Young inversion with multiple linked QTLs under selection in a hybrid zone

Abstract

Fixed chromosomal inversions can reduce gene flow and promote speciation in two ways: by suppressing recombination and by carrying locally favoured alleles at multiple loci. However, it is unknown whether favoured mutations slowly accumulate on older inversions or if young inversions spread because they capture pre-existing adaptive quantitative trait loci (QTLs). By genetic mapping, chromosome painting and genome sequencing, we have identified a major inversion controlling ecologically important traits in Boechera stricta. The inversion arose since the last glaciation and subsequently reached local high frequency in a hybrid speciation zone. Furthermore, the inversion shows signs of positive directional selection. To test whether the inversion could have captured existing, linked QTLs, we crossed standard, collinear haplotypes from the hybrid zone and found multiple linked phenology QTLs within the inversion region. These findings provide the first direct evidence that linked, locally adapted QTLs may be captured by young inversions during incipient speciation.

Fig. 1. Comparative Chromosome Painting of chromosome 1 in Bst1-std and Bst1-inv haplotypes

Shown are in situ chromosomal localization of painting probes on B. stricta pachytene chromosomes, their straightened images and diagrammatic representations of the Bst1-std and Bst1-inv haplotypes. Differential labeling of BAC contigs (F21M11, T21E18, T6B12, F27J15) identifies the breakpoints and extent of the paracentric inversion in the Bsi1-inv genotype. Scale bars, 10 μm. See Supplementary Fig. 3 for a detailed cytogenetic analysis.

Fig. 2. Inversion region and PCR genotyping

(a) Chromosome blocks (yellow and red) are shown for the derived (Bst1-inv, upper) and ancestral (Bst1-std, lower) haplotypes. Centromeres are indicated by white circles. Scaffolds (blue, above and below) are labeled with their name and size. Breakpoints are shown as vertical red lines within the scaffolds. Arrows indicate PCR primers. There are 1,591 annotated genes in the inversion region. (b) Gel image showing PCR products, allowing codominant identification of inversion genotypes.

Figure 3. Geographic locations of subspecies collections and inversion zone genotypes. Geographic locations of genotyped accessions used in this study

(a) Species-wide samples, with each dot indicating one genotype from a population. Subspecies assignments show EAST (blue) and WEST (red). Ellipse indicates the contact zone where subspecies overlap, and the star shows the inversion zone. N = 83. The collinear cross comes from within the hybrid zone (oval), outside the inversion zone (star). (b) Close-up of populations near the inversion zone (the star in Fig. 3a), with pie-diagrams indicating the inversion frequency (red) in each population, sized in proportion to sample size. Forested, unsuitable habitat likely limits gene flow. N = 122 genotypes. Individual populations range in size from 1 to 19.

Figure 4. Population genetic variation for inversion and standard groups

(a, b) Nucleotide diversity (π) and Tajima’s D of inverted genomic region (yellow), block D of chromosome 1 (red) and chromosome 2–7 (blue) in inversion (INV) and standard (STD) genotypes. Block D (south end of LG1) is treated separately because it has unusually high polymorphism in related species. In these box plots, the median is shown by a horizontal line, while the bottom and top of each box represents the first and third quartiles. The whiskers extend to 1.5 times the interquartile range. Outliers are represented by black dots. (c) Distribution of population genetic statistics along chromosome 1. Nucleotide diversity (π) in INV (red) and STD (blue) genotypes, with genome-wide averages as dashed horizontal lines. Linkage disequilibrium (R²) between the inversion and SNPs is shown in all INV and STD genotypes from the inversion zone, and the relationship between physical and linkage maps. The dashed vertical lines mark the inverted (golden) and block D (light blue) regions. N = 122, except four admixed individuals were excluded from LD analysis. LD for 10 comparator SNPs with derived allele frequencies similar to Bsi1-inv is shown in Supplementary Fig. 8b.

Figure 5. QTLs within the inversion

A collinear cross shows that several QTLs in the inversion region influence phenology and development traits. Inversion breakpoints are indicated by vertical lines. (a) Multivariate QTL mapping finds several QTL peaks (blue line) exceeding the significance threshold (horizontal dashed line); N = 1,714 F4 individuals in 153 families. (b) Testing the hypothesis that three linked QTLs have different pleiotropic effects. Plots for three QTLs (left, center, and right, shown in solid red, dotted blue, and dashed gray, respectively) quantify evidence that a locus with different patterns of pleiotropy occurs at each peak. Discriminant Function Analysis was used to identify new composite trait axes defined at each peak, and evidence for these composite traits was mapped across the inversion region. The left and center QTLs show little overlap, suggesting different patterns of pleiotropy. (c) Comparison of molecular divergence and time to coalescence of the East and West genotypes in the collinear QTL mapping cross. The horizontal red dashed line at 8.8 Kya is the upper confidence interval for age of the inversion. The vertical axes show nucleotide divergence (Dxy, left axis), and coalescence time (Kya, right axis). The blue dashed line indicates the mean genome-wide values of Dxy and coalescence time between these East and West alleles. The horizontal axis shows position across the inversion region in Mb, beginning at marker Scaf26675_2450000. Vertical blue shading indicates the QTL regions, +/− logP > 1.6 confidence intervals, with 408 annotated genes in these QTL regions. Divergence (Dxy) between these genotypes was calculated in 200kb non-overlapping windows, and the coalescence time was estimated using T = Dxy/2μ, where μ is 7E-9 per site per generation. Mean Dxy in the three QTLs (left to right) are 0.00206, 0.00169 and 0.00186, corresponding to coalescent times of 147 kya, 121 kya and 132 kya, respectively. The mean Dxy in the whole inversion region is 0.00191, and the average coalescence time is 136 kya.