Accuracy and precision in DM intake prediction models for lactating dairy cows

Area of Science:

• Animal Science
• Dairy Nutrition
• Nutrient Management

Background:

• Accurate prediction of daily dry matter intake (DMI) in lactating dairy cows is vital for optimizing diets, minimizing nutrient waste, and improving environmental sustainability.
• Existing DMI prediction models from various international research bodies (NRC, NASEM, INRA, GfE, CSIRO) require continuous evaluation against contemporary dairy herd data.
• DMI prediction accuracy directly impacts the precise formulation of energy, protein, and phosphorus, and influences estimations of nutrient excretion for crop fertilization.

Purpose of the Study:

• To assess and compare the accuracy and precision of six different DMI prediction models for lactating dairy cows.
• To evaluate the performance of the Swiss Agroscope (1994), North American NRC (2001) and NASEM (2021), French INRA (2018), German GfE (2023), and Australian CSIRO (2007) models.
• To determine the most reliable model for predicting DMI in Holstein cows under typical feeding conditions.

Main Methods:

• Utilized routine dairy herd data from Agroscope, collected between November 2015 and March 2021, encompassing 138 primiparous and 135 multiparous lactating Holstein cows (413 partial lactations).
• Collected daily data on DMI, milk yield (MY), and body weight (BW), alongside monthly milk and diet composition.
• Assessed model performance using Root Mean Square Error of Prediction (RMSEP) and Concordance Correlation Coefficient (CCC) analysis, including RMSEP decomposition.

Main Results:

• The NRC and NASEM models generally overestimated DMI, while the INRA and GfE models underestimated it.
• RMSEP for DMI prediction ranged from 2.50 to 4.37 kg/d in primiparous and 3.02 to 4.98 kg/d in multiparous cows across models.
• The Agroscope model exhibited the highest precision (Error of Central Tendency close to zero) and the highest CCC in primiparous cows (0.53), while INRA showed the highest CCC in multiparous cows (0.70).

Conclusions:

• The 30-year-old Agroscope model proved to be the most accurate and precise for predicting DMI in lactating dairy cows fed a diet based on mixed forage and concentrates.
• Model performance varied significantly, highlighting the need for ongoing validation with regional data.
• Accurate DMI prediction remains critical for efficient dairy cow nutrition, economic viability, and environmental stewardship.