There are two long‐standing biogeographic hypotheses regarding the glacial survival of plant species in the Qinghai–Tibetan Plateau (QTP): the in situ survival hypothesis and the tabula rasa hypothesis. We tested these two hypotheses in a phylogeographic study of Rhodiola sect. Prainia , a monophyletic section with ecologically divergent lineages. Molecular data from the nuclear internal transcribed spacer, six plastid markers and 13 nuclear microsatellite loci were analyzed for 240 individuals from 19 populations of this section. Environmental data were used to analyze the niches of major phylogenetic lineages within this section and to model changes in their distributions since the Last Glacial Maximum (LGM). We found that Rhodiola sect. Prainia consists of three evolutionary lineages: all populations of R. stapfii , R. prainii populations at the southern edge of the QTP, and R. prainii populations in the interior part of the QTP. During the LGM, the survival of R. prainii in the interior part of the QTP corresponded with the in situ survival hypothesis, while R. stapfii most probably survived the LGM in a manner corresponding with the tabula rasa hypothesis. The evolutionary history of different lineages of this section was shaped by topography, climate change, and lineage‐specific habitat preferences.

A ring species consists of two reproductively isolated forms connected by a chain of intergrading populations encircling a geographic barrier. The mountains encircling the Sichuan Basin in southwestern China harbor great species diversity and endemism, and they are candidate regions for ring species. Here, we examined a potential ring species complex with a ring distribution surrounding the Sichuan Basin and the reproductive barrier between sibling species Rhodiola yunnanensis and R. henryi of the R. yunnanensis complex. In this study, we test the hypothesis that R. yunnanensis and R. henryi diverged by the ring‐species model using an amplicon sequencing strategy targeting the introns of 27 single‐copy nuclear genes and 14 chloroplast DNA sequences. Our studies indicated that the R. yunnanensis complex is monophyletic, originates at the late Miocene, and forms its current ring distribution pattern after the LIG. In addition, clear genetic intergradation was not found among R. henryi populations within the distribution ring. All these findings suggest that the divergence of two sibling species was not driven by the geographic isolation, and they were not originated from the ring‐species model; however, the basin‐surrounding distribution pattern and reproductive barrier between them meet some criteria for being a ring species.