Resource allocation optimization with multi-trait genomic prediction for bread wheat (Triticum aestivum L.) baking...

Area of Science:

• Plant Breeding
• Genomics
• Quantitative Genetics

Background:

• Efficiently allocating phenotyping resources is crucial in plant breeding programs.
• Predicting complex, labor-intensive quality traits requires strategic approaches.

Purpose of the Study:

• To evaluate the predictive ability of multi-trait genomic prediction models.
• To assess the impact of varying phenotypic data depth from correlated traits on prediction accuracy.
• To optimize resource allocation in breeding programs for quality trait prediction.

Main Methods:

• Evaluated 495 wheat lines genotyped with genotyping-by-sequencing for eight baking quality traits.
• Compared single-trait and multi-trait genomic prediction models using cross-validation.
• Investigated effects of training population size, correlated trait phenotyping depth (50%, 100%), and number of correlated traits (1-3).

Main Results:

• No loss in predictive ability observed when reducing the training population by up to 30%.
• A multi-trait model using one highly correlated trait achieved optimal results considering resources and predictive gain.
• Reduced phenotyping of expensive traits is feasible using correlated trait data.

Conclusions:

• Multi-trait genomic prediction models enhance resource allocation by enabling targeted phenotyping of correlated traits.
• This strategy effectively predicts costly quality parameters, improving breeding program efficiency.
• Utilizing correlated traits is a strategic method to substitute phenotyping of high-cost, labor-intensive traits.