A comparison of the Schumacher with other functions for describing growth in pigs

This study was undertaken to introduce the Schumacher equation and compare it with established functions for describing growth in pigs. The relationship between live weight (LW) and cumulative feed intake was also investigated. A database was constructed from three independent trials in which LW, age and intake were measured from birth to 937 days. Three growth functions were used for analysis of growth versus age: Gompertz, Schumacher and Weibull; and the Richards and monomolecular equations were used for analysis of cumulative feed intake versus LW. The growth functions have different points of inflexion. Liveweight at inflexion is W-f/e for the Gompertz, where W-f is the final weight, W-f/e(2) for the Schumacher, and W-f - (W-f - W-0) exp[-(n - 1)/n] for the Weibull, where W-0 is initial body weight and n is a shape parameter. Meta-analysis of the data using mixed model and nonlinear regression procedures was conducted to identify the most appropriate growth function. Functions were compared using F-tests or Bayesian Information Criteria, which give a value based on best fit and number of parameters in the model. The three equations were fitted to five individual pig growth profiles and to the composite data. Although the Weibull had a lower residual mean square, it did not significantly improve the fit compared to the simpler models and appears to be over-parameterised. The results suggest that model selection should be based on the type and amount of data available for analysis. Residuals plot showed that Schumacher and Weibull better predict the initial growth phase, however, all models showed largest magnitude of residuals towards the end of the growth profile. The monomolecular equation was most appropriate for describing LW against cumulative feed intake and may be used to formulate diets based on the efficiency of conversion of feed to LW at various stages of the animal's life span. (C) 2007 Elsevier B.V. All rights reserved.