Image-based phenomic prediction can provide valuable decision support in wheat breeding

Area of Science:

• Agricultural Science
• Plant Breeding
• Genetics

Background:

• Traditional breeding relies on visual assessment in early generations.
• Advancements in genome sequencing and high-throughput phenotyping enable data-driven selection.
• Genotype-by-environment (GxE) interactions influence trait expression and breeding success.

Purpose of the Study:

• To investigate if GxE interactions of secondary traits are less complex than target traits (e.g., yield).
• To evaluate the efficacy of phenomic selection (PS) for identifying superior genotypes in early generations.
• To explore the physiological basis of target traits and identify beneficial ideotypes.

Main Methods:

• Analysis of 45 winter wheat varieties across 5 diverse environments.
• Utilized linear and factor-analytic (FA) mixed models to estimate GxE interactions.
• Employed drone-derived data (plant height, leaf area, tiller density) for dynamic trait assessment.

Main Results:

• Secondary traits and grain protein content exhibited minimal GxE interactions.
• Yield GxE modeling required a more complex FA model with two factors.
• A PS model achieved modest predictive correlations for yield performance (0.43), yield stability (0.30), and protein content (0.34).

Conclusions:

• Phenomic selection, leveraging less complex GxE interactions of secondary traits, is viable for early generation trials.
• PS provides physiological insights beyond simple prediction accuracy, complementing genomic selection (GS).
• An ideotype was identified that may mitigate negative pleiotropic effects between yield and protein content.