Prediction of cow diet composition from bulk milk spectra by Medium Infrared Reflectance Spectroscopy (MIRS)

The aim of this work was to develop equations using Medium Infrared Reflectance Spectroscopy (MIRS) spectra of bulk milk from commercial farms to predict dairy cow diet composition. A total of 429 bulk cow milk samples were collected from 176 commercial farms located in Northwest Italy from 2011 to 2017, and analysed by MIRS. Data on the production conditions were recorded on-farm at each sampling by compiling a detailed questionnaire. Calibrations for the predictions of cow diet composition were calculated with modified partial least square regression. Dairy cow feeding varied to a great extent. The total forages calculated including the 60% of DM of maize silage and the total concentrates including the 40% of DM of maize silage were successfully predicted by MIRS (R2CV > 0.80). The predictions for total concentrates (without part of DM of maize silage), the total forages (including the 100% of DM of maize silage), and the fresh herbage proposition in cow diet were found to be promising (R2CV > 0.75). These results can have a great interest in the authentication milk, especially for those dairy products having feeding restriction in their specification protocols.