Land-based classification of herb's origin based on supervised and unsupervised pattern recognition of plant and soil chemical profiling

摘要

In this work, a series of supervised and unsupervised pattern recognition techniques are assessed for the first time as a tool in order to elaborate a common approach able to classify different plant species based on a common pattern determined by their geographical origin. For this reason, soil samples and extracts from various parts of 7 self-sown indigenous Greek plants were collected in nature from two regions (highlands and insular) in NW Greece. Based on the chemical profiling of each species and the respective soil matrix composition, the identification of similarities and dissimilarities on a regional basis were pursued. The results indicate that hierarchical cluster classification and discriminant analysis using the chemical profile of inorganic elements and soil matrix composition provide the basis for the provision of an area-specific geographic marker that enables the classification of different herbs based on their origin with high discrimination efficiency.