Circannual pattern of milk crude protein and urea nitrogen and their relationship in Holstein cows in Swiss grassland-based dairy systems over a 30-year period.

The objectives of the study were to characterize the circannual variation in milk CP and MUN concentrations and to evaluate their relationship in Holstein cows managed under Swiss grassland-based dairy production systems. We hypothesized that milk CP and MUN exhibit distinct seasonal patterns, reflecting changes in forage quality and feeding practices, and that the strength of their association varies across seasons and production eras. A retrospective cohort study was conducted using historical milk recording data consisting of 2,313,348 cow-lactations from 897,854 unique Holstein cows across 22,830 Swiss dairy farms collected between 1994 and 2023. The dataset included primiparous (n = 817,638) and multiparous (n = 1,495,710) cow-lactations, with multiparous cows contributing a mean (±SD) lactation number of 3.04 ± 1.04. Monthly test-day records (January to December) within a 305-d lactation were used to assess temporal changes in milk CP (%) and MUN (mg/dL) and their relationship, resulting in approximately 18.3 million test-day observations. Mean (±SD) CP concentrations were 3.29% ±  0.34% for primiparous and 3.32% ± 0.36% for multiparous cows, and MUN concentrations were 10.8 ± 3.72 mg/dL and 10.5 ± 3.81 mg/dL, respectively. The study period was stratified into 2 eras: EraA (1994-2008) and EraB (2009-2023). Generalized linear mixed-effects models were fitted to assess temporal variation in milk CP and MUN and included fixed effects of test-day month (January to December), parity (primiparous vs. multiparous), era (EraA vs. EraB), their 2- and 3-way interactions, a natural cubic spline for DIM, and the interaction between parity and DIM, and random effects of farms and cow. To evaluate the relationship between milk CP and MUN, the model included fixed effects of linear and quadratic covariates of CP, parity, era, their 2- and 3-way interactions, a natural cubic spline for DIM and its interaction with parity, and random effects of farm nested within monthly test-day record and cow. A 3-way interaction between month, parity, and era was associated with milk CP and MUN concentrations, indicating that seasonal patterns differed between eras and that the magnitude and timing of these changes varied between primiparous and multiparous cows. Milk CP declined from winter to summer in both eras (EraA: primiparous 3.29-3.12, multiparous 3.34-3.18; EraB: primiparous 3.30-3.09, multiparous 3.36-3.12). In contrast, MUN concentrations increased from winter to late summer in both eras (EraA: primiparous 10.0-12.5, multiparous 9.82-12.2; EraB: primiparous 10.0-12.0, multiparous 9.71-11.8). A 3-way interaction between milk CP, parity, and era was observed for MUN, with MUN decreasing nonlinearly as CP increased from 2.40% to 4.20%, and steeper declines at lower CP. Across the milk CP range, multiparous cows generally exhibited slightly lower MUN than primiparous cows, with smaller parity differences in EraB. Overall, milk CP and MUN concentrations exhibited distinct circannual patterns across parity and production era, and MUN was associated with decline over time at similar CP levels, suggesting improved population-level N utilization, potentially associated with changes in nutrition, genetics, and forage management.