Modeling turkey growth with the relative growth rate

Area of Science:

• Animal Science
• Quantitative Biology
• Biostatistics

Background:

• Traditional sigmoidal models are widely used for biological growth analysis.
• Assessing the fit of these models to specific animal populations is crucial for accurate growth prediction.
• Turkey growth exhibits complex patterns that may not be captured by standard sigmoidal functions.

Purpose of the Study:

• To evaluate the fit of six common sigmoidal growth curves and two novel two-phase relative growth rate models.
• To identify limitations of existing models in describing turkey body weight data.
• To develop and validate improved growth models based on relative growth rate characteristics.

Main Methods:

• Comparison of root mean square error (RMSE) between sigmoidal models and a local regression smoother.
• Nonparametric estimation of relative growth rate (RGR) from experimental turkey body weight data.
• Derivation and application of two-phase segmented models for RGR analysis.

Main Results:

• All tested sigmoidal growth curves (logistic, Gompertz, von Bertalanffy, Richards, Weibull, Morgan-Mercer-Flodin) showed a significant lack of fit.
• Nonparametric RGR estimates revealed features not accommodated by sigmoidal models.
• Two derived two-phase models, particularly a linear-linear model with smooth transition, provided acceptable fits to the turkey growth data.
• Differences in model parameters (beta1, beta2, kappa) were observed between sexes and genetic lines (fast-growing vs. control).

Conclusions:

• Standard sigmoidal growth curves are inadequate for modeling turkey body weight.
• A segmented two-phase model effectively describes the relative growth rate dynamics in turkeys.
• The new models offer a more accurate representation of growth, with parameters reflecting differences between sexes and genetic lines.