Sorting by particle size in dairy cows fed different silages and concentrates in total mixed rations

Feed sorting should be minimised to assure that the nutrient supply predicted match what the animal actually ingests. The aim of this study was to quantify to what extent cows sort feed by particle size when different silages and concentrates are fed. Twenty-four lactating Danish Holstein dairy cows were used in a cross-over design with two periods of 21 d each. Each cow was fed two out of four diets. The four diets were two by two factorially arranged with two forages, either grass-clover or maize silage (G or M; 55.0 and 54.2% of diet DM, both chopped at 11 mm, respectively), and two concentrates, either rolled barley or rolled dried beet pulp pellets (B or D; 21.5% of diet DM). The diets were mixed as total mixed rations (TMR), and had an average DM content over the trial of 50.6% for GB, 51.0% for GD, 45.4% for MB and 45.6% for MD. In each period a sample from each TMR and the matching residue (aiming for 10% DM feed residue) from each cow after 24 h from a single feeding were run in a Penn State Particle Separator. The proportion of fractions (>19, >8, >1.18, and <1.18 mm) was calculated based on fresh weight. The proportions recovered for the TMR were for each sieve with a generalized linear mixed model assuming gamma distributed residuals, including the individual and all possible interactions as fixed effect of silage (G vs M), concentrate (B vs D), and random effect of testing day (1 to 4). The model means for the proportion by particle size from largest to smallest were 35.3, 20.2, 36.2, 8.26% for GB, 36.8, 26.7, 30.0, 6.44% for GD, 0.90, 44.2, 45.3, 9.66% for MB, and 0.91, 51.7, 37.9, 9.51% for MD. There was a significant effect of silage for particle >19 mm, effect of both silage and concentrate for particles >8 and >1.18 mm, and a significant interaction between silage and concentrate for particles <1.18 mm. The differences between the TMR and residue proportions for each sieve were analysed in a model similar as the previous, with addition of the random effect of the cow, and assuming normally distributed residuals. The fixed coefficients from the model were tested against zero. No coefficient was found different from zero, and therefore there is no indication of cows sorting by particle size.