Botanical authentication of honeys based on Raman spectra

Optical spectroscopy methods combined with multivariate statistical analysis for the classification of Cretan thyme, multi-floral and honeydew honey

BACKGROUND

The botanical origin of honey attracts both commercial and research interest. Consumers' preferences and medicinal uses of particular honey types drive the demand for the determination of their authenticity with regard to their botanical origin. This study presents the discrimination of thyme, multi-floral. and honeydew honeys by Fourier-transform infrared (FTIR) and ultraviolet (UV) absorption spectroscopy combined with multivariate statistical analysis. UV absorption spectroscopy was applied without any dilution of the sample using a custom-made cuvette. FTIR and UV absorption spectroscopic data were processed by means of the orthogonal partial least squares discriminant analysis.

RESULTS

The optimal classification of floral and honeydew honeys was accomplished with UV spectroscopy with a successful estimation of 92.65% for floral honey and 91.30% for honeydew honey. The discrimination of thyme versus the multi-floral honey was best achieved with FTIR, with a correct classification of 95.56% and 100% for multi-floral and thyme honey respectively. Furthermore, our findings revealed the region of 2400-4000 cm^-1 of the FTIR spectra as the most significant for this discrimination.

CONCLUSION

This work demonstrates that optical spectroscopic techniques in combination with multivariate statistical analysis can be a rapid, low-cost, easy-to-use approach for the determination of the botanical origin of honey without sample pretreatment. © 2021 Society of Chemical Industry.

Comparative Study of Several Machine Learning Algorithms for Classification of Unifloral Honeys

Unifloral honeys are highly demanded by honey consumers, especially in Europe. To ensure that a honey belongs to a very appreciated botanical class, the classical methodology is palynological analysis to identify and count pollen grains. Highly trained personnel are needed to perform this task, which complicates the characterization of honey botanical origins. Organoleptic assessment of honey by expert personnel helps to confirm such classification. In this study, the ability of different machine learning (ML) algorithms to correctly classify seven types of Spanish honeys of single botanical origins (rosemary, citrus, lavender, sunflower, eucalyptus, heather and forest honeydew) was investigated comparatively. The botanical origin of the samples was ascertained by pollen analysis complemented with organoleptic assessment. Physicochemical parameters such as electrical conductivity, pH, water content, carbohydrates and color of unifloral honeys were used to build the dataset. The following ML algorithms were tested: penalized discriminant analysis (PDA), shrinkage discriminant analysis (SDA), high-dimensional discriminant analysis (HDDA), nearest shrunken centroids (PAM), partial least squares (PLS), C5.0 tree, extremely randomized trees (ET), weighted k-nearest neighbors (KKNN), artificial neural networks (ANN), random forest (RF), support vector machine (SVM) with linear and radial kernels and extreme gradient boosting trees (XGBoost). The ML models were optimized by repeated 10-fold cross-validation primarily on the basis of log loss or accuracy metrics, and their performance was compared on a test set in order to select the best predicting model. Built models using PDA produced the best results in terms of overall accuracy on the test set. ANN, ET, RF and XGBoost models also provided good results, while SVM proved to be the worst.

Botanical origin discrimination of Greek honeys: physicochemical parameters versus Raman spectroscopy

BACKGROUND

The authenticity of honey is of high importance since it affects its commercial value. The discrimination of the origin of honey is of prime importance to reinforce consumer trust. In this study, four chemometric models were developed based on the physicochemical parameters according to European and Greek legislation and one using Raman spectroscopy to discriminate Greek honey samples from three commercial monofloral botanical sources.

RESULTS

The results of physicochemical (glucose, fructose, electrical activity) parameters chemometric models showed that the percentage of correct recognition fluctuated from 92.2% to 93.8% with cross-validation 90.6-92.2%, and the placement of test set was 79.0-84.3% successful. The addition of maltose content in the previous discrimination models did not significantly improve the discrimination. The corresponding percentages of the Raman chemometric model were 95.3%, 90.6%, and 84.3%.

CONCLUSION

The five chemometric models developed presented similar and very satisfactory results. Given that the recording of Raman spectra is simple, fast, a minimal amount of sample is needed for the analysis, no solvent (environmentally friendly) is used, and no specialized personnel are required, we conclude that the chemometric model based on Raman spectroscopy is an efficient tool to discriminate the botanical origin of fir, pine, and thyme honey varieties. © 2020 Society of Chemical Industry.

Differentiation of white architectural paints by microscopic laser Raman spectroscopy and chemometrics

A feasible, effective and non-destructive method that could be used to differentiate architectural paints was proposed by Microscopic laser Raman spectroscopy and chemometrics. A total of 252 white architectural paints from 7 different manufacturers were prepared for evaluating the potential of differentiating them. 5th Newton interpolation polynomial combined with Savitzky-Golay 7-point and 1st or 2nd polynomial smoothing under the 1st-order derivative were considered as the optimal pre-processing method for MLRM data. The Bayes discriminant analysis model realized 100% accuracy based on discriminant functions Z₁, Z₂ and Z₃, which was the more useful and practical method for differentiating white architectural paints than that of multilayer perceptron and radial basis function neural network models. All samples were differentiated exactly, which was rapid and non-destructive. The designed method demonstrated the potential of Microscopic Laser Raman spectroscopy in combination with pre-processing and chemometrics as a universal, confirmatory, and accurate method for the white architectural paint differentiation in forensic science.

Raman Spectroscopy and Chemometric Modeling to Predict Physical-Chemical Honey Properties from Campeche, Mexico

In this work, 10 chemometric models based on Raman spectroscopy were constructed to predict the physicochemical properties of honey produced in the state of Campeche, Mexico. The properties of honey studied were pH, moisture, total soluble solids (TSS), free acidity, lactonic acidity, total acidity, electrical conductivity, Redox potential, hydroxymethylfurfural (HMF), and ash content. These proprieties were obtained according to the methods described by the Association of Official Analytical Chemists, Codex Alimentarius, and the International Honey Commission. For the construction of the chemometric models, 189 honey samples were collected and analyzed in triplicate using Raman spectroscopy to generate the matrix data [X], which were correlated with each of the physicochemical properties [Y]. The predictive capacity of each model was determined by cross validation and external validation, using the statistical parameters: standard error of calibration (SEC), standard error of prediction (SEP), coefficient of determination of cross-validation (R2cal), coefficient of determination for external validation (R2val), and Student's t-test. The statistical results indicated that the chemometric models satisfactorily predict the humidity, TSS, free acidity, lactonic acidity, total acidity, and Redox potential. However, the models for electric conductivity and pH presented an acceptable prediction capacity but not adequate to supply the conventional processes, while the models for predicting ash content and HMF were not satisfactory. The developed models represent a low-cost tool to analyze the quality of honey, and contribute significantly to increasing the honey distribution and subsequently the economy of the region.

Establishing authenticity of honey via comprehensive Romanian honey analysis

Assessing the authenticity of honey is a serious problem that has gained much interest internationally because honey has frequently been subject to various fraudulent practices, including mislabelling of botanical and geographical origin and mixing with sugar syrups or honey of lower quality. To protect the health of consumers and avoid competition, which could create an unstable market, consumers, beekeepers and regulatory bodies are interested in having reliable analytical methodologies to detect non-compliant honey. This paper gives an overview of the different approaches used to assess the authenticity of honey, specifically by the application of advanced instrumental techniques, including spectrometric, spectroscopic and chromatographic methods coupled with chemometric interpretation of the data. Recent development in honey analysis and application of the honey authentication process in the Romanian context are highlighted, and future trends in the process of detecting and eliminating fraudulent practices in honey production are discussed.

A Screening Method Based on Headspace-Ion Mobility Spectrometry to Identify Adulterated Honey

Nowadays, adulteration of honey is a frequent fraud that is sometimes motivated by the high price of this product in comparison with other sweeteners. Food adulteration is considered a deception to consumers that may have an important impact on people's health. For this reason, it is important to develop fast, cheap, reliable and easy to use analytical methods for food control. In the present research, a novel method based on headspace-ion mobility spectrometry (HS-IMS) for the detection of adulterated honey by adding high fructose corn syrup (HFCS) has been developed. A Box-Behnken design combined with a response surface method have been used to optimize a procedure to detect adulterated honey. Intermediate precision and repeatability studies have been carried out and coefficients of variance of 4.90% and 4.27%, respectively, have been obtained. The developed method was then tested to detect adulterated honey. For that purpose, pure honey samples were adulterated with HFCS at different percentages (10-50%). Hierarchical cluster analysis (HCA) and principal component analysis (PCA) showed a tendency of the honey samples to be classified according to the level of adulteration. Nevertheless, a perfect classification was not achieved. On the contrary, a full classification (100%) of all the honey samples was performed by linear discriminant analysis (LDA). This is the first time the technique of HS-IMS has been applied for the determination of adulterated honey with HFCS in an automatic way.