Aim Our aim was to reconstruct the spatio-temporal genetic diversification of Androsace lactea, a widely but disjunctly distributed European mountain plant, to test the hypothesis that its distribution is the result of vicariance, in the late Tertiary or during the Pleistocene, or alternatively of long-distance dispersal. We also addressed the phylogeographic history of the Alps, emphasizing the role of Pleistocene refugia at their northern margin.
Location The central and southern European mountain ranges.
Methods We gathered amplified fragment length polymorphism (AFLP) data and plastid DNA sequences from one to four individuals of each of 26 populations spanning the entire distribution area. AFLP data were analysed with Bayesian clustering approaches, neighbour-joining analysis and NeighbourNet. Plastid sequences were used to depict relationships among haplotypes in a statistical parsimony network, to test for population expansions, and to obtain age estimates in a Bayesian framework.
Results The AFLP data suggested that many populations were genetically strongly differentiated. The internal structure, however, was weak, and only two major groups of populations, from the north-western Alps and adjacent regions and from the easternmost Alps, were supported in the neighbour-joining analysis. One of the Bayesian clustering approaches differentiated three groups of populations: Northern Alps, easternmost Alps and the remaining distribution area. Eleven closely related plastid haplotypes were found, separated by maximally four mutational steps, resulting in a star-like parsimony network. None of several estimators suggested statistically significant population expansions. The diversification age was inferred to be (mean/median) 0.135/0.08 Ma (95% highest posterior density interval 0.364–0.006 Ma).
Main conclusions We found no evidence that long-distance dispersal shaped the disjunct distribution range; our data rather favoured a vicariance scenario. However, in contrast to the hypothesis that wide but disjunct distributions are old, we conclude that range fragmentation probably happened in the Late Pleistocene, perhaps during the last glaciation. In the Alps, most populations are at least close to formerly unglaciated areas. Our data support distributional stasis and suggest that important refugia were situated at the north-eastern, but also at the northern and north-western edges of the Alps, thereby strengthening the evidence for glacial refugia in this strongly glaciated region.