Evaluating The Accuracy Of Models To Predict Rumemal Volatile Fatty Acids And Methane In Cattle

The rumen microbial degradation of forage produce the two important metabolites of volatile fatty acids(volatile fatty,acids,VFA)and methane,which VFA provides energy for the animal,and the molar ratio of VFA impact of health and reproduction of ruminant;methane as the important energy loss in the diet of emissions to the air to pollute the environment.This study mainly evaluated and analyzed the classical ruminal VFA models and methane models abroad,and discussed the reasons that affect the accuracy of model estimation,so as to provide reference for future development of models.Experiment 1:This study selected 3 classical models of rumen VFA stoichiometry,such as MUR(1982),DIJ(1992)and BAN(2006).The VFA data was selected from 27 articles published by Chinese scientists,including 17 SCI articles,9 chinese articles from Chinese Core Journals and 1 unpualished manuscript,and published includes that data of diet ingredients,body weight,dry matter intake,feed additives,total VFA concentration and the molar proportion of individual VFA.Mean Squared Prediction Error(MSPE)and Consistent Correlation Coefficient(CCC)methods were employed to evaluate the prediction accuracy,and factors influencing the accuracy were also discussed.Results:BAN model had the highest prediction accuracy of acetate molar proportion(R~2=0.33;P<0.001,RMSPE=6.39%),with overall bias being 59.3%.The accuracy of selected three models to predict molar proportion of propionate,butyrate and branched chain fatty acids were low with(R~2<0.20,P>0.001).Conclusion:BAN model had highest prediction accuracy to predict acetate molar proportion among the three models,but the prediction accuracy was still low with the error mainly coming from the overall bias.Three tested models were unable to accurately predict the molar proportion of propionate,butyrate and branched chain fatty acids.Therefore,it would be necessary to use Chinese dairy cow rumen VFA data to build new VFA stoichiometry model in future.Experiment 2:Seventeen beef cattle were selected in the Hunan Wangcheng beef farm to measure body weight,dry matter and nutrient intake,enteric methane emissions.The selected6 classic models included:equations developed by dry matter intake(DMI),such as model 1and 2;equations developed by fiber intake,such as model 3 and 4;equations developed by gross energy intake(GEI),such as model 5 and 6.Mean squared prediction error(MSPE)and consistent correlation coefficient(CCC)methods were employed to evaluate the prediction accuracy,and factors influencing the accuracy were also discussed.Results:Model 5(CCC=0.86)had the highest prediction accuracy among six models,next for Model 1(CCC=0.74)and 6(CCC=0.79),lowest for model 2(CCC=0.66),3(CCC=0.22)and 4(CCC=0.54).Model 1 and 2 had overall bias of 48.8%and 70.3%respectively.Model 3 had greatest deviation of regression slope from unity(47.6%),while model 4 had 28.6%deviation of regression slope from unity and 29.2%overall bias.The current results indicate that model 5developed based on world-wide data by IPCC(2006)Tier 2 is the most accurate model to predict the enteric methane emissions in beef cattle.