Optimization of a model to estimate residual feed consumption in the laying hen

Abstract

Multiple regression models for metabolizable energy consumption of White Leghorn laying hens were investigated. The animals were housed individually in cages during a 44-week laying period and fed either one of two diets of different energy density. Every 4 weeks, individual records on feed consumption, body weight, egg mass, egg number and number of abnormal eggs were available. Also age at first egg was recorded. Residual analysis of regression of metabolizable energy consumption ( MECD) on metabolic body weight ( MBW 3/4), egg mass ( EMD) and body weight gain ( BWG) showed significant effects of time and diet. Tests for these effects on the regression parameter estimates showed significant differences over time. In the regression model several alternatives for the energy demanding processes of maintenance, egg production and weight gain were compared. Additions of age at first egg and two-way interactions were also studied. Comparisons and additions were evaluated on the basis of F-test probabilities for improvement of goodness of fit. Only addition of an intercept (or non-linear regression with MBW α ) and using egg mass adjusted for abnormal eggs ( EMDc) instead of EMD gave significant improvement of fit. The important role for this optimization of the subdivision of the laying period in short time intervals is discussed. Finally, the trait residual feed consumption is defined as the regression residual from the optimum model of MECD on MBW 3/4, EMDc and BWG (including an intercept) within short time periods and within diet, and will be used for further study on the possibilities for selection.