Persistence of sub-genomes in paleopolyploid cotton after 60 million years of evolution

The importance of whole genome multiplication (WGM) in plant evolution has long been recognized. In flowering plants, WGM is both ubiquitous and in many lineages cyclical, each round followed by substantial gene loss (fractionation). This process may be biased with respect to duplicated chromosomes, often with over-expression of genes in less fractionated (LF) relative to more fractionated (MF) regions. This bias is hypothesized to arise through down-regulation of gene expression via silencing of local transposable elements (TEs). We assess differences in gene expression between duplicated regions of the paleopolyploid cotton genome and demonstrate that the rate of fractionation is negatively correlated with gene expression. We examine recent hypotheses regarding the source of fractionation bias and show that TE-mediated, positional down-regulation is absent in the modern cotton genome, seemingly excluding this phenomenon as the primary driver of biased gene loss. Nevertheless, the paleo sub-genomes of diploid cotton are still distinguishable with respect to TE content, targeting of 24nt-siRNAs and GC content, despite ~60 million years of evolution. We propose that repeat content per se and differential recombination rates may drive biased fractionation following WGM. These data highlight the likely importance of ancient genomic fractionation biases in shaping modern crop genomes.