[en] To make use of genetic variation of methane (CH4) emissions in dairy cattle, a large number of individual CH4 emissions across entire populations are needed. As direct measurement of animal CH4 production is very laborious, alternatives like the prediction of CH4 emissions are needed. A promising approach for this purpose is the use of milk mid-infrared (MIR) spectral data, which is readily available from routine milk recording analyses. A published MIR prediction equation was applied to spectral data from routine milk recording in Switzerland from August to October 2015. Swiss spectral data has been standardised to reduce spectral variability between instruments. In total, 933,307 spectral data were available from the time period. In a first step only data from Swiss Holstein cows in their 1st to 4th lactation and with less than 306 days in milk were considered in the study, which resulted in 88,974 predicted daily CH4 production values from 40,082 cows. Semiparametric regression analysis using regression splines was carried out using the R package mgcv. Without any correction for diet, animal, and management influences, increasing CH4 production from lactation 1 to 4 was found and changing emissions during lactation were identified. Combining the predicted CH4 emissions with additional herd book data allows a more detailed population-wide screening for potentially high or low emitting animals. The information can be used for management purposes and genetic analyses. In a joint project, the analyses will be extended to other breeds and adding new reference measurements will increase accuracy and robustness of the prediction equation.
