Evolutionary origins and dynamics of octoploid strawberry subgenomes revealed by dense targeted capture linkage maps

Abstract

Whole-genome duplications are radical evolutionary events that have driven speciation and adaptation in many taxa. Higher-order polyploids have complex histories often including interspecific hybridization and dynamic genomic changes. This chromosomal reshuffling is poorly understood for most polyploid species, despite their evolutionary and agricultural importance, due to the challenge of distinguishing homologous sequences from each other. Here, we use dense linkage maps generated with targeted sequence capture to improve the diploid strawberry (Fragaria vesca) reference genome and to disentangle the subgenomes of the wild octoploid progenitors of cultivated strawberry, Fragaria virginiana and Fragaria chiloensis. Our novel approach, POLiMAPS (Phylogenetics Of Linkage-Map-Anchored Polyploid Subgenomes), leverages sequence reads to associate informative interhomeolog phylogenetic markers with linkage groups and reference genome positions. In contrast to a widely accepted model, we find that one of the four subgenomes originates with the diploid cytoplasm donor F. vesca, one with the diploid Fragaria iinumae, and two with an unknown ancestor close to F. iinumae. Extensive unidirectional introgression has converted F. iinumae-like subgenomes to be more F. vesca-like, but never the reverse, due either to homoploid hybridization in the F. iinumae-like diploid ancestors or else strong selection spreading F. vesca-like sequence among subgenomes through homeologous exchange. In addition, divergence between homeologous chromosomes has been substantially augmented by interchromosomal rearrangements. Our phylogenetic approach reveals novel aspects of the complicated web of genetic exchanges that occur during polyploid evolution and suggests a path forward for unraveling other agriculturally and ecologically important polyploid genomes.

Keywords: Fragaria; genome assembly; introgression; phylogenetics; polyploidy; transposition.

Fig. 1.—

Schematic of our POLiMAPS approach. (A) Next-generation sequence reads from octoploids obtained through targeted capture are aligned to our reference genome assembly. LG SNPs occur at approximately one-eighth frequency in a single parent and show Mendelian segregation in the offspring. If an LG SNP shares a read with a marker that occurs on multiple linkage groups and/or diploid taxa, this marker can be used in phylogenetics. A majority of reads do not harbor both an LG SNP and a phylogenetic marker, and thus many linkage groups (e.g., dark blue) have missing data for many phylogenetic markers. (B) Identification of introgression. Phylogenetic markers can be classified as providing phylogenetic support, showing introgression-like homoplasy, or showing outgroup homoplasy. Multiple adjacent markers showing introgression-like homoplasy, interspersed with few or no phylogenetically supportive markers, are considered to be caused by true introgression.

Fig. 2.—

Mapping of scaffolds from FvH4 reference genome assembly to Fvb assembly. The path of every scaffold or scaffold segment is represented by a line (orange, inverted in diploid linkage maps; purple, not inverted in diploid linkage maps; black, no information about scaffold orientation in diploid linkage maps, retained default noninverted orientation).

Fig. 3.—

Prunus orthologs mapped onto Fragaria assemblies. Each horizontal line represents an orthologous gene, colored according to its Prunus chromosome, and with a width corresponding to its position on this Prunus chromosome (wider lines are close to the start of their respective Prunus chromosome). Its x axis position indicates the Fragaria pseudochromosome onto which it is assembled. Its y axis position indicates its position on this Fragaria pseudochromosome. Thus, regions showing lines of the same color and similar width indicate high synteny between the two genera. Overall, Fvb shows higher synteny than FvH4.

Fig. 4.—

Linkage maps for F. chiloensis and F. virginiana ssp. virginiana, aligned to the Fvb reference genome. Segregating LG SNPs are represented by small solid circles. For all LG SNPs, the y axis position corresponds to linkage map position (cM) and the x axis position corresponds to Fvb genome position (Mb). All seven Fvb chromosomes are represented on the x axis, with scale indicated by dark gray horizontal bars in all four corners (=10 Mb). All four parents are represented on the y axis, with scale indicated by light gray vertical bars in all four corners (=100 cM). Linkage groups are colored based on their phylogenetic position (red, Av; cornflower, B1; cyan, B2; blue, Bi). LG SNPs in black correspond to minor linkage groups (<5 LG SNPs) that could not be placed in the phylogeny. Short vertical black bars indicate junctures where smaller linkage groups were manually joined based on visual inspection of genotypes, phylogenetic position, and Fvb position. Introgression clusters are shown as light red circles behind LG SNPs, and their positions are again shown along the top of the figure (IC), colored according to subgenome. Interchromosome rearrangements are shown as squares around LG SNPs, and their positions are again shown along the top of the figure (IR), with numbers indicating the haploid chromosome of the linkage group.

Fig. 5.—

Phylogenies of all seven haploid Fragaria chromosomes. Octoploid subgenomes are named for the species (Fchil or Fvirg), haploid chromosome number (Roman numeral), subgenome (Av, B1, B2, or Bi), and parent (p or m), and are colored as in figure 4 with F. iinumae (blue) and F. vesca (FvH4, Fvb-m, Fvb-p, and Fvb-s; red) similarly colored. Other diploid taxa have distinct colors. Numbers on branches represent bootstrap values; bootstraps less than 50 or on branches with length less than 0.01 substitutions/informative site are not shown.

Fig. 6.—

Regions of 1 kb with phylogenetically informative markers. For a phylogeny consisting of an outgroup, F. vesca, F. iinumae, and a subgenome, there are three possible patterns (fig. 1B): Phylogenetic support, introgression-like homoplasy, and outgroup homoplasy. Regions with multiple markers may also show a mix of support and homoplasy. There is an excess of introgression-like homoplasy relative to outgroup homoplasy, but it can be accounted for entirely by the introgression clusters, which occur only on subgenomes B1, B2, and Bi.

Fig. 7.—

Relative normalized coverage in octoploids relative to diploid, for targeted regions showing dynamic evolutionary changes. Regions are binned on a log scale along the x axis with breakpoints at multiples of √2. (A) Targeted regions showing introgression do not have significantly different depth than regions not showing introgression. (B) Targeted regions showing interchromosome rearrangements have higher and more variable depth in octoploid relative to diploid, indicating that many of these regions undergo “copy-and-paste” transposition, resulting in unequal copy numbers across species.

Fig. 8.—

Ratio of F. vesca-like read depth to F. iinumae-like read depth for the two maternal octoploid parents in 1-Mb bins across the Fvb genome. Bins overlapping introgression clusters as determined from the linkage map data (red) show a much higher ratio. The ratio in all bins is likely skewed toward F. vesca because such reads are more likely to map to Fvb.