A comparison between FT-IR, NIR, and total fluorescence performance in honey adulteration detection

Honey is a natural, high value product, whose authenticity is often compromised by fraudulent practices, such as the addition of cheaper sugar syrups. Honey adulteration is one of the most widespread food frauds. This study evaluates the feasibility and performance of spectroscopic techniques to quantify the adulteration of citrus and wildflower honey. To this end, samples of authentic honey were adulterated with three different sugar syrups in increasing concentrations. FT-IR and NIR spectra, and fluorescence excitation-emission matrices (EEMs) were collected for pure honeys (n = 2), pure syrups (n = 3), and adulterated samples (n = 42). FT-IR and NIR spectra were explored by principal component analysis (PCA), then partial least squares (PLS) regression was used for quantify the level of adulteration. Considering NIR, a global PLS model showed promising predictive capacity of the level of adulteration, with an R2CV of 0.76 and an error in cross-validation (RMSECV) of 6%. However, the performance was affected by the type of adulterant. By splitting the dataset, the regression model improved significantly (R2CV 0.94; RMSECV 3%). PCA of FT-IR spectra highlighted differences in the samples according to the type of adulterant; nonetheless, the PLS model showed poor and insufficient predictive performance. Fluorescence EEMs revealed distinct bands in pure honeys while syrups had weaker signals. Parallel factor analysis (PARAFAC) identified two significant components, with Component 2 correlated with the level of adulteration. Linear regression based on Component 2 scores produced accurate estimates, but highlighted a bias between honey types. A second approach, based on NPLS trained on the whole dataset, provided interesting results (R2 0.92 in calibration, 0.90 in CV; RMSECV 4%). Overall, NIR and fluorescence spectroscopies, coupled with multivariate analysis, have proven to be suitable for honey authentication. NPLS applied on EEMs provided an effective global model for all the tested adulterants. In the other cases, it seems that a proper strategy rely on the development of adulterant-specific (NIR) or honey-specific (EEMs-PARAFAC based regression) models, which could be more challenging. This study was carried out within the Agritech National Research Center and received funding from the European Union Next-GenerationEU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR) – MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.4 – D.D. 1032 17/06/2022, CN00000022). This manuscript reflects only the authors’ views and opinions, neither the European Union nor the European Commission can be considered responsible for them.