Physico-chemical characterization and classification of honey from the territory of The Republic of Serbia according to regional and botanical origin using multivariate chemometric analysis

Cilj ove doktorske disertacije je karakterizacija srpskog meda i utvrđivanje veze između glavnih konstituenata ispitivanih uzoraka i njihovog botaničkog i geografskog porekla, u okviru koje su analizirana 372 uzorka meda, od toga: 201 monoflorni med (167 uzorka bagremovog meda, 11 uzoraka lipovog meda, 23 suncokretova meda) i 171 poliflorni med. Karakterizacija tri najvažnije monoflorne vrste meda zastupljene u Srbiji prikazana u ovom radu je, pre svega, doprinos sve ukupnom ispitivanju evropskog monoflornog meda, u koje srpski med do sada nije bio uključen. Analiza glavnih komponenata (PCA) i linearna diskriminantna analiza (LDA) su primenjene na dobijene reultate fizičkohemijskih parametara (sadržaj vode, električna provodljivost, ukupna kiselost, specifičnarotacija i pH-vrednost) sa ciljem otkrivanja osnovne sličnosti i obrazaca grupisanja među različitim botaničkim vrstama ispitivanih uzoraka meda. Na osnovu rezultata PCA, fizičko-hemijski parametri se mogu koristiti za razlikovanje ove tri botaničke vrste meda. Linearna diskriminantna analiza rezultirala je pouzdanim modelom klasifikacije sa visokom moći predviđanja koji se može uspešno primeniti za razlikovanje ove tri monoflorne vrste meda (suncokretovog, lipovog i bagremovog meda). Utvrđeno je da su varijable koje najviše utiču na razdvajanje uzoraka u skladu sa njihovim botaničkim poreklom: električna provodljivost, slobodna kiselost i pH. Dobijenim modelom je pokazano da osnovni fizičko-hemijski parametri mogu da se koriste kao brz i pouzdan alat, dostupan velikom broju laboratorija za potvrdu botaničkog porekla ovet tri vrste meda.The main goal of this PhD thesis is characterization of Serbian honey and establishing links between the main constituents of the samples and their botanical and geographical origin. A total of 372 honey samples were analyzed: 201 monofloral honey (167 acacia honey, 11 linden honey, 23 sunflower honey), and 171 polyfloral honey. Characterization of three most important unifloral honeys in Serbia presented in this studyis, firstly, contribution to the examination of European unifloral honeys, in which Serbian has not been included. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) were applied on the physico-chemical parameters (water content, electrical conductivity, free acidity, optical rotation and pH), in order to highlight basic similarities and patterns of clustering among samples of honey with different botanicl origin. According to PCA, studied physico-chemical parameters could be used for differentiation of three monofloral honey types. LDA resulted in a reliable classification model with a high predictive power, allowing distinction of these three monofloral classes (acacia honey, linden honey, sunflower honey). It is estimated that the variables that best discriminate the samples, according to their botanical origin are: electrical conductivity, free acidity and pH. The resulting model is showing that the basic physical-chemical parameters can be used as a fast and reliable tool, available to the most of laboratories, to confirm botanical origin of these three types of honey. In the study of polyfloral honey on the territory of Serbia, honey was characterized in terms of mineral composition, sugar content and basic physico-chemical parameters (water content, electrical conductivity, pH, titratable acidity and optical rotation). Total 164 samples of polyfloral honey were collected from different regions of Serbia