Genomic prediction for dry matter and chipping quality in tetraploid potato

Potato is one of the most space-efficient food crops and is of vital importance for global food security. The traditional “mate and phenotype” breeding approach is costly and time-consuming; however, genomic selection using genome-wide molecular markers is becoming increasingly applicable to crops
and is thus an attractive breeding alternative.
Genotyping-by-sequencing was used to genotype a large number of individuals from three panels. The main panel, the MASPOT population, consisted of 762 individuals derived from a larger population generated from biparental crosses of 18 tetraploid parents. Additionally, two test panels were established, one panel consisting of breeding clones from the same breeding station as the MASPOT population, named Test panel DK, and one panel from a breeding station in the UK, named Test panel UK.
Genomic prediction models were generated for dry matter and chipping quality. High cross-validated prediction accuracies of 0.72-0.82 were obtained for dry matter within each population, while prediction accuracies for chipping quality varied from 0.17 to 0.79. Similar prediction accuracies were obtained when combining the panels in one large training population, while across-population predictions were low or moderate.
Overall, the results suggest that genomic prediction and hence selection of breeding material can be obtained with good accuracies within tetraploid potato. Although the most optimal prediction accuracies were obtained when predicting within the same population, the results from combining training populations with genotypes from different populations suggest a promising approach for establishing a broad-application prediction model for the implementation of genomic selection in tetraploid potato breeding programmes.