Development of non-linear models for predicting enteric methane production

Abstract

The present study aimed to evaluate non-linear models (power and exponential) in predicting enteric methane (CH₄) production from a data-set of respiration chamber studies (207 treatment means). The developed power model was CH₄ (L/d) = 51.5(±4.5)×dry matter intake (DMI, kg/d)^{0.792(±0.034)} (adj. root mean square error, RMSE = 25.5 L/d), whereas the developed exponential model (Mitscherlich) was CH₄ (L/d) = 976(±95.3)×[(1 – e ^{(−0.0407(±0.00510)×DMI (kg/d))}], (adj. RMSE = 25.0 L/d). Adjusting the exponents for dietary concentration of ether extract, proportion of non-fibre carbohydrates of total carbohydrates and organic matter digestibility improved the models, and the effects of all these factors were significant (at least P<0.05) with both models. It is concluded that non-linear models can provide a better applicability over a wider range of intake, and are biologically more reliable for predicting CH₄ production as compared to linear models.

Keywords:

• Diet composition
• methane production
• non-linear modelling