Performance Comparison of the Prediction Models for Enteric Methane Emissions from Dairy Cattle

The enteric methane (CH₄) emission from dairy cattle is a significant factor contributing to anthropogenic climate change and the energy loss of animals. The objective of this study was to evaluate the prediction accuracy of the existing CH₄ estimation models from dairy cattle, and to identify the most reliable model for quantifying CH₄ emission. A database was compiled from 135 treatment means obtained from 81 peer-reviewed literatures, which included data on dietary composition, energy intake, and enteric CH₄ emission from dairy cattle. Forty existing dairy cattle prediction models were evaluated using this dataset based on the root mean square prediction error (RMSPE), concordance correlation coefficient (CCC), the ratio of RMSPE to standard deviation (RSR), and error decomposition indicators (ECT, ER, and ED). Results indicated that the RSR of model 38 was the lowest (0.71) but there were large prediction errors. Considering all evaluation indicators, model 21, which included dry matter intake (DMI), demonstrated the most robust predictive performance (RSR = 0.83, RMSPE = 14.41%, ECT = 3.42%, ER = 0.74%, ED = 96.75%, CCC = 0.58). Therefore, it is recommended for estimating enteric CH₄ emissions from dairy cattle. Future research will need to further improve the accuracy and robustness of enteric CH₄ prediction models by establishing a more comprehensive large-scale database, and expand the applicability of the model in various dairy farming systems.