Pan-genome and haplotype map of cassava cultivars and wild ancestors provide insights into its adaptive evolution and domestication.

Cassava is a highly resilient tropical crop that produces large, starchy storage roots and high biomass. However, how did cassava's remarkable environmental adaptability and key economic traits evolve from its wild species remain unclear. In this study, we obtained near complete telomere-to-telomere genome assemblies and their haplotype forms for the cultivar AM560, the wild ancestors FLA4047 and W14, constructed a graphic pan-genome of 30 representatives with a size of 1.15 Gb, and built a clarified evolutionary tree of 486 accessions. A comparison of structural variations and single-nucleotide variations between the ancestors and cultivated cassavas reveals predominant expansions and contractions of numbers of genes and gene families, which are mainly driven by transposons. Significant selective sweeping occurred in 122 footprints of genomes and affects 1,519 domesticated genes. We identify selective mutations in MeCSK and MeFNR2 that could promote photoreactions associated with MeNADP-ME in C photosynthesis in modern cassava. Coevolution of retard floral primordia and initiation of storage roots may arise from MeCOL5 variants with altered bindings to MeFT1, MeFT2, and MeTFL2. Mutations in MeMATE1 and MeGTR occur in sweet cassava, and MeAHL19 has evolved to regulate the biosynthesis, transport, and endogenous remobilization of cyanogenic glucosides in cassava. These extensive genomic and gene resources provided here, along with the findings on the evolutionary mechanisms responsible for beneficial traits in modern cultivars, lay a strong foundation for future breeding improvements of cassava.