Multienvironment genomic prediction in tetraploid potato.

Area of Science:

• Agricultural Science
• Genetics
• Plant Breeding

Background:

• Genomic prediction is crucial for accelerating crop improvement.
• Tetraploid potato breeding presents unique challenges due to its complex genome.
• Multienvironment testing is essential for understanding genotype-by-environment interactions.

Purpose of the Study:

• To evaluate multienvironment genomic prediction models for tetraploid potato.
• To optimize prediction strategies for breeders across different European regions and time points.
• To assess the impact of genetic variance heterogeneity on prediction accuracy.

Main Methods:

• Applied genomic prediction to 147 tetraploid potato varieties across 3 European locations over 2 years.
• Investigated three prediction scenarios for existing and new genotypes.
• Utilized mixed models to analyze variance-covariance matrices, accounting for heterogeneous genetic variances and correlations.

Main Results:

• Prediction accuracy varied by trait, with tuber weight benefiting significantly from models capturing heterogeneity.
• Dry matter prediction did not require complex multienvironment modeling.
• Including trials with low genetic correlation to the target decreased prediction abilities.

Conclusions:

• Effective genomic prediction in tetraploid potato requires clear scenarios, suitable training sets, and models reflecting G×E patterns.
• Model flexibility in handling genetic variance heterogeneity is key for traits like tuber weight.
• Strategic selection of training data is vital to avoid reduced prediction accuracy.