Metabolomics applied to proton nuclear magnetic resonance profile for the identification of seven floral origin of French honeys

A sample preparation protocol was developed for honey profiling by proton nuclear magnetic resonance (H-1 NMR). The honey samples (0.5 g each) were solubilized in 5 ml of oxalate buffer (0.25 mol.l(-1), pH 4.3). This solution (1 ml) supplemented with a lock solution (D2O, 0.1 ml) containing 3-(trimethylsilyl) propionic acid D4 sodium salt was used for analysis. The novelty of the proposed method consists in no need for subsequent pH adjustment. The developed method was used for the analysis of 212 French honeys split in seven floral origins (acacia, chestnut tree, scrub, colza, lavender, sunflower and honeydew). A soft independent modelling of class analogy (SIMCA) data treatment was applied to the H-1 NMR spectra after a range scale pre-processing. The obtained class distance indicated successful discrimination between the seven floral origins, confirmed by a 100% correct classification when a honey test set was used. The model was applied to commercial mono-floral honeys and the results were fully corroborated by sensory and pollen analyses.