Abstract: We compared the performances of the candidate loci for moss DNA barcoding and the primers used in amplifying the loci. Primers for three coded loci (matK, rps4 and rbcLa) and four noncoded loci (atpBrbcL, atpFH, psbKI and trnHpsbA) of the chloroplast genome, one from the mitochondrial genome (nad5), and one from the nucleus genome (ITS2) were evaluated. Seventyfour samples representing 14 species belonging to five genera of Trachypodoaceae (or Meteoriaceae) were screened. All primers for matK and a pair of primers for trnHpsbA failed. Low successes were encountered with the primers for atpFH and psbKI. The primers for psbKI produced several bands and the PCR products of atpFH were difficult to sequence. The powers of the remaining six loci were compared using the variability, identification success and the resolutions. It was found that ITS2 is the most promising candidate for DNA barcoding for mosses. Among the chloroplast genes, atpBrbcL exhibited the highest resolution. Although trnHpsbA is very variable, it is too short to be an ideal barcode alone. Combinations of chloroplast genes were also tried and Ps of both atpBrbcL+trnHpsbA and rbcLa++trnHpsbA were 64% using NJ method. More additions of loci did not increase the resolution. No barcoding gap exists for all these loci. Phylogenetic analyses were carried out prior to the DNA barcoding evaluation and some taxonomic problems do exist. This study exemplifies the necessity of correct species delimitation and the adoption of both plastid and nuclear loci in plant DNA barcoding.

Key words: DNA barcode