A decisive strategy for monofloral honey authentication using analysis of volatile compounds and pattern recognition techniques

In an effort to determine some undisputed markers of botanical origin, the volatile profile of Hellenic monofloral honeys was investigated using gas chromatography coupled to mass spectrometry (GC-MS) in combination with headspace solid phase microextraction (HS-SPME). Different varieties of honey (citrus, fir, pine, and thyme) were analyzed and semi-quantitative data of volatile compounds originating from 82 samples were subjected to multivariate analysis of variance (MANOVA). Results showed that by using only 9 volatiles [dill ether (0.012 mg kg(-1)), alpha-4-dimethyl-3-cyclohexene-1-acetaldehyde (0.030 mg kg(-1)), acetic acid ethyl ester (0.030-0.173 mg kg(-1)), octanoic acid ethyl ester (0.004-0.547 mg kg(-1)), methylanthranilate (0.030 mg kg(-1)), 2,2,4,6,6-pentamethyl-heptane (0.005-0.057 mg kg-1), phenylacetaldehyde (0.014-1.248 mg kg(-1)), cis-linalool oxide (0.007-0.010 mg kg(-1)), lilac aldehyde (isomer III) (0.032-0.437 mg kg(-1))] honeys could be classified correctly according to botanical origin. The correct prediction rates were 93.9%, 96.8%, and 89.5%, based on the cross-validation method and using stepwise linear discriminant analysis, training and holdout partitions (k-nearest neighbors analysis), respectively. The efficiency of the discrimination model was confirmed further by application to mixed floral honeys and the use of the aforementioned consecutive pattern recognition techniques. The use of specific volatile compounds in combination with statistical tools may constitute a fast, decisive and accurate strategy for monofloral honey authentication.