Production and nutrient use efficiency of lactating dairy cows fed concentrate feeds via an automatic feeding system

The objective was to investigate the effects of separate feeding of a partial mixed ration (PMR) and concentrate premixes (CMix) delivered at fixed or variable amounts using an automatic feeding system (AFS) on lactational performance, enteric CH₄ emissions, ruminal fermentation variables, digestibility of nutrients, and N excretion in lactating dairy cows. Forty-eight primi- and multiparous Holstein cows averaging 91 ± 26 DIM were enrolled in a 17-wk randomized complete block design experiment, with a 2-wk covariate, 5-wk adaptation, and 10-wk data and sample collection periods. Cows were blocked based on parity, DIM, and ECM. Cows within blocks were randomly assigned to 1 of 3 treatments: (1) CON = diet fed ad libitum as a TMR; (2) SF_F = diet fed as separate feeds (SF) with ad libitum PMR and fixed amounts of CMix via an AFS; and (3) SF_V = SF with variable amounts of CMix calculated by an algorithm derived based on balances of NEL and MP of individual cows and fed via an AFS. Partial mixed ration intake was increased, and milk yield was decreased by SF_V compared with CON but not with SF_F. Feed efficiency was decreased by both SF treatments compared with CON. Treatments did not affect ECM yield in the current study. Enteric gas emissions were not affected by SF_F, but CH₄ intensity tended to be increased by 10% in SF_V compared with CON. The concentration of ruminal VFA and VFA profile were not affected, except for an increase in butyrate concentration by both SF treatments and a tendency for increased ammonia concentration in SF_V compared with CON. Apparent DM and OM digestibility was increased by SF_V, whereas amylase-treated NDF and ADF digestibility was increased by both SF_F and SF_V. Treatments did not affect N excretion or secretion variables, except that urine N excretion (as % of N intake) was decreased by SF_F compared with CON. Total purine derivative excretion tended to be decreased by SF_V compared with CON but not by SF_F. Overall, SF_F did not affect lactational performance but decreased urinary N excretion compared with feeding a TMR. Feeding variable amounts of CMix calculated using an algorithm based on individual NEL and MP balances (i.e., SF_V) had a negative effect on lactational performance and CH₄ intensity, likely because of algorithm overfitting and lack of generalization.