Efficient ways to combine data from broiler and layer chickens to account for sequential genomic selection

Area of Science:

• Animal Breeding and Genetics
• Quantitative Genetics
• Poultry Science

Background:

• Traditional broiler breeding focuses on growth traits, with reproductive traits evaluated separately.
• Ignoring broiler performance data during reproductive trait evaluation can lead to inaccurate genetic predictions.
• Optimizing prediction accuracy and efficiency is crucial for effective genetic selection in broiler populations.

Purpose of the Study:

• To determine the most accurate, unbiased, and time-efficient method for jointly evaluating reproductive and broiler traits.
• To assess the impact of incorporating broiler performance data on the prediction quality of reproductive traits.

Main Methods:

• Utilized a large dataset including pedigree (577K birds), genotypes (146K), and phenotypes for reproductive (egg production, fertility, hatchability) and broiler traits (body weight, breast meat percent, fat percent, residual feed intake).
• Sequentially added broiler data (phenotypes and genotypes) to a baseline model of reproductive traits to evaluate prediction quality.
• Tested different core sizes (7K, 12K, 19K animals) within the prediction algorithm to identify the most time-efficient approach.

Main Results:

• Incorporating broiler data (RE_BR_GE) minimally affected accuracy for egg production and hatchability but improved fertility predictions.
• Including broiler data significantly reduced bias and improved dispersion for all reproductive traits, especially for low heritability traits like fertility.
• Ignoring broiler data led to substantial bias and dispersion issues, with accuracy losses up to 17.5% for low heritability traits.

Conclusions:

• Integrating broiler performance data is essential for maximizing accuracy and minimizing bias in genetic predictions for reproductive traits.
• The impact of including broiler data is more pronounced for traits with lower heritability.
• A random core of 7,000 animals in the prediction algorithm offers the most time-efficient approach for joint trait evaluation.