Evaluation of a Model (RUMINANT) for Prediction of DMI and CH(4) from Tropical Beef Cattle

SIMPLE SUMMARY: Methane (CH(4)) is a byproduct of the digestion of cattle; this gas has a greenhouse effect in the atmosphere, which contributes to global warming. As the direct measurement of methane demands time and resources, the objective of this work was to evaluate the ability of a mathematical model (RUMINANT) to predict methane emissions from livestock. With this objective, methane measurements were made in individual chambers, and the results were compared with methane emissions estimated by the RUMINANT model. The model showed a high capacity to predict dry matter intake. However, in the case of methane emissions, it did not. The model substantially underestimates methane emission in all diets (six) but one including Leucaena diversifolia. On diets without Leucaena, the precision of the model was adequate, but on diets with Leucaena, there was not a linear regression between the observed and simulated methane emission values. This may be an effect of the anti-methanogenic factors of Leucaena that are not accounted for by the RUMINANT model. This study contributes to improving national inventories of greenhouse gases from the livestock of tropical countries. ABSTRACT: Simulation models represent a low-cost approach to evaluating agricultural systems. In the current study, the precision and accuracy of the RUMINANT model to predict dry matter intake (DMI) and methane emissions from beef cattle fed tropical diets (characteristic of Colombia) was assessed. Feed intake (DMI) and methane emissions were measured in Brahman steers housed in polytunnels and fed six forage diets. In addition, DMI and methane emissions were predicted by the RUMINANT model. The model’s predictive capability was measured on the basis of precision: coefficients of variation (CV%) and determination (R(2), percentage of variance accounted for by the model), and model efficiency (ME) and accuracy: the simulated/observed ratio (S/O ratio) and slope and mean bias (MB%). In addition, combined measurements of accuracy and precision were carried out by means of mean square prediction error (MSPE) and correlation correspondence coefficient (CCC) and their components. The predictive capability of the RUMINANT model to simulate DMI resulted as valuable for mean S/O ratio (1.07), MB% (2.23%), CV% (17%), R(2) (0.886), ME (0.809), CCC (0.869). However, for methane emission simulations, the model substantially underestimated methane emissions (mean S/O ratio = 0.697, MB% = −30.5%), and ME and CCC were −0.431 and 0.485, respectively. In addition, a subset of data corresponding to diets with Leucaena was not observed to have a linear relationship between the observed and simulated values. It is suggested that this may be related to anti-methanogenic factors characteristic of Leucaena, which were not accounted for by the model. This study contributes to improving national inventories of greenhouse gases from the livestock of tropical countries.