Information of Growth Traits Is Helpful for Genetic Evaluation of Litter Size in Pigs

TL;DR

Using growth traits like age and lean meat percentage improves the genetic evaluation of litter size in pigs.

Contribution

This study is the first to show that combining production and reproduction traits in a multitrait model improves genetic evaluation accuracy for litter size in pigs.

Findings

The multitrait model increased prediction accuracy for litter size by up to 5.9 percentage points in validation scenarios.

Age at 100 kg had a positive impact on genetic evaluation, while lean meat percentage did not.

The multitrait model slightly reduced level bias in genetic evaluation.

Abstract

Litter size is an important trait in pig production. Selection accuracy for litter size is expected to be increased by genetic evaluation using data of production traits as additional information. This study investigated the improvement of genetic evaluation for litter size using a multitrait model including production traits. The multitrait model used in this study allows us to account for environmental correlation between litter size and production traits in the situation that one individual has only one record for a production trait while multiple records for litter size. The results show that the multitrait model including growth trait can improve genetic evaluation for litter size considerably. Therefore, it is recommended to use the multitrait model with data of both reproduction and production traits for routine genetic evaluation in pig breeding programs.

Litter size is an important trait in pig production. But selection accuracy for this trait is relatively low, compared with production traits. This study, for the first time, investigated the improvement of genetic evaluation of reproduction traits such as litter size in pigs using data of production traits as an additional information source. The data of number of piglets born alive per litter (NBA), age at 100 kg of body weight (Age100), and lean meet percentage (LMP) in a Yorkshire population were analyzed, using either a single-trait model or the multitrait model that allows us to account for environmental correlation between reproduction and production traits in the situation that one individual has only one record for a production trait while multiple records for a reproduction trait. Accuracy of genetic evaluation using single-trait and multitrait models were assessed by model-based accuracy (Rm) and validation accuracy (Rv). Two validation scenarios were considered. One scenario (Valid_r1) was that the individuals did not have a record of NBA, but Age100 and LMP. The other (Valid_r2) was that the individuals did not have a record for all the three traits. The estimate of heritability was 0.279 for Age100, 0.371 for LMP, and 0.076 for NBA. Genetic correlation was 0.308 between Age100 and LMP, 0.369 between Age100 and NBA, and 0.022 between LMP and NBA. Compared with the single-trait model, the multitrait model including Age100 increased prediction accuracy for NBA by 3.6 percentage points in Rm and 5.9 percentage points in Rv for the scenario of Valid_r1. The increase was 1.8 percentage points in Rm and 3.8 percentage points in Rv for the scenario of Valid_r2. Age100 also gained in the multitrait model but was smaller than NBA. However, LMP did not benefit from a multitrait model and did not have a positive contribution to genetic evaluation for NBA. In addition, the multitrait model, in general, slightly reduced level bias but not dispersion bias of genetic evaluation. According to these results, it is recommended to predict breeding values using a multitrait model including growth and reproduction traits.