Effect of feed restriction on dynamics of feed intake, milk yield, and energy balance in dairy cows.

Due to climate change, forage shortages may become more frequent, potentially leading to underfeeding in dairy cows. Limited knowledge exists on how milk yield and energy balance evolve during and, in particular, after the underfeeding phase. This study aimed to assess the impact of 2 feed restriction durations (4 and 6 weeks) on the changes in feed intake, milk yield, and energy balance in dairy cows. A total of 42 cows were first fed ad libitum for 3 weeks and then divided into 3 groups of 14 cows. Two groups were underfed by 5 kg of DM per day for 4 or 6 weeks (R4 and R6). After the restriction phase, cows were again fed ad libitum for 5 weeks. The control group was fed ad libitum throughout the trial. During the restriction phase, ECM yield dropped by 4.7 and 5.0 kg in the R4 and R6 groups, respectively, compared with the control group. During the refeeding phase, previously restricted cows exhibited higher DMI than the control group (+1.1 kg DM for R4; +0.80 kg DM for R6), while ECM yield remained similar across all groups. The energy balance during refeeding phase was higher in the restricted cows compared with the control group, suggesting a greater rate of body reserve reconstitution. The increase in feed intake during refeeding appeared primarily driven by the need to restore body reserves. Cows experiencing a more negative cumulative energy balance during restriction exhibited a greater proportional increase in intake during refeeding. Extending the duration of restriction from 4 to 6 weeks did not exacerbate milk yield losses nor enhance compensatory responses during refeeding, largely because cows progressively adapted their energy partitioning. During restriction, cows with initially high energy partitioning toward milk decreased the proportion of net energy secreted in milk relative to dietary energy supply, whereas those with lower partitioning increased it (e.g., from 0.60 to 0.66). During refeeding, this proportion continued to decrease in high-partitioning cows (e.g., from 0.80 to 0.71) but remained stable in low-partitioning cows (e.g., from 0.60 to 0.58), highlighting greater yield stability in these low-partitioning cows. Overall, the findings of present study provide useful perspectives for improving feeding system models by incorporating dynamic components, and highlight the need to consider different levels of restriction intensity in future studies.