Prediction of bulk milk fatty acid composition based on farming practices collected through on-farm surveys

International audience; The aim of this study was to predict the fatty acid (FA) composition of bulk milk using data describing farming practices collected via on-farm surveys. The FA composition of 1,248 bulk cow milk samples and the related farming practices were collected from 20 experiments led in 10 different European countries at 44 degrees N to 60 degrees N latitude and sea level to 2,000 m altitude. Farming practice-based FA predictions [coefficient of determination (R-2) >0.50] were good for C16:0, C17:0, saturated FA, polyunsaturated FA, and odd-chain FA, and very good (R-2 >= 0.60) for trans-11 C18:1, trans-10 + trans-11 C18:1, cis-9,trans-11 conjugated linoleic acid, total trans FA, C18:3n-3, n-6:n-3 ratio, and branched-chain FA. Fatty acids were predicted by cow diet composition and by the altitude at which milk was produced, whereas animal-related factors (i.e., lactation stage, breed, milk yield, and proportion of primiparous cows in the herd) were not significant in any of the models. Proportion of fresh herbage in the cow diet was the main predictor, with the highest effect in almost all FA models. However, models built solely on conserved forage-derived samples gave good predictions for odd-chain FA, branched-chain FA, trans-10 C18:1 and C18:3n-3 (R-2 >= 0.46, 0.54, 0.52, and 0.70, respectively). These prediction models could offer farmers a valuable tool to help improve the nutritional quality of the milk they produce.