Solving the phenotype challenge for heat tolerance in wheat is opening up a new way forward to improve yield resilience to rising temperature

The challenge

Weather and production statistics over the past 20 years point to temperatures increasing and environments becoming more variable. To meet this challenge, effective field-based strategies for assessing heat tolerance phenotypes are needed; one such strategy has been developed by GRDC scholar Rebecca Thistlethwaite at the University of Sydney.

The three-tiered approach starts with thousands of lines tested in the field and in replicated yield plots. These were sown at two times:

• the first at the optimal time for the control plots; and
• the second in replica plots sown late to expose them to heat stress at flowering and grain filling.

In the second phase, the best-performing lines from the field trials are screened again (at the optimal sowing time), this time in triplicate plots.

One set of plots receives no further treatment. The other two are subject to in-field heat chambers: one at ambient temperatures and the other is heated. The heated chambers are linked to in-field generators that can deliver heat shock for up to five days. This proved effective in validating findings from the larger ‘time of sowing’ experiments.

The final evaluation is conducted in a temperature-controlled glasshouse to confirm the expression of key traits observed in the field. Carbon dioxide treatments are also used in the glasshouse to further explore the future impacts of climate change.

This approach generated the phenotypes needed to underpin a genomic selection strategy.

This strategy starts with optimisation algorithms, which are used to assemble a large set of diverse lines. These are then genotyped and phenotyped in multiple locations under carefully monitored environmental conditions. The resulting data helps train the genomic selection models used to calculate the breeding values of parents that are subsequently crossed to derive new, more heat-tolerant progenies.

Increasingly, machine learning is being used to optimise this process as the size and complexity of data increases.

This work began in collaboration with Agriculture Victoria in 2016 under the project US00081 (2016–20) and produced some excellent heat-tolerant materials.

The translation phase of this research continued with Agriculture Victoria, the Western Australian Department of Primary Industries and Regional Development (DPIRD) and InterGrain within a national research program (project UOS2201-001RTX, which is due to finish in 2026).

In-field heat chambers being used to screen for heat tolerance in wheat. Photo: The University of Sydney

A way forward

A system has been developed in which wheat populations are sown at Narrabri, NSW – the ‘mother site’ – where large numbers of lines are phenotyped.

A subset is selected based on its heat tolerance characteristics and then sown at five sites across the country. The selections are based on genetic diversity and the breeding values assigned by the genomic selection modelling (also called genomic estimated breeding values or GEBVs that score and rank lines relative to their heat tolerance).

Field trials are then used to assess the accuracy of the predictions against actual yield under stress.

The similarity between predicted yields and actual yield (under heat stress) serves to assess the accuracy of the models.

This accuracy has been increasing steadily as the project team integrates more environmental variables into the genomic selection models. Accuracy is expected to improve further as artificial intelligence (AI) is integrated into these processes.

Once a line with potential is  identified (meaning a high GEBV for heat tolerance with confirmed key site performance), it is tested in InterGrain’s multi-environment trial network to provide final ‘road testing’ of the new genotypes.

The calculation of GEBVs has been a game changer in the rapid development of new heat-tolerant germplasm. Pre-breeders can now make crosses among materials in the glasshouse without phenotyping. Following several cycles of recombination, the new diversity is rapidly fixed using double haploidy (a genotype formed when haploid cells undergo chromosome doubling).

This is now the roadway for delivering research outcomes to growers.

AI plays a key role in creating today’s capabilities when dealing with complex traits such as heat tolerance.

AI increasingly assists in managing the enormous amount of pedigree, phenotypic, genotypic and environmental information that needs to be considered when making timely germplasm development decisions.

Increasingly, we are able to better predict the genotype (the genetic makeup) required to maximise yield under high temperatures in specific environments and regions. This helps us to better tailor our pre-breeding strategies so that more relevant genetic materials are delivered to wheat breeders.

A capability breakthrough

It is now possible to rapidly develop new heat-tolerant germplasm by intercrossing progenies on the basis of their genetic breeding values early in the breeding process.

This results in an ability to target heat tolerance genetics despite the complexity, and  it radically reduces breeding cycle times.

The process now involves two to three breeding cycles based solely on GEBV predictions. The new combination of genes with high breeding values for heat tolerance are then rapidly fixed using double haploidy. This new material comes with associated ‘metaQTL’ – markers to genomic regions that are hotspots for heat tolerance.

This pre-breeding pipeline is, therefore, capable of delivering to breeders fully fixed lines with confirmed heat tolerance for inclusion in breeding pipelines.

This is only possible because the GEBVs are based on high-quality, accurate and, importantly, relevant phenotypes.

This system also ensures that the new, fully fixed materials (with heat tolerance) are rust resistant since all progenies are screened for rust prior to delivery to breeders.

With rust-resistant lines with high GEBVs for heat tolerance undergoing testing at more than 30 sites across the country within InterGrain’s national multi-environment trials, this is a system that will continue to deliver increasingly relevant and better-adapted sources of heat tolerance to breeders.

The heat tolerance traits will arrive in agronomically superior backgrounds for inclusion in commercial pipelines.

Project snapshot

Genotype mapping: DNA marker-assisted profiles were generated for more than 8000 new wheat lines.

Phenotyping: 8000 lines were phenotyped at multiple sowing dates in multiple environments nationally.

Phenotyping measurements: capture of phenological, morphological, yield, grain weight and screenings data.

Hyperspectral data: drone-captured data (multispectral, hyperspectral and thermal images) on thousands of plots at multiple dates over multiple environments and years. Each plot comprises thousands of pixels, all managed and stored.

Environmental data: hourly weather data was captured at every location, including temperature, humidity and rainfall.

Analytics: heat-tolerant genomic prediction models based on the genotypes, site-specific environmental data and an extensive phenotypic reference population (comprising multiple traits, years, sites and environments). The integration of  these data streams significantly improved the calculation of GEBV breeding values and their prediction accuracies in multiple environments nationally.

More information: Richard Trethowan, richard.trethowan@sydney.edu.au