Genomic selection using information on multiple phenotypic traits and multiple growing environments

Plant breeders identify superior genotypes by collecting phenotypic data on multiple traits across field trials conducted in multiple environments. Traditionally, however, statistical analyses have been restricted to analysing a single trait across multiple environments to gauge genotype by environment (GE) interaction or multiple traits in a single environment to gauge genotype by trait (GT) interaction. There is currently no sensible approach to analyse multiple traits across multiple environments simultaneously, providing an informative model for the resultant genotype by environment by trait (GET) interaction. Hence, this research aimed to develop a new single-stage genomic selection (GS) approach within the factor analytic framework, which utilises information across multiple traits and multiple environments. Our research is an extension of Smith et al. (2001), which was adapted and applied for GS by Tolhurst et al. (2019). We demonstrate our new GS approach using a late-stage rice breeding dataset from Australia, which comprises ~1250 genotypes and 16 environments in the south-eastern rice-growing region of Australia. Phenotypic data were collected on several key traits, including days to flowering, plant height and grain yield. Finally, we extend the selection tools of Smith & Cullis (2018) to our new GS approach, which provide simple yet informative summaries of the genotype-specific responses across traits and environments. This work represents an important continuation in the advancement of plant breeding analyses which provide breeders with the best available information to improve genetic gain.