Figure 2. Linkage to the S-locus locally distorts the phylogenetic relationships.
(A) The two topologies of interest cluster haplotypes either by the S-allele to which they are linked (top) or by the populations where they came (bottom). Different S-alleles are represented by symbols of different colours, different populations of origin are represented by symbols of different shapes. (B) Difference in log likelihood of the two topologies of interest. Dots correspond to the difference in log likelihood for overlapping series of 50 SNPs around the S-locus for A. halleri (top panel) and A. lyrata (bottom panel). Positive values correspond to chromosomal positions where the topology by S-alleles explains the phylogeny of haplotypes better than the topology by populations. The right panels show the difference in log likelihood in the control regions. 2.5 and 97.5 percentiles of the distribution in the control regions are indicated by dashed lines.
Figure 2—figure supplement 1. Analysis of major components obtained for haplotypes of A. halleri of the Nivelle (black dots) and Mortagne (grey dots) populations based on SNPs in the first 5 kb, between 5 and 25 kb, and between 25 kb and 50 kb away from the S-locus.
The S-allele to which the SNPs are linked are represented by different symbols. The right panels show the analysis on control regions, each time matching the number of SNPs with that of the corresponding linked regions (left panels).
Figure 2—figure supplement 2. Analysis of major components (AMC) obtained for haplotypes of A. lyrata (of the PIN: grey dots, IND: red dots; and TSS: blue dots) populations based on the SNPs in the first 5 kb, between 5 and 25 kb, and between 25 kb and 50 kb away from the S-locus.
The S-allele to which the SNPs are linked are represented by different symbols. The right panels show the analysis on control regions, each time matching the number of SNPs with that of the corresponding linked regions (left panels).
Figure 2—figure supplement 3. Phylogenetic tree obtained by maximum likelihood for haplotypes of A. halleri (populations Nivelle and Mortagne) across the first 25 kb flanking the S-locus.
The Tamura-Nei model was used and the percentage of trees in which the associated haplotypes clustered together is shown next to the branches. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The black braces indicate haplotypes clustering by populations. The overall pattern shows a structure by S-alleles, with exceptions highlighted in yellow.
Figure 2—figure supplement 4. Phylogenetic tree obtained by maximum likelihood for haplotypes of A. halleri (populations Nivelle and Mortagne) based on the nucleotide positions between 25 kb and 50 kb away from the S-locus.
The Tamura-Nei model was used and the percentage of trees in which the associated haplotypes clustered together is shown next to the branches. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The black braces indicate haplotypes clustering by populations. The overall pattern shows a structure by S-alleles, with exceptions highlighted in yellow.
Figure 2—figure supplement 5. Phylogenetic tree obtained by maximum likelihood for haplotypes of A. lyrata (populations PIN, IND, TSS) across the first 5 kb flanking the S-locus.
The Tamura-Nei model was used and the percentage of trees in which the associated haplotypes clustered together is shown next to the branches. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The black braces indicate haplotypes clustering by populations. The overall pattern shows a structure by S-alleles, with exceptions highlighted in yellow.
Figure 2—figure supplement 6. Phylogenetic tree obtained by maximum likelihood for haplotypes of A. lyrata (populations PIN, IND, TSS) based on the nucleotide positions between 5 kb and 10 kb away from the S-locus.
The Tamura-Nei model was used and the percentage of trees in which the associated haplotypes clustered together is shown next to the branches. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The black braces indicate haplotypes clustering by populations. The overall pattern shows a structure by S-alleles, with exceptions highlighted in yellow.
Figure 2—figure supplement 7. The genetic structure of SNPs in the S-locus flanking regions in A. halleri and A. lyrata.
Left: mean FST (lines) and FST by pair (point) analysed among S-alleles (grey) or among populations (black) in 5 kb windows in the S-locus flanking regions. Right: distribution (count) of FST in the control regions analysed among S-alleles (grey bars) or among populations (black bars). The 95% percentiles of the distributions are represented by dotted lines and the medians by solid lines.
Figure 2—figure supplement 8. Patterns of genetic associations between S-alleles and SNPs across the genome.
Each dot corresponds to an SNP showing statistically significant association (top 0.1%) with a given S-allele. The left panel confirms that SNPs physically linked to the S-locus (in red) are considerably more likely to show statistical associations with S-alleles. The chromosomes are signified by alternance of black and grey. The middle panel shows a zoom on the 50 kb regions flanking the S-locus and shows that the statistical association extends over long distances, at least for some S-alleles. Each region of 25 kb is delimited by vertical lines. The right panel shows that the observed number of associated SNPs in the linked regions far exceeds that in regions of identical size from control regions. The histogram shows the distribution across control regions of the mean number of significantly associated SNPs per S-allele. The vertical lines correspond to the mean number of significantly associated SNPs in the first 25 kb (solid lines) and the 25–50 kb interval (dashed lines) away from S-locus.