Quality parameters and pattern recognition methods as a tool in tracing regional origin of multifloral honey

Multifloral honey was characterized in regards mineral to compo- sition, sugar content and basic physicochemical properties. A total of 164 honey samples were collected from different regions of Serbia during the har- vesting season 2009. Univariate data analysis (descriptive statistics and anal- ysis of variance), geographic information system and pattern recognition methods (principal component analysis and cluster analysis) were utilized in order to identify the geographical origin of honey. The content of Mg, K, and Cu, electrical conductivity and optical rotation were established as useful indicators in tracing regional differences between honey samples. Samples originating from Zlatibor region were clearly distinguished from those from the rest of Serbia, showing higher K and Mg contents, as well as higher values of optical rotation, electrical conductivity, and free acidity. The influence of the soil composition, and climate conditions, as well as the presence of particular flora on the honey composition is emphasized. The modeling of the geographic origin of honey was attempted by means of linear discriminant analysis.