The rapid conversion of Southeast Asian lowland rainforests into monocultures calls for the development of rapid methods for species identification to support ecological research and sustainable land-use management. Here, we investigated the utilization of DNA barcodes for identifying flowering plants from Sumatra, Indonesia. A total of 1,207 matK barcodes (441 species) and 2,376 rbcL barcodes (750 species) were successfully generated. The barcode effectiveness is assessed using four approaches: (a) comparison between morphological and molecular identification results, (b) best-close match analysis with TaxonDNA, (c) barcoding gap analysis, and (d) formation of monophyletic groups. Results show that rbcL has a much higher level of sequence recoverability than matK (95% and 66%). The comparison between morphological and molecular identifications revealed that matK and rbcL worked best assigning a plant specimen to the genus level. Estimates of identification success using best-close match analysis showed that >70% of the investigated species were correctly identified when using single barcode. The use of two-loci barcodes was able to increase the identification success up to 80%. The barcoding gap analysis revealed that neither matK nor rbcL succeeded to create a clear gap between the intraspecific and interspecific divergences. However, these two barcodes were able to discriminate at least 70% of the species from each other. Fifteen genera and twenty-one species were found to be nonmonophyletic with both markers. The two-loci barcodes were sufficient to reconstruct evolutionary relationships among the plant taxa in the study area that are congruent with the broadly accepted APG III phylogeny.
Keywords: DNA marker; EFForTS project; Sumatra; matK; molecular identification; rbcL.