Accuracy of methane emissions predicted from milk mid-infrared spectra and measured by laser methane detectors in Brown Swiss dairy cows

Abstract

Since heritability of CH 4 emissions in ruminants was demonstrated, various attempts to generate large individual animal CH 4 data sets have been initiated. Predicting individual CH 4 emissions based on equations using milk mid-infrared (MIR) spectra is currently considered promising as a low-cost proxy. However, the CH 4 emission predicted by MIR in individuals still has to be confirmed by measurements. In addition, it remains unclear how low CH 4 emitting cows differ in intake, digestion, and efficiency from high CH 4 emitters. In the current study, putatively low and putatively high CH 4 emitting Brown Swiss cows were selected from the entire Swiss herdbook population (176,611 cows), using an MIR-based prediction equation. Eventually, 15 low and 15 high CH 4 emitters from 29 different farms were chosen for a respiration chamber (RC) experiment in which all cows were fed the same forage-based diet. Several traits related to intake, digestion, and efficiency were quantified over 8 d, and CH 4 emission was measured in 4 open circuit RC. Daily CH 4 emissions were also estimated using data from 2 laser CH 4 detectors (LMD). The MIR-predicted CH 4 production (g/d) was quite constant in low and high emission categories, in individuals across sites (home farm, experimental station), and within equations (first available and refined versions). The variation of the MIR-predicted values was substantially lower using the refined equation. However, the predicted low and high emitting cows (n = 28) did not differ on average in daily CH 4 emissions measured either with RC or estimated using LMD, and no correlation was found between CH 4 predictions (MIR) and CH 4 emissions measured in RC. When individuals were recategorized based on CH 4 yield measured in RC, differences between categories of 10 low and 10 high CH 4 emitters were about 20%. Low CH 4 emitting cows had a higher feed intake, milk yield, and residual feed intake, but they differed only weakly in eating pattern and digesta mean retention times. Low CH 4 emitters were characterized by lower acetate and higher propionate proportions of total ruminal volatile fatty acids. We concluded that the current MIR-based CH 4 predictions are not accurate enough to be implemented in breeding programs for cows fed forage-based diets. In addition, low CH 4 emitting cows have to be characterized in more detail using mechanistic studies to clarify in more detail the properties that explain the functional differences found in comparison with other cows.