Sensory profiling and consumer acceptability of new dark cocoa bars containing Tuscan autochthonous food products

Abstract A new set of cocoa bars named Toscolata® were developed containing top‐quality extra virgin olive oil, dried apples cultivars, and chestnut flour. The present work has been conducted to define the sensory profile of these products through tasting by trained experts and consumers to study the acceptability, preference, and quality perception. The four sensorial profiles of the bars differed in the level of persistence, bitterness, aromaticity, acidity, astringency, and tastiness. In particular, the sour attribute could be traced to the presence of dried apple. Bars containing apple and chestnut flour obtained higher acceptance ratings, compared to those with extra virgin olive oil. The bar with chestnut flour was preferred by consumers who considered it to be sweeter due to the presence of natural sugars, which lowered the bitter sensation of cocoa. These results showed that the selection of the preferred bar by consumers was mainly based on the level of bitterness and, in particular, elderly consumers expressed a strong preference for the sweetest product. As far as we know, this is the first study comparing the results of a panel of expert tasters with that of consumers in the tasting of dark chocolate.


| INTRODUCTION
In recent years, consumers have become increasingly aware of foods of high in nutritional value with organoleptic qualities related to traditional production.
Tuscany is the regional base of several typical agricultural products; many of them are labeled under Protected Designations of Origin (PDO) or Protected Geographical Indications (PGI). Other traditional products are not so easy to be marketed because they do not meet market standards. For example, some fruits do not reach the desiderate size, color, and firmness or may have a very short season. Thus, it would be useful to find new strategies for their commercialization selecting new formulations, novel methods of transformation by which increase the economic value and the income of the producers.
One confectionary products preferred by consumers is cocoa and its derivatives. They are consumed for pleasure, acting as a stimulant, relaxant, or potentially an antidepressant (Parker, Parker, & Brotchie, 2006). Cocoa is appreciated not only for hedonistic properties (Beckett, 2000), but also for health benefits due to its high content of antioxidants. Many recent studies showed a correlation between consumption of dark chocolate and the reduction in cardiovascular risks with positive action against hypertension, free radicals, and lowdensity lipoproteins oxidation (Ariefdjohan & Savaiano, 2005;Arranz et al., 2013;Ding, Hutfless, Ding, & Girotra, 2006;Ellam & Williamson, 2013;Engler & Engler, 2006;Ried, Sullivan, Fakler, Frank, & Stocks, 2010).
Cocoa can be combined with fruits, vegetable oil, and many other ingredients. These combinations affect the texture and consequently consumer acceptability (Beckett, 1994) which is also influenced by variations in the proportions of ingredients or processing (Jackson, 1999). As reported by Rozin and Fallon (1987) "food acceptance are motivated by a combination of sensory-affective reasons ideational notions and safety concerns". Consumer opinion represents an effective quality-level assessment; however, for profiling a new product it is necessary to have the judgment of a trained panel which assures accuracy, sensibility, and repeatability.
In the 2013, the Tuscany region financed applicative research aimed at improving the agricultural economy and increasing the revenue of typical local foods. In collaboration with a private company, we developed a new set of cocoa bars containing top-quality extra virgin olive oil, dried apples produced by old autochthonous cultivars and chestnut flour. The prototypes of these new cocoa products named Toscolata ® have been subjected to several studies with different approaches. The present work has been performed to define the sensory profile made by trained experts of these new products and to study the consumer acceptability, preference, and quality perception.

| Samples
Seven cocoa bars prototypes were used for this study: five containing dried apples of different cultivars. Three of them "Mora," "Nesta," and "Ruggine" are traditional varieties from Tuscany while "Stayman" and "Golden Delicious" are internationally cultivated and were used as a comparison. Another bar was made using top-quality extra virgin olive oil, and the last bar by adding PGI Monte Amiata chestnut flour.
Each ingredient was obtained from local organic producers and selected for their high antioxidants levels (oil and apples) or organoleptic features (chestnut flour).
The samples were manufactured in the Vestri chocolate laboratory located in Arezzo (http://www.vestri.it) using the best mixture of cocoa beans produced directly by Vestri in their organic farm in Santo Domingo. All the production conditions (mixing, refining, conching, tempering, molding, and cooling) were set to maintain the highest organoleptic features of the original cocoa beans and ingredients. The composition of the bars is given in Table 1, where the abbreviations for each bar used in the text are also provided. Bars were packed in 40 g size and stored at 4°C.

| Experimental design
Several rapid techniques for the determination of consumer preference and food quality perception have been recently introduced (Varela & Ares, 2012) to speed up the process of sensory characterization and product profiling. Since our products had a similar composition (a base of 70% cocoa for all), we preferred to use a timeconsuming classic three-step approach that included (1) a tasting and discussion session with sensory professionals following the method

| Pretest
The aim of this stage was the identification of the descriptors to be used for the evaluation of each aspect of the products. A list of descriptors taken from the literature (Lanza et al., 2011;Sune, Lacroix, & Huon De Kermadec, 2002;Thamke, Durrschmid, & Rohm, 2009) was provided to the members of the round table.
Four preparatory sessions were conducted with selected standards and the assessors were asked to evaluate individually if the terms were suitable or whether it was necessary to introduce new terms.

| Panel test
Before the test, an introduction was delivered to the panelists to explain the objective of the research, the precise meaning of each cocoa attributes and to give examples of the grade of intensity. Several cocoa products were used as standards (dried cocoa beans; 100% dark chocolate; milk chocolate). All of the samples to be assessed were taken out of the refrigerator 24 hr before each session. Each cocoa bar was cut into 6.5 g servings, placed on to plastic plates codified with three-digit random numbers and served at room temperature (22°C) to the panelists without specifying the formulation. As recommended, water was used for cleansing the palate between samples.
The sensory evaluation was conducted in two replicate sessions by a panel of the Grosseto Chamber of Commerce composed of 10 experienced judges. In the first session, the products containing "Mora," "Nesta," "Ruggine," "Stayman," and "Golden Delicious" (DAM, DAN, DAR, DAS, and DAG, respectively) were evaluated to understand if there was any difference between local and international dried apples added to the cocoa. In the second session two apple bars were evaluated together with extra virgin olive oil (EVO) and chestnut flour (CHF). Attributes were expressed on a 9-cm line scale and quantified by measuring the distance of the mark from the origin (Dever, Macdonald, Cliff, & Lane, 1996). The consumers, both male and female, were selected by two main criteria: 18-80 years old and are frequent consumers of cocoa products.

| Consumer test
A total of 182 people took part in the study on voluntary basis. They were first asked to complete a questionnaire regarding their general sociodemographic information (see Table 5). Each sample consisted of 3.2 g of the cocoa bar, placed in transparent plastic bowl at room temperature, codified with a three-digit code and served in a random order.
Consumers were asked to taste cocoa bars and to express an acceptance score on a 1-9 hedonic scale from 1: "I completely dislike" to 9: "I like very much". Afterwards, the consumers were asked to focus on the sample they preferred the most and to select which attributes contributed to their choice. They could choose one or more sensorial descriptors among those considered to be discriminant after statistical analysis of the data provided by the professional panel and suggested during the interview.

| Statistical analysis
Geometric means were calculated using the level of intensity and frequencies of each descriptor during the pretest for selecting the at-

| Pretest Results
A total of 38 potential sensory terms were used for the preliminary description of cocoa prototypes (Table 2). During the round table sessions among the assessors another three attributes were introduced: two of them, "tastiness" and "aromaticity", were chosen to define the overall intensity of the respective taste and aroma without any precise qualitative definition. The third term "vegetal" was defined as the smell of green fruit. After tasting and the following discussion on cocoa bars, 13 terms listed in Table 2 were eliminated because they correlated with specific ingredients not used in our recipes (butter, cinnamon, vanilla, nutty, coffee, rice, alcoholic, spicy), were not suitable for the type of product (snappy, creamy, oily) or simply hedonistic (harmonic). The term "chocolate" was discarded because it was statistically correlated with the cocoa aroma and was not legally appropriate for defining EVO bar (European Union, 2000). Another nine terms were discarded because of their low frequency of use after calculation of the geometric mean as indicated in ISO 11035:1994 (data not showed). The final list of 16 descriptors selected to be introduced in the product evaluation sheet is shown in Table 3.

| Panel test results
Data from the two separated sessions, statistically elaborated by PCA, underlined the most discriminant attributes of the sensory profile of cocoa bars only containing dried apples. The total variance explained by the first two principal components was equal to 71% and the five attributes with the highest loadings were "sapidity" and "fruity" on the first component and "graininess," "astringency," and "persistence" on the second component (Figure 1).  Table 4. The "bitter" and "acidity" attributes contributed most in explaining the variation in the first component followed by "aromaticity" and "astringency" while in the second component "tastiness" and "persistence" were the most discriminant variables. The scatterplot of the scores of each sample on the first two PCs is reported in Figure 3. The four cocoa bars were well T A B L E 2 Complete vocabulary taken into consideration for the sensory description of Toscolata ® cocoa products and preparation of the evaluation Sheet

| Consumer test results
The 182 consumers who participated in the test were well distributed within the classes of gender and age (Table 5), and regarding their preference, 51% liked unsweetened cocoa products. Only 15% of the consumers associated the purchasing of cocoa to health benefits.
None of those interviewed disliked all the bars and only 16 (8.7% of the total) expressed to dislike (very or extremely) of one of the bars.
The number of "dislike" opinion were, respectively, 8 for EVO, 5 for the apple bars, and 3 for CHF with no statistical differences among the number of answers. The preference of the consumers ( Figure 5) was equally distributed between the bars containing chestnut flour (37%) and those containing dried apple (34%). This percentage takes into account both DAM and DAN as "apple" and the preference of the two was similar (43 vs. 57%). While 16% of the consumers did not express preference for any of the four (they liked all of them equally) only 13% of the total interviewed showed preference for the bar made with extra virgin olive oil.
We also tried to cluster the 182 consumers using the same employed in the PCA presented in Figure 1, and found that the three most discriminant variables for clustering were age, motivation, quantity of consumption. Only two clusters of consumers were obtained: one with both young and middle-aged consumers that buy high quantities of chocolate for pleasure (1-2/3-6 bars/month) and a second including all the elders buying low quantities (1-2 bars/month) that were conscious of their health (Table 6).
With regard to the distribution of preferences among groups of consumers, the only statistical difference was found to be "male over 61 years old" who preferred the cocoa bar with chestnut flour (Pearson's chi-square p = 0.030).
We performed a calculation based on the consumers' answers about the motivation for their preference, and the average score of the most highly used attribute for the four cocoa bars are reported in Table 7.

| DISCUSSION
As shown by Lanza et al. (2011), the selected specific descriptors for the products were able to profile the organoleptic features of the samples and to also discriminate among them. In contrast to "Modica" chocolate specialty, the appearance was not important in our case and only EVO was recognized as more "bright" than the other bars F I G U R E 3 Plot of cocoa bar scores on the first two principal components. The ellipses represent the 95% confidence limits of each cocoa prototypes centroid T A B L E 4 Loadings of each attribute on the first two principal components calculated using the data produced by the panel test on the four cocoa bars While similar research has been conducted for the assessment of apple quality by Gatti, Di Virgilio, Magli, and Predieri (2011)  Healthy benefits/food quality 13 Healthy benefits/food price 2 F I G U R E 5 Distribution of the consumer preference among the cocoa bars with different ingredients of 3-and 2-MCPD and glycidyl esters were highest in palm oil/fat, and that the level in some foods could cause health problem for young people. The EVO bar did not present any processing or conservation problem and has a peculiar organoleptic profile, different from other commercial products. Its action on human health is now under evaluation by our group. Since only 15% of the consumers declared that they buy cocoa for its effects on health, there is great potential for information strategies. The communication of scientific knowledge about these high-quality products could be more greatly explored and exploited to increase market penetration and price.

| CONCLUSION
The different ingredients used in the recipes directly influenced the acceptability of these novel bars to consumers. DAM, DAN, and CHF obtained higher acceptance ratings compared to EVO. This study demonstrated that novel food made with ingredients such as Tuscan autochthonous dried apples and PGI Monte Amiata chestnut flour had a high acceptance and preference by consumers. Pearson Chi-square preference content/sex p = .153. Pearson Chi-square age p = .418. Pearson Chi-square preference content/age p = .091. Pearson Chi-square preference content/age/sex p = .030 only for chestnut/male/61+.