Evaluation of models of enteric methane emissions in finishing steers

Accurate estimation of enteric CH4 emissions (i.e., MJ or g of CH4/ day)
in the ruminant sector is necessary for properly determining GHG emissions
and developing measuring, reporting, and verification programs. However,
measuring enteric CH4 emissions under commercial conditions presents
challenges due to technical and economic constraints. Thus, using
prediction models allows for estimating individual enteric CH4 emissions
according to animal and dietary characteristics. When evaluated in
independent datasets, there is limited information regarding the accuracy
and precision of the reported equations to predict enteric CH4 emissions
in steers fed a finishing diet. This study evaluated the predictive
performance of various reported equations for estimating enteric CH4
production in finishing steers. Data used to assess the prediction
equations came from 446 steers from five experiments during the finishing
phase. Gas flux, nutrient consumption, and animal growth performance were
evaluated in each experiment. Seventy-two equations were compared based on
the mean square prediction error (MSPE), the decomposition of the root
MSPE (RMSPE), and the concordance correlation coefficient (CCC).
Prediction equations for estimating enteric CH4 emissions showed lower
sensitivity with RMSPE (as a percentage of the observed mean) ranging from
17.79 to 99.21 and CCC ranging from -0.07 to 0.21. The decomposition of
the RMSPE showed mean bias (as a percentage of the RMSPE) ranging from
0.14 to 94.87 and slope bias (as a percentage of the RMSPE) ranging from 0
to 25.24. In addition, 49 equations underpredicted (ranging from 0.6 to
50.1%) and 23 overpredicted (ranging from 3.7 to 96.1%) enteric CH4
emissions. The prediction of enteric CH4 production showed greater CCC and
lower MSPE when the intake of dry matter and ether extract were included
as predictors. These results suggest a limited ability to predict enteric
CH4 by steers during the finishing phase. Further efforts are required to
generate sensitive models to accurately predict enteric CH4 emissions in
finishing steers.