Towards Precision Breeding: A comprehensive framework for integrating environmental and management covariates in the genetic evaluation of commercial maize lines

Plant breeding has traditionally focused on selecting for genotypic effects that improve the mean performance of individuals across the target population of environments (TPE). However, this approach often fails to capture the influence of environmental and management factors, resulting in gaps between predicted and realized performance. This PhD research aims to address the complex interplay between Genotype (G), Environment (E), and Management (M) in maize breeding, with the goal of finding innovative applications that help breeding programs exploit GxExM interactions to achieve higher genetic gain. Two main challenges will be addressed in this research. First, numerous GxExM models exist in the literature. Model performance is varied and context-dependent, which makes it difficult to choose the optimal approach. In the last decade, the incorporation of environmental and management covariates (ECs) in genomic models has increased. Using many ECs may explain higher proportions of trait variance but compromises interpretability of the results. Instead, focusing on fewer influential ECs together with strategies to combine them may help breeding for targeted goals. Additionally, ECs over the growing season may show large fluctuations, suggesting uncertainty around the information contained in ECs. Second, the optimal strategy to select germplasm with superior characteristics to meet market demands remains unclear. Despite extensive research, breeders continue to face the challenge of selecting lines within a narrow TPE definition or to expand the list of selection candidates to distinct but related TPEs. This choice has implications in the rates of germplasm exchange and long-term genetic gain. This research will pursue three approaches to address the challenges above. First, GxExM models that leverage large environmental and management datasets will be developed and assessed, focusing on feature selection techniques and cross validation. Second, selection strategies within or across TPEs will be compared to determine which approaches deliver higher rates of germplasm exchange and long-term genetic gain by means of a simulation study. Third, targeted modelling approaches to breed for yield under limited nitrogen conditions will be evaluated, where localized GxExM interactions may be particularly important.