Avaliação de equações para estimar o consumo de vacas leiteiras

Abstract

The present study aimed to perform a comparison of the predictive power of different equations to estimate intake by dairy cows. The work was conducted through collecting information in literature about body weight, milk yield, days in milk, dry matter intake and fat content of milk of dairy cows raised in Brazil. All this information, with the exception of dry matter intake, were used to predict the dry matter intake with equations of the model: National Research Council (NRC) 2001, Cornell Net Protein and Carbohydrate System (CNCPS), Agricultural and Food Research Council (AFRC) 1993 and De Freitas et al., (2006). Subsequently, the predictive power of the mathematical equations was compared through the graphical comparison of the standard as Draper and Smith (1966) and Montgomery (2005) and by means of Akaike's criterion (Akaike, 1974) waste. Models 2, 3 and 4 failed to represent the behavior of the observed data, as presented values of Δr > 10. Model NRC, 2001 (Model 1) was the best choice for presenting the ERr = 1, all others showed ERr much greater than 20 and therefore must be considered worse choices. Model 3 tended to underestimate the predicted values with 76% points above the line of invalidity. Model 1 is the one that best predicts the dry matter intake of dairy cows under the conditions studied