Despite substantial investments by agencies such as GRDC, integrating physiological traits into plant breeding for rainfed conditions remains challenging, prompting the exploration of cost-efficient methods such as phenomics to facilitate breeding.

A major push is underway to make new knowledge on the physiology of wheat – and how trait variation affects growth – more accessible for commercial breeders.

Around the globe, agencies such as GRDC and their science partners have put substantial effort and investment into understanding the physiology of wheat and how trait variation affects growth, yet there are still gaps in the knowledge and tools required to deploy physiological traits in a breeding program.

This is because measuring many physiological traits can be complicated, expensive and highly technical. Also, in many cases the link between these traits and improved yield hasn’t yet been demonstrated. Demonstration of this link can be particularly difficult if the trait is only relevant in a particular season or only measurable indirectly with yield.

Further, it is difficult to justify overhauling a breeding program to incorporate a new trait with already required traits where there is uncertainty about value and the expense of measuring this is high.

Hence, the challenge is to demonstrate the value of a wheat trait and to make measuring the trait cost-efficient, to facilitate uptake.

The need is to develop methods that allow breeders to rapidly screen variation in novel traits in ways that are cheaper and easier than previously possible. This is where ‘phenomics’ could play a significant role in next-generation measurement techniques that can be deployed to facilitate breeding.

Accelerated integration of physiology-based wheat traits

In 2021, GRDC invested in a five-year project with Australian Grain Technologies (AGT) to accelerate the adoption of physiology-based traits in wheat. This is being done through R&D partnerships between academic researchers and commercial wheat breeders.

The aim is to deliver wheat varieties to Australian growers with significantly enhanced yield potential under limited water conditions, given the challenges of consistent and profitable production in rainfed environments.

In Phase 1 of the project, researchers and breeders identified three questions that would need to be answered ‘yes’ for any physiological traits to be included in the project:

1. Does the trait have a clear value proposition for growers?
2. Can the trait be accurately and cost-efficiently measured across environments?
3. Does variation for the trait already exist within Australia’s commercial breeding germplasm or will it need to be introduced?

The project surveyed 25 Australian and international plant physiologists across nine institutions to identify relevant physiological traits for Australian environments. The survey identified 27 above-ground plant physiological traits.

Each trait was assessed with a Decision Tree for Physiological Traits, developed by AGT.

A trait’s path through the Decision Tree identified challenges for validation of each trait(s) in the context of a commercial breeding program and these were carefully documented. No traits fully satisfied all requirements for validation in a commercial setting, due to uncertainties in the adoptability of the methods, the genetics and/or the value proposition.

However, three traits were considered “most promising to validate”. These were:

• rate and duration of grain filling;
• respiratory traits; and
• photosynthetic capacity and efficiency.

There was evidence that all three traits might be predicted with hyperspectral measurements of the plants from the visible to the infrared.

Validation of traits

Phase 2 of the project engaged three ‘trait teams’ drawn from project partners with expertise in the development and measurement of these traits from the Australian National University (ANU), University of Western Australia (UWA) and CSIRO.

Wheats that represented the extremes of genetic diversity for the identified traits were brought together with elite germplasm suited to the Australian wheatbelt.

Recognising the promise of the hyperspectral measurements, a fourth team was assembled to attempt to measure them from a drone – an unmanned aerial vehicle (UAV).

This collection of 96 lines was grown at three field sites within the AGT trial network in 2023.

Genetic diversity for the traits was identified in the elite material. Airborne hyperspectral imaging of the crops was found to be a possibility for a commercial setting. However, new predictive models for the traits would need to be built and tested.

Building on investment

Early in 2024, GRDC invested in an add-on project – AcceleTraitPlus. This project, led by CSIRO, will use the AGT field trials combined with the expertise and ground-based measurements of UWA, plus results from the main project, to develop high-throughput, spectral-based UAV methods for these traits.

The project will develop predictive models for the physiological traits for use with UAV hyperspectral imagery.

The aim is to develop methods that are accurate, scalable and cost-effective. UWA’s Nic Taylor says the project is about taking research that has been developed in the laboratory and tested in research plots, and then providing a pathway to deploy it in a breeding program.

In 2024, validation work is underway at CSIRO’s purpose-built, state-of-the-art, 290-hectare research facility, Boorowa Agricultural Research Station. This facility is equipped with a full array of sensors that will provide further valuable insights.

CSIRO has invested in a $450,000 HySpex Mjolnir, a high-quality hyperspectral drone from Finland. This complements its existing fleet of cameras  and will provide data directly for the purpose of building the models. This drone will be flown more frequently at the Boorowa and the New South Wales AGT site to add additional data to the data collected during the 2023 field season.

Machine learning and artificial intelligence techniques will be used to build the predictive models for the traits. The models themselves will also be investigated to identify key spectral features for the development of adoptable and scalable imaging solutions as the cost of hyperspectral technology drops.

More information: Dr Anton Wasson, Anton.Wasson@csiro.au, 02 6246 24739