Multitrait and non-additive effects on genomic prediction for methane emission in Nelore cattle

• Grant number: 25/08156-7
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Start date: August 01, 2025
• End date: July 31, 2026
• Field of knowledge: Agronomical Sciences - Animal Husbandry - Genetics and Improvement of Domestic Animals
• Principal Investigator: Maria Eugênia Zerlotti Mercadante
• Grantee: Tainara Luana da Silva Soares
• Supervisor: Oscar Gonzalez-Recio
• Host Institution: Faculdade de Ciências Agrárias e Veterinárias (FCAV). Universidade Estadual Paulista (UNESP). Campus de Jaboticabal. Jaboticabal , SP, Brazil
• Institution abroad: University of Edinburgh, Scotland
• Associated to the scholarship: 23/14482-9 - Genomic prediction ability for methane emission traits in Nellore cattle, BP.DR

Abstract

Recently, livestock sectors have been exploring methods to reduce methane (CH4) emissions. The genetic selection of animals that emit less CH4 is both cumulative and permanent, making it an effective strategy to lower emissions in the livestock sector. Genomic multi-trait models have demonstrated improved prediction accuracies of genetic values when correlated traits are used. Research has indicated that enteric CH4 emission is moderate to high correlated to dry matter intake (DMI), average body weight (BW), and residual feed intake (RFI). In this context, the objective of this project is to evaluate the genomic prediction accuracy for CH4 emission using multi-trait models, including the traits DMI, BW, and RFI, using different statistical methods and validation strategies. A total of 1,162 Nellore animals genotyped with the BovineHD BeadChip (770k) and GGP Indicus (50k and 70k), will be evaluated. The animals genotyped with lower-density chips will be imputed to HD. The traits CH4, DMI, BW, and RFI will be included in the multi-trait model. The genomic prediction accuracy of the methods GBLUP, BayesA, and BayesB will be evaluated. The validation scenarios will be omitting all traits in the validation population, omitting CH4 and DMI, omitting CH4 and BW, and omitting CH4 and RFI. The results of the multi-trait models will be compared to those obtained using the single-trait model for CH¿ to determine if there is a difference in genomic prediction accuracy. The prediction ability will be given by the correlation (r) between the GEBV estimated in the validation population and the phenotype adjusted for fixed effects (y*). The prediction bias will be obtained by regressing y* on the GEBV, and the mean squared prediction error (MSPE) will be used to assess the model fit. (AU)