Objective Poisonous plants are a deadly threat to public health in China. The traditional clinical diagnosis of the toxic plants is inefficient, fallible, and dependent upon experts. In this study, we tested the performance of DNA barcodes for identification of the most threatening poisonous plants in China. Methods Seventy-four accessions of 27 toxic plant species in 22 genera and 17 families were sampled and three DNA barcodes （motK, rbcL, and ITS） were amplified, sequenced and tested. Three methods, Blast, pairwise global alignment （PWG） distance, and Tree-Building were tested for discrimination power. Results The primer universality of all the three markers was high. Except in the case of ITS for Hemerocollis minor, the three barcodes were successfully generated from all the selected species. Among the three methods applied, Blast showed the lowest discrimination rate, whereas PWG Distance and Tree-Building methods were equally effective. The ITS barcode showed highest discrimination rates using the PWG Distance and Tree-Building methods. When the barcodes were combined, discrimination rates were increased for the Blast method.
XIE LeiWANG Ying WeiGUAN Shan YueXIE Li JingLONG XinSUN Cheng Ye
