Background: Coffee is one of the most popular beverages around the world, consumed as an infusion of ground roasting coffee beans with a characteristic taste and flavor. Two main varieties, Arabica and Robusta, are produced worldwide. Furthermore, interest of consumers in quality attributes related to coffee production region and varieties is increasing. Thus, it is necessary to encourage the development of simple methodologies to authenticate and guarantee the coffee origin, variety and roasting degree, aiming to prevent fraudulent practices.
Results: C18 high-performance liquid chromatography with fluorescence detection (HPLC-FLD) fingerprints obtained after brewing coffees without any sample treatment other than filtration (i.e. considerably reducing sample manipulation) were employed as sample chemical descriptors for subsequent coffee characterization and classification by principal component analysis (PCA) and partial least squares regression-discriminant analysis (PLS-DA). PLS-DA showed good classification capabilities regarding coffee origin, variety and roasting degree when employing HPLC-FLD fingerprints, although overlapping occurred for some sample groups. However, the discrimination power increased when selecting HPLC-FLD fingerprinting segments richer in discriminant features, which were deduced from PLS-DA loading plots. In this case, excellent separation was observed and 100% classification rates for both PLS-DA calibrations and predictions were obtained (all samples were correctly classified within their corresponding groups).
Conclusion: HPLC-FLD fingerprinting segments were3 found to be suitable chemical descriptors for discriminating the origin (country of production), variety (Arabica and Robusta) and roasting degree of coffee. Therefore, HPLC-FLD fingerprinting can be proposed as a feasible, simple and cheap methodology to address coffee authentication, especially for developing coffee production countries. © 2020 Society of Chemical Industry.
Keywords: HPLC-FLD fingerprinting; coffee; food authentication; partial least square regression-discriminant analysis (PLS-DA); principal component analysis (PCA).
© 2020 Society of Chemical Industry.