Enhancing the accuracy of genetic values of tested early‐stage potato clones utilizing genotypic, pedigree, and drone‐retrieved spectral information.

Abstract The first generations of most clone breeding programs do not allow multi‐environment trials due to the lack of sufficient reproductive organs. We integrated the phenotypic records of such an early timepoint, called the A clone stage in potato ( Solanum tuberosum L.) breeding, into a genomic prediction framework and analyzed (i) if different relationship information can enhance the accuracy of the breeding values from such unreplicated trials, and (ii) whether a weighted combination of these enhanced breeding values with predicted breeding values calculated through genomic prediction can further improve accuracy. Compared to the breeding values derived without considering any relationship information among the clones, enhanced breeding values achieved a higher accuracy for 7–14 traits, depending on the relationship matrix used (multispectral, genetic, pedigree, or a combination). Combining these enhanced breeding values with predicted values from genomic prediction via a heritability index resulted in a weak increase of accuracy (0.5%) for five traits. However, a grid search to find the optimal combination of weights revealed more traits (8–12) with a higher mean increase of accuracy (0.5%–3.8%). Our findings show that the accuracies of unreplicated clones can be increased with easy‐to‐collect multispectral and pedigree data. The magnitude of this accuracy increase was negatively correlated with the per plot heritabilities, which makes this approach effective for lowly heritable traits. We see potential for this approach in the early stages of potato breeding programs, like the seedling or single hill stage, where the exact assessment of complex traits is impossible due to low heritability.