Diploidization in a wild rice allopolyploid is both episodic and gradual.

Polyploid organisms evolve from their initial doubled genomic condition through a number of processes collectively termed diploidization, whose tempo and mode remain poorly understood mainly due to the difficulty of discriminating de novo evolution subsequent to polyploidy from variation inherited from progenitors. Here, we generated chromosome-scale genome assemblies for the wild rice allopolyploid Oryza minuta and its two diploid progenitors, Oryza punctata and Oryza officinalis, and employed a population genomic approach to investigate the diploidization process in O. minuta at the sequence and transcriptomic level. We show that this wild rice allopolyploid originated around 0.7 Mya, and during subsequent diploidization, its two subgenomes have retained highly conserved synteny with the genomes of its extant diploid progenitors. This populational approach allowed us to distinguish parental legacy of inherited variation from postpolyploidy evolution, and our analyses revealed that whereas gene fractionation occurred in an early burst, accumulation of transposable elements (TEs) and homoeologous exchanges has been gradual. Patterns of homoeolog expression bias are highly variable across tissues, with no consistent subgenome expression bias. Our assessments of the impact of DNA methylation, TE distribution, and parental legacy on expression patterns provide some support for the TE load theory (the theory that the TE densities in flanking regions surrounding genes strongly influence expression levels), while also illustrating the complexity of transcription regulation.