摘要：The aim of this study is to characterise 8 different monofloral and multifloral types of Croatian honey (a total of 254 samples from 2003, 2004, and 2005 harvesting seasons) based on 11 common physicochemical parameters (water mass fraction, total reducing sugar mass fraction, sucrose mass fraction, ash mass fraction, electrical conductivity, acidity, diastase and invertase activity, hydroxymethylfurfural (HMF) mass fraction, proline mass fraction and optical rotation). Differences in the above-mentioned parameters, established among the honey samples, are influenced by different factors, such as botanical origin, climate and regional circumstances. After the sample characterisation, results obtained for 2 monofloral (acacia (Robinia pseudoacacia) and chestnut (Aesculus hippocastanum)), and 2 multifloral (floral and meadow) honey types were subjected to the pattern recognition procedures. In this regard, unsupervised methods such as cluster and principal component analyses were employed, with the goal of evaluating the possibility of differentiation of Croatian honey stemming from different botanical origins, based on their physicochemical profile. Cluster analysis (CA) revealed the existence of two clusters, in the first of which is acacia honey as the best grouped, and the second corresponds to the dispersed group constituted of the remaining three honey types under investigation (chestnut, floral, and meadow). Principal component analysis (PCA), i.e. its first two components, stood for the average of 50.5 % of the data variance. PCA and CA showed that physicochemical parameters are able to provide enough information to allow for the classification and distinction of the types of honey originating from four botanical origins under investigation (acacia, chestnut, floral and meadow).
