Clever chemometrics catches funny honey cheats

Skip to Navigation


  • Published: Nov 1, 2017
  • Author: Ryan De Vooght-Johnson
  • Channels: Laboratory Informatics / Chemometrics & Informatics
thumbnail image: Clever chemometrics catches funny honey cheats

The sources of honey samples need identifying

Honey can command different prices based on the source of the nectar. Typically, honey sourced largely from a particular flower, known as monofloral honey, sells for more than generic, blended honey. The honey from lime tree flowers, sometimes known as linden honey, is particularly sought-after and sometimes faked using cheaper products. The Chinese researchers examined honey samples from four different flowers: lime, false acacia, oilseed rape and Vitex negundo (Chinese chastetree). 87 samples of known origin were investigated, their identities being confirmed by taste, odour and pollen analysis.

SPME, GC-MS and chemometrics used to characterise honey

Honey samples were extracted by SPME (solid-phase microextraction), using a CTC Combi-PAL automated system and a divinyl benzene/carboxen/polydimethylsiloxane fibre. Samples were heated with brine in a closed vial, with the volatiles being collected in the headspace on the fibre, which was subsequently inserted into the GC injection port.

GC was carried out with an Agilent 7890A instrument fitted with an HP-5ms column. The temperature was taken from 50 to 250 °C in two ramps. Mass spectrometry employed an Agilent 5975C mass selective detector using electron ionisation (EI), the data being acquired in full-scan mode. Peak deconvolution and identification were carried out using Agilent’s MassHunter software. The processed data was then carried over into Agilent’s Mass Profiler Professional (MPP) software for further analysis.

Initially, there were a total of 2734 ‘entities’ detected by the mass spectrometer, but some of these were artefacts rather than real compounds. A series of software filters was used to cut these down to a more manageable 110 compounds, eliminating weak, doubtful or duplicate signals. Principal component analysis (PCA) was applied to the filtered data. Four principal components were used, giving a PCA diagram that separated the 87 samples into four distinct groups depending on their nectar source. The false acacia and Vitex groups were close together, but the lime and oilseed rape groups were both some distance away from the others.

Partial least squares discriminant analysis (PLA-DA) was also applied to the data, giving a model that could successfully classify honey samples according to nectar source. A naïve Bayes (NB) model was constructed; this too enabled successful classification of the honey samples. Finally, a back-propagation artificial neural network (BP-ANN) model was applied, again giving good classification. These three different methods were applied to 20 new authentic honey samples that had not been included in the initial model development. All three methods gave good classification, the NB model being the most successful, correctly identifying all 20 samples with a 1.00 confidence measure. The BP-ANN model also gave correct identification, with most of the confidence measures being 0.98 or 0.99. The PLA-DA gave correct identification, but some of the confidence measure values were low, the lowest being only 0.59.

As lime honey was the most expensive variety, a number of compounds that only appeared in it, such as cis-rose oxide, were identified by comparison to literature mass spectral data or mass spectral libraries. The absence of these compounds should help regulatory agencies detect cases of fraud.

Chemometric methods help distinguish monofloral honey samples

The application of GC-MS and chemometric methods clearly enabled the nectar source of honey samples to be identified. It is interesting that the relatively simple naïve Bayes model proved the best one for correctly classifying samples. Further extension of this work to other nectar sources and to honeys from mixed sources would be useful.

Related Links

Journal of Separation Science, Early View paper. Chen et al. Non-targeted volatile profiles for the classification of the botanical origin of Chinese honey by solid-phase microextraction and gas chromatography–mass spectrometry combined with chemometrics.

Comprehensive Reviews in Food Science and Food Safety, 2017, 16, 1072-1100. Soares et al. “A comprehensive review on the main honey authentication issues: production and origin.

Wikipedia, Naive Bayes Classifier

Article by Ryan De Vooght-Johnson

The views represented in this article are solely those of the author and do not necessarily represent those of John Wiley and Sons, Ltd.

Follow us on Twitter!

Social Links

Share This Links

Bookmark and Share


Suppliers Selection
Societies Selection

Banner Ad

Click here to see
all job opportunities

Most Viewed

Copyright Information

Interested in separation science? Visit our sister site

Copyright © 2019 John Wiley & Sons, Inc. All Rights Reserved