Longitudinal modeling of residual carbon dioxide and residual feed intake in the Nordic Red dairy cattle

Feed utilization efficiency is an important trait in dairy production playing a significant role in reducing feed costs and lowering methane emission. One of the metrics used to measure feed efficiency in dairy cows is residual feed intake (RFI). This metric requires routine measurement of feed intake. Since there is a positive high correlation between heat production and carbon dioxide (CO₂) production on the one hand and heat production and efficiency on the other hand, residual carbon dioxide (RCO₂) might be a useful metric to improve feed efficiency. The objectives of this study were to model the trajectories of RCO₂ and RFI as well as to estimate their repeatabilities and correlations at different stages of lactation. Daily CO₂ output and feed intake were recorded from 46 primiparous Nordic Red dairy cows using two Greenfeed Emissions Monitoring™ systems from 2 to 305 days in milk (DIM). Edited data comprised 5 995 daily averages. To calculate predicted values of CO₂ and DM intake (DMI), prediction models were developed by fitting multiple regression models to observations. Subsequently, RCO₂ and RFI were calculated by subtracting predicted values of CO₂ and DMI from their corresponding actual observations. A random regression bivariate model was fitted to estimate repeatabilities and animal correlations within lactation at different DIMs between RCO₂ and RFI traits. The model fitted included fixed effects of year-month of recording, lactation month, fixed regressions as well as random regressions for the animal effect. The residual variance was considered to be heterogeneous. Repeatabilities and animal correlations of RCO₂ and RFI between selected DIM (for every 30 DIM i.e., 6, 36,…, 246 and 276) were calculated. Repeatability of RCO₂ was high at the beginning of lactation (0.72 at DIM 6) and decreased around the peak of milk production (0.27 at DIM 96) and again increased gradually toward the end of lactation. Similarly, RFI also had high repeatability at the beginning (0.86 at DIM 6); however, it decreased in mid-lactation (0.37 at DIM 156) and then increased toward the end of lactation. Animal correlations between RCO₂ and RFI were moderate to high on the same DIM and ranged from 0.37 to 0.88. Overall, we found that animals with higher CO₂ production than expected also consume more DMI than expected, but the moderate correlation between RCO₂ and RFI found in this study calls for more research to assess the potential of RCO₂ to become a new feed efficiency metric.