Past and Present Trajectories of Grain Sorghum Improvement in Texas

Abstract

Understanding the rate of genetic gain and developing procedures that improve the efficiency of breeding for sorghum [Sorghum bicolor (L.) Moench] grain yield is paramount for delivering sustainable and climate-resilient genotypes to target environments. In this work, a historical assessment of five decades of sorghum hybrid improvement in Texas is conducted alongside a partial factorial approach to current hybrid testing in the Texas A&M AgriLife Research Sorghum Breeding program. In the retrospective investigation, data from the Texas A&M AgriLife Research Crop Testing program were evaluated from 1970 to 2021. Over these decades, grain yield increased at a rate of 0.03 t ha^-1 per year and an average rate of genetic gain of 0.001 t ha^-1 per year. A strong positive relationship between hybrid yield and plant height was present with the biggest yield increases occurring in the 1980’s and 1990’s. While latitude and production systems are factors in adaptation to a particular region, the results herein indicate that modest increases in height will increase yield and that available moisture is the preeminent delineator for environments and their effects on elite grain sorghum hybrid performance. To evaluate combining abilities for 94 elite inbreds, 399 hybrids were produced using a partial factorial mating scheme and tested in replicated trials across four total environments. Genotyping by sequencing data of the inbreds were used in phylogenetic and clustering analyses that depicted a clear split between heterotic groups and confirmed substantial genetic variation within heterotic groups. Combining abilities were calculated using a linear mixed model framework and reported for grain yield and other agronomic traits. For grain yield, general combining abilities ranged from -0.152 to 0.129 t ha^-1 among R-lines and -0.332 to 0.172 t ha^-1 among A-lines, while values for specific combining ability ranged from -0.088 to 0.071 t ha^-1. For all studied traits except panicle exsertion, analyses of combined environments detected highly significant interactions between genetic effects and the environment. Thus, a genotype x environment analysis was also conducted. These findings demonstrate strong potential for a partial factorial approach to hybrid testing which could later facilitate genomic and genomic x environmental prediction models.