Energy Balance Prediction in Japanese Holstein Cows by Using Milk Fatty Acid Composition.

Abstract

Negative energy balance (NEB) in early lactation adversely affects health and fertility in dairy cows, yet direct measurement of energy balance (EB) requires intensive monitoring of individual feed intake. We developed a practical EB prediction equation by augmenting routine milk yield and milk composition data with milk fatty acid (FA) information-specifically de novo and preformed FA proportions-and evaluated its external validity. By using 14,993 records from 93 Holstein cows from the Hokkaido Research Organization (6-103 days in milk [DIM]), stepwise regression based on the Schwarz Bayesian Criterion selected six traits: preformed FA, milk fat percentage, standardized DIM class, de novo FA, milk yield, and milk protein percentage. The resulting model achieved an R² of 0.65. External validation by using nationwide data from 18 institutions yielded an R² of 0.32-higher than that of our previous equation (R² = 0.17). Preformed FA, reflecting body fat mobilization, contributed most strongly to EB prediction, whereas de novo FA added complementary physiological information. Compared with equations based on milk yield and milk composition alone, incorporating FTIR-predicted milk FA composition improves the accuracy and robustness of EB prediction across diverse management conditions, supporting its application in metabolic monitoring and genetic evaluation.