Abstract
Various hypotheses have been proposed about the Quaternary evolutionary history of plant species on the Qinghai-Tibet Plateau (QTP), yet only a handful of studies have considered both population genetics and ecological niche context. In this study, we proposed and compared climate refugia hypotheses based on the phylogeographic pattern of Anisodus tanguticus (three plastid DNA fragments and nuclear internal transcribed spacer regions from 32 populations) and present and past species distribution models (SDMs). We detected six plastid haplotypes in two well-differentiated lineages. Although all haplotypes could be found in its western (sampling) area, only haplotypes from one lineage occurred in its eastern area. Meanwhile, most genetic variations existed between populations (F ST = 0.822). The SDMs during the last glacial maximum and last interglacial periods showed range fragmentation in the western area and significant range contraction in the eastern area, respectively, in comparison with current potential distribution. This species may have undergone intraspecific divergence during the early Quaternary, which may have been caused by survival in different refugia during the earliest known glacial in the QTP, rather than geological isolation due to orogenesis events. Subsequently, climate oscillations during the Quaternary resulted in a dynamic distribution range for this species as well as the distribution pattern of its plastid haplotypes and nuclear genotypes. The interglacial periods may have had a greater effect on A. tanguticus than the glacial periods. Most importantly, neither genetic data nor SDM alone can fully reveal the climate refugia history of this species. We also discuss the conservation implications for this important Tibetan folk medicine plant in light of these findings and SDMs under future climate models. Together, our results underline the necessity to combine phylogeographic and SDM approaches in future investigations of the Quaternary evolutionary history of species in topographically complex areas, such as the QTP.