Assessment of kernel characteristics to predict popping performance in grain sorghum

Growth in the niche market of popped grain sorghum [Sorghum bicolor (L.) Moench] has increased the demand for grain sorghum lines or hybrids with improved popping quality. While there is a clear morphological difference in kernel morphology between popcorn [Zea mays L. everta] and most other types of corn, most grain sorghum genotypes have kernels with generally similar morphological structure. The absence of a specific kernel morphology for sorghum makes it impossible to eliminate types of grain sorghum that will not pop based solely on that morphology. Consequently, screening of any sorghum genotype requires the actual popping of grain from that genotype. As such, the identification of traits or combinations thereof that effectively screen grain sorghum genotypes for popping efficiency (PE), expansion ratio (ER), flake size (FS), and popped density (PD) is necessary. Herein, grain from 78 diverse genotypes grown in two environments were characterized for physical (i.e., diameter, thousand kernel weight, kernel hardness index, test weight, and visual hardness rating), compositional (i.e., starch, fiber, fat, ash, and protein), and popping characteristics (i.e., PE, ER, FS, and PD). No single physical or compositional trait was sufficiently correlated to prediction of popping performance. Multi-trait models better predicted popping performance than the single trait correlations. Further, multi-trait models using compositional predictors increased prediction accuracies by 10.1% for PE, 42.9% for ER, 24.4% for FS, and 40.6% for PD compared with physical predictors. Among subgroups of genotypes, prediction accuracies varied considerably based on the criteria used to subdivide the genotypes. In conclusion, indirect selection for popping performance is possible by leveraging specific multi-trait models.

Automated summary

The study found that it is not possible to predict popping performance solely based on the morphology of sorghum kernels, and actual popping experiments are necessary to screen sorghum genotypes. Multi-trait models performed better in predicting popping performance, with compositional predictors showing higher prediction accuracies.