XX EuroFoodChem Conference
CA12 ORAL
27
Volatile fraction by HS-SPME-GC-MS and sensory evaluation of
more than 1200 Virgin Olive Oil samples: methods to support Panel
Test in Virgin Olive Oil classification
Lorenzo Cecchi
1, Marzia Migliorini
2, Elisa Giambanelli
2, Luca Calamai
3, Adolfo
Rossetti
2, Anna Cane
2, Fabrizio Melani
1, Nadia Mulinacci
1,*1Department of NEUROFARBA, and Multidisciplinary Centre of Research on Food Sciences
(M.C.R.F.S.- Ce.R.A),University of Florence, Via Ugo Schiff 6, 50019 Sesto F.no (Firenze), Italy
2 Carapelli Firenze S.p.A., Via Leonardo da Vinci 31, 50028 Tavarnelle Val di Pesa (Firenze), Italy 3 DISPAA, Università degli Studi di Firenze, Piazzale Cascine 28 50144 Firenze, Italy
* nadia.mulinacci@unifi.it
Extra virgin olive oil (EVOO) is considered as the highest quality product among the edible oils thanks to its pleasant taste and smell and the health properties given by the high content of phenolic compounds. Virgin olive oils are classified as extra virgin olive oil (EVOO), virgin olive oil (VOO) or lampante virgin olive oil (LVOO), based on their chemical and sensorial characteristics [1]. To date, the official method for assessing the sensorial properties is the Panel Test, carried out by a panel of 8-12 trained tasters and a head-panel. This analysis suffers of some drawbacks due to the use of humans, as emotionality, subjectivity, low reproducibility and high costs [2]. To date it is considered increasingly necessary to have availability of robust and reliable methods for support panel test in virgin olive oil classification, only based on chemical analysis and chemometric tools [3].
Aim of this work is to propose models for supporting panel test in virgin olive oil classification through predictive systems able to correlate chemical and organoleptic properties, developed working on more than 1200 virgin olive oil samples. Furthermore, we aimed at getting further light on the molecules able to discriminate between the different categories of samples.
To this aim, we analyzed the volatile fraction (VF) of all selected virgin olive oil samples by a recently validated HS-SPME-GC-MS method [1], which allowed us quantifying up to 73 volatile organic compounds using 9 internal standard for area normalization. Sensorial characteristics of the same same time, two oils with different fatty acid composition were stored for six months in several non-accelerated oxidative conditions and periodically analysed to better investigate the rancid defect. Different statistical approaches have been applied in order to create predictive models aimed at discriminating between EVOO and defective samples, to support panel test in virgin olive oil classification. We considered as a key-factor working with a high number of samples in order to have very robust statistical models and with almost all the selected samples belonging to EVOOs or VOOs with only a little number of LVOOs. The capability in discriminating between samples with oxidative and microbiological defects was also evaluated. After quantifying 73 volatile organic compounds for more than 1200 virgin olive oil samples, the obtained data were analyzed together with the sensorial data using statistic tools as t-test and Principal Component Analysis (PCA) for reducing the dimensionality of the data, and Linear Discriminant Analysis (LDA) to find combinations of variables and finding a linear fit able to separate categories of samples.
All the proposed approaches resulted able to predict correctly the category of approx. 80% of samples. The main defect of almost all the defective samples was correctly identified. Finally, those VOCs able to better discriminate between different categories were identified and resulted the same identified in in the oil samples stored in non-accelerated oxidative conditions.
References:
[1] M. Fortini, M. Migliorini, C. Cherubini, L. Cecchi, L. Calamai, Talanta, 165 (2017) 641 [2] I. Romero, D.L. Garcia-Gonzalez, R. Aparicio-Ruiz, M.T. Morales, Talanta, 134 (2015) 394. [3] C. Sales, T. Portoles, L.G. Johnsen, M. Danielsen, J. Beltran, Food Chemistry, 271 (2019) 488