Lead Chief Investigator: Mark Cooper, The Universty of Queensland

Collaborating Chief Investigators: Christine Beveridge, David Jordan, Graeme Hammer, Kevin Burrage, Diane Donovan, Daniel Ortiz-Barrientos, Ian Wright

Objectives

Plants are hierarchical organisms. Their success is largely dependent on dynamic internal signalling regulated by the interplay of genes, environment and management practices.  

Yet, prediction methods used in plant breeding ignore much of the details of this hierarchical and dynamic structure.  

Our aim is to develop integrated prediction models that combine hierarchical trait models and prior knowledge of this internal signalling (gene and physiological networks).  

Our approach

We are taking an iterative experimental-modelling approach.  

Our research will target: 

  • identifying the genome-genetic architecture of traits  
  • defining traits across levels of the hierarchy from genome to phenome  
  • developing appropriate high-throughput phenotyping methods  
  • modelling the genome-to-phenome (G2P) map.  

Early iterations will focus on branching, then expand across plant development and physiology, particularly flowering, symbioses, water, and source-sink relations.  

We will integrate G2P multi-trait models from quantitative genetics for discovery and prediction applications, and implement them across cycles of selection for multiple species—sorghum, Arabidopsis and Senecio.  

 

Investigating G2P relationships for multiple traits across selection cycles will allow us to investigate how well we can predict trait phenotypes across environments for individual plants and their progeny.  

Thus, within quantitative genetics we can explore new modelling methods to predict an individual’s genotypic value and breeding value.  

The selection treatments will increase the power to detect underlying quantitative trait loci and resolve functional genome variation determining trait phenotypic variation at different levels within the G2P trait hierarchy.  

We will design new prediction methods for breeding by combining mathematical methods for describing and predicting trajectories on complex response surfaces with genetic simulations and experimental results; this will allow us to refine and improve our abilities to discover important trait G2P relationships, and predict trajectories in genetic and phenotypic space.  

Emerging mathematical methods, such as Approximate Bayesian Computation, will help us discover new G2P modelling and prediction principles that can be applied broadly across species, and for diverse trait adaptation scenarios applicable to breeding and the evolution of natural systems.  

The models and experimental resources we develop will provide gene-editing targets for further testing and improving the G2P model. 

TOP