Abstract
Abstract
Recombination suppression is widely regarded as a barrier to genetic exchange, yet whether it can instead preserve shared ancestry across ploidy barriers and subsequently facilitate ecological adaptation has remained unknown. Combining chromosome-scale population genomics and transcriptomics from 126 individuals spanning multiple ploidy levels and salinity ecotypes of the common reed (Phragmites australis), we show that ancestry shared between an octoploid and a tetraploid lineage is disproportionately concentrated in centromeric and pericentromeric regions across the genome, where large shared haploblocks persist despite extensive recombination-driven erosion elsewhere and are enriched for functionally coherent gene clusters. At one locus, this retained haploblock has been independently recruited by selective sweeps in geographically isolated coastal and inland saline populations, linking centromeric ancestry retention to repeated ecological adaptation. Remarkably, the same recombination-suppressed genomic architecture produces the opposite evolutionary outcome at other loci, where haplotypic inversions promote rapid, lineage-specific centromere divergence rather than ancestry retention. Together, these findings reframe the classical centromere paradox, in which centromeric sequences turn over rapidly despite conserved centromere function, by showing that recombination suppression can drive either long-term ancestry retention or accelerated centromeric divergence, depending on local structural context. Our findings identify a centromeric region that has been repeatedly recruited for ecological adaptation across ploidy barriers, and suggest that centromeres more broadly may act as recurrent reservoirs of adaptive variation wherever polyploidization and hybridization coincide.