Genomic prediction of hybrid performance in grain sorghum (Sorghum bicolor L.)

Authors: MAULANA, FRANK - Kansas State University; PERUMAL, RAMASAMY - Kansas State University; Serba, Desalegn; TESSO, TESFAYE - Kansas State University

Submitted to: Frontiers in Plant Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/22/2023
Publication Date: 4/24/2023
Citation: Maulana, F., Perumal, R., Serba, D.D., Tesso, T. 2023. Genomic prediction of hybrid performance in grain sorghum (Sorghum bicolor L.). Frontiers in Plant Science. 14. Article 1139896. https://doi.org/10.3389/fpls.2023.1139896.
DOI: https://doi.org/10.3389/fpls.2023.1139896

Interpretive Summary: Conventional plant breeding is time consuming and expensive. In hybrid breeding, identifying a prominent hybrid that can be released as a cultivar involves parental development, crossing among selected parents, and a lengthy process of multi-environment evaluation of hybrids along the parents. The advent of molecular markers, especially the next-generation sequencing technology that enabled simultaneous marker development and genotyping, opened a new horizon for enhancing breeding efficiency and selection accuracy. It has become a faster method that allows selection of inbred parents with enhanced hybrid performance. The efficacy of predicting the performance of grain sorghum hybrids using genomic information of parental genotypes was assessed in three grain sorghum populations using genomic, pedigree, and environmental covariables interaction models. The outcome of the study suggests that extensive hybrid synthesis and evaluation schemes in sorghum breeding can be replaced by less expensive, fast, and high accuracy genome-based prediction. Conventional plant breeding is time consuming and expensive. In hybrid breeding, identifying a prominent hybrid that can be released as a cultivar involves parental development, crossing among selected parents, and a lengthy process of multi-environment evaluation of hybrids along the parents. The advent of molecular markers, especially the next-generation sequencing technology that enabled simultaneous marker development and genotyping, opened a new horizon for enhancing breeding efficiency and selection accuracy. It has become a faster method that allows selection of inbred parents with enhanced hybrid performance. The efficacy of predicting the performance of grain sorghum hybrids using genomic information of parental genotypes was assessed in three grain sorghum populations using genomic, pedigree, and environmental covariables interaction models. The outcome of the study suggests that extensive hybrid synthesis and evaluation schemes in sorghum breeding can be replaced by less expensive, fast, and high accuracy genome-based prediction.

Technical Abstract: Genomic selection is expected to improve selection efficiency and genetic gain in breeding programs. The objective of this study was to assess the efficacy of predicting the performance of grain sorghum hybrids using genomic information of parental genotypes. One hundred and two public sorghum inbred parents were genotyped using genotyping-by-sequencing. The sequence analysis generated 66,265 SNP markers that were used to predict the performance of 204 F1 hybrids resulted from crosses between the parents. Both additive (partial model) and additive and dominance (full model) were constructed and tested using various training population (TP) sizes and cross-validation procedures. Increasing TP size from 41 to 163 increased prediction accuracies for all traits. With the partial model, the five-fold cross validated prediction accuracies ranged from 0.03 for thousand kernel weight (TKW) to 0.58 for grain yield (GY) while it ranged from 0.06 for TKW to 0.67 for GY with the full model. The results suggest that genomic prediction could become an effective tool for predicting the performance of sorghum hybrids from parental genotypes.