DNA barcoding ferns in an unexplored tropical montane cloud forest area of southeast Oaxaca, Mexico.

DNA barcoding can be useful for species identification and phylogenetic analysis, but its effectivity has not been verified in most neotropical cloud forest plants. We tested three plastid barcodes, rbcLa, matK, and trnH-psbA, in selected pteridophytes, a well-represented group in these forests, from a little-explored area in Oaxaca, Mexico, applying the CBOL criteria for barcoding. We used BLASTn, genetic distance, and monophyly tree-based analyses employing neighbor-joining (NJ), maximum likelihood (ML), and Bayesian inference methods. Universal primers for rbcLa and trnH-psbA were successfully amplified and bi-directionally sequenced, but matK could not be amplified for most species. rbcLa showed the highest species discrimination in BLASTn (66.67%). trnH-psbA exhibited higher significant interspecific divergence values than rbcL and rbcLa + trnH-psbA (two-sample sign test, P value < 2.2e-16). Using NJ and ML phylogenetic trees, monophyletic species were successfully resolved (100%), differing only in support values and displaying full agreement with the most recent fern classification. ML trees showed the highest mean support value (80.95%). trnH-psbA was the only barcode that could detect the Elaphoglossoideae subfamily. Species discrimination did not increase using rbcLa + trnH-psbA. rbcLa is useful for fern barcoding, trnH-psbA is most helpful for phylogenetic analyses, and matK may not work as a universal barcoding marker.