first component represents mainly the following variables: TTA, F/SO, F, T/FI, T/CH and T/FN. The second component would basically represent F/SW, F/GR, F/FR, F/FL. Finally the third component would represent

Fig. 1. Scree plot of principal component analysis.

Table 7 Percentage of variance and % accumulated variance for each principal component after VARIMAX transformation.

Variance (%) Accumulated variance (%)

D1

D2

D3

30,885 30,885

29,594 60,479

16,072 76,551

Table 8 Contribution of each variable (%) to the different principal components after VARIMAX transformation.

SSC TTA F F/AF F/FL F/FR F/GR F/OF F/SO F/SW O/FL O/FR O/GR O/OS T/CH T/CR T/FN T/FI T/JU T/ME T/TP

D1

D2

D3

2,535 12,510 12,232 3,047 0,309 0,945 0,195 4,343 12,421 0,370 1,544 2,634 0,434 6,183 10,120 4,982 9,319 11,198 0,150 2,809 1,720

8,950 0,023 0,660 8,097 10,510 11,760 12,174 8,554 0,664 14,061 4,981 8,523 5,222 3,800 0,112 0,067 0,011 0,193 0,587 0,779 0,273

3,559 0,102 1,285 1,732 4,821 2,379 1,831 0,075 0,086 0,021 0,031 1,321 6,932 1,066 2,588 17,653 0,016 5,325 25,794 18,988 4,396

SSC: Soluble Solid Contents %, TTA: Total Titrable Acidity, g/L, F: Firmness, kg, F/AF: Flavour-Apple Flavour, F/FL: Flavour-Flowery, F/FR: Flavour-fruitness, F/GR: Flavour: grassiness, F/OF: Flavour-Overall Flavour, F/SO: Flavour-Sourness, F/SW: Flavour-Sweetness, O/FL: Odour-Flowery, O/FR:Odour-Fruity, O/GR: Odour-Grassiness, O/OS: Odour-Overall Smell, T/CH: Texture-Chewiness, T/CR: Texture-Crispness, T/FN: Texture-Firmness, T/FI: Texture-Fineness, T/JU: Texture-Juiciness, T/ME: Texture-Mealiness, T/TP: Texture-Toughness Peel.

T/JU, T/ME/ and T/CR. Therefore the sensory map will be represented by 3 major dimensions. The first component or dimension (ACIDIY–FIRMNESS) is representing more acid and firm fruit in the positive range and less acid and less firm in the negative range. The second dimension (SWEETNESS) is correlated with sweetness (F/SW) and fruity flavour/odour (F/FR, F/FL) in the positive range and grassiness, unripe or not matured fruit in the negative range (F/GR). Finally the third dimension (JUICINESS–CRISPNESS), most probably, is representing juiciness and crispness (T/JU, T/CR) in the positive range and mealiness in the negative range (T/ME). In Fig. 2 (depicting first and second components) and Fig. 3 (first and third dimension), the position of the 11 cultivars tested in the 3 dimensional sensory map can be observed. ARIANE cov is well distinguished since it situated in an area in the map indicating high values for the first component (acidity and firmness) while situated in the mid range values for perceived sweetness (first dimension). PINK LADYÒ Cripps Pink cov, GOLDCHIEFÒ Goldpink cov and RUBENSÒ CIVni cov are varieties that are situated in the centre region for acidity (both instrumental TTA and perceived F/SO) and Firmness (F) while at the same time are situated in the high values region for the second dimension of perceived sweetness, fruity flavour and low values of perceived grassiness. As it concerns the third component positively related to T/JU and T/CR and negatively correlated to F/GR, these varieties are located in the middle region. KANZIÒ Nicoter cov, JUNAMIÒ Milwa cov, WELLANTÒ and LIGOL are located in the middle range values of the acidity–firmness dimension and in the negative range of the second dimension

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

323

Fig. 2. Biplot of instrumental quality and sensory variables and cultivars in the first and second component sensory map after VARIMAX transformation. (SSC: Soluble Solid Contents%, TTA: Total Titrable Acidity, g/L, F: Firmness, kg, F/AF: Flavour-Apple Flavour, F/FL: Flavour-Flowery, F/FR: Flavour-fruitness, F/GR: Flavour: grassiness, F/OF: Flavour-Overall Flavour, F/SO: Flavour-Sourness, F/SW: Flavour-Sweetness, O/FL: Odour-Flowery, O/FR:Odour-Fruity, O/GR: Odour-Grassiness, O/OS: Odour-Overall Smell, T/CH: Texture-Chewiness, T/CR: Texture-Crispness, T/FN: Texture-Firmness, T/FI: Texture-Fineness, T/JU: Texture-Juiciness, T/ME: Texture-Mealiness, T/TP: Texture-Toughness Peel) (ARI: Ariane; FUJ: Fuji; GC: Gold Pink; GD: Golden Delicious; JON: Jonagold; JUN: Milwa; KAN: Nicoter; LIG: Ligol; PL: Cripps Pink; RUB: CIVni; WEL: Cpro-47).

(sweetness), indicating relatively low sweetness values. At the same time, KANZIÒ Nicoter cov is located in the high end values of third dimension (positively correlated to T/JU, T/CR), JUNAMIÒ Milwa cov, WELLANTÒ in the middle range whereas LIGOL on the lower range of this component. Finally, although GOLDEN DELICIOUS, JONAGOLD and FUJI are situated in the negative range for the first dimension (acidity and firmness) as regards to their classification for sweetness, GOLDEN DELICIOUS is located in the mid range for sweetness; JONAGOLD is falling in the negative part and FUJI in the positive range of it. They also fall in different categories regarding third component (T/JU; T/CR). Whereas FUJI falls into the category of high values for third dimension (juicy varieties), JONAGOLD is falling in the middle range for this component, whereas GOLDEN DELICIOUS is falling into the category of low values for juiciness and crispness. A summarized description of the projection of the varieties in the sensory map can be found in Table 9. 3.3. Cluster analysis. Grouping consumers Consumers were grouped in clusters based in similarity of the scores given to the cultivars tested in common as it has been described in the methodology. Cluster analysis (SAS FASTCLUS procedure) was performed based on the Euclidean distances between consumers calculated through the SAS MDS procedure. Six clusters were selected based on the analysis of the percentage of explained variance from the total for each additional cluster considered. Frequencies in each cluster are described in Table 10. Four major

clusters, considering the size of each cluster, were detected: Cluster 2, 4, 1 and 3 with two other minor clusters: Clusters 5 and 6. Average hedonic scores for each cultivar for all consumers in each cluster and the cultivar preferences within each cluster can be observed in Table 11. Cluster 1 (21% of consumers), Cluster 2 (38%) and Cluster 5 (3%) show the same varieties for the 4 most preferred varieties: GOLDCHIEFÒ Goldpink cov, RUBENSÒ CIVni cov, FUJI, PINK LADYÒ Cripps Pink cov although with changes in the order of preference among them. Cluster 6 (6%) is showing also FUJI, GOLDCHIEFÒ Goldpink cov and PINK LADYÒ Cripps Pink cov within the first four positions but includes GOLDEN instead of RUBENSÒ CIVni cov in the four most preferred. These clusters have also in common that they show JONAGOLD and ARIANE cov as the least preferred varieties, with the exception of Cluster 6 that is KANZIÒ Nicoter cov and ARIANE cov the worst appreciated cultivars for the consumers in this group. Flavour acceptance scores for Cluster 1, 2, 5 and 6 for cultivars assessed show a similar pattern. However, consumers in cluster 5 tend to score the varieties lower than consumers in clusters 1 or 2. Consumers in cluster 6, although showing a similar pattern to those in cluster 1, 2 and 5 are somehow different in the sense that consumers in this cluster seem to show a higher preference for GOLDEN DELICIOUS than consumers in cluster 1, 2 or 5. These four clusters of similar pattern (Cluster 1, 2, 5, 6) are considered to be part of a larger group of consumers called megacluster A with 68% of consumers that took part in the test. On the other hand, Cluster 3 (10% of consumers) and Cluster 4 (22%) show also the same four varieties in the same order of

324

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

Fig. 3. Biplot of instrumental quality and sensory variables and cultivars in the first and third component sensory map after VARIMAX transformation. (SSC: Soluble Solid Contents %, TTA: Total Titrable Acidity, g/L, F: Firmness, kg, F/AF: Flavour-Apple Flavour, F/FL: Flavour-Flowery, F/FR: Flavour-fruitness, F/GR: Flavour: grassiness, F/OF: Flavour-Overall Flavour, F/SO: Flavour-Sourness, F/SW: Flavour-Sweetness, O/FL: Odour-Flowery, O/FR:Odour-Fruity, O/GR: Odour-Grassiness, O/OS: Odour-Overall Smell, T/CH: Texture-Chewiness, T/CR: Texture-Crispness, T/FN: Texture-Firmness, T/FI: Texture-Fineness, T/JU: Texture-Juiciness, T/ME: Texture-Mealiness, T/TP: Texture-Toughness Peel) (ARI: Ariane; FUJ: Fuji; GC: Gold Pink; GD: Golden Delicious; JON: Jonagold; JUN: Milwa; KAN: Nicoter; LIG: Ligol; PL: Cripps Pink; RUB: CIVni; WEL: Cpro-47).

preference as the most four preferred cultivars: KANZIÒ Nicoter cov, ARIANE cov, JUNAMI and RUBENS. They also coincide with the fact that is GOLDEN DELICIOUS the least preferred variety. Both groups of consumers show a similar pattern of acceptance of varieties with small differences in the order of preference. The main difference is that Cluster 4 tends to score higher the varieties than Cluster 3 (Table 11). These two clusters are considered to be part also of a larger group called megacluster B (32% of consumers).

3.4. Preference Map Average preferences for each cultivar were modelled for each cluster as a function of the main components (D1, D2 and D3) using PREFMAP method included in XLSTAT software. Linear, circular, elliptical and quadratic model were tested. For cluster 1, circular model was retained whereas for the other clusters a linear or vector model was the only model statistically significant. A statistical summary for each cluster can be found in Table 12. In Figs. 4 and 5, bi-dimensional Preference Maps can be observed. These maps include position of the ideal point and isopreference curves for the circular model for cluster 1 and vectors indicating the direction of maximum preference for each one of the linear models for clusters 2, 3, 4, 5 and 6. The position or direction for each one of the clusters, confirm the existence of two main groups or megaclusters as already stated previously. One group of consumer clusters, megagroup A includes C1, C2, C5 and C6 whereas the other group is made of C3 and C4 clusters.

Within the first megagroup, models of acceptance prediction for clusters C2 and C5, accounting for 41% of consumers, are linear models. These models are represented in the different bi-dimensional plots as vectors indicating direction of maximum acceptance. As it can be observed in Fig. 4, the vector models are quite aligned with D2 (sweetness component) pointing to the positive range of it, indicating that consumers in these clusters prefer apple cultivars with high values of attributes positively correlated (F/SW, F/FR, F/FL) and negatively correlated (F/GR) being relatively independent of D1 (acidity–firmness) and D3 (juiciness–crispness) and its correlated variables TTA, F/SO, F, T/FI and T/CH for D1 and T/JU, T/ME/ and T/CR for D3 respectively. The varieties which projection into the bi–dimensional preference map plot (D1, D2) is closer to the vector of preference for these consumer clusters are GOLDCHIEFÒ Goldpink cov, RUBENSÒ CIVni cov, FUJI, PINK LADYÒ Cripps Pink cov. These two clusters of consumers prefer varieties with high values of sweetness (F/SW and SSC), fruitiness (F/FR) and flowery (F/FL) and low values of grassiness (F/GR). At the same time their preference is relatively independent of component D1 (acidity–firmness) and D3 (juiciness–crispness) provided these values are in a middle range values. Comparing the two clusters C2 and C5, in fact C5 is very minor cluster regarding number of consumers (3%), and besides a slight change in the preference direction compared to C2, consumers in C5 scored on average much lower all varieties compared to consumers in C2. Cluster C1 with of 21% of consumers is the only cluster where the retained model is a circular model with the optimum

325

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329 Table 9 Classification of each of the 11 varieties included in the consumer test regarding their position in the sensory map. First component-dimension (D1) Total Tritable Acidity/TTA (+) Firmness (F) (+) Flavour Sourness F/SO (+) Texture Fineness (F/FI) (+) Texture Chewiness (+)

Second component-dimension (D2) Flavour Sweetness (F/SW) (+), Flavour Fruitness (F/FR) (+) Flavour Floweriness (F/FL) (+) Flavour Grassiness (F/GR) ( )

Third component-dimension (D3) Texture Juiciness (T/JU) (+) Texture Crispness (T/CR) (+) Texture Mealiness (T/ME) ( )

+

+

+ ±

ARIANE cov

±

ARIANE cov

+ ±

ARIANE cov

+ ± ±

PINK LADYÒ Cripps Pink cov RUBENSÒ CIVni cov KANZIÒ Nicoter cov JUNAMIÒ Milwa cov WELLANTÒ CPRO47 LIGOL GOLDCHIEFÒ Goldpink cov

PINK LADYÒ Cripps Pink cov GOLDCHIEFÒ Goldpink cov RUBENSÒ CIVni cov

+ ±

KANZIÒ Nicoter cov JUNAMIÒ Milwa cov WELLANTÒ CPRO47

+ ±

LIGOL

+ ±

+

FUJI

+ ±

±

GOLDEN DELICIOUS

+ ±

JONAGOLD

+ ±

+

±

FUJI GOLDEN DELICIOUS JONAGOLD

PINK LADYÒ Cripps Pink cov GOLDCHIEFÒ Goldpink cov RUBENSÒ CIVni cov KANZIÒ Nicoter cov JUNAMIÒ Milwa cov WELLANTÒ CPRO47

FUJI

GOLDEN DELICIOUS

Table 10 Cluster analysis results. Cluster

Frequency

%

RMS

Closest cluster

1 2 3 4 5 6

902 1631 436 944 121 256

21 38 10 22 3 6

0.0624 0.0725 0.0709 0.0660 0.0700 0.0760

2 6 2 2 3 2

situated in high values of D2 (sweetness), low values for D1 (acidity–firmness) and mid range values for D3 (juiciness–crispness). For these consumers, contrarily to C2 and C5 clusters where the higher the values for sweetness component higher the acceptance,

JONAGOLD

there is an optimal value after which the acceptance diminishes. The same comment is also valid for the acidity–firmness component. In this case, GOLDCHIEFÒ Goldpink cov, FUJI, and PINK LADYÒ Cripps Pink cov and RUBENSÒ CIVni cov are the varieties with projected values on the bi-dimension (D1, D2) preference map closer to preference vector for this cluster. Regarding consumer cluster C6 (6% of consumers), the vector of maximum preference has a significant projection in the positive range of D2 (sweetness component) and in the negative range of component D1 (acidity–firmness) being relatively independent of D3 (juiciness–crispness). FUJI and GOLDEN DELICIOUS are the varieties with a projection closer to this vector of preference. As it concerns the clusters included in megagluster B, clusters 3 and 4 with a total of 32% of consumers, the directions of maximum preference are completely different from the previously described

Table 11 Average score of each cultivar for all consumers in the different clusters. Cultivar

Clusters C1

ARIANE cov FUJI JONAGOLD WELLANTÒ CPRO-47 RUBENSÒ CIVni cov GOLDEN DELICIOUS GOLDCHIEFÒ Gold Pink cov JUNAMIÒ Milwa cov LIGOL KANZIÒ Nicoter cov PINK LADYÒ Cripps Pink cov a

6.42 7.87 7.07 7.60 7.92 7.76 8.00 7.49 7.40 7.21 7.79

C2 a

f ab e abcd a abc a bcd cde de abc

5.08 6.78 5.10 5.93 6.57 6.51 6.98 6.21 5.85 5.58 6.66

C3 f ab f cd b b a c de e b

6.05 4.29 3.91 4.41 5.74 3.85 5.67 5.85 5.12 6.67 4.75

C4 b efg fg ef b g bc b cd a de

7.35 5.53 5.90 6.41 7.04 5.01 6.90 7.33 7.02 7.78 6.52

C5 b f f e bc g cd b bc a de

3.11 5.09 2.93 3.38 4.85 4.72 5.26 4.84 3.41 3.89 4.86

C6 c a c c ab ab a ab c bc ab

3.12 7.73 5.10 5.65 6.22 7.12 6.48 4.92 4.90 3.59 6.32

g a ef de cd ab bc ef f g cd

LSMEANS in the same column correspond to comparisons of the different varieties in the same cluster. Means with different letters are significantly different (a = 0.05).

326

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

Table 12 Statistical parameters of models obtained with PREFMAP (XLSTAT Software) method. Cluster C1 C2 C3 C4 C5 C6

Model a

Circular Linearb Linear Linear Linear Linear

df

SS

MS

R2

F

Pr > F

F-ratio

Pr > F

4 3 3 3 3 3

9.114 7.709 6.757 7.342 7.443 8.365

2.279 2.570 2.252 2.447 2.481 2.788

0.911 0.771 0.676 0.734 0.744 0.836

15.434 7.851 4.861 6.446 6.790 11.936

0.003 0.012 0.039 0.020 0.018 0.004

13.716

0.010

a Circular model: Y = a0 + a1 D1 + a2 D2 + a3 D3 + a4 D12 + a2 D22 + a3 D32, where Y is average acceptance score for each variety and D1, D2 and D3, main components from Principal Component Analysis. Ideal point at D1 = 1,505, D2 = 1,109, D3 = 0,314. b Linear model : Y = a0 + a1 D1 + a2 D2 + a3 D3, where Y is average acceptance score for each variety and D1, D2 and D3, main components from principal component analysis.

Fig. 4. Bidimensional preference map of components D1 and D2 with representation of each cultivar, the vector of maximum preference for clusters C2, C3, C4, C5 and C6 and isopreference curves according to circular model with ideal point for cluster C1. Additionally contour plot of estimated percentage of clusters above average preference is depicted. (ARI: Ariane; FUJ: Fuji; GC: Gold Pink; GD: Golden Delicious; JON: Jonagold; JUN: Milwa; KAN: Nicoter; LIG: Ligol; PL: Cripps Pink; RUB: CIVni; WEL: Cpro-47).

clusters. The main characteristic of these clusters is that the direction of preference is in the direction of the positive range for D1 (acidity–firmness), i.e. high values of the correlated variables TTA, F/SO, F, T/FI and T/CH. These clusters are basically independent of D2 (sweetness) and relatively positively correlated with D3 (juiciness–crispness). KANZIÒ Nicoter cov, ARIANE cov, JUNAMIÒ Milwa cov and RUBENSÒ CIVni cov are the varieties with projections more aligned with the preference vector in these clusters. Besides an slight difference in the direction of preference, C3 cluster consumers tend to score lower the varieties than consumers in C4. Bi-dimensional representations (Figs. 4 and 5) of preference map show also contour areas of estimated percentage of clusters with above average preference. Observing the figures it can be noticed an area of maximum percentage (80–100%) of clusters with preference above average. This area covers all positive range of D2 (sweetness component) and small positive range of D1 (acidity–firmness). In the D1 (acidity–firmness), D3

(juiciness–cripsness) plane, area of maximum preference covers a portion of the positive range of D3 and short portion of negative range of D3. Finally, on the D2, D3 plane, the area is defined by the whole positive range of D2 (sweetness) and a significant portion of D3 (juiciness–crispness). GOLDCHIEFÒ Goldpink cov, and PINK LADYÒ Cripps Pink cov and RUBENSÒ CIVni cov fall into the area of maximum acceptance.

4. Discussion The sensory map with 3 main components explained 76.6% of the total variance of the sensorial and instrumental variables used to describe the 11 varieties under test. First component is about acidity and firmness and shows a high correlation with both instrumental and sensorial acidity and firmness. The second component is related with sweetness and it is highly positively correlated with sensorial sweetness, flowery, fruitiness scores and to a

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

327

Fig. 5. Bidimensional preference map of components D1 and D3 with representation of each cultivar, the vector of maximum preference for clusters C2, C3, C4, C5 and C6 and isopreference curves according to circular model with ideal point for cluster C1. Additionally contour plot of estimated percentage of clusters above average preference is depicted. (ARI: Ariane; FUJ: Fuji; GC: Gold Pink; GD: Golden Delicious; JON: Jonagold; JUN: Milwa; KAN: Nicoter; LIG: Ligol; PL: Cripps Pink; RUB: CIVni; WEL: Cpro-47).

lesser extent to Soluble Solids content and negatively correlated with sensorial grassiness. Finally, the third component is positively related to juiciness and crispness and negatively with mealiness. These findings are at large, in concordance with previous preference maps for apple published so far on apple preference map (Carbonell, Izquierdo, Carbonell, & Costell, 2008; DaillantSpinnler et al., 1996; Jaeger et al., 1998; Popper, 1998). Attributes on sweetness, firmness and juiciness or related conform the main axis of the preference maps available. Daillant-Spinnler et al., 1996 in a study involving 12 varieties and 120 consumers in the UK, found two main components. First component was associated to juiciness, crispness, hardness, green apple and fresh flavor and second dimension was related to sweetness and acidity (latter one also related to the 1st component). Jaeger et al., 1998, in a study including Danish (117) and English (127) consumers with 3 varieties at various levels of mealiness, found that the first dimension was related flavor (sweetness, acidity) differences and second to texture differences mostly mealiness level. Popper, 1998, found that two dimensions accounted for a significant part of the variability. The first component combined both taste attributes like sweetness and acidity with texture attributes like hardness (interpreted as firmness) while the second dimension was related to juiciness and crispness. Finally, Carbonell et al., (2008) in a methodological trial involving 3 apple varieties and 3 levels of 2 months storage temperature in Spain with 99 consumers, found also two main dimensions. All attributes of taste and texture measured where associated to both dimensions except for juiciness, flouriness, granularity and taste intensity that were basically related to only the second dimension. Across different countries with different varieties, most of preference maps dimensions are associated with flavor, sweetness and acidity, texture like hardness or firmness and also juiciness and crispness. When only two dimensions are considered, sweetness and acidity might be part of the same dimension although in opposite directions. Also acidity

and firmness are often part of the same axis since there are biological reasons for this association (Harker et al., 2003). Clustering of consumers yielded 6 consumer segments. Four of these clusters, megacluster A (cluster 1, 2, 5 and 6) accounting for 68% of the consumers that took part of the test have in common for the preference of a sweet apple whereas only 32% preferred a more acidic variety. Clusters 2 and 5 are very similar in preference directions and account 41% of the population. Consumers in these clusters are characterized by better acceptance for varieties with higher sweetness, fruitiness and flowery scores, relatively independent of the acidity and firmness scores and mid range for juiciness and crispness. Cluster 1 (21% of consumers) is defined by an optimal point of preference on the range of high values for sweetness, relatively low values for acidity and firmness and mid range for juiciness and crispness. Cluster 6 (6% of consumers) is characterized by a group of people that dislikes acidity but maintain preference for high sweetness and mid range to high values for crispness and juiciness. On the other hand in megacluster B, (clusters 3 and 4) representing 32% of consumers prefers an acidic, firm, juicy and crisp variety with mid range values for sweetness. These results are quite difficult to compare with previously reported preference maps because lack of reporting on relative importance of consumer segments, differences of countries and varieties. Nevertheless, Daillant and Spinnler et al., 1996, reported two segments of consumers, one of the groups preferring sweet, hard and juicy and the other one liking more juicy and acidic apple. These two groups could be roughly assimilated to megagroup A and B respectively. However, no report on relative importance of each group as percentage of consumers is given. Results about consumer segments described in Popper, 1998 and Carbonell et al., 2008, are also compatible with results described in this paper in the sense that they both describe a group of consumers liking more acidic, hard and juicy products and other groups with more prevalence of sweetness, aromatic, fruity products even if they are less firm and juicy.

328

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329

Carbonell et al., 2008 reports that the acidic, juicy and hard preferred variety group accounts to 29% of the people in Spain. Results on consumer segments reported by Jaeger et al., 1998 are less comparable since the two consumer segments are based on one liking most fresh apples (sweet, floral, fruity, hard, juicy and crisp) and the other segment preferring also sweet but mealier apples. No report on percentages and relative importance of each consumer segment is given either. Preference mapping can be of practical use in different parts of the fruit value chain. With 62% of consumers, Cluster 1 (21% of consumers), Cluster 2 (38%) and Cluster 5 (3%), showing the same 4 varieties as the most preferred varieties: GOLDCHIEFÒ Goldpink cov, RUBENSÒ CIVni cov, FUJI, PINK LADYÒ Cripps Pink cov, and 32% (Clusters 3 and 4), liking ARIANE cov and KANZIÒ Nicoter, it indicates very clearly the segmentation of the market in different countries in Europe. This would allow the positioning of these varieties or varieties with equal characteristics in the supermarket shelf and it will need a communication effort regarding their characteristics of flavour and texture according to consumer needs. Another practical implication is related to fruit breeding programs. Either the breeding aims are concentrated in an area of the preference map with higher percentage of groups showing acceptance above average (Figs. 4 and 5) described as having positive values of dimension 1 (sweetness, fruitness, flowery), slightly positive values for dimension 2 (acidity and firmness) and moderate positive values for dimension 3 (juiciness and crispness) or else the breeding aims are concentrated in producing varieties with characteristics for each one of the consumer clusters, specially the larger ones. Although, in this analysis instrumental measurements have been used as descriptors of variety products together with sensorial evaluations by panelists, the preference dimensions are linear combinations of both and in spite of good correlations between instrumental and sensorial evaluations, it is not a simple task to translate the preference map dimension values for preferred products into practical standard measurements of quality (Harker et al., 2003). On the other hand, the preference map has been obtained from a consumer test developed at a specific time in the season. Certainly, results would have changed if the test would have been carried out earlier or later in the season although scale and direction of change remains to be observed and it does not preclude utility of this point in time measurement.

5. Conclusions A preference map for apple in different European countries has been developed. The preference map encompasses 3 main dimensions (1-sweetness, fruitiness, flowery attributes, 2-acidity, firmness, 3-juiciness and crispness). The eleven varieties included in the test have been placed into the map and consumers have been segmented in 6 clusters according to their preferences. The 6 clusters are grouped into two main mega clusters A (68% of consumers) and B (32% of consumers). Megacluster A (Clusters 1, 2, 5 and 6) is characterized by preferring sweet apples. Clusters 2 and 5 (41% of consumers) liked sweet apple independently of acidity and firmness and moderate positive values on dimension of juiciness and crispness. Cluster 1 (21% of consumers) has an optimal point in positive values of sweetness dimension, moderate negative value for acidity and firmness and moderate positive value for juiciness and crispness. Cluster 6 (6% of consumers) besides preferring sweet varieties dislike acid-firm varieties. As to regard to megacluster B (32% of consumers), they prefer varieties that are acidic-firm and juiciy and crisp with values in the mid range of sweetness dimension. In spite of the difficulties in translating preference dimensions into standard practical values for fruit quality and

the fact of beign a punctual measurement of consumer behaviour, this preference map can be of practical use of different actors on the fruit value chain likle marketers and breeders. Acknowledgements This work was supported through the ISAFRUIT project, funded by the European Commission under the Thematic Priority 5 – Food Quality and Safety of the 6th Framework Programme of RTD (Contract No. FP6-FOOD-CT-2006-016279). The views and opinions expressed in this publication are purely those of the writers and may not, in any circumstances, be regarded as stating an official position of the European Commission. References Allan-Wotjas, P., Sanford, K. A., McRae, K. B., & Carby, S. (2003). An integrated microstructural and sensory approach to describe apple texture. Journal of the American Society for Horticultural Science, 128, 381–390. Barendsee, V. (1993). Effect of harvesting time and handling period on quality of apple. Fruit Belge, 48(392), 275–278. Bonany, J., Buehler, A., Carbó, J., Codarin, S., Donati, F., Echeverria, G., et al. (2013). Consumer eating quality acceptance of new apple varieties in different European countries. Food Quality and Preference, 30(2), 250–259. ISSN 0950-3293, http://dx.doi.org/10.1016/j.foodqual.2013.06.004, . Briz, T., Sijtsema, S. J., Jasiulewicz, A., Kyriakidi, A., Guardia, M. D., van Berg, I., et al. (2008). Barriers to fruit consumption: Driving forces behind consumer behaviour. Scripta Horticulturae, 8(1), 7–18. Callier, P., & Schlich, P. (1997). La cartographie des preferences incompletes – Validation par simulation. Sciences des Aliments, 17, 155–172. Carbonell, L., Izquierdo, L., Carbonell, I., & Costell, E. (2008). Segmentation of food consumers according to their correlations with sensory attributes projected on preference spaces. Food Quality and Preference, 19, 71–78. Center for Disease Control (2002). National 5 a day partnership plan 2001–2004. Atlanta (GA), USA: National Center for Chronic Disease Prevention and Health Promotion. . Cohen, N. L., Stoddard, A. M., Sarouhkhanians, S., & Sorensen, G. (1998). Barriers toward fruit and vegetable consumption in a multiethnic worksite population. Journal of Nutrition Education, 30(6), 381–386. Daillant-Spinnler, B., MacFie, H. J. H., Beyts, P. K., & Hedderley, D. (1996). Relationships between perceived sensory properties and major preference directions of 12 varieties of apples from the Southern Hemisphere. Food Quality and Preference, 7, 113–126. Department of Health. The NHS Plan – A plan for investment (2000). A plan for reform. London: The Stationery Office. Ellinger, W. (1987). By which criteria are apples bought and how fruit from the South Tyrol evaluated? An analysis from the view point of the german market. Obstbau-Weinbau, 241, 4–8. Gilmer, R. (2005). A new day for 5-A-day (vol. 98(2)). Willoughby: Florida Grower, p. 31. Harker, F. R., Gunson, F. A., & Jaeger, S. R. (2003). The case for fruit quality: An interpretive review of consumer attitudes, and preferences for apples. Postharvest Biology and Technology, 28, 333–347. Höhn, E., & Güggenbühl, B. (1999). Golden Delicious: Festigkeit, Zuckergehalt und Akzeptanz bei der Konsumentenschaft. Schweizer Zeitschrift für Obst-und Weinbau, 14, 339–342. Jaeger, S. R., Andani, Z., Wakeling, I. N., & MacFie, H. J. H. (1998). Consumer preferences for fresh and aged apples: A cross-cultural comparison. Food Quality and Preference, 9, 335–366. Jaeger, S. R., Lund, C. M., Lau, K., & Harker, F. R. (2003). In search of the ‘ideal’ pear (Pyrus spp.): Results of a multidisciplinary exploration. Journal of Food Science, 1108–1117. Jaeger, S. R., Rossiter, K. L., Wismer, W. V., & Harker, F. R. (2003). Consumer-driven product development in the kiwifruit industry. Food Quality and Preference, 14, 187–198. Kaiser, H. F. (1958). The varimax criterion for analytic rotation in factor analysis. Psychometrika, 23, 187–200. Kellerhals, M., Duerr, P., Wittmann, P., Goerre, M., Hoehn, E., & Daetwyler, D. (1999). Qualität neuer Apfelsorten: Sensorische und analytische Aspekte. Agrarforschung, 6(2), 53–56. Konopacka, D., Jesionkowska, K., Rutkowski, K., Płocharski, W., & Tomala, K. (2006). A comparison of the quality of scab resistant apples in expert and consumer evaluation. Vegetable Crop Research Institute, 65, 185–194. Lock, K., Pomerleau, J., Causer, L., Altmann, D. R., & McKee, M. (2005). The global burden of disease attributable to low consumption of fruit and vegetables: Implications for the global strategy on diet. Bulletin of the World Health Organization, 83(2), 100–108. , ISSN 0042-9686, http://dx.doi.org/10.1590/S0042-96862005000200010. Mante, W. (1973). Consumer preferences in apple purchase by 762 Berlin households. Erwerbobstbau, 157, 101–102.

J. Bonany et al. / Food Quality and Preference 32 (2014) 317–329 PanelCheck software. (2006). Nofima Mat, A ° s, Norway. . Popper, R. (1998). Consumer segmentation in sensory research. In Measurement of Consumer Attitudes and their Influence on Food Choice, Report series from EU project AIR-CAT, pp. 34–38. Robertson, A., Tirado, C., Lobstein, T., Jermin, M., Knai, C., Jensen, H. J., et al. (2004). Food and health in Europe: A new basis for action. WHO regional publications, European series (No. 96). Copenhagen: World Health Organization. . Schlich, P., & McEwan, J. A. (1992). Cartographie des Préférences. Un outil statistique pour l’industrie agro-alimentaire. Science des Aliments, 12, 339–355. Subar, A. F., Heimendinger, J., Patterson, B. H., Krebs-Smith, S. M., Pivonka, E., & Kessler, R. (1995). Fruit and vegetable intake in the United States: The baseline survey of the five a day for better health program. American Journal of Health Promotion, 9(5), 352–360. Tomala, K., Barylko-Pikielna, N., Jankowski, P., Jeziorek, K., & Wasiak-Zys, G. (2009). Acceptability of scab-resistant versus conventional apple cultivars by Polish adult and young consumers. Journal of the Science of Food and Agriculture, 89, 1035–1045.

329

van de Abeele, P., & Reijnders, C. (1980). Survey on consumption of apples in Belgium during a promotion campaign in schools. Fruit Belge, 48(392), 271–274. Villatoro, C., López, M. L., Echeverria, G., Graell, J., & Lara, I. (2009). Effect of controlled atmospheres and shelf life period on concentrations of volatile substances released by ‘Pink LadyÒ’ apples and on consumer acceptance. Journal of the Science of Food and Agriculture, 89, 1023–1034. Wakeling, I. N., & MacFie, H. J. H. (1995). Designing consumer trials balanced for first and higher orders of carry-over effect when only a subset of k samples from t may be tested. Food Quality and Preference, 6, 299–308. WHO. (2003). Diet, nutrition and the prevention of chronic diseases. Report of a Joint WHO/FAO Expert Consultation. WHO technical report series (Vol. 916). World Health Organization, Geneva. Yeh, M., Ickes, S. B., Lowenstein, L. M., Shuval, K., Ammerman, A. S., Farris, R., et al. (2008). Understanding barriers and facilitators of fruit and vegetable consumption among a diverse multi-ethnic population in the USA. Health Promotion International, 23(1), 42–51.