Recent studies indicate that the tobacco contraband market in Canada could be as high as 20% which translates to billions of dollars in lost revenue for the Government and legitimate tobacco industry.

While it is known that the majority of the contraband cigarettes sold in Canada are manufactured within Canada and northern New York state, the provenance of the tobacco is unknown. This information could help reduce the supply of tobacco to the contraband cigarette producers. For this reason, a project was initiated to see if it was possible to determine the geographic origin of tobacco leaves.

Approximately 1000 samples of tobacco leaves from the top ten flue-cured producing countries including Canada were collected. Over 60 components including trace elements (metals), organic acids and isotopic ratios of nicotine were quantified using different analytical techniques. A Partial Least Square Discriminant Analysis (PLSDA) model and a Support Vector Machine-Classification (SVM-C) model were developed.

The PLSDA model was able to predict if leaves originated from Canada, USA or another country (international). The SVM-C model was able to predict if a leaf originated from Canada or USA but was also able to discriminate between the other countries.