To reduce the risk of losses from extreme heat and frost, particularly during flowering, it is important to match canola phenology – the annual sequence of plant development – to the optimal flowering window. Decision-support tools help growers achieve this – to better match canola varieties to their production environments. But such tools first need considerable data and ground-truthing by experts.

New knowledge and methods are now coming onstream to make these tools even more effective for growers pursuing improved canola productivity.

Phenology is driven by genetics and environment. In canola, thermal (temperature) and vernal (cold) responses in phenology across representative environments can be reduced to a set of parameters unique to each variety.

Representing flowering time in this way exploits existing biological knowledge that can be linked to process-based simulation models (for example, the Agricultural Production Systems sIMulator, or APSIM) to predict when a variety will flower in a given environment.

The Canola Flowering Calculator is an example of this.

However, up to now, such phenology modelling across different environments has required time-costly field-based assessments of new varieties as they are released, delaying variety optimisation. A better understanding and integration of phenology genetics into the modelling process is needed to support the development of a flexible and widely applicable tool.

Multidisciplinary approach

In response to this need and building on prior GRDC investments with CSIRO, the Optimising Canola Phenology project was initiated in 2019 to increase understanding of phenology genetics. The objective was to use this knowledge to shorten the cycle for evaluation of key phenology parameters of new varieties based on their genome. This would reduce the dependence on field-based assessments and enable growers to use the latest available germplasm more effectively.

The knowledge being developed in this project will help canola breeders develop new varieties with targeted phenology genetics.

The project has tackled this challenge with a collaborative, multidisciplinary approach that has brought together expertise in crop process modelling, agronomy, genomics, phenomics and machine learning from across CSIRO.

The team has surveyed phenology, genomic and transcriptomic variation in 350 diverse canola varieties in field and controlled environments, amassing more than 300,000 measurements from 14 field environments spanning Australian canola growing regions.

Using this data, genomic-based models were developed to predict phenology traits and derive parameters, which will be used in crop process models (in APSIM) to predict flowering time for varieties across environments.

Assessments of field and controlled-environment experiments have revealed considerable diversity in canola phenology under Australian conditions.

Using machine-learning-based, genome-wide association analysis, genome variation influencing key phenology developmental stages has been identified. This has confirmed the involvement of a number of known phenology genes, but also potentially novel candidates that will be further explored for marker development.

Phenology traits have been reliably predicted for varieties observed in field experiments from their genome, consistent with high heritability of phenology traits detected in the study.

To build a genomic model that is applicable to a broad range of environments, this genomic prediction step will be linked with the APSIM simulation model and a prototype should be available for further field-based validation in 2022.

More information: Shannon Dillon, 02 6246 4834, shannon.dillon@csiro.au; Chris Helliwell, chris.helliwell@csiro.au