Defining growth curves with nonlinear models in seven sheep breeds in Mexico

Characterizing growth in livestock is important when making management, marketing and genetic improvement decisions. Nonlinear models were tested to identify those with the best fit for growth curves in seven sheep breeds [Blackbelly (n= 19,084); Pelibuey (n= 39,025); Dorper (n= 35,814); Katahdin (n= 74,154); Suffolk (n= 10,267); Hampshire (n= 7,561); and Rambouillet (n= 7,384)]. Using breed registry databases, live weight was assessed from birth to 230 d of age. The SAS program was applied to test six nonlinear models: Brody, Verhulst, von Bertalanffy, Gompertz, Mitscherlich and logistic. The criteria for selecting the best-fit model were the average prediction error; the prediction error variance; the Durbin-Watson statistic; the coefficient of determination; the root-mean-square error; and the Akaike and Bayesian information criteria. For the Hampshire, Pelibuey and Suffolk breeds the best-fit model was the von Bertalanffy, with a sigmoid curve and an inflection point age between 40 and 57 d. For the Katahdin, Blackbelly, Dorper and Rambouillet breeds the best-fit models were the Brody and Mitscherlich models, with a continuous growth curve, no inflection point and constant growth rate. Marked differences were observed in adult Rev Mex Cienc Pecu 2019;10(3):664-675 665 weight between breeds, with average values (kg) of 44.6 for Blackbelly, 49.2 for Rambouillet, 52.9 for Pelibuey, 55.6 for Hampshire, 60.2 for Katahdin, 64.7 for Suffolk and 65.2 for Dorper; values tended to be highest in the Brody and Mitscherlich models, and lowest in the logistic and Verhulst models.