Evaluation of nonlinear models in fitting growth curve for one of the buffalo herds in Khuzestan province

The growth is defined as the change of live weight per unit of time and can be studied by monitoring its changes at different ages, using mathematical describing growth models. So, in this research to study the growth curve, weight records of 37 male and 42 female buffaloes were collected from birth to five years in one herd of Khuzestan province (totally 2694 weight records) were used. Surveyed functions were included of five Gompertz, Logistic, Brody, VanBertalanfy and Richards growth models. Regression models were perfumed, using NLIN procedure of SAS 9.1 software. To determine the best model the criteria including of adjusted coefficient of determination, Mean Squared Error, Absolute Mean Residual Deviation and second order bias correction Akaike Information Criterion were used. The results showed that the VanBertalanfy, with high adjusted determination coefficient and the lowest error criteria, was more suitable than other models. Models fitted for both male and female buffaloes had the same results and Vanbrtalanfy, Richards, Gompertz, Brody and logistics were ranked from 1 to 5 for both groups, respectively. The maturity weight and the maturation rate based on appropriate model were 631 and 857 kg, and 0.052 and 0.050 kg per month for female and male buffaloes, respectively. Correlation between the maturity weight and the maturation rate was also obtained -0.92 and -0.91 for male and female buffalo. The age of female and male buffalo in inflection point were estimated 11.62 and 12.42 months, respectively. The proposed model can be used in prediction of the growth.