Interest, limitations and perspectives of using milk mid-infrared spectra for management and breeding of dairy cattle

The increasing consumer’s concern about the relationship between food and human health as well as the environmental impact of food production combined with the increase of production cost and the variability of milk price must be take into account by dairy herd improvement organisations to develop future management and breeding tools for dairy farmers. In this context, the use of Fourier transform mid-infrared (FTMIR) spectrometry is relevant to record phenotypes related to the nutritional quality and process ability of milk, the cow health and well-being and the environment. Those new phenotypes obtained from different equations have not the same accuracy and reliability due to the use of different calibration sets and/or a partial uncovering of the spectral variability. So, even if the computational facility to implement midinfrared (MIR) prediction equations is a reality, some points of attention must be considered to ensure high quality MIR predictions and spectra. This is related to a well-designed calibration set often reached through international collaborations, an appropriate standardization of the spectral data used to perform the prediction and an open communication about the equation accuracy and the representation of the spectral variability of a milk sample to be analysed compared to the used calibration set. More and more investigations are conducted currently to use directly the spectral data to limit the prediction bias. However, although a confirmation of the genetic background related to those particular spectral phenotypes, additional researches must be conducted to define the best methodology to include those records in breeding indices.