1H nuclear magnetic resonance spectra of chloroform extracts of honey for chemometric determination of its botanical origin

Abstract

In this work, we present a new NMR study, coupled with chemometric analysis, on nonvolatile organic honey components. The extraction method is simple and reproducible. The 1H NMR spectra of chloroform extracts acquired with a fast and new pulse sequence were used to characterize and differentiate by chemometric analysis 118 honey samples of four different botanical origins (chestnut, acacia, linden, and polyfloral). The spectra collection, processing, and analysis require only 30 min. The 1H spectrum provides a fingerprint for each honey type, showing many characteristic peaks in all spectral regions. Principal component analysis (PCA) and projection to latent structures by partial least squares-discriminant analysis (PLS-DA) were performed on selected signals of the spectra to discriminate the different botanical types and to identify characteristic metabolites for each honey type. A distinct discrimination among samples was achieved. According to the distance to model criterion, there was no overlap between the four models, which proved to be specific for each honey type. The PLS-DA model obtained has a correlation coefficient R2 of 0.67 and a validation correlation coefficient Q2 of 0.77. The discriminant analysis allowed us to classify correctly 100% of the samples. A classification index can be calculated and used to determine the floral origin of honey as an alternative to the melissopalinology test and possibly to determine the percentage of various botanical species in polyfloral samples. Preliminary data on the identification of marker compounds for each botanical origin are presented. © 2009 American Chemical Society.