Identification of Tobacco Leaf Cultivation Zones Using Energy Dispersive X‐Ray Fluorescence Spectra and Multivariate Statistical Analysis

ABSTRACT

The authentication of tobacco products by geographical origin is crucial for combating counterfeiting, especially for high‐value products such as Cuban Premium cigars. This study presents a simplified method to identify the cultivation zone of tobacco leaves using energy dispersive X‐ray fluorescence (EDXRF) spectra combined with multivariate statistical analysis, without prior determination of elemental concentrations. Tobacco leaves from three Cuban cultivation zones were analyzed by EDXRF. From the original 17 spectral variables, nine (S, K, Zn, WL, SrK, BrL, Cl, Ca, Mn) were retained following rigorous selection criteria. Principal component analysis (PCA) on these nine variables yielded a two‐component model explaining 68% of the total variance, with all samples within 95% confidence limits. MANOVA confirmed that cultivation zone significantly affects the spectral variables. A linear discriminant analysis (LDA) model was built on 28 training samples and validated internally (leave‐one‐out cross‐validation) and externally on six independent samples collected at different times by different operators; the LDA model achieved 100% classification accuracy on both validation sets. When applied to 11 additional samples of unknown origin (due to a labeling error), the LDA model provided zone assignments. Comparison with unsupervised PCA clustering showed 73% coincidence between the two independent methods, providing internal consistency evidence. Two principal achievements are highlighted. First, the method successfully discriminates intra‐national cultivation zones using direct EDXRF spectra, offering a rapid, non‐laborious, and cost‐effective screening tool to combat counterfeiting of Cuban tobacco products. Second, because inter‐country variability in pedoclimatic conditions and agronomic practices is expected to be substantially larger than the intra‐national differences resolved here, this approach is highly promising for further development into an inter‐national tobacco authentication method. The scalability of this strategy opens the door to global origin verification using simple, widely available EDXRF instrumentation.