Predicting enteric methane emission in lactating Holsteins based on reference methane data collected by the GreenFeed system

R Liu; D Hailemariam; T Yang; F Miglior; F Schenkel; Z Wang; P Stothard; S Zhang; G Plastow

doi:10.1016/j.animal.2022.100469

Predicting enteric methane emission in lactating Holsteins based on reference methane data collected by the GreenFeed system

Animal. 2022 Mar;16(3):100469. doi: 10.1016/j.animal.2022.100469. Epub 2022 Feb 18.

Authors

R Liu¹, D Hailemariam², T Yang³, F Miglior⁴, F Schenkel⁴, Z Wang³, P Stothard³, S Zhang⁵, G Plastow³

Affiliations

¹ Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton T6G 2R3, Canada; Key Laboratory of Animal Breeding and Reproduction of Ministry of Education, Hauzhong Agricultural University, Wuhan 430070, China.
² Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton T6G 2R3, Canada. Electronic address: hailemar@ualberta.ca.
³ Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton T6G 2R3, Canada.
⁴ Center for Genetic Improvement of Livestock, University of Guelph, Guelph, ON, Canada.
⁵ Key Laboratory of Animal Breeding and Reproduction of Ministry of Education, Hauzhong Agricultural University, Wuhan 430070, China.

PMID: 35190321
DOI: 10.1016/j.animal.2022.100469

Abstract

Methane emission is not included in the current breeding goals for dairy cattle mainly due to the expense and difficulty in obtaining sufficient data to generate accurate estimates of the relevant traits. While several models have been developed to predict methane emission from milk spectra using reference methane data obtained by the respiration chamber, SF6 and sniffer methods, the prediction of methane emission from milk mid-infrared (MIR) spectra using reference methane data collected by the GreenFeed system has not yet been explored. Methane emission was monitored for 151 cows using the GreenFeed system. Prediction models were developed for daily and average (for the trial period of 12 or 14 days) methane production (g/d), yield (g/kg DM intake (DMI)) and intensity (g/kg of fat- and protein-corrected milk) using partial least squares regression. The predictions were evaluated in 100 repeated validation cycles, where animals were randomly partitioned into training (80%) and testing (20%) populations for each cycle. The best performing model was observed for average methane intensity using MIR, parity and DMI with validation coefficient of determination (R²_val) and RMSE of prediction of 0.66 and 4.7 g/kg of fat- and protein-corrected milk, respectively. The accuracy of the best models for average methane production and average methane yield were poor (R²_val = 0.28 and 0.12, respectively). A lower accuracy of prediction was observed for methane intensity and production (R²_val = 0.42 and 0.17) when daily records were used while prediction for methane yield was comparable to that for average methane yield (R²_val = 0.16). Our results suggest the potential to predict methane intensity with moderate accuracy. In this case, prediction models for average methane values were generally better than for daily measures when using the GreenFeed system to obtain reference methane emission measurements.

Keywords: Dairy cows; Methane intensity; Prediction; Production; Yield.

MeSH terms

Animals
Cattle
Diet / veterinary
Female
Intestine, Small
Lactation*
Methane*
Milk / chemistry
Pregnancy

Substances

Methane