Utilization of Multivariate Analysis Technique (Chemometry) in Assessing Quality of Yemeni Honeys

Abstract

In this work, 38 honey samples, collected from different area in Yemen, were analyzed by means of physicochemical and spectroscopic techniques. In respect to physicochemical parameters, all samples were of good quality and almost every sample complied with international standards. In this study, an attempt was made to discriminate botanical origin of unifloral Sidr honey using minimum physicochemical characteristics or its spectroscopic data. Two discriminant models were developed either, using conductivity, pH, free acidity and the percentage of water or using the UV spectra of the honey samples as variables. Both models allow to perfectly reaching this goal, in a very short time and at low cost. Pattern recognition techniques appeared as a valuable classification tool to discriminate the botanical origin of Sidr honey from non Sidr honey samples. In fact, PLS-DA resulted in a predictive ability of 100%, on validation, using only a minimum number of experimental indices (which can be determined by all quality control laboratories). Additionally, the use of the class-modeling technique SIMCA, based on spectroscopic data, was successfully able to classify the authentic Sidr honey from non-Sidr honey samples with a clear demarcation. Also, the model was able to detect admixture of genuine Sidr honey with closely related polyfloral honey by detection limits of 20% or higher which can be used as possible way for detection of adulteration. A high correlation was shown between the results of PLS-DA and SIMCA models in classification of marketed samples, though the both models are based on different analytical data, the same samples were classified by both models as being Sidr honey. However, chemometric technique in combination with spectroscopic data found to be superior to the physicochemical data analysis, because of minimal reagent requirement and provide a great deal of information with only one test, also it is fast, relatively low-cost and require very small amount, about 2g., of sample. The result from this study verified that differences exist between the honeys from different botanical origin, and suggested that the spectroscopy coupled with multivariate data 60 analysis method hold the necessary information for successful classification of honey according its botanical origin.