Distangaling genotype-by-environment and maternal effects in breed-specific genomic predictions for growth traits
Authors: Sara Nilson 1; Troy Rowan 1,2,3,4; Robert Schnabel 1,5; Jared Decker 1,5
Affiliations: 1) Division of Animal Sciences, University of Missouri, Columbia, MO; 2) Genetics Area Program, University of Missouri, Columbia, MO; 3) Department of Animal Science, University of Tennessee, Knoxville, TN; 4) College of Veterinary Medicine, Large Animal Clinical Science, University of Tennessee, Knoxville, TN; 5) Institute for Data Science and Informatics, University of Missouri, Columbia, MO
Keywords: Genomic selection/prediction
Genotype-by-environment interactions influence the productivity of beef cattle as one of the last agricultural species remaining outside for the entirety of their lives. Yet, maternal effects shape the potential of calves’ development before they are directly exposed to their environment and sets the baseline for lifetime growth. Currently, national genomic evaluation models do not include interactions despite their potential being acknowledged. Novel models including a direct or maternal genotype-by-environment interaction in addition to a genetic maternal effect, will disentangle the source and amount of variance that is contributed to production traits: birth weight, weaning weight, and yearling weight. Approximately 45,000 Red Angus and 98,000 Simmental cattle distributed across the United States with ~850,000 imputed SNPs were included in developing breed-specific genomic predictions. To test if prediction accuracy was further improved, SNPs were reduced to the top hits from analyses of selection. The top associations from envGWAS (associations between SNP genotypes and environmental variables indicative of local adaptation) were used to estimate genotype-by-environmental effects and GPSM (associations between SNP genotypes and time indicative of directional selection) for direct effects. Variance components were estimated with GCTA v1.93.2 and breeding values were estimated and predicted with the BLUPF90 suite. Utilizing a validation set of the youngest 10% of individuals, two measures of accuracy were calculated, accLR and accPA, to compare the genotype-by-environment inclusive models and the current national evaluation. The amount of variance explained by the effects is expected to shift among sources between the growth traits reflecting changes in environmental interactions among life stages. These genotype-by-environment models will influence selection decisions in the beef industry allowing for performance to be accurately predicted in a rapidly changing climate.