Phylogeny-driven pangenome analysis uncovers the genomic landscape of domesticated and wild Armeniaca species.

Long-read sequencing and pangenomics are revolutionizing crop research by providing more complete genome information and revealing crucial structural variations (SVs) linked to important agricultural traits. Building on recent advances in intraspecific pangenome construction, this study addresses the challenge of creating broader, cross-taxon pangenomes, using the Armeniaca taxonomic section as a model. Leveraging a diverse panel of genome assemblies as well as completing it with seven more genome assemblies generated for this study, we constructed a pangenome graph and cataloged the associated genetic variation, identifying approximately 25 million single nucleotide polymorphisms and over 537 000 structural variants. We characterized the diversity of these variants and assessed the extent to which different taxa contribute to overall pangenome expansion. Additionally, we evaluated the performance of low-depth sample mapping to the graph-based reference, highlighting key technical limitations that may affect the quality of downstream analyses. We further identified specific subsets of SVs that exhibit associations with particular classes of transposable elements (TEs). We showed that TEs are a major driver of SV, particularly insertions and deletions, with distinct size and distribution patterns (peaking in the 200- to 400-bp indel bin). They are also nonrandomly positioned in the genome, showing a tight concentration near coding genes, which suggests a role in gene regulation. As a case study illustrating the potential functional relevance of graph-derived SVs, we examined the genomic configuration of the Dormancy-Associated MADS box locus within the Armeniaca pangenome. These findings provide a framework to investigate adaptation in perennial fruit trees of the Armeniaca section.