Predicting Carotenoid Breeding Values and Kernel Color in Maize
Mary LaPorte, University of California, Davis
The biofortification of maize grain (Zea mays ssp. mays) with carotenoids that have provitamin A activity has been effective in alleviating vitamin A deficiency in communities with maize-based diets (Bouis and Welch 2010; Graham et al. 2001; Prasanna et al. 2020; Palmer et al. 2016). Increasing concentrations of carotenoids, which impart a yellow, orange, or red hue, can also influence maize kernel color. Consumer dispreference for yellow maize in certain geographic regions has created the impetus to breed orange biofortified maize grain (De Groote and Kimenju 2008; Pillay et al. 2011; Muzhingi et al. 2008). Genomic prediction strategies, particularly in tandem with genome-wide association studies, can be used to shorten breeding time and/or increase genetic gain for quantitative traits such as carotenoid concentrations or kernel color, depending on the method of their application (Xu et al. 2021; Crossa et al. 2017). In this study, linear and Random Forest regression models were constructed to predict kernel color given carotenoid concentrations. Additionally, several genomic prediction (GP) models were tested within the Carotenoid Association Mapping panel, developed by the International Maize and Wheat Improvement Center (CIMMYT), to predict breeding values for grain carotenoid traits. The predictive ability of Ridge Regression, LASSO, Elastic Net, and Reproducing Kernel Hilbert Space GP models was compared for maize lines in the same environment. Additionally, the ability of the overall best performing model to predict carotenoid trait levels for lines grown in different environments was tested. Further experimentation will utilize known carotenoid-related markers and additional populations to move towards the routine implementation of genomic prediction/selection for grain carotenoid traits.
Abstract Author(s): Mary-Francis LaPorte, Akiyoshi Koide, Willy Suwarno, Jose Crossa, Natalia Palacios-Rojas, Christine Diepenbrock