Genomic prediction of sweet sorghum agronomic performance under drought and irrigated environments in Haiti

Abstract

Over the past decade, genomic selection (GS) has gained significant traction as a valuable tool for predicting the phenotypic performance in plant breeding populations and for expediting the development of new cultivars. Diverse statistical models and approaches have been developed to facilitate the integration of GS into plant breeding practices, with a growing emphasis on strategies that enhance accurate and resource-efficient prediction. Since its inception in 2010, the sweet sorghum [Sorghum bicolor (L.) Moench] breeding program at Centre Haïtien d'Innovation en Biotechnologies et pour une Agriculture Soutenable has taken the lead in endeavors to cultivate and introduce varieties that exhibit resilience against both abiotic and biotic stresses. Among these challenges, drought stress holds particular prominence, given the reliance of growers on unpredictable rainfall patterns for successful sorghum production. The central objective of this study was to assess the predictive ability of genomic prediction models across varying environmental conditions in Haiti, employing two statistical methods. Our assessment encompassed 12 distinct sorghum traits, with genomic predictions conducted both within and across irrigated and water-stress treatments, executed at different planting dates. Overall, the two methods showed similar results. Prediction accuracy was notably higher for within-environment scenarios (ranging from 0.30 to 0.71) as opposed to across-environment scenarios (ranging from 0.08 to 0.68). Furthermore, there was considerable variation in the prediction accuracy for all traits, with “total soluble solids” displaying the highest mean value (0.71), while “total stem number” exhibited the lowest (0.38). The attained genomic prediction accuracies in this study offer encouraging insights for the integration of GS strategies in small-scale breeding programs, particularly those aimed at enhancing drought tolerance.