The theme is focused on developing and testing hierarchical genome to phenome (G2P) models for multiple traits to provide a framework for enhancing genomic prediction to accelerate genetic gain for yield and yield stability achieved by breeding programs. Crop Growth Models (CGMs) are used as a multi-trait framework for hierarchical integration of trait-by-trait-by-environment interactions to investigate and predict crop adaptation to agricultural environments. The APSIM modelling platform (see Theme 3) is utilised for crop modelling. Implementations of APSIM and ensemble prediction methods are enabled through the development of software pipelines within the high-performance computing capacities of the Australian National Computing Infrastructure.

A wholistic breeding cycle approach is considered to investigate the interplay between the components of the breeder’s equation for investigation of selection trajectories and genetic gain; including, (1) the applied and realised selection intensity throughout the G2P trait hierarchy, (2) the theoretical and realized prediction accuracy for the genomic, environomic and phenotyping capabilities of a breeding program, (3) the alignment of the breeding program testing capacities with the Genotype-by-Environment-by-Management (GxExM) context of the Target Population of Environments, (4) trait genetic architecture, genetic diversity and the accessible standing genetic variation of the reference population of genotypes of the breeding program, and (5) effective cycle time of the target breeding program.

Experimental investigations are designed to explore achievable selection trajectories and predict genetic gain over breeding cycles. Experiments are conducted for the model species Arabidopsis thaliana and the important crop species Sorghum bicolor. Collaborations with industry partners are investigating the potential to operationalise enhanced prediction capabilities; Maize and Sorghum with Corteva, and Soybean with BASF. Projects are conducted to consider the complex relationships between predictability and interpretability of the genetic architecture for complex traits. Experiments are conducted to map traits and investigate multi-trait strategies contributing to crop performance and adaptation for current and projected future environments. The branching network and its influences on plant architecture in combination with plant development and flowering time are priority traits used to test G2P modelling capabilities and prediction-based breeding strategies. In addition, traits related to the water availability and temperature conditions of agricultural environments are also priority areas of focus. All experiments are designed to leverage advances in genomics and pangenomics capabilities, together with improvements in trait phenotyping.