Estimation of Parameters of Linear and Nonlinear Growth Curve Models at Early Growth Stage in California Turkeys

Abstract

This study was designed to investigate growth characteristics of totally 194 California turkeys (98 male, 96 female) using their weight-age data at early growth stage (during first 10 weeks). Linear, Logistic with three parameters and Richards' growth models were applied to estimate growth parameters. Results of analyses showed that both Logistic and Richard's growth models fitted weight and age data very well with R-2 = 99.3% and 99.4% for males and R-2 = 99.2% and 99.3% for females. Mature weights of males and females estimated using Logistic growth (2,228 +/- 55.3 g and 2,011 +/- 54.4 g) model and Richard's growth model (2,195 +/- 62.5 g and 1,973 +/- 59.9 g) was similar. In the case of the estimate on the change of body weights of male and female turkeys by equations W-t = -196.22 + 275.98t and W-t = -169.50 + 246.28t using the linear model, the k2 values of the estimates were 93.8% and 94.0%. R2 values in the linear model were smaller than those in the Logistic and Richards's models. When comparison is done based on the other criteria (Jp, AIC, BIC, DW, and asymptotic correlation), Logistic and Richard's growth models can be assumed to be more appropriate to estimate growth characteristics of young turkeys. On the other hand, among these growth curve models the logistic model can give the most reliable estimates of growth parameters of turkeys at early stage of life.