Assessment of segregation variance estimates from derivation, simulations, and empirical data in autotetraploid species exemplified in potato

The optimal choice of parents and crosses and, therefore, the prediction of the segregation variance are of high relevance to maximize genetic gain in breeding programs. Several methods have been developed for the prediction of segregation variance, includingcorrelation with genotypic diversity, progeny simulations, or algebraic derivations in case of a diploid inheritance. To the best of our knowledge, no algebraic derivation using parental genotypic information is available to predict segregation variance for autotetraploid species. The objectives of our study were to (1) derive algebraic derivation based on linkage disequilibrium (LD) between linked loci to predict the segregation variance in autotetraploid species; (2) compare the performance of segregation variance estimated based on simulated progenies and the algebraic derivations; (3) investigate by simulations how experimental parameters affect the accuracy of segregation variance prediction; and (4) compare the segregation variance estimated in empirical data of potato and the one based on the algebraic derivations. The segregation variance estimated by the developed derivations showed very high correlations with the one observed in large simulated progenies, but those were lower when phased parental haplotypes were not available or family size decreased. The correlation between segregation variance estimated by the developed derivation and the empirical data was low. This could be attributed to the small family size used in the study, which we could show to increase LD between unlinked loci. The proposed algebraic derivations promise to be a precise alternative to simulations to help breeders in optimizing their family choices and sizes considering the segregation variance.