Chemometric methods for botanical classification of Chinese honey based on the volatile compound profile

Significance of the topic

Honey is a widely consumed natural product prized for its nutritional and medicinal benefits. Differentiating botanical origins is essential to ensure product authenticity, prevent fraud, and maintain consumer trust. Conventional methods can be time-consuming and require complex sample preparation, motivating the need for rapid, reliable analytical strategies.

Objectives and study overview

This work aimed to establish a non-targeted chemometric classification approach based on solid-phase microextraction coupled with gas chromatography-mass spectrometry (SPME-GC/MS) to discriminate Chinese honey samples from acacia, linden, vitex, and rape sources. A training set of 87 authentic samples underwent blind analysis, followed by an independent validation with 20 additional honey samples to assess model robustness.

Methodology and instrumentation

Headspace volatile compounds were extracted using a DVB/CAR/PDMS SPME fiber at 80 °C for 30 min in vials containing honey, water, and salt. Thermal desorption occurred in the GC inlet at 250 °C for 2 min. GC separation employed an HP-5MS column (30 m×0.25 mm, 0.25 µm), with oven programming from 50 °C to 250 °C under helium flow. Mass spectra were acquired in full-scan mode (m/z 40–600, EI 70 eV).

Used instrumentation

• CTC autosampler with 2 cm DVB/CAR/PDMS SPME fiber
• Agilent 7890A gas chromatograph with HP-5MS column
• Agilent 5975C mass spectrometer (electron ionization, 70 eV)

Data processing and chemometric workflow

Raw GC/MS data were converted from ChemStation to MassHunter format, deconvoluted, and imported into Mass Profiler Professional. A four-step filtering procedure (reliability flags, frequency, ANOVA p<0.05, and fold-change threshold) reduced 2,734 initial entities to 70 variables. Principal component analysis (PCA) explained 79 % of total variance with the first four components, clearly separating linden from non-linden honeys.

Main results and discussion

Classification models based on partial least squares discriminant analysis (PLS-DA), naïve Bayes (NB), and back-propagation artificial neural network (BP-ANN) achieved 100 % recognition and cross-validation accuracy on the training set. Independent validation of 20 samples (five per origin) also yielded 100 % correct classification. Confidence measures exceeded 0.7 for most samples, confirming model reliability. A Venn diagram approach identified eight volatile markers unique to linden honey, including pulegone and cis-rose oxide, which align with literature reports on lime blossom volatiles.

Benefits and practical applications

The combined SPME-GC/MS and chemometric strategy offers a rapid, solvent-free, and reproducible protocol for botanical authentication of honey. It supports quality control laboratories and industry stakeholders in verifying origin claims, detecting adulteration, and standardizing product labeling.

Future trends and possibilities

Integration of higher-resolution mass spectrometry, two-dimensional GC, and machine learning algorithms may further enhance sensitivity and generalizability across diverse honey types. Real-time portable GC/MS devices combined with cloud-based chemometric platforms could enable on-site screening and continuous supply chain monitoring.

Conclusion

The study demonstrates that non-targeted volatile fingerprinting using SPME-GC/MS, coupled with robust chemometric filtering and classification models, reliably discriminates honey of different botanical origins. Validation with independent samples confirms the method’s practicality for routine authenticity testing.

Reference

1. Cuevas-Glory L F et al. A review of volatile analytical methods for determining the botanical origin of honey. Food Chemistry. 2007;103:1032–1043.
2. Chen H, Jin L, Fan C. Non-targeted volatile profiles for classification of Chinese honey by solid-phase microextraction and gas chromatography–mass spectrometry combined with chemometrics. Journal of Separation Science. 2017;40:4377–4384.
3. Chudzinska M, Baralkiewicz D. Estimation of honey authenticity by multielements characteristics using ICP-MS combined with chemometrics. Food and Chemical Toxicology. 2010;48:284–290.
4. Wang M et al. An integrated approach utilising chemometrics and GC/MS for classification of chamomile flowers, essential oils and commercial products. Food Chemistry. 2014;152:391–398.
5. Lušic D et al. Volatile Profile of Croatian Lime Tree (Tilia sp.), Fir Honeydew (Abies alba) and Sage (Salvia officinalis) Honey. Food Technology and Biotechnology. 2007;45:156–165.
6. Blank I, Fischer K H, Grosch W. Intensive neutral odorants of linden honey: Differences from honeys of other botanical origin. Zeitschrift für Lebensmittel-Untersuchung und Forschung. 1989;189:426–433.
7. Piasenzotto L, Gracco L, Conte L. Solid phase microextraction (SPME) applied to honey quality control. Journal of the Science of Food and Agriculture. 2003;83:1037–1044.
8. Špánik I et al. Characterisation of VOC composition of Slovak monofloral honeys by GC×GC-TOF-MS. Chemical Papers. 2013;67:127–134.