Evaluation of methane prediction equations for beef cattle fed high forage or high concentrate diets

Monday, July 21, 2014
Exhibit Hall AB (Kansas City Convention Center)
Paul Escobar , Agriculture and Agri-Food Canada, Lethbridge, AB, Canada
Karen A. Beauchemin , Lethbridge Research Centre, Agriculture and Agri-Food Canada, Lethbridge, AB, Canada
Masahito Oba , University of Alberta, Edmonton, AB, Canada
Abstract Text:

Enteric methane (CH4) emission is the major contributor to greenhouse gases from beef cattle farms. Many equations are available to predict enteric CH4 emissions from beef cattle, but the predictions vary substantially amongst equations. The aims of this study were to: 1) construct a database of enteric CH4 emissions for beef cattle fed forage and grain based diets from published literature, and 2) identify the most precise and accurate extant CH4 prediction models for beef cattle fed diets varying in forage content. The database was comprised of treatment means of enteric CH4 production from in vivo beef studies published from 2000 to 2013. Criteria for selecting data to include in the database were: animal description, intake, diet composition, and measurement of enteric CH4 production. Missing values were estimated using feed composition tables, nutritional software or by calculation from the diet description. Fifty-one equations that predict CH4 production from diet composition were evaluated. Precision and accuracy of the equations was evaluated using the concordance correlation coefficient (rc), bias correction factor (Cb) and root mean square prediction error (RMSPE, g/d), and then ranked highest to lowest based on rc. Statistical analysis was performed using JMP and SAS. The final database contained 39 studies and 163 treatment means that were divided into 2 subsets: a subset comprised of data from diets containing 40% or more forage and a subset comprised of data from diets containing less than 40% forage (dry matter basis). Using the complete database, equations with highest rc were: G (Ellis et al., 2009), IPCC (2006) and J (Ellis et al., 2009), with rc: 0.71, 0.67, 0.63, Cb: 0.98, 0.98, 0.98, and RMSPE: 55.0, 57.7, 61.4, respectively. For the high forage dataset, equations with highest rc were IPCC (2006), G (Ellis et al., 2009), and Non-linear 2 (Mills et al., 2003) with rc: 0.75, 0.75, 0.71, Cb: 0.96, 0.99, 0.95, and RMSPE: 47.9, 51.7, 52.5, respectively. For the low forage dataset, equations with highest rc were 9b (Ellis et al., 2007), G and P (Ellis et al., 2009), with rc: 0.52, 0.48, 0.47, Cb: 0.67, 0.81, 0.81, and RMSPE: 56.0, 64.0, 62.0, respectively. Ranking of extant CH4 prediction equations for their accuracy and precision differed with forage content of the diet. When used for cattle fed low forage diets, extant CH4 prediction models were generally imprecise and lacked accuracy. 

Keywords: beef cattle, enteric methane emission, models