Abstract
Environmental differences between the breeding (B) and commercial production (C) environments may lead to genotype-by-environment interactions (GxE) i.e. re-ranking of breeding values of animals in the two environments. A substantial re-ranking implies genetic progress achieved in breeding programs is not realized in performance of production animals. The issues of GxE are not new and several solutions exist, however, there has not been much focus on solutions for breeding programs for poultry. This PhD-project investigated GxE interactions in breeding programs for poultry and solutions to improve genetic progress in these breeding programs.
A strong GxE interaction for body weight (BW) traits was found in broilers that were raised in B and C environments. Indications of GxE were significant re-ranking of breeding values, heterogeneous variances and different heritability for BW under B and C conditions. The genetic correlations between BW traits measured in B and C environments were in the range 0.48-0.54. Genetic variances of C traits were more than 2 times higher than those of B traits. Heritability of C traits (0.31-0.37) were higher than those of B traits (0.27-0.30).
In this thesis, several approaches to improve genetic gains of the poultry breeding programs in the presence of GxE have been investigated: phenotyping strategies, optimal modelling of traits, use of group records, and the use of genomic information. Different phenotyping strategies were compared in a breeding program for broilers that used genomic selection. It was found that when the genetic correlations between traits measured in B and C were 0.5 and 0.7, allocation of 70% and 30% hatched birds to B and C environments, respectively, for phenotype testing led to the highest genetic gains among the compared phenotyping strategies. When the genetic correlation was 0.9, moving birds to C did not improve genetic gains of the breeding scheme due to reduced selection intensity. Increasing proportion of birds moved to C (from 15 to 45%) could reduce rate of inbreeding of the breeding program.
Optimal modelling of traits was explored in a genetic analysis that was carried out for BW in broilers at different ages raised in a commercial environment. A statistical model was developed with the aim to increase predictive ability of the model for the traits affected by maternal effects. A criterion for the development of the statistical model was based on correlation between EBVs and corrected phenotypes of half-sib individuals. The statistical model also accounted for heterogeneous variances between sexes.
In breeding programs for village chicken, where strong GxE interactions are expected, the use of group records was a good option to increase genetic gain of the breeding programs. The use of group records from villages significantly improved genetic gains compared to the scheme without birds tested in the village although group records led to a slightly lower genetic gain compared to individual records.
In addition, the use of genomic information was exploited to improve genetic gain of poultry breeding programs in the presence of GxE. Compared to pedigree, genomic information increased accuracy of the prediction from individual records. The use of combined pedigree and genomic information in the ssGBLUP prediction from individual records substantially increased accuracy of EBVs of C traits by 31-37%, and reduced bias of prediction for genotyped selection candidates. Genomic information was also utilized to form groups, so that accuracy of the prediction from group records increased compared to the use of pedigree information.
Overall, differences between the breeding and production environments can lead to substantial GxE interactions. In the presence of GxE interactions, a breeding program for poultry should establish recording systems under the production environments in either individual or group records in order to ensure maximum genetic gains and provide customers with genotypes well adapted to the production environments. In addition, an optimal cross-validation procedure for the choice of statistical models is needed for genetic evaluations in poultry breeding programs as better modelling of traits is a low-cost approach to improve accuracy of selection.
A strong GxE interaction for body weight (BW) traits was found in broilers that were raised in B and C environments. Indications of GxE were significant re-ranking of breeding values, heterogeneous variances and different heritability for BW under B and C conditions. The genetic correlations between BW traits measured in B and C environments were in the range 0.48-0.54. Genetic variances of C traits were more than 2 times higher than those of B traits. Heritability of C traits (0.31-0.37) were higher than those of B traits (0.27-0.30).
In this thesis, several approaches to improve genetic gains of the poultry breeding programs in the presence of GxE have been investigated: phenotyping strategies, optimal modelling of traits, use of group records, and the use of genomic information. Different phenotyping strategies were compared in a breeding program for broilers that used genomic selection. It was found that when the genetic correlations between traits measured in B and C were 0.5 and 0.7, allocation of 70% and 30% hatched birds to B and C environments, respectively, for phenotype testing led to the highest genetic gains among the compared phenotyping strategies. When the genetic correlation was 0.9, moving birds to C did not improve genetic gains of the breeding scheme due to reduced selection intensity. Increasing proportion of birds moved to C (from 15 to 45%) could reduce rate of inbreeding of the breeding program.
Optimal modelling of traits was explored in a genetic analysis that was carried out for BW in broilers at different ages raised in a commercial environment. A statistical model was developed with the aim to increase predictive ability of the model for the traits affected by maternal effects. A criterion for the development of the statistical model was based on correlation between EBVs and corrected phenotypes of half-sib individuals. The statistical model also accounted for heterogeneous variances between sexes.
In breeding programs for village chicken, where strong GxE interactions are expected, the use of group records was a good option to increase genetic gain of the breeding programs. The use of group records from villages significantly improved genetic gains compared to the scheme without birds tested in the village although group records led to a slightly lower genetic gain compared to individual records.
In addition, the use of genomic information was exploited to improve genetic gain of poultry breeding programs in the presence of GxE. Compared to pedigree, genomic information increased accuracy of the prediction from individual records. The use of combined pedigree and genomic information in the ssGBLUP prediction from individual records substantially increased accuracy of EBVs of C traits by 31-37%, and reduced bias of prediction for genotyped selection candidates. Genomic information was also utilized to form groups, so that accuracy of the prediction from group records increased compared to the use of pedigree information.
Overall, differences between the breeding and production environments can lead to substantial GxE interactions. In the presence of GxE interactions, a breeding program for poultry should establish recording systems under the production environments in either individual or group records in order to ensure maximum genetic gains and provide customers with genotypes well adapted to the production environments. In addition, an optimal cross-validation procedure for the choice of statistical models is needed for genetic evaluations in poultry breeding programs as better modelling of traits is a low-cost approach to improve accuracy of selection.
| Original language | English |
|---|
| Publisher | Århus Universitet |
|---|---|
| Number of pages | 199 |
| ISBN (Print) | 978-87-93787-80-3 |
| ISBN (Electronic) | 978-94-6395-209-5 |
| DOIs | |
| Publication status | Published - Nov 2019 |