In Brazil, Cannabis sativa is the most commonly trafficked illicit drug, with supply chains originating from both domestic and international sources, often involving sophisticated trafficking methods. In response to the Brazilian Federal Police (BFP) eradication efforts targeting large-scale plantations, alternative trafficking strategies have emerged, such as dispatching C. sativa seeds via postal services for indoor cultivation. In contemporary forensic genetics, microsatellites remain the gold standard for identification and have gained recognition as valuable tracking tools in investigative contexts. However, currently available microsatellite panels are not optimally informative for Brazilian samples. This study aimed to identify and evaluate highly informative microsatellites for origin tracking, using genomic data from 38 C. sativa samples sourced from Paraguay (PG), Colombia (CO), the Marijuana Polygon (MP) region, and a foreign group (FG) of commercial strains that constitute a substantial portion of national trafficking. Various combinations of Machine Learning (ML) Feature Selection (FS) techniques and classification models were used. To this end, diverse metrics for evaluating the performance of classification models were considered alongside the execution of principal coordinate analyzes (PCoA) and hierarchical clustering. Generally, Support Vector Classifiers (SVC) combined with different FS strategies demonstrated superior performance in predicting sample origin, achieving perfect classification of all samples using only 9 SSRs. These markers also effectively differentiated the four groups of seizures based on their biogeographic origin in PCoA and their hierarchical clustering. This study successfully identified informative SSR markers for distinguishing C. sativa samples associated with significant trafficking routes. Further research involving new and larger sample groups can be important to validate their application in practical investigative settings.
Bettim et al. (Fri,) studied this question.