Prediction of enteric methane production, yield and intensity of beef cattle using an intercontinental database

Henk van Lingen*, Mutian Niu, Ermias Kebreab, Sebastião Valadares Filho, JA Rooke, C-A Duthie, Angela Schwarm, Michael Kreuzer, Phil Hynd, Mariana Caetano, Maguy Eugène, Cécile Martin, Mark McGee, Padraig O'Kiely, Martin Hünerberg, Tim McAllister, Telma Berchielli, Juliana Messana, Nico Peiren, Alex ChavesEd Charmley, Andy Cole, Kristin Hales, Sang-Suk Lee, Alexandre Berndt, Christopher Reynolds, Les Crompton, Ali-Reza Bayat, David Yáñez-Ruiz, Zhongtang Yu, André Bannink, Jan Dijkstra, David Casper, Alexander Hristov

*Corresponding author for this work

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Enteric methane (CH4) production attributable to beef cattle contributes to global greenhouse gas emissions. Reliably estimating this contribution requires extensive CH4 emission data from beef cattle under different management conditions worldwide. The objectives were to: 1) predict CH4 production (g d¬-1 animal-1), yield [g (kg dry matter intake; DMI)-1] and intensity [g (kg average daily gain)-1] using an intercontinental database (data from Europe, North America, Brazil, Australia and South Korea); 2) assess the impact of geographic region, and of higher- and lower-forage diets. Linear models were developed by incrementally adding covariates. A K-fold cross-validation indicated that a CH4 production equation using only DMI that was fitted to all available data had a root mean square prediction error (RMSPE; % of observed mean) of 31.2%. Subsets containing data with ≥ 25% and ≤ 18% dietary forage contents had an RMSPE of 30.8 and 34.2%, with the all-data CH4 production equation, whereas these errors decreased to 29.3 and 28.4%, respectively, when using CH4 prediction equations fitted to these subsets. The RMSPE of the ≥ 25% forage subset further decreased to 24.7% when using multiple regression. Europe- and North America-specific subsets predicted by the best performing ≥ 25% forage multiple regression equation had RMSPE of 24.5 and 20.4%, whereas these errors were 24.5 and 20.0% with region-specific equations, respectively. The developed equations had less RMSPE than extant equations evaluated for all data (22.5 vs. 23.2%), for higher-forage (21.2 vs. 23.1%), but not for the lower-forage subsets (28.4 vs. 27.9%). Splitting the dataset by forage content did not improve CH4 yield or intensity predictions. Predicting beef cattle CH4 production using energy conversion factors, as applied by the Intergovernmental Panel on Climate Change, indicated that adequate forage content-based and region-specific energy conversion factors improve prediction accuracy and are preferred in national or global inventories.
Original languageEnglish
Article number106575
Number of pages19
JournalAgriculture, Ecosystems and Environment
Volume283
Early online date17 Jun 2019
DOIs
Publication statusPrint publication - 1 Nov 2019

Fingerprint

methane production
beef cattle
forage
methane
prediction
energy conversion
multiple regression
specific energy
Intergovernmental Panel on Climate Change
South Korea
greenhouse gas emissions
average daily gain
dry matter intake
dry matter
greenhouse gas
linear models
climate change
diet
fold
Brazil

Keywords

  • Empirical modeling
  • Geographical region
  • Forage content
  • Dietary variables
  • Methane emission

Cite this

van Lingen, Henk ; Niu, Mutian ; Kebreab, Ermias ; Valadares Filho, Sebastião ; Rooke, JA ; Duthie, C-A ; Schwarm, Angela ; Kreuzer, Michael ; Hynd, Phil ; Caetano, Mariana ; Eugène, Maguy ; Martin, Cécile ; McGee, Mark ; O'Kiely, Padraig ; Hünerberg, Martin ; McAllister, Tim ; Berchielli, Telma ; Messana, Juliana ; Peiren, Nico ; Chaves, Alex ; Charmley, Ed ; Cole, Andy ; Hales, Kristin ; Lee, Sang-Suk ; Berndt, Alexandre ; Reynolds, Christopher ; Crompton, Les ; Bayat, Ali-Reza ; Yáñez-Ruiz, David ; Yu, Zhongtang ; Bannink, André ; Dijkstra, Jan ; Casper, David ; Hristov, Alexander . / Prediction of enteric methane production, yield and intensity of beef cattle using an intercontinental database. In: Agriculture, Ecosystems and Environment. 2019 ; Vol. 283.
@article{c01a225529ee478f85138bd7c2920507,
title = "Prediction of enteric methane production, yield and intensity of beef cattle using an intercontinental database",
abstract = "Enteric methane (CH4) production attributable to beef cattle contributes to global greenhouse gas emissions. Reliably estimating this contribution requires extensive CH4 emission data from beef cattle under different management conditions worldwide. The objectives were to: 1) predict CH4 production (g d¬-1 animal-1), yield [g (kg dry matter intake; DMI)-1] and intensity [g (kg average daily gain)-1] using an intercontinental database (data from Europe, North America, Brazil, Australia and South Korea); 2) assess the impact of geographic region, and of higher- and lower-forage diets. Linear models were developed by incrementally adding covariates. A K-fold cross-validation indicated that a CH4 production equation using only DMI that was fitted to all available data had a root mean square prediction error (RMSPE; {\%} of observed mean) of 31.2{\%}. Subsets containing data with ≥ 25{\%} and ≤ 18{\%} dietary forage contents had an RMSPE of 30.8 and 34.2{\%}, with the all-data CH4 production equation, whereas these errors decreased to 29.3 and 28.4{\%}, respectively, when using CH4 prediction equations fitted to these subsets. The RMSPE of the ≥ 25{\%} forage subset further decreased to 24.7{\%} when using multiple regression. Europe- and North America-specific subsets predicted by the best performing ≥ 25{\%} forage multiple regression equation had RMSPE of 24.5 and 20.4{\%}, whereas these errors were 24.5 and 20.0{\%} with region-specific equations, respectively. The developed equations had less RMSPE than extant equations evaluated for all data (22.5 vs. 23.2{\%}), for higher-forage (21.2 vs. 23.1{\%}), but not for the lower-forage subsets (28.4 vs. 27.9{\%}). Splitting the dataset by forage content did not improve CH4 yield or intensity predictions. Predicting beef cattle CH4 production using energy conversion factors, as applied by the Intergovernmental Panel on Climate Change, indicated that adequate forage content-based and region-specific energy conversion factors improve prediction accuracy and are preferred in national or global inventories.",
keywords = "Empirical modeling, Geographical region, Forage content, Dietary variables, Methane emission",
author = "{van Lingen}, Henk and Mutian Niu and Ermias Kebreab and {Valadares Filho}, Sebasti{\~a}o and JA Rooke and C-A Duthie and Angela Schwarm and Michael Kreuzer and Phil Hynd and Mariana Caetano and Maguy Eug{\`e}ne and C{\'e}cile Martin and Mark McGee and Padraig O'Kiely and Martin H{\"u}nerberg and Tim McAllister and Telma Berchielli and Juliana Messana and Nico Peiren and Alex Chaves and Ed Charmley and Andy Cole and Kristin Hales and Sang-Suk Lee and Alexandre Berndt and Christopher Reynolds and Les Crompton and Ali-Reza Bayat and David Y{\'a}{\~n}ez-Ruiz and Zhongtang Yu and Andr{\'e} Bannink and Jan Dijkstra and David Casper and Alexander Hristov",
year = "2019",
month = "11",
day = "1",
doi = "10.1016/j.agee.2019.106575",
language = "English",
volume = "283",
journal = "Agriculture, Ecosystems and Environment",
issn = "0167-8809",
publisher = "Elsevier",

}

van Lingen, H, Niu, M, Kebreab, E, Valadares Filho, S, Rooke, JA, Duthie, C-A, Schwarm, A, Kreuzer, M, Hynd, P, Caetano, M, Eugène, M, Martin, C, McGee, M, O'Kiely, P, Hünerberg, M, McAllister, T, Berchielli, T, Messana, J, Peiren, N, Chaves, A, Charmley, E, Cole, A, Hales, K, Lee, S-S, Berndt, A, Reynolds, C, Crompton, L, Bayat, A-R, Yáñez-Ruiz, D, Yu, Z, Bannink, A, Dijkstra, J, Casper, D & Hristov, A 2019, 'Prediction of enteric methane production, yield and intensity of beef cattle using an intercontinental database', Agriculture, Ecosystems and Environment, vol. 283, 106575. https://doi.org/10.1016/j.agee.2019.106575

Prediction of enteric methane production, yield and intensity of beef cattle using an intercontinental database. / van Lingen, Henk; Niu, Mutian; Kebreab, Ermias ; Valadares Filho, Sebastião ; Rooke, JA; Duthie, C-A; Schwarm, Angela ; Kreuzer, Michael ; Hynd, Phil ; Caetano, Mariana ; Eugène, Maguy ; Martin, Cécile ; McGee, Mark ; O'Kiely, Padraig ; Hünerberg, Martin ; McAllister, Tim ; Berchielli, Telma ; Messana, Juliana ; Peiren, Nico ; Chaves, Alex ; Charmley, Ed ; Cole, Andy ; Hales, Kristin ; Lee, Sang-Suk ; Berndt, Alexandre ; Reynolds, Christopher ; Crompton, Les ; Bayat, Ali-Reza ; Yáñez-Ruiz, David ; Yu, Zhongtang ; Bannink, André ; Dijkstra, Jan ; Casper, David ; Hristov, Alexander .

In: Agriculture, Ecosystems and Environment, Vol. 283, 106575, 01.11.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Prediction of enteric methane production, yield and intensity of beef cattle using an intercontinental database

AU - van Lingen, Henk

AU - Niu, Mutian

AU - Kebreab, Ermias

AU - Valadares Filho, Sebastião

AU - Rooke, JA

AU - Duthie, C-A

AU - Schwarm, Angela

AU - Kreuzer, Michael

AU - Hynd, Phil

AU - Caetano, Mariana

AU - Eugène, Maguy

AU - Martin, Cécile

AU - McGee, Mark

AU - O'Kiely, Padraig

AU - Hünerberg, Martin

AU - McAllister, Tim

AU - Berchielli, Telma

AU - Messana, Juliana

AU - Peiren, Nico

AU - Chaves, Alex

AU - Charmley, Ed

AU - Cole, Andy

AU - Hales, Kristin

AU - Lee, Sang-Suk

AU - Berndt, Alexandre

AU - Reynolds, Christopher

AU - Crompton, Les

AU - Bayat, Ali-Reza

AU - Yáñez-Ruiz, David

AU - Yu, Zhongtang

AU - Bannink, André

AU - Dijkstra, Jan

AU - Casper, David

AU - Hristov, Alexander

PY - 2019/11/1

Y1 - 2019/11/1

N2 - Enteric methane (CH4) production attributable to beef cattle contributes to global greenhouse gas emissions. Reliably estimating this contribution requires extensive CH4 emission data from beef cattle under different management conditions worldwide. The objectives were to: 1) predict CH4 production (g d¬-1 animal-1), yield [g (kg dry matter intake; DMI)-1] and intensity [g (kg average daily gain)-1] using an intercontinental database (data from Europe, North America, Brazil, Australia and South Korea); 2) assess the impact of geographic region, and of higher- and lower-forage diets. Linear models were developed by incrementally adding covariates. A K-fold cross-validation indicated that a CH4 production equation using only DMI that was fitted to all available data had a root mean square prediction error (RMSPE; % of observed mean) of 31.2%. Subsets containing data with ≥ 25% and ≤ 18% dietary forage contents had an RMSPE of 30.8 and 34.2%, with the all-data CH4 production equation, whereas these errors decreased to 29.3 and 28.4%, respectively, when using CH4 prediction equations fitted to these subsets. The RMSPE of the ≥ 25% forage subset further decreased to 24.7% when using multiple regression. Europe- and North America-specific subsets predicted by the best performing ≥ 25% forage multiple regression equation had RMSPE of 24.5 and 20.4%, whereas these errors were 24.5 and 20.0% with region-specific equations, respectively. The developed equations had less RMSPE than extant equations evaluated for all data (22.5 vs. 23.2%), for higher-forage (21.2 vs. 23.1%), but not for the lower-forage subsets (28.4 vs. 27.9%). Splitting the dataset by forage content did not improve CH4 yield or intensity predictions. Predicting beef cattle CH4 production using energy conversion factors, as applied by the Intergovernmental Panel on Climate Change, indicated that adequate forage content-based and region-specific energy conversion factors improve prediction accuracy and are preferred in national or global inventories.

AB - Enteric methane (CH4) production attributable to beef cattle contributes to global greenhouse gas emissions. Reliably estimating this contribution requires extensive CH4 emission data from beef cattle under different management conditions worldwide. The objectives were to: 1) predict CH4 production (g d¬-1 animal-1), yield [g (kg dry matter intake; DMI)-1] and intensity [g (kg average daily gain)-1] using an intercontinental database (data from Europe, North America, Brazil, Australia and South Korea); 2) assess the impact of geographic region, and of higher- and lower-forage diets. Linear models were developed by incrementally adding covariates. A K-fold cross-validation indicated that a CH4 production equation using only DMI that was fitted to all available data had a root mean square prediction error (RMSPE; % of observed mean) of 31.2%. Subsets containing data with ≥ 25% and ≤ 18% dietary forage contents had an RMSPE of 30.8 and 34.2%, with the all-data CH4 production equation, whereas these errors decreased to 29.3 and 28.4%, respectively, when using CH4 prediction equations fitted to these subsets. The RMSPE of the ≥ 25% forage subset further decreased to 24.7% when using multiple regression. Europe- and North America-specific subsets predicted by the best performing ≥ 25% forage multiple regression equation had RMSPE of 24.5 and 20.4%, whereas these errors were 24.5 and 20.0% with region-specific equations, respectively. The developed equations had less RMSPE than extant equations evaluated for all data (22.5 vs. 23.2%), for higher-forage (21.2 vs. 23.1%), but not for the lower-forage subsets (28.4 vs. 27.9%). Splitting the dataset by forage content did not improve CH4 yield or intensity predictions. Predicting beef cattle CH4 production using energy conversion factors, as applied by the Intergovernmental Panel on Climate Change, indicated that adequate forage content-based and region-specific energy conversion factors improve prediction accuracy and are preferred in national or global inventories.

KW - Empirical modeling

KW - Geographical region

KW - Forage content

KW - Dietary variables

KW - Methane emission

U2 - 10.1016/j.agee.2019.106575

DO - 10.1016/j.agee.2019.106575

M3 - Article

VL - 283

JO - Agriculture, Ecosystems and Environment

JF - Agriculture, Ecosystems and Environment

SN - 0167-8809

M1 - 106575

ER -