Short communication: Variation in feed efficiency hampers use of carbon dioxide as a tracer gas in measuring methane emissions in on-farm conditions.

Breeding cows for low CH4 emissions requires that the trait is variable and that it can be recorded with low cost from an adequate number of individuals and with high precision, but not necessarily with high accuracy if the trait is measured with high repeatability. The CH4:CO2 ratio in expired breath is a trait often used as a tracer with the production of CO2 predicted from body weight (BW), energy-corrected milk yield, and days of pregnancy. This approach assumes that efficiency of energy utilization for maintenance and production is constant. Data (307 cow-period observations) from 2 locations using the same setup for measuring CH4 and CO2 in respiration chambers were compiled, and observed production of CH4 and CO2 was compared with the equivalent predicted production using 2 different approaches. Carbon dioxide production was predicted using a previously reported model based on metabolic BW and energy-corrected milk production and a currently developed model based on energy requirements and the relationship between observed CO2 and heat production (models 1 and 2, respectively). Animals used were categorized (low, medium, and high efficiency) according to (1) residual feed intake and (2) residual milk production. Model 1 underestimated CH4 production by 15%, whereas model 2 overestimated CH4 by 1.4% for the whole database. Model 1 underestimated CO2 production by 2.8 and 0.9 kg/d for low- and high-efficiency cows, respectively, whereas model 2 underestimated CO2 production by 0.9 kg/d for low-efficient animals but overestimated it by 1.2 kg/d for high-efficiency cows. Efficient cows produce less heat, and consequently CO2, per unit of metabolic body weight and energy-corrected milk than inefficient cows, challenging the use of CO2 as a tracer gas. Because of biased estimates of CO2 production, the models overestimated CH4 production of high-efficiency cows by, on average, 17% relative to low-efficiency cows, respectively. Selecting low CH4-emitting cows using a CO2 tracer method can therefore favor inefficient cows over efficient cows.