DNA barcodes successfully identified Macaronesian Lotus (Leguminosae) species within early diverged lineages of Cape Verde and mainland Africa

Plant DNA barcoding currently relies on the application of a two-locus combination, matK + rbcL. Despite the universality of these two gene regions across plants, it is suspected that this combination might not have sufficient variation to discriminate closely related species. In this study, we tested the performance of this two-locus plant barcode along with the additional plastid regions trnH-psbA, rpoC1 and rpoB and the nuclear region internal transcribed spacer (nrITS) in a group of 38 species of Lotus from the Macaronesian region. The group has radiated into the five archipelagos within this region from mid-Miocene to early Pleistocene, and thus provides both early divergent and recent radiations that pose a particularly difficult challenge for barcoding. The group also has 10 species considered under different levels of conservation concern. We found different levels of species discrimination depending on the age of the lineages. We obtained 100 % of the species identification from mainland Africa and Cape Verde when all six regions were combined. These lineages radiated >4.5 Mya; however, in the most recent radiations from the end of the Pliocene to the mid-Pleistocene (3.5-1.5 Mya), only 30 % of the species were identified. Of the regions examined, the intergenic region trnH-psbA was the most variable and had the greatest discriminatory power (18 %) of the plastid regions when analysed alone. The nrITS region was the best region when analysed alone with a discriminatory power of 26 % of the species. Overall, we identified 52 % of the species and 30 % of the endangered or threatened species within this group when all six regions were combined. Our results are consistent with those of other studies that indicate that additional approaches to barcoding will be needed in recently evolved groups, such as the inclusion of faster evolving regions from the nuclear genome.