Genomic prediction ability for feed efficiency traits using different models and pseudo-phenotypes under several...

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Animal Breeding

Background:

• Improving feed efficiency (FE) in cattle is crucial for sustainable livestock production.
• Genomic selection offers a promising approach to enhance FE traits, which are challenging to measure directly.
• Limited research exists on implementing genomic selection for FE in indicine cattle at a large commercial scale.

Purpose of the Study:

• To evaluate the feasibility of genomic selection for FE traits in Nelore cattle.
• To compare different prediction models and pseudo-phenotypes under various validation strategies.
• To identify optimal genomic approaches for predicting FE-related traits in Nelore cattle.

Main Methods:

• Utilized phenotypic and genotypic data from 4,329 and 3,467 Nelore cattle, respectively.
• Assessed six prediction methods including single-step genomic best linear unbiased prediction and Bayesian approaches (Bayes A, B, Cπ, BLASSO, R).
• Employed pseudo-phenotypes: adjusted phenotypes (Y*), estimated breeding values (EBV), and deregressed EBV (DEBV), with three validation strategies (random, age, EBV accuracy).

Main Results:

• Prediction methods yielded similar results for prediction ability (PA) and bias.
• Random cross-validation demonstrated the highest PA (0.17) compared to EBV accuracy (0.14) and age (0.13).
• PA was significantly higher using Y* (30.0-34.3% higher) than EBV or DEBV as pseudo-phenotypes.

Conclusions:

• Random validation is recommended for populations with predominantly young animals and traits with limited historical data.
• Age-based validation is suitable for high heritability traits, enabling prediction of future genetic merit.
• These findings provide valuable insights for breeders to select effective genomic prediction strategies for FE traits in Nelore cattle.