Optimization of olive-fruit paste production using a methodological proposal based on a sensory and objective color analysis

The food industry is a dynamic system that needs to innovate itself almost constantly in order to meet consumers’ demands. During the development and marketing of a new foodstuff it is important to consider many aspects related to its quality from different angles (psychological, technological, legal, etc.). Concerning legal matters, it is crucial to consider food safety as well its nutritional, sensory and commercial quality, among other aspects (Moskowitz et al., 2005). Nonetheless, there are also other factors like flavor, texture, and appearance which play a prominent role in the acceptability of foods and must be therefore assessed (Coleman, 1990; Lawless and Michelle, 1993; Pecore and Kellen, 2002; González et al, 2007). In this regard, it is indisputable that the consumer’s opinion is critical (Moldão-Martins et al., 2004; Muñoz et al., 1992;Yantis, 1992). Consequently, modern food companies pay great attention to the evaluation of the sensory attributes of foodstuffs by both trained panels and potential consumers, which is essential in the development, maintenance, optimization, quality improvement and assessment of new products (Coello et al., 1999; Abu-Salem and Abou-Arab, 2008). For example, appropriate preference tests have been performed to optimize a reduced-calorie tropical mixed-fruit jam (Abdullah and Cheng, 2001), the texture of spaghettis enriched with resistant starch (Sozer et al., 2007), a chocolate-flavored peanut-soy beverage (Deshpande et al., 2008) or the formulation of an apple pomace and a quince jelly (Royer et al., 2006), among others. Olive fruit is a typical product from the Mediterranean basin. Its praised lipid content is characterized by the predominance of unsaturated fatty acids (mainly oleic acid) over the saturated ones, hence it is considered quite healthy. In addition, it is a good source of other compounds GRASAS Y ACEITES, 60 (4), JULIO-SEPTIEMBRE, 396-404, 2009, ISSN: 0017-3495 DOI: 10.3989/gya010509


INTRODUCTION
The food industry is a dynamic system that needs to innovate itself almost constantly in order to meet consumers' demands.During the development and marketing of a new foodstuff it is important to consider many aspects related to its quality from different angles (psychological, technological, legal, etc.).Concerning legal matters, it is crucial to consider food safety as well its nutritional, sensory and commercial quality, among other aspects (Moskowitz et al., 2005).Nonetheless, there are also other factors like flavor, texture, and appearance which play a prominent role in the acceptability of foods and must be therefore assessed (Coleman, 1990;Lawless and Michelle, 1993;Pecore and Kellen, 2002;González et al, 2007).In this regard, it is indisputable that the consumer's opinion is critical (Moldão-Martins et al., 2004;Muñoz et al., 1992;Yantis, 1992).Consequently, modern food companies pay great attention to the evaluation of the sensory attributes of foodstuffs by both trained panels and potential consumers, which is essential in the development, maintenance, optimization, quality improvement and assessment of new products (Coello et al., 1999; Abu-Salem and Abou-Arab,  2008).For example, appropriate preference tests have been performed to optimize a reduced-calorie tropical mixed-fruit jam (Abdullah and Cheng, 2001), the texture of spaghettis enriched with resistant starch (Sozer et al., 2007), a chocolate-flavored peanut-soy beverage (Deshpande et al., 2008) or the formulation of an apple pomace and a quince jelly (Royer et al., 2006), among others.
Olive fruit is a typical product from the Mediterranean basin.Its praised lipid content is characterized by the predominance of unsaturated fatty acids (mainly oleic acid) over the saturated ones, hence it is considered quite healthy.In addition, it is a good source of other compounds This paper deals with the optimization of the conditions to formulate paste based on olive fruits.The processing stages included: washing, blending, oil addition and preservation.Pair-comparisons and ranking tests were carried out by both a trained panel and a consumer panel to single out the preferred sample at each stage.The sensory attributes considered were taste, visual texture, texture in the mouth and overall preference.The results of the sensory analyses were processed statistically by means of ANOVA and the Friedman test to select the most appropriate processing conditions: washing for 30 minutes three times, blending at 2000 r.p.m, and the addition of 10 mL of oil.Several pasteurization conditions were assayed (62.5, 75, 85 and 95 °C for 15 minutes).No significant color differences (p Ͻ 0.05) were found for the objective color measurement (CIELAB) of the pastes submitted to the different pasteurization conditions.Based on the results of the with nutritional importance like dietary fiber, vitamins, minerals and antioxidants.According to the IOOC ( 2004) "table olives are the sound fruit of varieties of the cultivated olive tree (Olea europaea L.) that are chosen for their production of olives, whose volume, shape, flesh-to-stone ratio, fine flesh taste, firmness and ease of detachment from the stone make them particularly suitable for processing; treated to remove its bitterness and preserved by natural fermentation; or by heat treatment, with or without the addition of preservatives; packed with or without covering liquid" (Sánchez Gómez et al., 2006).Due to the positive image of the olive fruits and their derived products, the demand for new products elaborated with them and olive oil has increased markedly over the past years.
According to the prevailing legislation, olive oil can be used as an ingredient in the elaboration of table olive products.In relation to their preservation, olive fruits fit for consumption can be subjected to different processes.One of them is pasteurization, which leads to the destruction of the vegetative forms of pathogenic and banal nature microorganisms.In the case of table olives, the microorganism of reference is the propionic bacteria (González, 1982).The effect of thermal treatments on the color and texture of pickled green olives has been assessed by Sánchez et al. (1991), who concluded that such organoleptic attributes did not change appreciably as a result of the pasteurization treatments necessary to guarantee the product's stability.This kind of analysis is important as it is indisputable that the color of foods, which depends on their composition and therefore can be affected by industrial practices, dramatically influences consumer preferences (Calvo et al., 2001;Hutchings, 1994).Furthermore, the color of a food is the first contact point of the consumers with it at the market and the first attribute considered in sensory analyses.
The objective of this study was to develop a new product fulfilling suitable levels of hygienic-sanitary and organoleptic quality at the request of a company that markets olive-based products.The experimental design was established after considering and evaluating statistically the results of the sensory analyses of different formulations.In summary, it can be claimed that this undertaking is a good example of collaboration between the industry and researchers in Food Science and Technology for the sensible a meaningful design of a new product.

MATERIALS AND METHODS
The raw materials used for the elaboration of pastes were of a standard quality to minimize season to season variations.Pitted table olives from the Manzanilla variety stored in highly salted brine were used due to its high pulp to stone relation and its acceptable oil content.The covering virgin olive oil was obtained from Hojiblanca olives.

Paste formulation
The flow diagram of the different stages leading to the formulation of the final product is shown in Figure 1.Two main stages can be distinguished: -The first one included the procedures followed to obtain the final quantitative formulation (washing, blending, oil addition and final formula).
-The second one corresponded to the different preservation conditions tested.
The codes of the samples are numbered and explained in Table 1.

Washing
In order to obtain olives with a suitable salt content it was necessary to establish washing conditions (time and number of water changes).Four paste samples ([T00], [T30], [T60], [T90]) were elaborated (Table 1).

Texture
The pastes were prepared with washed pitted olives.The texture depended on the granularity of the olives.In some cases, virgin olive oil (always the same amount) was added during the blending.The blending was performed with either a 600 watt MMB2000 domestic blender (Bosch, Gerlingen-Stuttgart, Germany) or an Abencor laboratory oil mill (Abengoa, Seville, Spain).Three different granularities were obtained: fine-grained [F], medium-grained [M] and large-grained or coarse [L] (Table 1).

Added oil/Covering oil
The pastes with added oil were produced adding 50 g of virgin olive oil/kg of olive fruits.The samples with covering oil contained 15 mL of virgin olive oil /250 g of paste.

Preservation conditions
The four pasteurization methods tested (Table 1) were performed using a temperature controlled water bath (Selecta, Barcelona, Spain).For this purpose 100 g glass containers filled with paste and covered with olive oil were used.After the thermal treatments the containers were kept in a fridge at 4-8 °C.

Experimental design
Three different variables were taken into account for the formulation of the paste: salty taste, texture and the addition or not of oil.Each variable comprised 3-4 levels.Once the most appropriate formulation was singled out, four different pasteurization temperatures were assayed, as well  as their impact on the sensory quality of the final product.In order to facilitate the selection process, an approach based on a phase by phase choice of attributes by sensory analysis was followed.Paired comparison tests (ISO 5495, 1983) and preference ranking tests (ISO 8587, 2001) were conducted at each stage in order to decide the formula that passed to the next optimization phase.The final experimental design is summarized in Table 1.

Sensory analyses
Either a trained or a consumer panel was used depending on the purpose of the analysis.A comprehensively trained 10-judge panel took part in the tests in the initial formulation stages.The members of this panel were chosen from trained academics with long expertise in the sensory analysis of food for research and quality control purposes.In the final stage, the sensory analyses were performed by a consumer panel of 40 people randomly recruited at the university campus.This gender-balanced panel consisted of students and university staff lacking training in sensory analysis and with ages ranging between 25 and 55.The judges were asked to rank the samples according to their overall preference considering the following attributes: appearance (mainly related to color), texture and both taste and aroma quality and intensity.
The assessments were carried out in individual booths (70 ϫ 70 ϫ 55 cm) at room temperature and under white light.Twenty-gram samples of paste at 25 °C were presented in white plastic plates labeled simultaneously to the judges, who were asked to taste and rank them according to their degree of preference once the different attributes tested had been considered.This test was applied in the different stages of the process (Figure 1).

Statistical analysis
The software Statistica® v 6.0 (StatSoft, 2001) was used for the statistical treatment of the data.Analysis of variance (ANOVA) were carried out to ascertain whether there were significant differences (p Ͻ 0.05) between samples (Norman and Streiner, 1996).Additionally, significant post hoc analyses were performed (Tukey HSD test if variances in the different groups were identical or Games-Howell test if they were not).The homogeneity of variance was tested by Levene's Test.
Friedman's two-way analysis of variance was carried out when the variables were measured in terms of categories.As recommended by the ISO 8587, this test (Friedman, 1937) was considered to establish the statistical significance of the differences that the panelists might detect between samples.The differences among the Friedman's rank sums were compared to critical values calculated according to Hollander et al. (1973).When according to the Friedman's test statistically significant differences among samples were found, the pairs of differing samples were identified using an analogue of Fisher's least significant difference (Fisher, 1998).

Objective color measurement
The reflectance color measurements were made using a CAS 140 B spectroradiometer (Instrument Systems, Munich, Germany) fitted with a Top 100 telescope optical probe (Instrument Systems, Munich, Germany) and a Tamron zoom model SP with three-digit random codes.A glass of water was offered to the judges between samples to rinse their mouths.To prevent biases related to the serving order, this was determined by random permutation.Furthermore the codes were different for each test.

Paired comparison test
Paired comparison tests were used to determine preferences regarding texture and oil addition.They judge were asked to answer the following questions: 1. "Is there any difference between the samples?" 2. "What sample is more [attribute]?"3. "What sample do you prefer in terms of [attribute]?"When differences were noticed, one point was assigned to the favorite sample and 0 points to the other.When no differences between samples were detected, 0 points were assigned to both samples.All the points were summed up and divided by the number of judges to obtain the final score of each sample.When three samples had to be tested, three comparison tests were carried out and the maximum score of the sample was 2. To test 4 samples, six comparison tests were carried out and the maximum score of the sample was 3.
Concerning texture, the attributes considered were visual appearance and sensation in the mouth.In the case of the addition of oil the attributes considered were visual texture, sensation in the mouth and taste.

Rank-order test
This test allowed the differences among several samples to be estimated in terms of overall liking.The samples, in random order, were presented minimal significant differences of Fisher, indicated that [T00] was significantly different from [T60] and [T90] (p ϭ 0.01) and that [T30] and [T90] also differed significantly (p ϭ 0.05).In light of these results it was concluded that the favorite sample was the one soaked in water for 90 minutes [T90] with water changes every 30 minutes.This washing procedure was followed for the production of all the pastes hereafter.

Texture paste selection
In a second stage, it was necessary to determine the appropriate texture for the production of the paste.For this purpose two attributes were taken into consideration separately: texture in the mouth and visual texture.After performing a paired comparison test, 98.3% of the judges found differences among the pastes assessed.These differences proved statistically significant (p Ͻ 0.0001) on the basis of the results of the corresponding ANOVA test.
To identify the pairs of samples that were significantly different a post hoc comparisons Tukey HSD test was applied.For this purpose it was assumed that the variances in the different groups were identical (Levene test p Ͼ 0.05 In both groups (pastes with and without oil) significant differences (p Ͻ 0.01) in granularity were found for all the pairs considered.
As can be observed in Figure 2a, when the samples prepared without added oil were considered, the medium-grained texture got the best scores for mouth texture.Concerning visual texture, both the fine and medium-grained pastes were scored alike.In contrast, considering the samples prepared with added oil, the fine-grained sample was the best scored when both the visual and the mouth texture were considered (Figure 2b).The ANOVA test revealed that only the pastes produced without oil differed significantly for mouth texture (p ϭ 0.0083) and visual texture (p ϭ 0.0000).Contrarily no significant differences were found in the samples with added oil.
To identify the pairs of samples that were significantly different in terms of mouth texture, a post hoc comparisons Tukey HSD test was applied.In this regard, it was assumed that the variances in the different groups within the design were identical, because the Levene test for [F]/[M]/[L] resulted in p ϭ 0.790 for mouth texture.In light of the outcome of the test, it was concluded that only the mediumgrained paste differed significantly (p ϭ 0.0294) from the coarse paste.
When the studied attribute was the visual texture, the Games-Howell test was applied to compute the post hoc comparisons (variances were not identical, since the Levene test for [F]/[M]/[L] was p Ͻ 0.05).The large-grained paste differed significantly (p ϭ 0.018) from the fine-grained and medium-grained pastes.
The color parameters corresponding to the uniform color space CIELAB (L*, a*, b*, C* ab and h ab ) (CIE, 1986) were obtained directly from the apparatus.L* is an approximate measure of luminosity and its values oscillate between 0 (black) and 100 (white).The color coordinate a* is higher than zero for reddish colors and lower than zero for greenish ones.On the other hand, b* takes positive values for yellowish colors and negative values for bluish ones.Additionally, two psychological parameters, namely hue (hab) and chroma (C ab *) are defined, which are related to a* and b* as follows: C* ab is regarded as the quantitative attribute of colorfulness and reflects the degree of difference relative to a grey color with the same lightness for each hue.Hue angle takes values from 0°to 360°a nd is the qualitative attribute that allows any color to be described as bluish, reddish, etc.This parameter allows any color to be differentiated from a grey one with the same lightness.The color differences (∆E* ab ) between two colors in the CIELAB space are calculated as the Euclidean distance between their locations in the threedimensional space defined by L*, a* and b*:

Paste formulation
The Manzanilla table olives provided by the industry were preserved in highly salted brine.In the industry, the characteristic salty taste of the final product is obtained by washing the olives until the desired salt content is obtained.Consequently, our starting point to formulate the paste was to establish the appropriate washing procedure leading to a salt content that was appealing for the consumer.In this regard, up to three different washing procedures were assayed.Thus, pastes were produced with the initial raw olives [T00] and with the three different washing procedures, which were used for a rank order test.The sum scores (R) from all the judges were 10, 20, 30 and 40 for the samples [T00], [T30], [T60] y [T90] respectively.After analyzing the results with the Friedman's test, it was concluded that there were statistically significant differences among them (p Ͻ 0.01) (F ϭ 30 Ͼ F critical ϭ 11.34, ␣ ϭ 0.01) (Table 2).The pair comparison tests, computed according to the OPTIMIZATION OF OLIVE-FRUIT PASTE PRODUCTION USING A METHODOLOGICAL PROPOSAL BASED ON A SENSORY AND… not lead to marked changes in the color of the pastes.These results are in agreement with those reported by Sánchez et al. (1991) for pickled green olives.
At any rate, considering the goal of the study, it was far more important to ascertain whether the pasteurization conditions tested led to color changes visually discernible by consumers.The rank sum scores (R) obtained for the thermally treated pastes were 14, 25, 27 and 24, for the samples [P6], [P7], [P8] and [P9], respectively.It was clearly seen that the highest scored sample was the one pasteurizated at 85 °C and that the sample heated at 62.5 °C was by far the least appealing.According to the Friedman's test the samples were not significantly different (F ϭ 6.73 Ͻ F critical ϭ 7.8).However, when pairs of samples were considered significant (p Ͻ 0.05) differences were found between [P6] and [P7] and [P8] (Table 2).In view of these results, 85 °C was selected as the most appropriate pasteurization temperature.

CONCLUSIONS
The main objective of this study was to develop an optimized and experimental protocol based on sensory evaluation to select a final formulation of an olive-fruit based product.The selected paste formulation was characterized by a fine grain texture without added olive oil in the whipping stage  but covered with olive oil.According to the sensory panels this paste was preferred due its moist but not greasy texture and its delicate and smooth taste.
Figure 1 Flow diagram of the stages leading to the development of the new product.

Figure 5 Figure 6
Figure 5 Summary of the judges' answers in the rank-order tests for the selection of the oil content.a. Preferences for the medium-grained samples.b.Preferences for the fine-grained samples MARÍA LUISA ESCUDERO-GILETE, A.J. MELÉNDEZ-MARTÍNEZ, F.J. HEREDIA AND ISABEL M. VICARIO