Genomic selection to optimize doubled haploid-based hybrid breeding in maize

Abstract

Crop improvement, as a long-term endeavor, requires continuous innovations in technique from multiple perspectives. Doubled haploid (DH) technology for pure inbred production, which shaves years off of the conventional selfing approach, has been widely used for breeding. However, the final success rate of in vivo maternal DH production is determined by four factors: haploids induction, haploids identification, chromosome doubling, and successful selfing of the fertile haploid plants to produce DH seeds. Traits in each of these steps, if they can be accurately predicted using genomic selection methods, will help adjust the DH production protocol and simplify the logistics and save costs. Here, a hybrid population (N=158) was generated based on an incomplete half diallel design using 27 elite inbred lines. These hybrids were induced to create F1-derived haploid families. The hybrid materials, as well as the 27 inbreds, the inbred-derived haploids (N=200), and the F1-derived haploids (N=5,000) were planted in the field to collect four DH-production traits, three yield-related traits, and three developmental traits. Quantitative genetics analysis suggested that in both diploids and haploid families, most of the developmental traits showed high heritability, while the DH-production and developmental traits exhibited intermediate levels of heritability. By employing different genomic selection models, our results showed that the prediction accuracy ranged from 0.52 to 0.59 for the DH-production traits, 0.50 to 0.68 for the yield-related traits, and 0.44 to 0.87 for the developmental traits. Further analysis using index selection achieved the highest prediction accuracy when considering both DH production efficiency and the agronomic trait performance. Furthermore, the long-term responses through simulation confirmed that index selection would increase the genetic gain for targeted agronomic traits while maintaining the DH production efficiency. Therefore, our study provides an optimization strategy to integrate GS technology for DH-based hybrid breeding.