CH4 estimated from milk MIR spectra: model on data from 7 countries and 2 measurement techniques

Availability of a robust proxy to estimate individual daily methane (CH4) emissions from dairy cows would be valuable especially for large scale studies, for instance with genetic purpose. Milk mid infrared (MIR) spectroscopy presents potential to meet this aim as spectra can be obtained routinely at reasonable cost through milk recording process. Development of a prediction equation requires as much variability as possible in the reference data set to improve the accuracy and ensure the robustness of the model. So, two datasets including CH4 measurements and corresponding milk MIR spectra have been merged: the first contained 532 data from 156 cows of Ireland and Belgium with CH4 measurements obtained with SF6 tracer technique; the second reached 584 data from 147 cows of Switzerland, United Kingdom, France, Denmark and Germany with CH4 measurements obtained with respiration chambers. In addition to the Partial Least Squares (PLS) equation using the raw CH4 values, a second equation was performed with a reduction of 8% to CH4 values from chambers to evaluate the need to correct the potential method bias in accordance with literature. A 5-group cross-validation was performed to test the robustness of the models. R2 and the standard error of cross-validation were 0.63 and 62 g/day from raw data and 0.65 and 59 g/day when CH4 respiration chamber values were adjusted. This slight improvement due to the adjustment of chamber measurement does not permit to conclude that this correction is needed. The study of residuals showed a non-significant effect due to the CH4 measurement technique. In conclusion this new equation combining CH4 from 2 different techniques covered more variability (cows, diets and country specific information) and shows potential as a proxy especially for genetic evaluation.