Forensic analysis of Salvia divinorum using multivariate statistical procedures. Part II: association of adulterated samples to S. divinorum.

Salvia divinorum, a plant with hallucinogenic properties due to its active ingredient salvinorin A, was studied to see if it could be identified when mixed with other plant materials. The plant was extracted and added to samples of S. divinorum, Salvia officinalis, Cannabis sativa, and Nicotiana tabacum to mimic adulterated forensic samples. After extraction and gas chromatography-mass spectrometry analysis, the data were preprocessed and examined with principal components analysis (PCA). Visual inspection of PCA scores plots allowed association of adulterated extracts with unadulterated S. divinorum. Additional statistical tests—Euclidean distance, hierarchical cluster analysis, t tests, Wilcoxon rank-sum tests, and Pearson correlation—provided further evaluation, with their forensic advantages and limitations compared.

Study at a glance

Abstract

Salvia divinorum is a plant material that is of forensic interest due to the hallucinogenic nature of the active ingredient, salvinorin A. In this study, S. divinorum was extracted and spiked onto four different plant materials (S. divinorum, Salvia officinalis, Cannabis sativa, and Nicotiana tabacum) to simulate an adulterated sample that might be encountered in a forensic laboratory. The adulterated samples were extracted and analyzed by gas chromatography-mass spectrometry, and the resulting total ion chromatograms were subjected to a series of pretreatment procedures that were used to minimize non-chemical sources of variance in the data set. The data were then analyzed using principal components analysis (PCA) to investigate association of the adulterated extracts to unadulterated S. divinorum. While association was possible based on visual assessment of the PCA scores plot, additional procedures including Euclidean distance measurement, hierarchical cluster analysis, Student's t tests, Wilcoxon rank-sum tests, and Pearson product moment correlation were also applied to the PCA scores to provide a statistical evaluation of the association observed. The advantages and limitations of each statistical procedure in a forensic context were compared and are presented herein.