260 Journal homepage: www.fia.usv.ro/fiajournal Journal of Faculty of Food Engineering, Ştefan cel Mare University of Suceava, Romania Volume XIX, Issue 4 - 2020, pag. 260 - 266 TILIA HONEY’S FRUCTOSE, GLUCOSE AND SUCROSE CONTENT PREDICTION USING FT-IR SPECTRA WITH PARTIAL LEAST SQUARES REGRESSION Daniela PAULIUC 1 , Paula CIURSA1, Florina DRANCA 1 , Sorina ROPCIUC 1 , *Mircea OROIAN 1 1Faculty of Food Engineering, Stefan cel Mare University, Suceava, Romania, m.oroian@fia.usv.ro, *Corresponding author Received 5th October 2020, accepted 29th December 2020 Abstract: The aim of this study was to assess the usefulness of Fourier transform infrared (FT- IR) spectroscopy coupled with partial least squares regression (PLS-R) to predict the fructose, glucose and sucrose content of tilia honeys. In order to achieve the aim of this study, 22 samples of tilia honey were purchased from Suceava, Neamt and Iasi County in the year of 2020. The fructose, glucose and sucrose content was determined prior the PLS-R prediction using high- performance liquid chromatography coupled with refractive index detector (HPLC-RID). The fructose content of tilia honeys ranged from 31.94 to 35.22%, the glucose content ranged between 26.76 and 33.15%, while sucrose content ranged between 0 and 2.20%. For the prediction of fructose, glucose and sucrose it was used the 3000 - 2800 + 1700 - 1600 +1540 - 700 cm-1 spectral range. The spectral data was submitted to different mathematical pretreatments in order to reduce the noise and to improve the prediction of results. For the prediction of fructose content the suitable pretreatment was Multiplicative Scatter Correction – MSC, for glucose prediction the suitable pretreatment was Standard Normal Variate – SNV, while for the prediction of sucrose the suitable pretreatment was 1st derivate. Keywords: tilia honey, fructose, glucose, sucrose, FT-IR, prediction 1. Introduction Honey is the oldest sweetener used by humankind and is worldwide known by its taste, color, aroma, and medicinal and nutritional properties as antioxidant, antiinflamatory, antitumor, anticancer and antimicrobial [1]. It is considered that honey has more than 200 substances as: sugars, organic acids, proteins, aminoacids, enzymes, vitamins, minerals, phenolic compounds and some solid particles (e.g. wax, pollen or other particles originating from its harvesting) [2,3]. The sugar composition of honey is a complex mixture of mono-, di-, tri-, oligo- and polysaccharides, but the main sugars present are fructose and glucose [4]. The main method used to determine the sugar composition of honey is high-performance liquid chromatography coupled with refractive index detector (HPLC-RID) [5]. Fourier transform infrared (FT-IR) spectroscopy presents information from frequencies of fundamental molecular vibrations and mostly reveals sharp peaks and different spectral characteristics [6,7]. Fourier transform infrared spectroscopy was used in many non-destructive methods for the characterization of honey and to determine different parameters, as follows: moisture content, sugar composition, free acidity, authentication of the botanical origin of honeys or to estimate the phenolic compounds and antioxidant capacity [7-10]. Partial least squares regression (PLS-R) is a multivariate regression method used for the complex data (e.g. FT-IR spectra, Raman http://www.fia.usv.ro/fiajournal Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University - Suceava Volume XIX, Issue 4 – 2020 Daniela PAULIUC, Paula CIURSA, Florina DRANCA, Sorina ROPCIUC, Mircea OROIAN, Tilia honey’s fructose, glucose and sucrose content prediction using FT-IR spectra with partial least squares regression, Food and Environment Safety, Volume XIX, Issue 4 – 2020, pag. 260 - 266 261 spectra, large number of data, UV-VIS spectra) and physicochemical data [11-13]. The PLS-R models are able to provide reduced models constructed on the original data in order to reduce data complexity [13, 14]. In this study is presented the prediction of fructose, glucose and sucrose from tilia honeys using FT-IR data with partial least squares regression (PLS-R). 2. Materials and methods Materials 22 samples of tilia honeys were collected from Suceava, Neamt and Iasi County from 2020 production. Methods The fructose, glucose and sucrose content were determined using a high-performance liquid chromatography (HPLC) method based on the protocol of Bogdanov & Baumann [4]. FT-IR analysis An ATR-Nicolet iS-20 spectrometer (Thermo Scientific, Karlsruhe, Dieselstraße, Germany) was used for honey spectra acquisition in the mid-infrared region of 4000-650 cm-1 with a resolution of 4 cm-1. The sample was placed on the ATR crystal and the spectra were collected using OMNIC software (version 32, Thermo Scientific). Statistical analysis Partial least squares regression (PLS-R) was realized using the Unscrambler X 10.1 software (Camo, Norway). 3. Results and discussion Sugar content In Table 1 are presented the mean, minimum and maximum concentrations of fructose, glucose and sucrose from the analyzed tilia honey samples. As it can be observed, the fructose concentration was higher in all samples, while sucrose content ranged between 0 and 2.20 %. The concentrations of fructose + glucose were higher than 60% in all the honey samples therefore meeting the threshold established by the EU legislation [15]. The fructose/glucose ratio was higher than 1 which confirmed the liquid state of the analyzed samples. Table 1. Fructose, glucose and sucrose content in tilia honeys Parameter Mean Min Max Fructose (%) 33.14 31.94 35.22 Glucose (%) 28.79 26.76 33.15 Sucrose (%) 0.36 0 2.20 F/G 1.15 Fig. 1. FT-IR spectra of tilia honeys Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University - Suceava Volume XIX, Issue 4 – 2020 Daniela PAULIUC, Paula CIURSA, Florina DRANCA, Sorina ROPCIUC, Mircea OROIAN, Tilia honey’s fructose, glucose and sucrose content prediction using FT-IR spectra with partial least squares regression, Food and Environment Safety, Volume XIX, Issue 4 – 2020, pag. 260 - 266 262 Table 2. Regression parameters of the calibration and validation procedure calculated for the FT-IR spectral data submitted to partial least square regression (PLS-R) analysis in order to predict glucose, fructose and sucrose in tilia honey Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University - Suceava Volume XIX, Issue 4 – 2020 Daniela PAULIUC, Paula CIURSA, Florina DRANCA, Sorina ROPCIUC, Mircea OROIAN, Tilia honey’s fructose, glucose and sucrose content prediction using FT-IR spectra with partial least squares regression, Food and Environment Safety, Volume XIX, Issue 4 – 2020, pag. 260 - 266 263 Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University - Suceava Volume XIX, Issue 4 – 2020 Daniela PAULIUC, Paula CIURSA, Florina DRANCA, Sorina ROPCIUC, Mircea OROIAN, Tilia honey’s fructose, glucose and sucrose content prediction using FT-IR spectra with partial least squares regression, Food and Environment Safety, Volume XIX, Issue 4 – 2020, pag. 260 - 266 264 Fig. 2. Reference vs predicted fructose, glucose and sucrose content using FT-IR spectra and PLS-R FT-IR spectra The FT-IR spectra collected for the analyzed honey samples are shown in Figure 1. According to the scientific literature, the spectral range typical for sugars in honeys are 3000-2800 + 1700-1600 +1540-700cm-1 As it can be observed in Figure 1, the signal for the spectral range presented above is high in the tilia honey samples that were analyzed in this study. PLS-R prediction The PLS-R model was used for the prediction of fructose, glucose and sucrose based on FT-IR spectral data. As it can be seen in Figure 1, the noise recorded for the honey samples was high and to reduce it the spectra must be submitted to different mathematical procedures, as follows: Multiplicative Scatter Correction – MSC, Standard Normal Variate – SNV, baseline, normalization, 1st derivate, 2nd derivate, smoothing (Table 2). As it can be observed in Table 2, each parameter predicted has a different pretreatment: fructose – Multiplicative Scatter Correction – MSC, glucose – Standard Normal Variate – SNV and sucrose – 1st derivate. The suitable model for each parameter is highlighted in Table 2 with different colours. The calibration and validation coefficients for the fructose and sucrose were higher than 0.861 proving the suitability of the proposed models, and in the case of glucose, the calibration parameters have a good magnitude, however, the validation parameters cannot be considered satisfactory. Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University - Suceava Volume XIX, Issue 4 – 2020 Daniela PAULIUC, Paula CIURSA, Florina DRANCA, Sorina ROPCIUC, Mircea OROIAN, Tilia honey’s fructose, glucose and sucrose content prediction using FT-IR spectra with partial least squares regression, Food and Environment Safety, Volume XIX, Issue 4 – 2020, pag. 260 - 266 265 For improving the correlation coefficients of the glucose prediction it would be necessary a higher number of samples. In Figure 2 is presented the evolution of reference and predicted values for fructose, glucose and sucrose. 4. Conclusion In this study it was determined the concentration of fructose, glucose and sucrose in 22 samples of tilia honeys from Suceava, Neamt and Iasi County. As it was observed, the concentration of fructose was higher than that of glucose in all the samples, while the concentrations of fructose + glucose met the threshold established by the UE legislation. The FT- IR spectra of the tilia honey samples were used for the prediction of glucose, fructose and sucrose content. The pretreatment of the FT-IR spectra increased the values of the prediction parameters. The fructose and sucrose prediction and validation parameters were satisfactory (R2 > 0.860), while the glucose parameters where not satisfactory in the validation step. For improving the correlation coefficients of the glucose prediction, a higher number of samples would be necessary. 5. Acknowledgment This work was supported by a grant of Romanian Ministry of Education and Research, CNCS-UEFISCDI, project number PN-III-P1-1.1-TE-2019-0583, and by the Romania National Council for Higher Education Funding, CNFIS, project number CNFIS-FDI-2020-0615. 6. References [1]. FERNANDES, L., RIBEIRO, H., OLIVEIRA, A., SILVA, A. S., FREITAS, A., HENRIQUES, M., & RODRIGUES, M. E. Portuguese honeys as antimicrobial agents against Candida species. Journal of Traditional and Complementary Medicine, (2020). [2]. DA SILVA, P. M., GAUCHE, C., GONZAGA, L. V., COSTA, A. C. O., & FETT, R. Honey: Chemical composition, stability and authenticity. Food Chemistry, 196, 309-323, (2016). [3]. PEREIRA, J. R., CAMPOS, A. N. D. R., DE OLIVEIRA, F. C., SILVA, V. R., DAVID, G. F., DA SILVA J. G., ... & DENEDAI, Â. M. Physical- chemical characterization of commercial honeys from Minas Gerais, Brazil. Food Bioscience, 100644, (2020). [4] PASCUAL-MATÉ, A., OSÉS, S. M., MARCAZZAN, G. L., GARDINI, S., MUIÑO, M. A. F., & SANCHO, M. T. Sugar composition and sugar-related parameters of honeys from the northern Iberian Plateau. Journal of food composition and analysis, 74, 34-43, (2018). [5] BOGDANOV, S., MARTIN, P., & LULLMANN, C. Harmonised methods of the international honey commission. Swiss Bee Research Centre, FAM, Liebefeld. (2002). [6] SCHÖNBICHLER, S. A., FALSER, G. F. J., HUSSAIN, S., BITTNER, L. K., ABEL, G., POPP, M., ... & HUCK, C. W. Comparison of NIR and ATR-IR spectroscopy for the determination of the antioxidant capacity of Primulae flos cum calycibus. Analytical Methods, 6(16), 6343-6351, (2014). [7] TAHIR, H. E., XIAOBO, Z., ZHIHUA, L., JIYONG, S., ZHAI, X., WANG, S., & MARIOD, A. A. Rapid prediction of phenolic compounds and antioxidant activity of Sudanese honey using Raman and Fourier transform infrared (FT-IR) spectroscopy. Food Chemistry, 226, 202-211, (2017) [8] ANJOS, O., CAMPOS, M. G., RUIZ, P. C., & ANTUNES, P., Application of FTIR-ATR spectroscopy to the quantification of sugar in honey. Food chemistry, 169, 218-223, (2015) [9] ZHU, X., LI, S., SHAN, Y., ZHANG, Z., LI, G., SU, D., & LIU, F., Detection of adulterants such as sweeteners materials in honey using near-infrared spectroscopy and chemometrics. Journal of Food Engineering, 101(1), 92-97, (2010). Food and Environment Safety - Journal of Faculty of Food Engineering, Ştefan cel Mare University - Suceava Volume XIX, Issue 4 – 2020 Daniela PAULIUC, Paula CIURSA, Florina DRANCA, Sorina ROPCIUC, Mircea OROIAN, Tilia honey’s fructose, glucose and sucrose content prediction using FT-IR spectra with partial least squares regression, Food and Environment Safety, Volume XIX, Issue 4 – 2020, pag. 260 - 266 266 [10] JANDRIĆ, Z., HAUGHEY, S. A., FREW, R. D., MCCOMB, K., GALVIN-KING, P., ELLIOTT, C. T., & CANNAVAN, A., discrimination of honey of different floral origins by a combination of various chemical parameters. Food Chemistry, 189, 52-59, (2015). [11] JOVIĆ, O., SMOLIĆ, T., PRIMOŽič, I., & HRENAR, T., Spectroscopic and chemometric analysis of binary and ternary edible oil mixtures: qualitative and quantitative study. Analytical chemistry, 88(8), 4516-4524, (2016). [12] MONFREDA, M., GOBBI, L., & GRIPPA, A. Blends of olive oil and seeds oils: characterisation and olive oil quantification using fatty acids composition and chemometric tools. Part II. Food chemistry, 145, 584-592., (2014). [13] LI, S., NG, T. T., & YAO, Z. P., Quantitative analysis of blended oils by matrix-assisted laser desorption/ionization mass spectrometry and partial least squares regression. Food Chemistry, 334, 127601, (2020). [14] WOLD, S., SJÖSTRÖM, M., & ERIKSSON, L., PLS-regression: a basic tool of chemometrics. Chemometrics and intelligent laboratory systems, 58(2), 109-130, (2001). [15] Council Directive 2001/110/EC of 20 December 2001 relating to honey 1. Introduction