Honey is a sweet, naturally occurring substance produced by bees from nectar or from secretions originating from the living parts of plants, enriched with secretions from their salivary and hypopharyngeal glands. It is a food particularly prone to adulteration, with a global economic impact estimated in the billions of dollars each year. To counter the spread of fraudulent practices, various analytical techniques are employed, including high-performance liquid chromatography (HPLC) and Ion Chromatography (IC) for determining chemical composition, as well as melissopalynological analysis to verify botanical origin. Despite their accuracy, these methods are costly, time-consuming, destructive, and require specialized personnel. In this context, spectroscopic techniques such as visible spectroscopy (VIS), near-infrared spectroscopy (NIR), and energy-dispersive X-ray fluorescence (ED-XRF) may offer a rapid, non-destructive, and easily implementable analytical alternative. The aim of this study is to evaluate the effectiveness of spectroscopic techniques in predicting parameters related to honey composition (such as sugar, anion, and cation profiles) and in identifying the botanical origin of different varieties of Italian honey (acacia, chestnut, honeydew, multifloral, lavender, and linden). A total of 35 samples of Italian honey of different botanical origins were analyzed. HPLC analyses enabled the determination of glucose, fructose, trehalose, and maltose concentrations, while IC analyses provided the content of chlorides, nitrites, nitrates, phosphates, sulfates, sodium, ammonium, potassium, magnesium, and calcium. Moisture was measured using a refractometer, while pH and electrical conductivity were measured with a pH-conductivity meter. The spectra obtained with the aforementioned instruments were used to develop multivariate models based on PLS, PCA, and PLS-DA, with the aim of assessing predictive capability for qualitative parameters and botanical origin. The results obtained from PLS highlighted the ability of NIR, VIS, and XRF spectra to accurately predict nitrate, sulfate, and potassium content as well as electrical conductivity. PCA was unable to correctly discriminate the different botanical varieties, whereas PLS-DA showed that ED-XRF achieved sensitivity and specificity values close to unity in differentiating the honeys. Although the limited number of samples affected the performance of some models, the proposed approach shows potential as a rapid screening tool. Future work should increase the number of samples in order to strengthen the robustness of the methods.
Il miele è una sostanza dolce di origine naturale prodotta dalle api a partire dal nettare o dalle secrezioni provenienti da parti vive delle piante, arricchito dalle secrezioni delle loro ghiandole salivari e ipofaringee. Si tratta di un alimento particolarmente esposto a fenomeni di adulterazione, con un impatto economico globale stimato nell’ordine dei miliardi di dollari ogni anno. Per contrastare il dilagare di pratiche fraudolente, vengono impiegate diverse tecniche analitiche, tra cui la cromatografia liquida ad alte prestazioni (HPLC) e la Cromatografia Ionica (IC) per la determinazione della composizione chimica e l’analisi melissopalinologica per la verifica dell’origine botanica. Nonostante l’accuratezza, questi metodi risultano economicamente costosi, lunghi, distruttivi e richiedono personale specializzato. In quest’ottica, le tecniche spettroscopiche, come la spettroscopia nel visibile (VIS), la spettroscopia nel vicino infrarosso (NIR) e la fluorescenza a raggi X a dispersione di energia (ED-XRF), possono rappresentare un’alternativa analitica rapida, non distruttiva e facilmente implementabile. Il presente studio ha l’obiettivo di valutare l’efficacia delle tecniche spettroscopiche nella predizione dei parametri relativi alla composizione del miele (come il profilo zuccherino, quello anionico e quello cationico) e di identificare l’origine botanica di diverse varietà di miele italiano (acacia, castagno, melata, millefiori, lavanda e tiglio). A tal fine, sono stati analizzati 35 campioni di miele italiano di diversa origine botanica. Le analisi HPLC hanno permesso di determinare le concentrazioni di glucosio, fruttosio, trealosio, maltosio e le analisi IC il contenuto di cloruri, nitriti, nitrati, fosfati, solfati, sodio, ammonio, potassio, magnesio e calcio. L’umidità è stata misurata tramite rifrattometro; mentre pH e conducibilità elettrica sono stati misurati usando un pHmetro-conduttimetro. Gli spettri ottenuti con gli strumenti sopracitati sono stati utilizzati per costruire modelli multivariati basati su PLS, PCA e PLS-DA, con lo scopo di valutare la capacità predittiva rispetto ai parametri qualitativi e all’origine botanica. I risultati ottenuti dalla PLS hanno evidenziato le capacità degli spettri NIR, VIS e XRF di predire con buona accuratezza il contenuto di nitrati, solfati, potassio e conducibilità elettrica. L’analisi PCA non è stata in grado di discriminare correttamente le diverse varietà botaniche, mentre la PLS-DA dimostra che l’ED-XRF ha valori di sensibilità e specificità prossimi all’unità nella discriminazione dei mieli. Sebbene il numero limitato di campioni abbia influenzato le prestazioni di alcuni modelli, l’approccio proposto mostra un potenziale come strumento di screening rapido. Futuri dovranno implementare il numero di campioni al fine di rafforzare la robustezza dei metodi.
UTILIZZO DELLA SPETTROSCOPIA PER LA CLASSIFICAZIONE DELL'ORIGINE BOTANICA E LA PREDIZIONE DELLA COMPOSIZIONE CHIMICA DI MIELE ITALIANO
SCARPA, MATTEO
2024/2025
Abstract
Honey is a sweet, naturally occurring substance produced by bees from nectar or from secretions originating from the living parts of plants, enriched with secretions from their salivary and hypopharyngeal glands. It is a food particularly prone to adulteration, with a global economic impact estimated in the billions of dollars each year. To counter the spread of fraudulent practices, various analytical techniques are employed, including high-performance liquid chromatography (HPLC) and Ion Chromatography (IC) for determining chemical composition, as well as melissopalynological analysis to verify botanical origin. Despite their accuracy, these methods are costly, time-consuming, destructive, and require specialized personnel. In this context, spectroscopic techniques such as visible spectroscopy (VIS), near-infrared spectroscopy (NIR), and energy-dispersive X-ray fluorescence (ED-XRF) may offer a rapid, non-destructive, and easily implementable analytical alternative. The aim of this study is to evaluate the effectiveness of spectroscopic techniques in predicting parameters related to honey composition (such as sugar, anion, and cation profiles) and in identifying the botanical origin of different varieties of Italian honey (acacia, chestnut, honeydew, multifloral, lavender, and linden). A total of 35 samples of Italian honey of different botanical origins were analyzed. HPLC analyses enabled the determination of glucose, fructose, trehalose, and maltose concentrations, while IC analyses provided the content of chlorides, nitrites, nitrates, phosphates, sulfates, sodium, ammonium, potassium, magnesium, and calcium. Moisture was measured using a refractometer, while pH and electrical conductivity were measured with a pH-conductivity meter. The spectra obtained with the aforementioned instruments were used to develop multivariate models based on PLS, PCA, and PLS-DA, with the aim of assessing predictive capability for qualitative parameters and botanical origin. The results obtained from PLS highlighted the ability of NIR, VIS, and XRF spectra to accurately predict nitrate, sulfate, and potassium content as well as electrical conductivity. PCA was unable to correctly discriminate the different botanical varieties, whereas PLS-DA showed that ED-XRF achieved sensitivity and specificity values close to unity in differentiating the honeys. Although the limited number of samples affected the performance of some models, the proposed approach shows potential as a rapid screening tool. Future work should increase the number of samples in order to strengthen the robustness of the methods.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi_MS.pdf
accesso aperto
Dimensione
1.79 MB
Formato
Adobe PDF
|
1.79 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/101202