30B-6 |
Prediction of sensory descriptors of tomato flavor with and without partitioning as a function of volatile and non-volatile components. |
E. G. ABEGAZ1, R. L. Shewfelt1, K. S. Tandon1, and E. A. Baldwin2. (1) Dept. of Food Science & Technology, Univ. of Georgia, Food Science Bldg., Athens, GA 30602-7610, (2) USDA-ARS-Citrus & Subtropical Products Laboratory, PO Box 1909, Winter Haven, FL 33883 Lack of characteristic tomato flavor is a common complaint with supermarket tomatoes. A complex mixture of soluble sugars, free acids, amino acids, minerals, and aroma volatile compounds contribute to characteristic flavor of tomato fruits. Although more than 400 aromatic compounds have been identified in tomato fruits, less than 20 volatile compounds are considered to be important toward flavor based on their odor thresholds. Descriptive sensory analysis has been used by sensory professionals to characterize tomato flavor, but there is little definitive research relating specific compounds directly to sensory perception of flavor. We tested the hypothesis that partitioning of taste and aromatic notes using nose clips would improve predictability of mathematical models. Our objective was to determine the effect of partitioning during descriptive analysis on prediction of sensory descriptors as a function of volatile and non-volatile components. Nine panelists evaluated six breeding line tomatoes and four supermarket tomatoes using modified spectrum technique for descriptive analysis. Panelists were calibrated based on reference standards using 150-mm line scale. Chemical analysis of Soluble Solids (SS), Sucrose Equivalents (SE), pH, Titratable Acidity (TA) were conducted. Volatile analysis was achieved using Perkin Elmer GC, equipped with HS-6 headspace sampler and FID detector. Our results showed linear regression models developed from the six breeding lines using partitioning had higher coefficients of determination (R2 ) for most tomato flavor descriptors than those developed without partitioning. Greater differences in the two techniques were observed for sour, salty, tomato-like and astringency descriptors. These model equations were validated with four supermarket selections. The models tended to over predict for sweet, fruity and green/grassy descriptors while under predict bitter and bite notes. Partitioning appears to reduce interference between volatile and non-volatile components.
Session 30B, Fruit & Vegetable Products: Sensory, Product Development, Fresh-Cut, and Storage
|