Quality assessment of mānuka honeys using non-invasive Near Infrared systems

Abstract

Mānuka honey comes mainly from Leptospermum scoparium nectar. Methylglyoxal (MGO) is the primary anti-bacterial in mānuka honeys and other chemical markers contribute to its value. MGO forms gradually in the honey matrix from dihydroxyacetone (DHA) that uniquely hyperaccumulates in L. scoparium nectar. Leptosperin plus four polyphenolic compounds are used for botanical verification of mānuka honey. All these markers are unique to mānuka and indirectly indicate mānuka honey potency (measured by UMF™ scores) and purity (verification of botanical origin). This study aimed to develop techniques for Visible-Near Infrared spectroscopy for grading honey before extraction to avoid inadvertent mixing of high and low value honeys. A set of 1451 mono-floral, multi-floral mānuka and other (non-mānuka) honeys from eight geographic regions across New Zealand was investigated containing naturally variable levels of chemical marker. Chemometric models using Partial Least Squares (PLS), Support Vector Machine (SVM) and discriminant analysis, PLS-DA and SVM-DA (RBF) were developed to predict mānuka honey potency and purity. Prediction of potency gave 74 % accuracy by a PLS regression model. PLS-DA and SVM-DA classifiers for botanical origin gave 89 % overall accuracy. In particular, mono-floral mānuka honeys could be classified with 92–97 % accuracy.