A character-based approach in the Mexican cycads supports diverse multigene combinations for DNA barcoding

A DNA barcoding study was conducted to determine the optimal combination of loci needed for successful species-level molecular identification in three extant cycad genera-Ceratozamia, Dioon, and Zamia-that occur in Mexico. Based on conclusions of a previous multigene study in representative species of all genera in the Cycadales, we tested the DNA barcoding performance of seven chloroplast coding (matK, rpoB, rpoC1, and rbcL) and non-coding (atpF/H, psbK/I, and trnH-psbA) regions, plus sequences of the nuclear internal transcribed spacer. We analysed data under the assumptions of the "character attributes organization system" (CAOS), a character-based approach in which species are identified through the presence of 'DNA diagnostics'. In Ceratozamia, four chloroplast regions and one nuclear region were needed to achieve > 70% unique species identification. In contrast, the two-gene combination atpF/H + psbK/I and the four-gene combination atpF/H + psbK/I + rpoC1 + ITS2 were needed to reach 79% and 75% unique species identification in Dioon and Zamia, respectively. The combinations atpF/H + psbK/I and atpF/H + psbK/I + rpoC1 + ITS2 include loci previously considered by the international DNA barcoding community. However, none of the three combinations of potential DNA barcoding loci found to be optimal with a character-based approach in the Mexican cycads coincides with the 'core barcode' of chloroplast markers (matK + rbcL) recently proposed for universal use in the plant kingdom.