Optimizing whole-genomic prediction for autotetraploid blueberry breeding

Area of Science:

• Plant breeding
• Genomics
• Quantitative genetics

Background:

• Blueberry (Vaccinium spp.) is a complex autopolyploid crop with significant health benefits.
• Genomic prediction is feasible in blueberry, offering reduced breeding cycles and increased genetic gain.
• High sequencing costs impede genome-based breeding in polyploid crops like blueberry.

Purpose of the Study:

• To evaluate the impact of training population size, composition, marker density, and sequencing depth on phenotype prediction accuracy in blueberry.
• To identify optimal resource allocation strategies for genomic prediction in blueberry breeding.

Main Methods:

• Utilized genotypic data from 86,930 markers across a breeding population of 1804 blueberry individuals.
• Assessed prediction accuracy for three traits: fruit firmness, fruit weight, and total yield.
• Analyzed the effects of varying training population sizes, marker densities, and sequencing depths.

Main Results:

• Marker density, sequencing depth, and training population size can be significantly reduced with minimal impact on prediction model accuracy.
• Resource allocation for genotyping and phenotyping can be optimized to maximize prediction accuracy.
• The study demonstrates substantial resource reduction for applying genomic prediction in blueberry.

Conclusions:

• Optimized genomic prediction strategies can accelerate the release of improved blueberry varieties.
• The findings provide a framework for efficient resource allocation in blueberry breeding programs.
• The proposed pipeline is applicable to optimizing genomic prediction in other diploid and polyploid species.