Genetic parameters for methane production, intensity, and yield predicted from milk mid-infrared spectra throughout lactation in Holstein dairy cows

J Dairy Sci. 2024 Oct 4:S0022-0302(24)01192-5. doi: 10.3168/jds.2024-25231. Online ahead of print.

Abstract

Genetic selection to reduce methane (CH4) emissions is a promising solution for reducing the environmental impact of dairy cattle production. Before such a selection program can be implemented, however, it is necessary to have a better understanding of the genetic determinism of CH4 emissions and how this might influence other traits of interest. In this study, we performed a genetic analysis of 6 CH4 traits predicted from milk mid-infrared spectra. We predicted 4 CH4 traits in g/d (MeP, calculated using different prediction equations), one in g/kg of fat- and protein-corrected milk (MeI), and one in g/kg of dry matter intake (MeY). Using an external data set, we determined these prediction equations to be applicable in the range of 70 to 200 DIM. We then estimated genetic parameters in this DIM range using random regression models on a large data set of 829,025 spectra collected between January 2013 and February 2023 from 167,514 first- and second-parity Holstein cows. The 6 CH4 traits were found to be genetically stable throughout and across lactations, with average genetic correlations within a lactation ranging from 0.93 to 0.98, and those between lactations ranging from 0.92 to 0.98. All 6 CH4 traits were also found to be heritable, with average heritability ranging from 0.24 to 0.45. The average pairwise genetic correlations between the 6 CH4 traits ranged from -0.15 to 0.77, revealing that they are genetically distinct, including the 4 measurements of MeP. Of the 6 traits, 2 measures of MeP and MeI did not present antagonistic genetic correlations with milk yield, fat and protein contents, and SCS, and can probably be included in breeding goals with limited impact on other traits of interest.

Keywords: genetic correlations; heritability; methane prediction; random regression models.