Breaking Free: The Genomics of Allopolyploidy-Facilitated Niche Expansion in White Clover

Abstract

The merging of distinct genomes, allopolyploidization, is a widespread phenomenon in plants. It generates adaptive potential through increased genetic diversity, but examples demonstrating its exploitation remain scarce. White clover (Trifolium repens) is a ubiquitous temperate allotetraploid forage crop derived from two European diploid progenitors confined to extreme coastal or alpine habitats. We sequenced and assembled the genomes and transcriptomes of this species complex to gain insight into the genesis of white clover and the consequences of allopolyploidization. Based on these data, we estimate that white clover originated ∼15,000 to 28,000 years ago during the last glaciation when alpine and coastal progenitors were likely colocated in glacial refugia. We found evidence of progenitor diversity carryover through multiple hybridization events and show that the progenitor subgenomes have retained integrity and gene expression activity as they traveled within white clover from their original confined habitats to a global presence. At the transcriptional level, we observed remarkably stable subgenome expression ratios across tissues. Among the few genes that show tissue-specific switching between homeologous gene copies, we found flavonoid biosynthesis genes strongly overrepresented, suggesting an adaptive role of some allopolyploidy-associated transcriptional changes. Our results highlight white clover as an example of allopolyploidy-facilitated niche expansion, where two progenitor genomes, adapted and confined to disparate and highly specialized habitats, expanded to a ubiquitous global presence after glaciation-associated allopolyploidization.

Figure 1.

The Range of White Clover and Extant Relatives of its Progenitors. The present-day ranges of white clover (T. repens, Daday, 1958; green), and extant relatives of its diploid progenitors T. occidentale (blue) and T. pallescens (orange). T. occidentale is found within ∼100 m of the seashore while T. pallescens grows in alpine regions between 1,800 and 2,700 m.

Figure 2.

White Clover Genetic Map and Synteny with Extant Progenitor and Model Forage Legume Genomes. (A) Genetic map based on LD analysis of ∼7,300 scaffold-mapped markers from a F₁ biparental mapping population (n = 93). Red color indicates high level of pairwise LD (r²) between markers. Black dots indicates SLHS microsatellite markers from an existing, colinear genetic linkage map (Griffiths et al., 2013; Supplemental Figure 2). Progenitor origin of LG homeologs/subgenomes is denoted by “O” (Trifolium occidentale) and “P” (T. pallescens). (B) Circos (circos.ca) diagram showing inter-pseudomolecule relationships between the subgenomes of white clover (Tr_To; Tr_Tp) and their progenitors (To; Tp) in these assemblies. The outer ring (green-shaded) represents pseudomolecules in megabases (Mb), and the inner ring (blue) depicts gene density as a proportion along the pseudomolecules (%). Colored lines represent synteny blocks constructed by whole genome alignment using the program LASTZ (Harris, 2007) comprising matches >33 kb in length. Blocks within 100-kb windows were merged and represented by a single line. Cross-progenitor links indicate regions of high conservation between the subgenome and both progenitors, not putative recombination events. (C) A matrix plot assessment of synteny between white clover subgenomes (“O” and “P”) and the reference genome of M. truncatula, a model forage legume with eight pseudomolecules (Mt 1 to Mt 8). Synteny was based on alignment of the 3,364 LD-ordered white clover anchor scaffolds with M. truncatula genome Mt4.0 (Tang et al., 2014).

Figure 3.

The White Clover Allopolyploidization Event. (A) Schematic phylogenetic tree for white clover (Tr) and its progenitors T. occidentale (To) and T. pallescens (Tp). A common ancestor, “A,” gave rise to To and Tp (gray line), which hybridized (Hyb, blue line) to give rise to the two subgenomes Tr_To and Tr_Tp in allopolyploid white clover. (B) Blue dots indicate split time estimates based on sets of 100 alignment blocks derived from the pairwise genome comparisons indicated, e.g. To versus Tp. Dots are superimposed on box-and-whiskers plots, where the median is indicated by a black vertical line. Using a mutation rate of 1.8 × 10⁻⁸, split time estimates suggest To and Tp diverged from a common ancestor ∼192 Kya, whereas white clover Tr_To and Tr_Tp subgenomes split from their extant progenitors 15 Kya and 17 Kya, respectively, suggesting genesis of white clover ∼16 Kya. (C) White clover progenitor speciation (gray block) and hybridization giving rise to white clover (blue block) aligned with global temperature variation relative to present-day average temperature based on ice-core data (Jouzel et al., 2007). The blocks indicate the extent of possible divergence times using mutation rates in the range 1.1–1.8 × 10⁻⁸ (Supplemental Table 9). (D) PSMC curve based on whole-genome resequencing data from each of four white clover individuals. Each curve indicates inferred population size history through time. The analysis depicted here was performed using a mutation rate of 1.8 × 10⁻⁸. See Supplemental Table 11 for results with lower mutation rates. (E) MSMC figure based on the same data as the PSMC figure. The number of haplotypes corresponds to the number of individuals included. Eight haplotypes comprise all four. Six haplotypes comprise individuals 81, 122, and 183—all of the individuals with the most pronounced bottlenecks. Four of the haplotypes have three combinations of two individuals: [1] includes 81 and 122; [2] includes 122 and 183; [3] includes 81 and 183. The results were scaled to a mutation rate of 1.8 × 10⁻⁸. See Supplemental Figure 10 for MSMC using different mutation rates. (F) SFS of simulations across 20k generations under the different demographic scenarios indicated on the right (Simulated Effective Population Size (EPS). The observed SFS is scaled to match total polymorphism count for the simulations. Dashed line represents the expected SFS under a constant EPS. Green numbers indicate simulated SNP densities (%) and goodness of fit between observed and simulated data. The goodness of fit was calculated by first dividing the simulated and observed values for each bin with the maximum value of the two and then using the formula: Σ(observed-simulated)²/simulated. Lower values indicate better fit. The observed SNP density was 7.0%.

Figure 4.

Homeolog-Specific Expression Analysis. (A) Genes classified based on the number of tissues in which the majority of reads were derived from the Tr_To or the Tr_Tp homeolog. The numbers underneath the graph indicate the number of tissues for which the statement on the left is true. Most of the genes (69%) show the same direction of bias across all four tissues. (B) Log(Tr_To/Tr_Tp) expression ratios in flowers versus leaves. (C) Spearman correlation coefficients for cross-tissue comparisons of the total expression level for both homeologs (Tr_To+Tr_Tp) and for the homeolog expression ratios (Tr_To/Tr_Tp). Fl, flower; Le, leaf; St, stolon/shoot; Ro, root. Error bars = 95% confidence intervals calculated using the formula: tanh(arctanh[r²] ± 1.96/sqrt[n−3]), where r² is the Pearson correlation estimate and n = 19,954 is the number of observations. (D) Venn diagram showing the overlaps between the expression outlier genes detected in the “Pooled,” “S9,” and “S10” experiments. (E) Log(Tr_To/Tr_Tp) difference from the mean values from all three experiments (Pooled, S9, and S10) plotted for a randomly selected gene and an expression ratio outlier. The gray horizontal line indicates the mean. m-leaf, mature leaf; leaf, emerging young leaf. (F) Clustered heatmap showing row-normalized log(Tr_To/Tr_Tp) values for all 909 expression outliers detected using the ANOVA method for all three experiments (P [pooled], S9 and S10) across four tissues: m-leaf, mature leaf; leaf, emerging young leaf. (G) Smoothed scatterplot showing the log(Tr_To/Tr_Tp) difference from the mean for all 19,954 genes and the 909 ANOVA outliers, respectively. m-leaf, mature leaf; leaf, emerging young leaf. (H) Expression ratio outliers in flavonoid metabolism. Blue text highlights enzymes found as expression ratio outliers. Numbers above enzyme names are EC identifiers.