Search

Geographical and floral origin determines the composition of honey. Melissopalynology analysis is the traditional method for determining the source of the pollen in honey. This method of analysis is difficult, time-consuming and needs to be executed by an expert. Methods that are rapid and easily performed to determine the geographical origin of honey are therefore important and necessary. This study aimed to evaluate NIR spectroscopy, together with partial least squares discriminant analysis (PLS-DA) and different pre-processing methods to classify South African honey from three different geographical locations in the Western Cape. Two different NIR instruments, i.e. a laboratory instrument and a handheld device (glass tube and Teflon cup), were used in transflectance and reflectance mode respectively. Principal component analysis score plots showed separation between the Hopefield and Stanford classes. The Stellenbosch class overlapped with both the Hopefield and Stanford classes due to the Stellenbosch vegetation, which served as a nectar source for the bees, being a combination of fynbos and Eucalyptus, while for Hopefield and Stanford, it was predominantly fynbos and Eucalyptus respectively. The best prediction accuracy (84,21%) to classify between the three geographical origins was achieved with the handheld device (glass vial) together with PLS-DA and external parameter orthogonalisation as pre-processing method. From the data sets used in this study, the nectar source contributed greatly to the geographical classification. NIR spectroscopy could thus be considered as a supportive screening method for traditional melissopalynology. [ABSTRACT FROM AUTHOR]