Modelling growth curves of Nigerian indigenous normal feather chicken using bayesian nonlinear model

Abstract

This study was conducted to predict the growth curve parameters using Bayesian Gompertz and logistic models and also to compare the two growth function in describing the body weight changes across age in Nigerian indigenous normal feather chicken. Each chick was wing-tagged at day old and body weights were recorded at the end of each two-week period up to 16 weeks of age. A non-informatics prior was used for the Bayesian inference. The two models were fitted to the weight-age data from day old to sixteen weeks of age for each individual chicken by non-linear Bayesian regression using Winbugs. Models were compared using the mean value, MCE, convergence and the estimate weight when the parameters are fitted to each model. Differences were observed in the growth parameters of chickens. Gompertz having a mean value of 1.70kg and rate of growth (k) of 0.03 for the Asymptotic value (A) while Logistic A mean value (1.88) and k (1.62) which indicate that Logistic has higher growth rate and also attain a mature weight than the Gompertz. MCE for Gompertz ranges from 0.001 to 0.56 and in Logistic from 0.05 to 0.39. Based on all the criteria used for comparing these models, it can be concluded that the mean value for Gompertz model is closer the observed value (1.60kg) than the logistic, also Gompertz model have a lower value of MCE which makes the model more reliable. The Credible interval for Gompertz is more close to the mean value than Logistic function. Gompertz gave the best fit for the age-body relationship for the Nigerian indigenous normal feather chicken, although Logistic function is equally good in predicting the growth curves of the chickens.

Keywords: Normal feather, growth curves, modelling, Nigerian indigenous chicken.