Maintenance of Species Differences in Closely Related Tetraploid Parasitic Euphrasia (Orobanchaceae) on an Isolated Island

Abstract

Polyploidy is pervasive in angiosperm evolution and plays important roles in adaptation and speciation. However, polyploid groups are understudied due to complex sequence homology, challenging genome assembly, and taxonomic complexity. Here, we study adaptive divergence in taxonomically complex eyebrights (Euphrasia), where recent divergence, phenotypic plasticity, and hybridization blur species boundaries. We focus on three closely related tetraploid species with contrasting ecological preferences that are sympatric on Fair Isle, a small isolated island in the British Isles. Using a common garden experiment, we show a genetic component to the morphological differences present between these species. Using whole-genome sequencing and a novel k-mer approach we call "Tetmer", we demonstrate that the species are of allopolyploid origin, with a sub-genome divergence of approximately 5%. Using ∼2 million SNPs, we show sub-genome homology across species, with a very low sequence divergence characteristic of recent speciation. This genetic variation is broadly structured by species, with clear divergence of Fair Isle heathland Euphrasia micrantha, while grassland Euphrasia arctica and coastal Euphrasia foulaensis are more closely related. Overall, we show that tetraploid Euphrasia is a system of allopolyploids of postglacial species divergence, where adaptation to novel environments may be conferred by old variants rearranged into new genetic lineages.

Keywords: allopolyploidy; divergence with gene flow; incipient speciation; k-mer spectrum; taxonomic complexity; tetraploid.

Figure 1

Three Focal Euphrasia Species and Their Distributions on Fair Isle. (A) Geographic distributions of E. arctica, E. foulaensis, and E. micrantha based on a field survey recording 282 presence (colored) or absence (white) census points. Labels indicate the populations from which seeds were sourced and traits were measured. (B–D) (B)E. arctica and (C)E. micrantha. The smaller form is typical for Fair Isle. (D)E. foulaensis.

Figure 2

Morphological Trait Differentiation between Three Species of Euphrasia in a Common Garden Experiment and in Natural Populations. (A) Significance levels of trait differences between species (left) and populations (right) from field measurements in natural populations (bottom right triangles) and in the common garden (top left triangles, see Supplemental Text 1 for the magnitudes of differences and Supplemental Table 2 for the means and standard errors). Comparisons within rows are corrected for multiple testing. Within columns, the color scale is the p value corrected for the number of traits tested (seven in natural populations and 14 in the common garden). While significant trait differences are rare between species, they are numerous between populations (see Figure 1A or Supplemental Table 1 for population codes). (B) PCA of trait measurements from natural populations shows separate clusters per species. (C) PCA of trait measurements from plants grown in the common garden shows little grouping by species. (D) LDA separates species in the common garden.

Figure 3

Evolutionary Relationships of British Euphrasia Plastid Genomes and rDNA Sequences. (A) Phylogenetic analysis of plastid genomes performed using a maximum-likelihood approach implemented in IQ-TREE with the K3Pu+F+I substitution model. Haplogroups are indicated by numbered vertical lines. (B) Phylogenetic analysis of nuclear rDNA sequences using a neighbor-joining approach implemented in Geneious. There were two rDNA haplotypes in sample F3. In both A and B, colored circles indicate species identity: orange, E. arctica; blue, E. foulaensis; gray, E. micrantha; black, putative hybrid individuals; and purple, diploid species. Scale bars indicate branch lengths. Numbers on branches indicate bootstrap support ≥70%.

Figure 4

Estimates of Heterozygosity and Sub-genome Divergence in Allotetraploids Based on k-mer Spectra. (A) Schematic of the shapes of k-mer spectra of diploids, autotetraploids, and allotetraploids. Spectra of low-diversity species are shown in blue and high-diversity species in red. A general feature is that the higher the genetic diversity, the higher the 1× peak. Our app, Tetmer, allows these models to be fitted to empirical k-mer spectra. (B) Heterozygosity estimates for Euphrasia individuals based on k-mers (dark bars) and SNPs (light bars). (C) Estimates of sub-genome divergence for tetraploid Euphrasia individuals.

Figure 5

Diploid–Tetraploid Scaffold Homology and the Clustering of Euphrasia Populations. (A) Relative mapping depth in the tetraploid (2× depth in tetraploids) and conserved (2× depth all individuals) scaffold sets. Colors represent mapping coverage (see inset). Tetraploid scaffolds not contained in the conserved set have low mapping depths in diploids, indicative of absence (red). (B) PCA of genomic data (conserved scaffolds) separates Fair Isle E. micrantha individuals from other Euphrasia. PC2 separates tetraploids and diploids. The analysis was based on 3454 SNPs, with one SNP per scaffold. (C) STRUCTURE analysis shows Fair Isle E. micrantha as a separate genetic cluster. Analysis based on the same SNP data as in (B). The plot shows the results for three genetic clusters (K = 3), with individuals on the x-axis and admixture proportions on the y-axis. The mainland tetraploids, M0 and A0, are inferred to be admixed. Colored outlines in B and dots in C represent taxa to match Figure 3.

Figure 6

Considerable Genetic Structure with Little Differentiation between Euphrasia Species. Histograms show per-scaffold statistics with population means indicated by dashed lines at the bottom of each graph. The top row shows results based on the tetraploid data set for non-hybrid individuals from Fair Isle, while the bottom row is based on the conserved set, and it includes all non-hybrid individuals (with all diploids treated as one group). (A) Nucleotide diversity on Fair Isle is slightly lower in E. micrantha than in E. arctica and E. foulaensis, and overall, these values are considerably higher than per-individual heterozygosity estimates (Figure 4B), consistent with high levels of selfing. (B) On Fair Isle, the net divergence shows very similar patterns in the comparisons involving E. micrantha, while divergence between E. arctica and E. foulaensis is much lower. With wider sampling, the divergence between E. micrantha and the other species is lower, an indication that mainland E. micrantha carries alleles shared with the other species. The net divergences estimated here are of similar magnitude or lower than the nucleotide diversities shown in (A). (C) Although net divergence tends to be low between species of Euphrasia, the fixation index can be extreme for some scaffolds, for example, Fair Isle comparisons including E. micrantha (arrow). This genetic differentiation disappears when samples from additional populations are included, indicating that allelic frequency divergence is greater at the population level than at the species level.

Figure 7

Complex Evolutionary Relationships and Extensive Discordance in Euphrasia. (A) ASTRAL consensus tree based on 3454 per-scaffold trees from the conserved scaffold set. The numbers are a node's posterior probability and its age in coalescent units. The tree is rooted at the longest branch (between Fair Isle E. micrantha and all other individuals). (B) Overlaid gene trees of the conserved scaffold set show no single clear species relationship. (C) Topological weighting of 3454 trees of all Fair Isle non-hybrids (FI) and all non-hybrid tetraploids (UK), both using the diploids as an outgroup, carried out with Twisst. While blue topology 1 tends to receive the highest weighting, few trees have a very high weight (near 1) for any one topology. The two alternative topologies receive similar levels of support. Colored dots represent taxa to match Figure 3.