Borrowed alleles and convergence in serpentine adaptation

Abstract

Serpentine barrens represent extreme hazards for plant colonists. These sites are characterized by high porosity leading to drought, lack of essential mineral nutrients, and phytotoxic levels of metals. Nevertheless, nature forged populations adapted to these challenges. Here, we use a population-based evolutionary genomic approach coupled with elemental profiling to assess how autotetraploid Arabidopsis arenosa adapted to a multichallenge serpentine habitat in the Austrian Alps. We first demonstrate that serpentine-adapted plants exhibit dramatically altered elemental accumulation levels in common conditions, and then resequence 24 autotetraploid individuals from three populations to perform a genome scan. We find evidence for highly localized selective sweeps that point to a polygenic, multitrait basis for serpentine adaptation. Comparing our results to a previous study of independent serpentine colonizations in the closely related diploid Arabidopsis lyrata in the United Kingdom and United States, we find the highest levels of differentiation in 11 of the same loci, providing candidate alleles for mediating convergent evolution. This overlap between independent colonizations in different species suggests that a limited number of evolutionary strategies are suited to overcome the multiple challenges of serpentine adaptation. Interestingly, we detect footprints of selection in A. arenosa in the context of substantial gene flow from nearby off-serpentine populations of A. arenosa, as well as from A. lyrata In several cases, quantitative tests of introgression indicate that some alleles exhibiting strong selective sweep signatures appear to have been introgressed from A. lyrata This finding suggests that migrant alleles may have facilitated adaptation of A. arenosa to this multihazard environment.

Keywords: adaptation; gene flow; plant; population genomics.

Fig. 1.

A. arenosa populations sampled for this study. (A) Locations of the 29 A. arenosa populations sampled. Orange dot gives the location of the focal Gulsen (GU) serpentine population along with other highlighted populations at Hochlantsch (HO) and Kasparstein (KA). Note: one Swedish location is not pictured (see SI Appendix, Table S1 for global positioning system locations). (B) PCA of A. arenosa range-wide showing relatedness between highlighted populations. (C) Lineage topology highlighting the major introgression events (green arrows), with MLEs for introgression in lineages per generation and MLEs for divergence times in generations.

Fig. 2.

Serpentine A. arenosa is an extreme outlier for the accumulation of many elements. Elemental profiling of 29 A. arenosa populations. Green distributions represent plant tissue data, brown distributions represent data from soil collected at plant sites. Orange dots indicate position in distribution where the serpentine autotetraploid Gulsen sample lies. (A) S, sulfur; K, potassium; Ca/Mg, calcium-to-magnesium ratio. (B) Ni, nickel; Cu, copper; Zn, zinc; Cd, cadmium. We normalized all values to the 0–1 range using feature scaling, where x′ = (x − x_min)/(x_max − x_min).

Fig. 3.

Measures of differentiation. (A) Watterson estimator θ_W diversity in resequenced populations over genome windows. The vertical dashed line for each population gives the mean. (B) θ_H, a diversity metric sensitive to extreme frequency SNPs (double asterisk signifies that Gulsen distribution is highly significantly different (P < 2.2e-16) from Hochlantsch or Kasparstein populations). (C) Mean number of fixed differences relative to Austrian A. lyrata in windows across the genome in each population (double asterisk signifies that Gulsen distribution is highly significantly different [P < 2.2e-16] from HO or KA populations). (D) D_xy, absolute net divergence between Gulsen and nonserpentine A. arenosa over genomic windows. (E) Relationship of diversity and differentiation in windows, indicating 0.1% empirical outliers in yellow. (F) DD residual values, indicating outliers with lower diversity for their given level of differentiation, a classic selective sweep signature. (G) F_ST distribution with outliers marked. (H) Overlap of outlier gene loci by all tests. (I) Positive f_d values from four taxon ABBA-BABA test with outliers marked and blue rug indicating each window value.

Fig. 4.

Selective sweeps on A. lyrata-like alleles in serpentine A. arenosa. (A) Allele frequency differences in example differentiated regions. Dots represent polymorphic SNPs. The x axis gives chromosome location; y axis gives degree of differentiation calculated by plotting the difference in allele frequencies between serpentine and nonserpentine populations. Arrows indicate gene models. Black arrow indicates sweep candidate with localized differentiation. (B) Linear plot showing the proportion of SNPs shared between the three pairwise population comparisons in the same region as in A. (C) Sequence similarity at the same regions among A. lyrata, Gulsen, and Kasparstein visualized using a color triangle. Areas where two rows show the same color (yellow) indicate localized high similarity specifically between Gulsen and A. lyrata, but not Kasparstein. (D) Genomic view of divergence and gene flow metrics at a postive ABBA-BABA outlier and top sweep candidate locus. D_xy gives net divergence, Div_diff, a selective sweep signature (relatively reduced diversity specifically in Gulsen vs. other A. arenosa; more negative values indicate specifically low diversity in Gulsen), f_d gives ABBA-BABA outlier status, ZengE_diff, negative values give localized negative excesses of rare variants in Gulsen (also see SI Appendix, Section S5). Dashed lines represent 1% outlier levels.