Extensive crop-wild hybridization during Brassica evolution and selection during the domestication and diversification of Brassica crops

Abstract

Adaptive genetic diversity in crop wild relatives (CWRs) can be exploited to develop improved crops with higher yield and resilience if phylogenetic relationships between crops and their CWRs are resolved. This further allows accurate quantification of genome-wide introgression and determination of regions of the genome under selection. Using broad sampling of CWRs and whole genome sequencing, we further demonstrate the relationships among two economically valuable and morphologically diverse Brassica crop species, their CWRs, and their putative wild progenitors. Complex genetic relationships and extensive genomic introgression between CWRs and Brassica crops were revealed. Some wild Brassica oleracea populations have admixed feral origins; some domesticated taxa in both crop species are of hybrid origin, while wild Brassica rapa is genetically indistinct from turnips. The extensive genomic introgression that we reveal could result in false identification of selection signatures during domestication using traditional comparative approaches used previously; therefore, we adopted a single-population approach to study selection during domestication. We used this to explore examples of parallel phenotypic selection in the two crop groups and highlight promising candidate genes for future investigation. Our analysis defines the complex genetic relationships between Brassica crops and their diverse CWRs, revealing extensive cross-species gene flow with implications for both crop domestication and evolutionary diversification more generally.

Keywords: Brassica; Plant Genetics and Genomics; crop wild relatives; domestication; introgression; phylogenomics.

Fig. 1.

Phylogenetic relationships and hybridization within and between Brassica oleracea (a, c, e) and Brassica rapa (b, d, f), and their wild relatives. a, b) Maximum likelihood phylogenetic relationships based on single-nucleotide polymorphisms (SNPs) filtered by linkage disequilibrium for samples mapped to a) B. oleracea and b) B. rapa. Polyphyletic groups are identified by colored dots (see legend), and bootstrap values > 70 are indicated. c, d) RNDmin, a measure of the pairwise distance relative to an outgroup calculated in 50 kb windows for c) domesticated B. oleracea versus wild relatives and d) domesticated B. rapa (excluding ssp. rapa) versus wild relatives. e, f) Most likely phylogenetic networks identified for e) B. oleracea (BOL) and f) B. rapa (BRP), with dotted lines indicating admixture. Posterior probabilities (PP) with 95% confidence intervals are given below each network.

Fig. 2.

Population genetic statistics and population structure of wild and domesticated Brassica oleracea (a–e) and Brassica rapa (f–j). a, f) Distribution of population genetic statistics across the genome; b, g) linkage disequilibrium decay; c, h) Evanno's delta K for STRUCTURE analyses; d, i) STRUCTURE analysis, with colors representing the proportional assignment of each individual to each of the K clusters; e, j) demographic history inference of effective population size over time.

Fig. 3.

Signatures of selection in Brassica oleracea and Brassica rapa. a, b) Overlap between genomic regions targeted by positive selection in a) wild (bottom) and domesticated (top) B. oleracea (excluding B. oleracea var. alboglabra) and b) overlap between regions targeted by positive selection in combined wild (bottom, including B. rapa ssp. rapa) and domesticated (top) B. rapa. CLR values in the top 1% of both the CLR (Sweed) and ω-statistic (OmegaPlus) are highlighted as red or blue points. Shaded boxes define windows around these points that maximize CLR. Bars at the top show the location of these windows affected by selection (light) and the likely targets of selection within them (dark). Overlapping regions are indicated with arrows and numbers indicate the number of genes in the overlap and the number with AT annotations. c) Size (Mb) of genomic regions targeted by positive selection and their overlap between domesticated B. oleracea varieties and between B. rapa subspecies. d) Overlap in gene ontology categories that were enriched in regions targeted by positive selection.

Fig. 4.

Evidence for parallel positive selection between pairs of Brassica oleracea domesticates (left) and Brassica rapa domesticates (right) with similar phenotypes. CLR values in the top 1% CLR values and top 1% of ω-statistic values highlighted as blue and red points for B. oleracea and B. rapa, respectively. Shaded boxes define windows around these points that maximize CLR. The top bars show the location of these windows affected by selection (light) and the candidate target regions of selection within them (dark). Positions of putative rapa-oleracea orthologues in target selection regions according to reciprocal BLAST are indicated (full gene information is given in SI Appendix, Supplementary Table 11). Tajima's D is plotted below with negative values highlighted in blue.