In pig and poultry breeding programs, the animals that provide food for consumers are usually crossbred, while selection takes place in their purebred parental lines. Breeders therefore need to estimate breeding values of purebred (PB) animals for the performance of their crossbred (CB) offspring.
The accuracy of such breeding values may be higher when CB instead of PB data is used, especially when the genetic correlation between PB and CB performance is low. In this study, we used data from broiler chicken to show that using CB instead of PB data was beneficial when the correlation was 80%, but not when the correlation was 96%.
Breeding values can be estimated with a technique called genomic prediction. Genomic prediction makes use of a reference population that consists of animals that have both phenotypes and marker genotypes available. This information is used to estimate breeding values of selection candidates that only have marker genotypes available. Genomic prediction is especially useful for traits that cannot be measured on the selection candidates themselves. For example, the performance of crossbred offspring cannot be measured on purebred selection candidates.
Genomic prediction for CB performance
Estimating breeding values for CB performance with genomic prediction can be done in several ways. One important decision breeders have to make, is whether to use CB animals in the reference population, or PB animals. The expectation is that the use of a CB animals is preferred over PB animals when the genetic correlation between PB and CB performance is substantially lower than 100%. When a CB reference population is used, the model can be improved even further by separating genotypes of CB into separate components, one for each parental line.
Benefits of CB data over PB data
Our results showed that, a CB reference population resulted in similar or higher accuracy than a PB reference population when the correlation between purebreds and crossbreds was 80%. In contrast, when the correlation was 96%, accuracy was lower with a CB reference population than with a PB reference population. In conclusion, this study suggests that the benefit of using CB instead of PB data for the accuracy of GEBVCB is larger when the genetic correlation between purebreds and crossbreds is lower. However, breeders need to determine whether this difference in accuracy is large enough to justify the collection of data on CB animals, which may be costly and time consuming.