Comparison of Some Nonlinear Functions for Describing Broiler Growth Curves of Cobb500 Strain

This study was conducted to compare some nonlinear functions to describe the broiler growth curve of the Cobb500 strain. A flock of fifty one-day-old chicks were randomly selected from a henhouse of 2500 chicks. Our goal was to establish a growth curve using weighting data using mathematical solutions of time-dependent differential functions. In total, six equations were subjected to a statistical calibration by a sequential quadratic programming under the non-linear regression procedure of the SPSS program. The results showed that the heterogeneity rate between individuals of the same batch increases with the age of the chicks, from more than 10% an early age to less than 30% at the slaughter age. The goodness of fit for six dynamic models showed that the number of iterations required increases with the number of parameters of the model. However, the three parameter models were the best model for describing growth curve (the greatest efficiencies and the lowest error components). The asymptomatic values ​​(3500g to 7500g) and their estimation errors (2% to 12%) are relatively acceptable for the three-parameter models compared to those of four parameters (more than 8000g and up to 100% error). Finally, the comparison between actual and predicted values by models shows that the Gompertz model was the most suitable till up to the four weeks of age. After 1 month of age, the Gompertz has a lower precision and the logistics, Von Bertalonffy and WLS models accurately described the growth curve.