Magnitude of change in prepartum feed intake: Estimations using multiple classes of predictors and associations with transition metabolism, health, and milk production

The objectives of this study were to identify factors associated with the relative change in prepartum DMI (RCDMI) of 273 cows fed individually, evaluate the performance of linear models to estimate RCDMI using different classes of predictors, and characterize the implications of RCDMI to transition metabolism, health, and subsequent milk production. Two periods of interest were established. Period 1 comprised d −21 to −12 relative to calving, when DMI was stable. Period 2 comprised d −4 to −1, when average DMI was distinctly declined. The RCDMI from period 1 to 2 was calculated as a percentage value, which ranged from −75% to 15% and averaged −18.1% (± 15.0%). Season, parity, body fatness, body weight, milk production in the previous lactation and at dry-off, and length of dry period were associated with RCDMI and explained 11% of the variation in all cows, and 19% when only parous cows were considered. Performance of linear models to predict RCDMI was improved when data on rumination and physical activities and target blood metabolites were added. The adjusted R² increased to values between 0.45 and 0.55, and selected models performed consistently in cross-validation analyses. To evaluate the implications of RCDMI, cows were ranked within parity according to RCDMI and classified into terciles as large decline (LDec), moderate decline (MDec), or small decline (SDec). By design, DMI did not differ between tercile groups in period 1 (13.3 ± 0.2 kg/d), but differed substantially in period 2 (LDec = 8.8; MDec = 11.2; SDec = 12.7 ± 0.2 kg/d), creating important differences in RCDMI among groups (LDec = −33.8; MDec = −16.2; SDec = −3.4% ± 0.8%). At enrollment, cows in the LDec and MDec groups were heavier (LDec = 788; MDec = 775; SDec = 750 ± 7 kg), and the proportion of cows with BCS >3.5 was higher in LDec (LDec = 63%; MDec = 47%; SDec = 38%). An interaction between group and time was observed for postpartum DMI, which started lower in LDec than in SDec cows, equaled by the end of transition, and inverted at wk 13 and 14 after calving. Yields of ECM were greater in LDec than in SDec cows, and both groups did not differ from MDec (LDec = 41.0; MDec = 40.3; SDec = 39.0 ± 0.5 kg/d). The LDec cows had decreased energy balance and greater concentrations of nonesterified fatty acids, β-hydroxybutyrate, and aspartate aminotransferase in serum, and greater glutathione peroxidase activity in plasma than SDec cows. Larger declines in prepartum DMI were also associated with increased risk for postpartum disease, although the associations were only weak to moderate. In conclusion, a large decline in prepartum DMI of cows fed ad libitum was associated with important adjustments in the energy metabolism and antioxidants activities, and greater milk production in the subsequent lactation. These findings indicate that feed intake decline close to parturition is likely a normal response to physiological adaptations at the onset of lactation when cows are fed ad libitum.