Comparative evaluation of equations predicting methane production of dairy cattle from feed characteristics

Techniques that allow direct measurements on animals to quantify methane (CH4) emissions are costly and difficult to transfer to herd level. Mathematical approaches have been developed to predict methane emissions of cattle based on diet and intake characteristics, which were calibrated against largely varying calorimetry data. In this study, nine CH4 prediction equations were applied to five typical Central European dairy cow diets in order to compare their applicability. The five diets differed in respect of forage proportion and type. In a first attempt, regression equations were selected containing easily accessible data such as dry matter intake (DMI, kg/d) forage proportion, as well as neutral and acid detergent fibre that can also be extracted from on-farm datasets. Smallest differences to mean values were observed with the application of equations using neutral detergent fibre, while standard deviations were highest. Therefore, the best capability to differentiate between diets was shown, when using equations that operated with forage proportion and DMI. Nevertheless, the role of CH4 prediction equations should not be overestimated. The differences in CH4 estimates show that frequently used equations are still inaccurate and may only serve as implications to locate trends. It should be taken into consideration to expand datasets, involving future CH4 measurements, on animal and herd level, feeding typical, up-to-date regional diets in order to get more precise equations, suitable for a greater range of estimations. To ease and simplify the future applications, the prediction equations could be classified into groups, clearly stating by which data they were derived, for example, regional origin and diet composition.