Rapid differentiation of Khat (Catha edulis Vahl. Endl.) using single point and imaging vibrational spectroscopy

• Vibrational spectroscopy methods were developed to differentiate khat samples.
• Hyperspectral imaging (HSI) data was acquired from powdered leaves.
• Chemometric models with good model statistics were obtained.
• Results from UPLC–MS validated our OPLS-DA models.
• Leaf maturity and geographical origin can be distinguished using HSI.

The leaves of Catha edulis Vahl. Endl., known as Khat, are widely used as a “natural amphetamine” stimulant in Africa and in the Arabian Peninsula. In recent years, there has been an increase in the amounts exported to the USA, Australia and Europe. Although the dried, powdered form of the young and mature leaves, as well as samples from different geographical origins, have similar physical appearance, they have not been proven to exhibit equipotent stimulatory activity. In this study, powdered young and mature leaves were differentiated using vibrational spectroscopy, which included mid-infrared (MIR) and near-infrared (NIR) spectroscopy and hyperspectral imaging (HSI), and ultra performance liquid chromatography-mass spectrometry (UPLC–MS), in combination with chemometric data analysis. Samples originating from Ethiopia or South Africa could also be distinguished using chemometric models constructed from HSI data. Principal component analysis (PCA) and orthogonal projection to latent structures discriminant analysis (OPLS-DA) was successfully applied to classify powdered Khat samples according to their geographical origin and level of maturity. The OPLS-DA model constructed from the MIR and UPLC–MS data displayed the ability to clearly distinguish between young and mature leaves, as illustrated by the model statistics (R2Xcum and Q2cum > 80% and R2Y > 90%). Shortwave infrared hyperspectral imaging could distinguish Khat samples from two distinct geographical regions with good model statistics (R2Xcum = 0.91 and Q2cum = 0.92) and discriminate between the maturity levels (R2Xcum = 0.85 and Q2cum = 0.80). These models could prove valuable as rapid techniques for identifying confiscated Khat in forensic environments.