Blackwell Publishing Ltd.Oxford, UK and Malden, USAJFQJournal of Food Quality0146-94282004 Blackwell Publishing2004270264273Original Articles CUMIN FLAVOR, PROFILE, PCA AND MAPPINGA. DATTATREYA, M. PRAKASH and K.K. BHAT SENSORY FLAVOR PROFILING AND MAPPING OF MARKET SAMPLES OF CUMIN (CUMINUM CYMINUM L.) ANUPAMA DATTATREYA, MAYA PRAKASH 1 and K.K. BHAT Department of Sensory Science Central Food Technological Research Institute Mysore-570 013, India Accepted for Publication August 8, 2003 ABSTRACT Eight market samples of cumin (R1, R2, R3, R4, R5, R6, R7 and R8) from different regions of India were examined for sensory quality by conducting threshold tests, time-intensity (TI) profiling and flavor profiling. Principal component analysis (PCA) was carried out to group the samples. Threshold values ranged from 0.006 to 0.017% with R7 and R8 lots showing lower values for their thresholds (0.006%). Higher intensity of aroma of R7 and R8 was further confirmed by more of a lingering aroma as shown by the TI study. Flavor profiling by quantitative descriptive analysis showed that the market samples of cumin did not differ significantly (P 0.05). Mapping of samples using PCA technique showed, based on intensity of attributes, four distinct groups comprising a) R1 and R3, b) R7, c) R2 and R5 and d) R4 and R8. R6 occupied a position in between a and b. INTRODUCTION Spices are the major flavoring agents used in Indian prepared foods. Consumers associate the flavor of many food items with the aroma characteristics of spices used in the preparation. Types of spices and the method of using them depend on consumer preferences, which show a wide variation across the world. Contribution of flavor by a spice is the combined effect of variety, region of its cultivation, its processing conditions, level of addition, storage history and possible synergistic effects with other ingredients of the food. Cumin is well recognized for its aromatic and spicy 1 Contact author: Dr. (Mrs.) Maya Prakash, Central Food Technological Research Institute, Mysore-570 013, Karnataka, India. TEL: +91-821-2515842; FAX: +91-821-2517233, EMAIL: mayaprksh@yahoo.co.in. Journal of Food Quality 27 (2004) 264 273. All Rights Reserved. Copyright 2004, Blackwell Publishing 264
CUMIN FLAVOR, PROFILE, PCA AND MAPPING 265 nature. Differences in its odor quality and sensory responses to cumin subjected to different processing methods as well as from different regions have been reported by many researchers. Granville et al. (1971) studied the chemical and sensory properties of cumin and have reported that flavor and odor of heated cumin differed from that of fresh seeds. Bera et al. (2001) have optimized the grinding, packaging and storage of ground cumin with respect to sensory properties and shelf-life. Cliff (1994) has characterized pungency differences by undertaking four experiments to evaluate oral pungency and quantify the TI response of oral irritant flavor compounds in cumin (cuminaldehyde). Bandoni et al. (1991) studied six samples of cumin seeds produced in Argentina for their physico-chemical and sensory characteristics. Subbulakshmi et al. (1991) studied the sensory attributes and some quality indices of irradiated cumin and they concluded that there was no significant difference in sensory qualities of irradiated and nonirradiated samples. Though work on analytical and sensory aspects of cumin has been studied, no information is available on psychometric studies and sensory flavor profiling of Indian varieties coupled with multivariate analysis of the sensory data. Therefore, an attempt was made to study the psychometric aspects such as threshold and TI and also sensory flavor profiling of market samples of cumin procured from different regions of India. The results are reported here. MATERIALS AND METHODS Cumin samples were procured from the local markets of northern, western and southern parts of India [Two of northern (R7 and R8), one of western (R2) and five from the southern belt (R1, R3, R4, R5 and R6)]. The samples were packed in airtight containers and stored in a cold room (4 6C) during the course of study. Panel Training A group of 10 panelists was trained over three sessions for psychometric (threshold and TI studies) and Quantitative Descriptive Analysis (QDA) tests. The members of the panel were drawn from the scientific staff familiar with sensory analysis techniques and who had earlier experience in the sensory evaluation of spices. Samples were evaluated in booth room maintained at a temperature of 22 ± 2C under fluorescent lighting equivalent to daylight.
266 A. DATTATREYA, M. PRAKASH and K.K. BHAT Psychometric Studies Preparation of Stock Solution for Threshold and Time Intensity Tests. Cumin samples were ground to a particle size passing through 60-mesh sieve. Ground cumin powder (2 g) was refluxed with 100 ml of distilled water for 15 min, cooled to room temperature and then centrifuged for 10 min at 5000 rpm. The supernatant was taken in a 100 ml volumetric flask and the volume was made up to 100 ml using distilled water. The extracts obtained served as the stock solution for the psychometric studies and it contained 2% cumin. Threshold Test. Threshold tests were conducted as per the method given in IS: 5126 (1969) and ASTM Manual series (1996). Preliminary trials had shown that 0.3% sodium choride solution was a suitable medium for threshold determination of spice aroma and therefore different concentrations of cumin extract were prepared in 0.3% sodium chloride solution. From the 2% stock solution of cumin extract, a series of dilutions were made representing increasing concentration of the cumin aroma. Both geometric (0.0 to 0.16%) and arithmetic series (0.0 to 0.02%) of solutions were evaluated. The panelists were asked to taste the series arranged in increasing order of concentrations and mark zero if no stimulus was perceived,? if the stimulus was perceived to be different from the blank but not recognizable and X for threshold i.e. the lowest concentration at which the panel could perceive and recognize the cumin aroma. Sodium chloride solution (0.3%) served as the blank. The threshold for cumin aroma is reported here as the arithmetic mean of the X values. Time Intensity Study. A TI test was carried out to identify the samples having higher scores for lingering cumin aroma which serves as an indication of the quality of the spice with respect to the volatile compounds. The test was performed as per the guidelines given by ASTM manual series (1996) with slight modifications in the line scale. In the present study, panelists were trained to mark the scores on a structured quantitative descriptive analysis scale of 15 cm with anchoring at low (1.25 cm) and high (13.75 cm) thresholds of 0.02% and 2.0% cumin extracts, respectively. The panelists were asked to mark the perceived intensity of cumin aroma on the scale at regular intervals of 10 s, starting from the onset of aroma perception up to a total period of 60 s. The panelists evaluated 1% cumin extract. Each panelist was given 10 ml of the sample in a 25 ml beaker and asked to mark the intensity of the perceived sensation on the scorecard as soon as he/she took the sample into his/her mouth (without swallowing). This marked the onset of perception of cumin aroma. Then the panelists were asked to hold the sample in their
CUMIN FLAVOR, PROFILE, PCA AND MAPPING 267 mouths for 10 s and then swallow it and mark the intensity. After swallowing the panelist continues to mark the intensity every 10 s up to a total time of 60 s. TI profiles were obtained by plotting mean scores of cumin aroma versus time in seconds. Flavor Profiling of Cumin Samples Sample Preparation. Samples were prepared using cornstarch gruel as a medium. The gruel was prepared by dispersing 4 g of starch in 100 ml water to which 0.5% salt was added and heated until a homogeneous gelatinized liquid was obtained. To this gruel, 0.5% cumin powder (60 mesh size) was added and stirred thoroughly to get a uniform dispersion. The QDA method (Stone et al. 1974) was used for flavor profiling of cumin. As a prerequisite for flavor profiling of cumin, panelists were asked to give as many descriptors as applicable. Following this, open discussion was held to reach an agreement on appropriate descriptors, especially for aroma and flavor. For this, the guidelines described by Jowitt (1974), Dravnieks (1985) and Jellinek (1964) were used. The common descriptors selected by at least one third of the panel and a few important descriptors cited in the literature were utilized in the development of the scorecard. The trained panelists, who were involved in the earlier two tests, participated in this evaluation. The descriptors used for flavor profiling of cumin and the reference compounds used are given in Table 1. A QDA which consisted of 15 cm line scale wherein 1.25 cm was anchored as low and 13.75 cm as high was used for quantification of sensory the attributes. The panelists were asked to mark a vertical line on the scale and write the code of the sample close to the line. TABLE 1. DESCRIPTORS USED FOR FLAVOR PROFILING OF CUMIN S. No. Descriptor Reference Compound 1. Pleasant Cinnamyl alcohol, Myrcene 2. Citrusy Citral, D-limonene 3. Oily 2-4,Trans-trans-decadienal 4. Herbal Isopropyl quinoline 5. Cooling Menthol, Spearmint 6. Earthy Isopropyl quinoline 7. Starchy Rice starch gruel, corn starch gruel 8. Spicy Cumin oil, cumin powder 9. Salty Sodium chloride 10. Astringent Raw banana, tannins, grape juice
268 A. DATTATREYA, M. PRAKASH and K.K. BHAT Samples (15 ml) were warmed to 40C and were served one by one in a random order in 25 ml beakers bearing 3 digit codes. The profilogram was obtained by plotting mean scores of attributes for the samples versus the attributes. Statistical Analysis Principal component analysis (PCA) was carried out for the data obtained by QDA, using statistical software STATISTICA 99 from Stat Soft, USA. Threshold Test RESULTS AND DISCUSSION The threshold values for cumin aroma were reported as the arithmetic mean of X values (Table 2). It may be seen from the table that the least threshold, 0.006%, was found for R7 and R8 and highest threshold, 0.017%, for R5 indicating R7 and R8, which belonged to the northern region, had higher intensity of cumin aroma while R5 had the least. Univariate analysis of variance was carried out to compare the threshold values for the eight samples. The results showed that among the market samples of cumin significant differences existed (P 0.001) in aroma thresholds indicating that a threshold test may be used to detect subtle differences in cumin aroma. Time Intensity Study The TI profiles for cumin flavor at 1.0% concentration is given in Fig. 1. The lingering aroma of R7 and R8 was more than that of the other samples. TABLE 2. THRESHOLD VALUES OF CUMIN SAMPLES Sample Threshold Value Range Standard Error R1 0.0069 0.006 0.010 0.00048 R2 0.0111 0.010 0.012 0.00035 R3 0.0127 0.010 0.016 0.00065 R4 0.0093 0.008 0.010 0.00041 R5 0.0174 0.016 0.020 0.00043 R6 0.0120 0.010 0.014 0.00049 R7 0.0060 0.004 0.008 0.00054 R8 0.0060 0.004 0.008 0.00065
CUMIN FLAVOR, PROFILE, PCA AND MAPPING 269 FIG. 1. TIME INTENSITY PROFILE FOR CUMIN AROMA 1.0% In other words, they had higher intensities of cumin aroma, which correlated well with the results of the threshold tests. R7 had an intensity score of 10.7 at 0 s which declined to 1.19 after 60 s. R8 had an intensity score of 1.24 after 60 s. However, the results of the TI study showed that the market samples of cumin did not differ significantly from each other (P 0.05). The standard errors for the TI curves of the eight samples were for R1 0.43; R2 and R3, 0.49; R4, 0.37; R5 and R6, 0.48; R7, 0.31 and R8, 0.39. Flavor Profiling The flavor profiles of cumin samples (Fig. 2) showed that the samples did not differ significantly from each other in all the attributes. Mapping of market samples of cumin was carried out using a PCA technique. When samples were grouped based on the intensity of attributes (Fig. 3) the first and second principal component, PC1 and PC2, accounted
270 A. DATTATREYA, M. PRAKASH and K.K. BHAT FIG. 2. FLAVOR PROFILE OF CUMIN SAMPLES for 62.38% and 16.78% of the variance, respectively, making a total of 79.16% together. The factor pattern is given in Table 3. The first axis was loaded positively with the attributes pleasant, citrusy, oily, herbal, cooling, earthy, starchy and salty and negatively with the attributes spicy and astringent. The second axis was loaded positively with the attributes pleasant, citrusy, oily, herbal, cooling, earthy, starchy, spicy and astringent and negatively with the attribute salty. The major attributes contributing to the grouping of samples were citrusy, oily, herbal, starchy, salty, earthy, astringent and spicy. Though salty and starchy are not directly related with cumin aroma, they influenced the intensity of cumin aroma; i.e. if the sample had lower intensity of cumin aroma, starchiness was perceived to a higher extent, similarly if the sample was more astringent, saltiness was perceived to a lower extent. The samples
CUMIN FLAVOR, PROFILE, PCA AND MAPPING 271 TABLE 3. FACTOR LOADINGS FOR PRINCIPAL COMPONENTS Attributes Factors 1 2 Pleasant 0.78 0.25 Citrusy 0.93 0.17 Oily 0.85 0.19 Herbal 0.94 0.18 Cooling 0.66 0.02 Earthy 0.25 0.85 Starchy 0.95 0.10 Spicy -0.53 0.72 Salty 0.87-0.38 Astringent -0.85 0.33 Eigenvalue 6.24 1.68 * Underlined vectors load the factors maximally. FIG. 3. PCA OF CUMIN SAMPLES
272 A. DATTATREYA, M. PRAKASH and K.K. BHAT occupied different quadrants based on the intensity of attributes. R7 occupied the first quadrant and was found to have higher intensity of pleasantness, cooling, citrusy, oily, earthy and herbal notes. R1 and R3 occupied the second quadrant and had higher intensity of spicy and astringent notes. R2 and R5 were grouped in the third quadrant and R4 and R8 occupied the fourth quadrant. R6 occupied a position in between the first and the second quadrants. CONCLUSIONS Market samples of cumin procured from three different regions of India (R1 to R8) evaluated for aroma by threshold and TI tests showed variations in threshold and TI values with samples from northern India being more intense in cumin aroma. Univariate analysis of variance showed that among the cumin samples significant differences existed (P 0.001) in aroma threshold suggesting that a threshold test may be used to detect subtle differences in cumin aroma. The TI study and flavor profiling showed that the cumin samples did not differ significantly from each other (P 0.05) though there were some variations in the intensity of attributes. Mapping of market samples of cumin, using PCA indicated the grouping of samples with a northern India specimen (R7) having higher intensity of pleasantness, cooling, citrusy, oily, earthy and herbal notes. Though one or more samples were selected from the Northern, Southern and Western parts of India, the results have been generalized for the different regions because the samples procured were market samples and there is a tendency to pool up the samples in the different regions before they are sent to the local market. Hence the samples from each region may be expected to be homogenous and representative of that region. The threshold test which employs lower concentrations of cumin extract could identify the subtle differences in cumin aroma while the time intensity and flavor profiling methods could not, probably due to the use of the higher concentration of the sample which might have masked subtle differences. Also it is possible that the different combinations of the specific qualities create synergistic or suppressive effects, a sum that is probably not identifiable by the rating of its individual parts. Hence it can be concluded that threshold test seems to be a promising method to identify the subtle differences in cumin aroma. REFERENCES ASTM. 1996. Sensory Testing Methods, 2nd ed. ASTM, Manual 26 (E. Chamber IV and M.B. Wolf, eds.) pp. 54 72, ASTM, West Conshohocken, PA.
CUMIN FLAVOR, PROFILE, PCA AND MAPPING 273 BANDONI, A.L., JUAREZ, M.A. and MIZRAHI, I. 1991. Essential oil of cumin (Cuminum cyminum L.). Essenze Derivatti Agrumari 61(1), 32 39. BERA, M.B., SHRIVASTAVA, D.C., SINGH, C.J., KUMAR, K.S. and SHARMA, Y.K. 2001. Development of cold grinding process, packaging and storage of cumin powder. J. Food Sci. Technol. India 38(3), 257 259. CLIFF, M.A. 1994. Sensory characterization of oral irritants. Dissertation Abstracts International, B54(7) 3411, Order No. DA9400014, Univ. of Missouri, Columbia, MO. DRAVNIEKS, A. 1985. Atlas of odor character profiles, ASTM Data Series DS 61, ASTM, West Conshohocken, PA. GRANVILLE, C.M., RUSSELL, G.F. and HEINZ, D.E. 1971. Chemical and sensory properties of cumin. AGFD 34, Abstracts of Papers. American Chemical Society 162, Washington, DC. IS: 5126. 1969. Glossary of general terms for sensory evaluation of foods, Part I, Methodology, Indian Standards Institute, New Delhi. JELLINEK, G. 1964. Introduction to and critical review of modern methods of sensory analysis (odour, taste and flavour evaluation) with emphasis on descriptive sensory analysis (flavour profile method). J. Nutrition Dietetics 1, 219 260. JOWITT, R. 1974. The terminology of food texture. J. Texture Studies 5(3), 351 358. STONE, H., SIDEL, J., OLIVER, S., WOOLSEY, A. and SINGLETON, R.C. 1974. Sensory evaluation by quantitative descriptive analysis. Food Technol. 28(11), 24 34. SUBBULAKSHMI, G., UDIPI, S. and RAHEJA, R. 1991. Evaluation of sensory attributes and some quality indices of irradiated spices. J. Food Sci. Technol. India 28(6), 396 397.