Comparing carbon dioxide, oxygen, and methane exchanges, and heat production measured using GreenFeed and respiration chambers.

The aim of this study was to evaluate the precision and accuracy of GreenFeed (GF) Emission Monitoring system in measuring O, CO, and CH exchanges, and heat production (HP) when compared with respiration chambers (RC). Thirty-two lactating Nordic Red cows (634 ± 60.4 kg BW, 145 ± 63.1 DIM, 28.9 ± 6.7 kg milk/d, and parity 1.6 ± 1.16) allocated to 8 blocks based on BW, milk yield, DIM, and parity were used to compare CO, O, and CH exchanges measured by GF units (n = 2) versus RC (n = 4) in a switch-back design. The experimental period for every block was 5 wk; the cows were measured by GF during 2 wk, the third wk in RC, and another 2 wk by GF. The cows were fed a grass silage-based diet with constant forage-to-concentrate ratio for each cow during the study, ranging from 42:58 to 67:33, depending on their milk production at the beginning of study. For GF data, the records outside of the range of mean ± 2.5 × SD of all measurements over 4 wk for each cow were identified as outliers. Altogether, 214 out of 3,115 data points were removed as the outliers. Average CO, O, and CH exchanges, and HP in addition to their SD measured by GF during 4-wk periods (12,778 ± 1,137, 8,795 ± 773, 462 ± 57 g/d, and 131 ± 11 MJ/d) were close to those measured by RC (13,103 ± 1,421, 8,817 ± 941, and 453 ± 56 g/d, and 132 ± 14 MJ/d), respectively. Repeatability of O, CO, and CH exchanges, and HP for GF was 0.88, 0.88, 0.82, and 0.88, respectively, when calculated based on the averages of 2-wk periods. Accuracy of 0.90 was achieved in 6, 8, and 12 d for O, CO, and CH, respectively. Based on mixed model linear regression analysis, GF (averages of 4-wk data) estimated O exchanges of cows with a strong relationship with RC data (averages of 3-d; R = 0.89, root mean square prediction error [RMSPE] = 3.1% of mean, concordance correlation coefficient [CCC] = 0.94), followed by CO (R = 0.84, RMSPE = 3.5% of mean, CCC = 0.88) and CH (R = 0.79, RMSPE = 4.6% of mean, CCC = 0.87). Heat production estimated by GF was strongly related to the estimations made by RC (R = 0.88, RMSPE = 3.2% of mean, CCC = 0.93). The intercept and slope were not biased for O, CO, and HP, whereas for CH, intercept and slope were greater than 0 and less than 1, respectively. Therefore, GF can be used as an alternative to measure respiration gas exchanges and HP of dairy cows, which paves the way for large-scale measurements of energy metabolism and feed efficiency.