Tracking the ancestry of known and ‘ghost’ homeologous subgenomes in model grass Brachypodium polyploids

Unraveling the evolution of plant polyploids is a challenge when their
diploid progenitor species are extinct or unknown or when genome sequences
of known progenitors are unavailable. Existing subgenome identification
methods cannot adequately infer the homeologous genomes that are present
in the allopolyploids if they do not take into account the potential
existence of unknown progenitors. We addressed this challenge in the
widely distributed dysploid grass genus Brachypodium, which is a model
genus for temperate cereals and biofuel grasses. We used a
transcriptome-based phylogeny and newly designed subgenome detection
algorithms coupled with a comparative chromosome barcoding analysis. Our
phylogenomic subgenome detection pipeline was validated in Triticum
allopolyploids, which have known progenitor genomes, and was used to infer
the identities of three subgenomes derived from extant diploid species and
four subgenomes derived from unknown diploid progenitors (ghost
subgenomes) in six Brachypodium polyploids (B. mexicanum, B. boissieri, B.
retusum, B. phoenicoides, B. rupestre, and B. hybridum), of which five
contain undescribed homeologous subgenomes. The existence of the seven
Brachypodium progenitor genomes in the polyploids was confirmed by their
karyotypic barcode profiles. Comparative phylogenomics of nuclear vs
plastid trees allowed us to formulate hypothetical homoploid
hybridizations and allo- and autopolyploidization scenarios that could
have generated the six Brachypodium polyploids.