Table S4: References for Figure 1B

Table S1: PCR primers used to amplify the ten microsatellite loci used to check paternity in the progeny. Markers are grouped according to 4 PCR-multiplexes (M1 to 4).

Table S2: Three couples of PCR primers used to amplify the MatK gene. Asterisks indicate primers described in Fénart et al. (2006)

Table S3: Detail of sampled subpopulations: location, estimated size (2008), sampling size and nomenclature according to the European Life project.

Table S4: References for Figure 1B.

TableS5: List of the 34 taxa used in the estimation of divergence between B. neustriaca MatK haplotypes with BEAST software.

Figure S1:Chloroplastic sequences alignment found in 80 Biscutella neustriaca individuals (length: 2495 pb). Only polymorphic sites are indicated and reported through dotted lines on a schematic map of the MatK gene and flanking regions. Bars (–) indicate when a nucleotide at a given position is the same than for reference sequence (Ia). Stars (*) indicate deletions as compared to the reference sequence. The 35pb insertion in haplotype II is indicated in grey. Black arrows represent positions of sequence fragments obtained from each primer used in this study.

FigureS2: a) Relationship between Nei's gene diversity (He) and estimates of subpopulation size in 2008; b) relationship between Allelic richness and estimates of subpopulation size. Values of gene diversity and allelic richness were estimated as averages over nine nuclear microsatellite loci. Estimated subpopulation sizes are represented in log scale. Each circle correspond to a subpopulation and is labelled with the name of the corresponding population and colored according to the group: north (white) and south (black). Correlations were tested by considering logarithm of population size (Spearman coefficient).

FigureS3: Determination of the number K of genetic clusters in B. neustriaca populations, according to the Δ(K) method developed by Evanno et al. (2005), represented for three Bayesian-based clustering analyses (STRUCTURE): global analysis (a), north analysis (b) and south analysis (c). Top: mean (circles) likelihood LnP(D) and standard deviation (boxplots) estimated over 20 runs, according to the number K of clusters assumed. Bottom: Δ(K) (log scale) according to K in the corresponding analysis. Standard deviations are indicated by boxplots. Black circles indicate K values with highly similar results obtained over 20 runs.

FigureS4: GlobalBayesian-based clustering analyses (STRUCTURE): the large variation in likelihood LnP(D) observed for K>3 mainly depend of the number of clusters found in southern group. Diamonds represent runs according to : one (white), two (grey) or three (black) clusters found in southern group. Unstable runs after 1,000,000 iterations of the MCMC chain are indicated by crosses.