Assessing the CO2-prediction equation used in association with sniffers to estimate daily methane emissions in dairy cows with different feed efficiencies

Absolute methane (CH4) production (g/d) from sniffer systems can be estimated by multiplying the measured CH4 to carbon dioxide (CO2) ratio with predicted CO2 output (g/d) based on body weight and energy-corrected milk yield. However, this method may introduce bias in CO2 prediction due to differences in energy use efficiency, affecting CH4 estimates. Using respiration chamber data from 1,060 individual records across 37 studies, we simulated CH4 predicted from sniffers to examine the relationship of CH4 and CO2 estimates with feed efficiency, categorized by residual feed intake (RFI). Carbon dioxide production was overestimated in the low-RFI (efficient) cows (12.2 vs. 11.6 kg/d) and underpredicted in the high-RFI (inefficient) cows (12.3 vs. 13.4 kg/d). Consequently, CH4 emissions were overpredicted in efficient cows (380 vs. 362 g/d) and underpredicted in inefficient cows (384 vs. 419 g/d; P < 0.01). When cows were grouped by CH4 prediction error, observed CH4 emissions and yields increased across error groups, while predicted values showed the opposite trend (P < 0.01). These results indicate that CH4 estimates based on predicted CO2 may be systematically biased, which could affect animal ranking and genetic progress if used in selection indices without adjusting for feed intake or efficiency.