Impact of Weighing Data Structure on the Adjustment of the Local Kids’ Growth Curve Under Pastoral Conditions in Southern Tunisia

Six non-linear models; Brody, Logistic, Gompertz, Janoschek, Von Bertalanffy and Polynomial equations were tested to identify the suitable model to adjust the growth curve of local kid's under pastoral conditions in Southern Tunisian. Data base issued from 10 years periodical weighing was used to adjust growth model. All models’ parameters were estimated by an iterative non linear regression procedure. Various data subsets were used to test models convergence behavior and accuracy towards data structure variation. Models evaluation criteria were the iterations number (NI), the coefficient of determination (CD) and the residuals means squares (RMS). Gompertz model allowed the best kid's growth model fitting by its convergence rapidity and parameters estimation accuracy with all data subsets. Using all observed data, the convergence point (10⁻⁸) was reached after 9 iterations at a CD equal to 67%. So, arid environmental factors affect both quantitatively and qualitatively the kids’ growth curve assessment and shape through their impacts on the data file structure. The suitable model remains a crucial step to handle local goat characterization and genetic improvement under harsh conditions.