Extensive genetic differentiation between recently evolved sympatric Arctic charr morphs

Abstract

The availability of diverse ecological niches can promote adaptation of trophic specializations and related traits, as has been repeatedly observed in evolutionary radiations of freshwater fish. The role of genetics, environment, and history in ecologically driven divergence and adaptation, can be studied on adaptive radiations or populations showing ecological polymorphism. Salmonids, especially the Salvelinus genus, are renowned for both phenotypic diversity and polymorphism. Arctic charr (Salvelinus alpinus) invaded Icelandic streams during the glacial retreat (about 10,000 years ago) and exhibits many instances of sympatric polymorphism. Particularly, well studied are the four morphs in Lake Þingvallavatn in Iceland. The small benthic (SB), large benthic (LB), planktivorous (PL), and piscivorous (PI) charr differ in many regards, including size, form, and life history traits. To investigate relatedness and genomic differentiation between morphs, we identified variable sites from RNA-sequencing data from three of those morphs and verified 22 variants in population samples. The data reveal genetic differences between the morphs, with the two benthic morphs being more similar and the PL-charr more genetically different. The markers with high differentiation map to all linkage groups, suggesting ancient and pervasive genetic separation of these three morphs. Furthermore, GO analyses suggest differences in collagen metabolism, odontogenesis, and sensory systems between PL-charr and the benthic morphs. Genotyping in population samples from all four morphs confirms the genetic separation and indicates that the PI-charr are less genetically distinct than the other three morphs. The genetic separation of the other three morphs indicates certain degree of reproductive isolation. The extent of gene flow between the morphs and the nature of reproductive barriers between them remain to be elucidated.

Keywords: Lake Thingvallavatn; RNA‐Seq; Salvelinus alpinus; genetic variability; pooled sequencing.

Figure 1

The phenotypically distinct sympatric Arctic charr from Lake Þingvallavatn and the sampling strategy. (a) The four sympatric morphs are Small benthic (SB), Large benthic (LB), Planktivorous (PL), and Piscivorous (PI) charr. They differ in size (size bars = 5 cm), head and feeding morphology and pigmentation. Adapted from Sandlund et al. (1992) ©Wiley‐Blackwell, drawings by Eggert Pétursson. (b) Sampling of charr for transcriptome sequencing (circles) and genotyping (squares) of population samples. The top 3 morphs were mined for genetic variation in the transcriptome and population samples were studied from all four, to confirm genetic variants. The transcriptome samples came from embryos at six developmental stages prior to hatching, from 100τs to 200τs, in the three morphs (circles) (Guðbrandsson et al., 2018). The sampling of each morph and developmental timepoint combination was replicated three times (biological replicates), each sample is a pool of mRNA from three embryos. Six timepoints were sampled of SB‐charr, and five of LB‐ and PL‐charr embryos. The population samples (squares) were obtained by gill netting on the spawning grounds, see Section 2. The morph coloring scheme (SB: blue, LB: green, PL: red, and PI: purple) will be retained throughout the article

Figure 2

(a) Genetic separation of samples from three sympatric Arctic charr morphs, based on principal component analysis of 19,252 transcriptome variants. The first and second principal components are depicted with the proportion of variance explained. Colors and shapes indicate which morphs (SB: blue, LB: green and PL: red) and crosses the embryos originated from (see legend). Note, while the samples from the three SB crosses do not overlap entirely, then they are very distinct from the PL and LB specimens. Overlaid are 68% normal data ellipses for each morph. (b) Separation of morphs based on the 2,331 variants with F _ST > 0.2. Each column represents a sample, name below indicates morph, developmental timepoint, and biological replicate. The white‐blue scale depicts allele frequency, higher frequency of the alternative allele with darker blue. Hierarchical clustering grouped samples and variants. This separated the morphs (abscissa) and similar variants by allele frequency (ordinate). A total of 1,174 variants had a higher frequency in PL‐charr, 552 in LB‐, and 605 in SB‐charr. Missing values are indicated by pink. Coloring of individuals by morph, SB: blue, LB: green and PL: red

Figure 3

(a) F _ST values plotted by position of variants on the Arctic charr genome from Christensen, Rondeau, et al. (2018). The colors indicate which morph differs most strongly in allele frequency from the other two for variants with F _ST above 0.2 (SB: blue, LB: green, and PL: red) and gray represents variants with F _ST below 0.2. (b) Proportion of each variant group on each linkage group (MG: mitochondrial chromosome). Unplaced scaffolds and contigs are represented by “linkage” group 41 and in (b) NA refers to unmapped markers

Figure 4

(a) Genetic separation of the four morphs sympatric charr morphs depicted with principal component analyses of 22 variants genotyped in population samples. Individuals are graphed according to scores of the first two PC's, along with 68% normal data ellipses for each morph (SB: blue, LB: green, PL: red and PI: purple). (b) Heatmap of genetic variation in population samples of four morphs, depicting association of variants with morphology and linkage of markers. Genotypes of the 22 variants were clustered by genes. F _ST values (shades of red) for morphs are graphed for each marker. The morphs are color coded as in (a), and genotypes, homozygous reference allele (white), heterozygous (light‐blue), and homozygous alternate allele (blue), with pink indicating missing data

Figure 5

Detailed view of chromosomal regions and variants showing high divergence in the KASP assay. (a) Region on LG4q.2 that includes calm1 and many other variants with high F _ST values (not validated). (b) A large region on LG26 that includes Kiaa1324 and eif4g2b. (c) A region on LG11 that includes msi1 and tcf15. (d) A region on LG18 that includes lrcc1, cox11 and adk. (e) A region on LG10 that includes dennd5a, wee1. The location of the second best blast hit for the variant in gas1 is shown with a vertical line. (f) A region on LG15 that includes tmem9b and gas1l. The location of the best blast hit for the variant in gas1 is shown with a vertical line. The regions on LG10 and LG15 shown in (e) and (f) are ohnologous and some of the markers might belong in the opposite linkage group as we suspect is the case for gas1l. Colored dots are values from the transcriptome as in Figure 3 and validated variants are marked with a black circle (○). The triangles (▽) show the F _ST value from the KASP assay for three morphs (PL, SB, and LB) and the diamonds (♢) F _ST for all morphs (including PI)