Title: A multivariate statistical analysis approach to analyze gas chromatography-olfactometry data of tangerine hybrids Authors
Submitted to: Gordon Research Conference Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: June 28, 2010
Publication Date: N/A
Technical Abstract: Gas chromatography (GC) hyphenated with olfactometry (O) when a human subject smells the effluent of the GC is a useful technique to identify aroma activity of volatile compounds in a food. Many techniques have been developed, based on olfactory thresholds (CHARM analysis, AEDA), or based on psychophysics laws. Methods based on olfactory thresholds involve smelling samples in successive dilutions until no odor is perceived. These methods require multiple GC runs and can be time consuming; therefore, results are often reported from one or two subjects only. However, thresholds can vary widely among the population, and relying on one or two subjects many not be accurate. Therefore, methods that use more than one subject, and obtain an intensity response may provide more information. Five tangerine hybrid samples were analyzed by GC-O using the OSME method. Three trained subjects smelled the GC effluents of the 3 samples in triplicate runs. A consensus of 45 odorants was obtained with descriptors of green/grassy, fruity, mushroom, green/metallic, green/fresh, floral and fatty. Compounds with such odors were hexanal, ethyl 2-methylbutanoate, 1-octen-3-one, ß-myrcene, 1,8-cineole, linalool and (E,E)-2,4-nonadienal, respectively. Differences between samples were analyzed by analysis of variance of individual odor-active peaks. However, panelists did not always have the same response to individual compounds. Therefore, a general Procrustes analysis (GPA), which standardizes panelist responses, gave a better overall picture of samples profiles by odor-active peaks.