Near infrared spectroscopy applied to the rapid prediction of the floral origin and mineral content of honeys

Abstract

Consumers demand to know the floral origins of honeys. Therefore, the use of simple and reliable techniques for differentiating among honeys by their origins is necessary. Multivariate statistical techniques and near infrared spectroscopy applied to palynological and mineral characteristics make it possible to differentiate among the types of honey collected from Northwestern Spain. Prediction models using a modified partial least squares regression for the main pollen types (Castanea, Eucalyptus, Rubus and Erica) in honeys and their mineral composition (potassium, calcium, magnesium and phosphorus) were established. Good multiple correlation coefficients (higher than 0.700) and acceptable standard errors of cross-validation were obtained. The ratio performance deviation exhibited a good prediction capacity for Rubus pollen and for Castanea pollen, whereas for minerals, for Eucalyptus pollen and for Erica pollen the ratio performance deviation was excellent. Near infrared spectroscopy was established as a rapid and effective tool to obtain equations of prediction that contribute to the honey typification.