Mathematical characterization of broiler carcass yield dynamics

Area of Science:

• Agricultural Science
• Animal Science
• Quantitative Biology

Background:

• Broiler chicken supply chain economics relies on precise biological growth and yield models.
• Accurate prediction of carcass and carcass part weights is crucial for economic modeling.

Purpose of the Study:

• To statistically evaluate eight dynamic nonlinear models for predicting broiler carcass and carcass part weights.
• To identify the most suitable model for estimating broiler carcass composition and yield.

Main Methods:

• Evaluated four sigmoidal (S) models (Gompertz, modified Gompertz, Richards, Lopez) based on age.
• Assessed three diminishing returns (DR) models (Lopez, Mitscherlich, log linear) and a log-linear proportional yield (PY) model based on feather- and fat-free empty body mass (FFEBM).
• Statistical analysis focused on prediction accuracy, error, and bias.

Main Results:

• Sigmoidal models with flexible inflection points outperformed the fixed inflection Gompertz model and DR models.
• Log-linear models demonstrated 100% convergence in evaluations.
• The allometric proportional yield (PY) model provided the most accurate predictions for most carcass parts with the least bias compared to DR log-linear models.

Conclusions:

• Flexible sigmoidal and log-linear proportional yield models are superior for broiler carcass part weight prediction.
• The allometric proportional yield model is recommended for its accuracy and low bias in predicting broiler carcass weights.
• Coefficient estimates for the log-linear PY model and carcass chemical composition data are provided for 12 key carcass parts.