Figure 4 Genetic structure and demographic modeling of Saxifraga facchinii. a) Sampled populations and their admixture proportions atK = 4 for derived from STRUCTURE analyses. Sizes of pie charts reflect the number of private alleles (PA). Numbers correspond to population identifiers. Coloured curves indicate groups of populations used for demographic modeling. Light-blue areas indicate the maximum extent of glaciers during the LGM 21 ka (based on Ehlers et al., 2011, Van Husen 1987). The map was generated with UTM 32 N projection, with datum WGS84. b) STRUCTURE bar plots depicting the cluster memberships of each individual at the two best K s (2 and 5). c) Neighbor-Nets with populations colour-coded according to the most likely clustering solution in STRUCTURE analyses at K = 4. Populations with <70% assignment to a group are indicated with a black dotted line. d) Two- and three population demographic modeling results; color coding of population groups match those presented in panel a). The left column shows a graphical representation of the selected two-population models in δaδi; “2nd best model” indicates that the depicted model does not correspond to the best model according to the Akaike Information Criterion (see the main text for an explanation). The right column shows the distribution of the estimated divergence time after the goodness-of-fit test with 100 replicates. The histogram represents the distribution, the blue curve shows the density distribution, and the red line represents the LGM (20 ka). In case of three-population models, the goodness-of-fit test could not be performed, instead a graphical representation of the second-best three-population model is shown.