Abstract: It is crucial for the studies of taxonomy and biodiversity by using DNA barcode technology to fast and accurately make species identification in the forests across tropics and temperate zones. In this study, the 183 plant species in a 20 hm² subtropical forest plot in Dinghushan (DHS) National Nature Reserve of South China were sampled and sequenced, and the matK, rbcL, and psbA-trnH were employed to generate multilocus barcodes. For the plot, the psbA-trnH possessed the highest integral success rate, i.e., the product of sequencing recovery and correct species identification (75%), followed by matK (70%), and rbcL (56%). A combination of three-locus barcode (matK,rbcL and psbA-trnH) could identify greater than 87% of the total species, followed by two-locus barcode (85% for matK+psbA-trnH, 83% forrbcL+psbA-trnH, and 81% formatK+rbcL). A comparison was made with the previously published results from one subtropical forest plot (LFDP in Puerto Rico, 143 species) and two tropical forest plots (BCI in Panama, 296 species; and NRS in French Guiana, 254 species) to evaluate the universality and species identification correctness of the proposed DNA barcodes for these four forest plots. For the plots in tropics and subtropics, the sequencing success rate ofrbcL,psbA-trnH and matK were 93% and 95.1%, 91.5% and 94.6%, and 68.5% and 79.7%, respectively. The combination ofmatK+rbcL showed a high identification capacity in geographically restricted regions  in taxonomic groups, whereas the three-locus barcode had a high rate of correct species identification both in tropics (84%) and in subtropics (90%).

Key words: DNA barcode technology, species identification, local community, biodiversity, tropical forest, subtropical forest.