Graph pangenome reveals structural variation dynamics during cucumber breeding

Structural variants (SVs) represent a significant yet underexplored component of plant genome diversity. Analyzing SV dynamics is essential for understanding their contributions to phenotypic variation and their influence on genome evolution. Here, we present a graph-based pangenome for cucumber, constructed from 39 reference-level genome assemblies. The pangenome captures 171,892 high-confidence SVs and enables genotyping of these SVs across 443 cucumber accessions, representing diverse geographic origins and spanning wild progenitors, landraces, and modern cultivars. Our analyses reveal that during cucumber domestication, a substantial portion of mildly deleterious SNPs are retained, whereas SVs are consistently purged, highlighting their potentially highly deleterious nature. During the geographic expansion of cucumber, a reduction in SV burden and a younger age of SVs than SNPs were observed, suggesting stronger purifying selection acting on SVs. Additionally, notable gene flow from wild population to African and European populations was detected, resulting in an increased SV burden, potentially due to hitchhiking effects. Importantly, incorporating SV burden into genomic prediction models improved the prediction accuracy for several agronomically important traits. This study illuminates the dynamics of SVs during cucumber domestication and improvement, highlights the complex interplay between SVs and selection pressures, and underscores the extensive implications of SVs for future cucumber breeding.