Genomic constraint and hypervariability correlate with fitness effects in elite tetraploid potato breeding material

Tetraploid potato is the most important vegetable food crop in the world. However, genetic gains from modern breeding efforts have been limited due to tetrasomic inheritance, the accumulation of deleterious mutations during domestication, and subsequent clonal propagation during modern breeding. In this study, we analyzed previously identified evolutionarily constrained genomic regions in the potato genome relative to a breeding panel of tetraploid potato clones derived from elite cultivars (the MASPOT panel). We demonstrate a significant association between derived allele frequency (DAF) and both positive (constrained) and negative (excess of substitutions relative to neutrality, i.e., hypervariable) nucleotide conservation scores (GERP scores) in the breeding panel, indicating that negative GERP scores can annotate sites affected by positive selection or other non-neutral processes affecting population allele frequency. The deep phylogeny approach can be leveraged to alleviate inbreeding in potatoes through the qualification of causative alleles in tetraploids, offering a valuable resource for potato genetic improvement by molecular breeding. Estimates of genetic load within the tetraploid clones reveal that heterozygous loads vastly exceed homozygous loads for deleterious alleles, consistent with masking of recessive mutations. Deleterious load was significantly correlated with several trait phenotypes but did not improve prediction accuracy of any traits in the tetraploid panel. This could be due to the increased complexity of tetrasomic inheritance, including more effective load complementation, and/or a higher proportion of heterozygosity in tetraploid compared to diploid potato. Intriguingly, the hypervariable load correlated significantly with several phenotypes and improved genomic prediction accuracy of yield by 5% relative to a negative control (permuted loads). This indicates that the inferred hypervariable sites from Solanaceae evolutionary information may contain genetic information relevant for breeding, and that, in contrast to current practice, it is relevant to include and further explore sites with negative GERP scores in general.