Background: Although the main method for authentication of monofloral honey is pollen analysis, other classification approaches have been also applied. However, the majority of the existing classification models so far have utilized a few honey types or a few honey samples of each honey type, which can lead to inaccurate results. Aiming at addressing this, the goal of the present study was to create a classification model by analysing in total 250 honey samples from 15 different monofloral honey types in ten physicochemical parameters and then, multivariate analysis [multivariate analysis of variance (MANOVA), principal component analysis (PCA) and multi-discriminant analysis (MDA)] was applied in an effort to distinguish and classify them.
Results: Electrical conductivity and colour were found to have the highest discriminative power, allowing the classification of monofloral honey types, such as oak, knotgrass and chestnut honey, as well as the differentiation between honeydew and nectar honeys. The classification model had a high predictive power, as the 84.4% of the group cases was correctly classified, while for the cases of chestnut, strawberry tree and sunflower honeys the respective prediction was correct by 91.3%, 95% and 100%, allowing further determination of unknown honey samples.
Conclusion: It seems that the characterization of monofloral honeys based on their physicochemical parameters through the proposed model can be achieved and further applied on other honey types. The results could contribute to the development of methodologies for the determination of honey's botanical origin, based on simple techniques, so that these can be applied for routine analysis. © 2021 Society of Chemical Industry.
Keywords: honey; monofloral honeys; multi-discriminant analysis (MDA); multivariate analysis; physicochemical parameters; principal component analysis (PCA).
© 2021 Society of Chemical Industry.