Nonlinear mixed models to study metabolizable energy utilization in broiler breeder hens

Area of Science:

• Animal Science
• Nutritional Physiology
• Mathematical Modeling

Background:

• Linear models of energy partitioning in hens have limitations in accurately predicting metabolizable energy intake (MEI).
• Existing models may not fully account for individual variations in energy expenditure, leading to estimation biases.
• Broiler breeder hen nutrition requires precise energy partitioning models for optimal feed allocation and performance.

Purpose of the Study:

• To develop and compare mathematical models for energy partitioning in broiler breeder hens.
• To overcome the limitations of linear models by introducing nonlinear approaches.
• To improve the accuracy of predicting metabolizable energy intake (MEI) based on body weight, growth, and egg mass.

Main Methods:

• Compared 1 linear and 2 nonlinear mixed-effects models using empirical data from 288 broiler breeder hens (20-60 wk).
• Models analyzed MEI as a function of body weight (BW), average daily gain (ADG), egg mass (EM), and temperature.
• Model fit evaluated using Bayesian information criterion; bias analyzed via linear regressions of observed vs. expected values.

Main Results:

• The linear model showed the poorest fit (R(2) = 0.64) with significant slope bias.
• The first nonlinear model (R(2) = 0.71) indicated MEI partitioned to ADG and EM were exponential factors.
• The second nonlinear model (R(2) = 0.75) provided the best fit, showing BW-dependent changes in energy requirements for ADG and EM.

Conclusions:

• Nonlinear mixed models significantly improve the accuracy of energy partitioning analysis in hens compared to linear models.
• These advanced models reduce estimation bias by incorporating individual variation in maintenance energy expenditure.
• The developed nonlinear models offer valuable tools for predicting MEI, evaluating feed efficiency, and assessing diet impacts on energy requirements.