Abstract
BACKGROUND AND AIMS: Hosting several global biodiversity hotspots, the region of the Qinghai-Tibetan Plateau (QTP) is exceptionally species-rich and harbours a remarkable level of endemism. Yet, despite a growing number of studies, factors fostering divergence, speciation and ultimately diversity remain poorly understood for QTP alpine plants. This is particularly the case for the role of hybridization. Here, we explored the evolutionary history of three closely related Gentiana endemic species, and tested whether our results supported the mountain geo-biodiversity hypothesis (MGH). METHODS: We genotyped 69 populations across the QTP with one chloroplast marker and 12 nuclear microsatellite loci. We performed phylogeographical analysis, Bayesian clustering, approximate Bayesian computation and principal components analysis to explore their genetic relationship and evolutionary history. In addition, we modelled their distribution under different climates. KEY RESULTS: Each species was composed of two geographically distinct clades, corresponding to the south-eastern and north-western parts of their distribution. Thus Gentiana veitchiorum and G. lawrencei var. farreri, which diverged recently, appear to have shared at least refugia in the past, from which their range expanded later on. Indeed, climatic niche modelling showed that both species went through continuous expansion from the Last Interglacial Maximum to the present day. Moreover, we have evidence of hybridization in the northwest clade of G. lawrencei var. farreri, which probably occurred in the refugium located on the plateau platform. Furthermore, phylogenetic and population genetic analyses suggested that G. dolichocalyx should be a geographically limited distinct species with low genetic differentiation from G. lawrencei var. farreri. CONCLUSIONS: Climatic fluctuations in the region of the QTP have played an important role in shaping the current genetic structure of G. lawrencei var. farreri and G. veitchiorum. We argue that a species pump effect did occur prior to the Last Interglacial Maximum, thus lending support to the MGH. However, our results do depart from expectations as suggested in the MGH for more recent distribution range and hybridization dynamics.