41 Using digital phenotypes for genomic evaluations in pigs.

Abstract As digital phenotyping techniques continue to showcase their potential for delivering large-scale, cost-effective, and accurate data in livestock production, expectations for their use in genomic selection for complex traits have risen. For instance, conformation traits, traditionally assessed using categorical scoring systems, can now be measured continuously by combining computer vision and machine learning. This approach could eliminate the inherent bias of subjective human assessments and provide more reliable data. This study aimed to estimate genetic parameters for complex traits in pigs recorded using visual and digital phenotyping methods and to investigate the impact of digital phenotypes in genetic evaluation models. The number of records for the visual (digital) traits was 577,056 (32,864) for the front leg, 563,567 (32,864) for the rear leg, and 605,867 (20,580) for weight measured at the end of a growth test. The pedigree included about one million animals. Heritabilities were estimated using single-trait models. Genetic correlations were obtained using two-trait animal models. Heritability estimates for the visual (digital) traits were 0.18 ± 0.00 (0.60 ± 0.02) for the front leg, 0.12 ± 0.00 (0.62 ± 0.02) for the rear leg, and 0.36 ± 0.03 (0.36 ± 0.03) for weight. Genetic correlations among visual traits ranged from -0.08 ± 0.02 to 0.25 ± 0.09, among digital traits from -0.19 ± 0.05 to -0.74 ± 0.02, and between visual and digital traits from 0.83 ± 0.03 to 0.97 ± 0.00. In general, heritability and genetic correlations were lower among visual traits than digital traits. High genetic correlations between visual and digital traits suggest substantial genetic overlap, demonstrating they are genetically the same trait. Our findings indicate that digital phenotyping—especially for subjective traits—increases heritability, which may lead to more accurate genetic evaluations and, ultimately, better response to selection. This study provides a foundation for integrating digital phenotyping into swine breeding programs.