Wei WANG, Hong-Lei LI, Xiao-Guo XIANG, Zhi-Duan CHEN
2014, 52 (5): 551–565
Previous studies based on different molecular datasets have generated conflicting topologies for Ranunculeae. Here, we revisit the phylogeny of Ranunculeae by analyzing the individual matK/trnK, psbJ-petA, and internal transcribed spacer (ITS) data, the combined matK/trnK, psbJ-petA, and ITS dataset, and the combinedrbcL, trnL-F, matK/trnK, psbJ-petA, and ITS dataset. Based on the tree-based comparisons, blast searches against NCBI of the sequences, and close examination of the alignment, we found that 10 psbJ-petA sequences previously used were questionable (erroneous or problematic) and responsible for previous conflicting topologies. After omitting these questionable sequences, we provide a new phylogeny for Ranunculeae, in which Beckwithia–Cyrtorhyncha, Kumlienia, andPeltocalathos were replaced. These new replacements are supported by corresponding morphological characters. Moreover, three previously proposed intercontinental disjunct distributions within Ranunculus were also refuted. In our framework, our divergence time and biogeographic analyses indicate that divergence time estimates and the ancestral areas reconstructed for 10 of the 15 nodes in the genus-level phylogeny were influenced by elimination of the questionable sequences. The most recent common ancestor of Ranunculeae was inferred to be present in Europe and North America during the late Eocene. Clades I and II began to diversify in Europe and North America, respectively, and subsequently migrated to other continents. This study shows that it is necessary to analyze individual chloroplast DNA region datasets separately to detect questionable sequences early in the study. The combined dataset including the questionable sequences resulted in an erroneous phylogenetic tree, and the use of this tree subsequently affected age estimates and biogeographic analyses.