Using milk spectral data for large-scale phenotypes linked to mitigation and efficiency | Utilisation de l'information spectrale pour du phénotypage à large échelle lié à la mitigation et à l'efficience

Inglés. peer reviewed

Inglés. Even if producing milk efficiently has always been a major concern for producers, the direct environmental impact of their cows is becoming a novel one. Traits linked to this issue were identified as methane emission (CH4), dry matter intake (DMI) and feed efficiency (FE); however they are available on a small scale. Researches showed that CH4 could be predicted from milk mid-infrared (MIR) spectra, allowing large-scale recording at low cost. The main objective of this study was to show, using a modelling approach, that DMI and FE could be derived from milk MIR spectra. For that, knowledge of body weight (BW) is required; however it was unknown in this study. Derived procedure was based on milk yield and composition, MIR CH4, and modelled standard animal requirements, allowing the prediction of expected BW. An external validation was conducted based on 91 actual records. 95% confidence limit for the difference ranged between -0.66 and 18.84 kg for BW, from -0.02 to 0.26 kg/day for DMI, and from -0.02 to 0.002 kg of fat corrected milk/kg DM for FE. Root mean square errors were 39.66 kg, 0.56 kg/d, and 0.03 kg/DM for the 3 studied traits. P-value for the t-test was not significant for BW and DMI. This suggests the possibility to obtain expected BW and therefore DMI from MIR spectra. Single trait animal test-day models used 1,291,850 records to assess the variability of studied traits. Significant variations were observed for the lactation stage, parity, genetics, and age. These findings were in agreement with the literature except for early lactation. This suggests in conclusion that the MIR information gave similar results for DMI and CH4 for the major part of lactation. The use of this novel method to predict expected BW offers new possibilities interesting for the development of genomic and genetic tools.