Gene-based prediction of heading time to target real-time and future climate adaptation in wheat

Abstract

Spring wheat production systems in Australia require fine-tuning of heading time in order to maximise the efficient use of resources (radiation, water, fertiliser) across the season, while minimising the risk of crop failure due to frost, heat and drought. Therefore heading time adaptation is critical to develop varieties that have the right combination of vernalisation, photoperiod and temperature-responsive genes for different regions. Additionally, dryland cropping depends on rainfall at sowing, so that sowing time varies across a wide window and must also be accounted for. Breeding programs have an interest in managing heading time in their research for targeting adaptation in current climates, for predicting genotype flowering time and for considering adaptation to future climates, given that mean winter season temperatures have already increased by 1 to 2°C over the last 50 years.

We have validated a model of heading time that accounts for the effects of the spring vs winter alleles for the three major VRN genes, and for the sensitive vs insensitive allele for PpdD1. The residual thermal time and residual VRN or PPD response of a new genotype with known alleles for VRN and PpdD1, can inferred either by undertaking a photoperiod and vernalisation experiment, or by a reverse calculation from a set of observed heading times with a new genotype. This has allowed construction of a robust model, incorporated into the APSIM cropping systems model, that is able to predict heading dates across the entire Australian wheatbelt, with high reliability and low error (RMSE < 4 days). In additional analyses, we applied the above method to an association panel to estimate SNP effects on these residual thermal time and VRN or PPD responses.

In this paper, we demonstrate how the model can be used to:

1. Target characterised genotypes to different regions where the heading time can be optimised on suitable criteria, e.g. maximising average yield, minimising risk of failure.
2. Estimate heading time of regional testing trials, from weather data alone.
3. Compare different genotypes for ‘broad’ adaptation across the wheatbelt in current of future climates to determine needs for introgression of novel alleles
4. Accurately predict the ‘regional’ flowering date of new genotypes, based solely on SNP data, e.g. based on DNA from DH progeny, or from characterised parents.

In conjunction, with CIMMYT, the model is now being tested in global applications for prediction.