Classification of 7 monofloral honey varieties by PTR-ToF-MS direct headspace analysis and chemometrics

• The headspace of 70 unifloral honeys (7 classes) were characterized by PTR-ToF-MS.
• Headspace analysis is rapid, non-invasive and completely automated.
• Supervised and unsupervised methods provide low classification errors.
• Classification success ranges from 81 to 99% in the internal validation sets.

Honey, in particular monofloral varieties, is a valuable commodity. Here, we present proton transfer reaction-time of flight-mass spectrometry, PTR-ToF-MS, coupled to chemometrics as a successful tool in the classification of monofloral honeys, which should serve in fraud protection against mispresentation of the floral origin of honey. We analyzed 7 different honey varieties from citrus, chestnut, sunflower, honeydew, robinia, rhododendron and linden tree, in total 70 different honey samples and a total of 206 measurements. Only subtle differences in the profiles of the volatile organic compounds (VOCs) in the headspace of the different honeys could be found. Nevertheless, it was possible to successfully apply 6 different classification methods with a total correct assignment of 81–99% in the internal validation sets. The most successful methods were stepwise linear discriminant analysis (LDA) and probabilistic neural network (PNN), giving total correct assignments in the external validation sets of 100 and 90%, respectively. Clearly, PTR-ToF-MS/chemometrics is a powerful tool in honey classification.

Figure optionsDownload as PowerPoint slide