Leveraging additive and dominance effects, and inbreeding control in sweetpotato mate allocation

Genomic selection of genotypes with high genetic merit, often closely related, leads to reduced genetic variability, which can make long-term genetic gain in breeding programs unfeasible. A strategy to overcome this problem is to control the level of inbreeding in the population while simultaneously exploiting markers’ additive and dominance effects. This work aimed to perform mate allocation of a sweet potato population from NARO-Uganda by means of a convex optimization algorithm. A total of 527 clones were phenotyped for key sweet potato traits over nine location-season combinations, and 304 clones were ultimately genotyped using 2.5k DArTag-derived markers. The solution to select high-merit crosses, assuming 2% inbreeding at most, resulted in 31 putative parents selected based on the additive-dominant model, whereas 21 parents were selected based on an exclusive additive model. For some traits, the merit of the best crosses from the dominance-additive model is greater than that from the additive model. In model 3 (additive-dominant), we selected more parents, in all traits together, and may be able to avoid cross incompatibility in sweet potato.