Genetic change of feed intake curves in growing pigs using non‐linear two‐stage genetic analysis and linear random regression models

Abstract

SummaryLongitudinal data of daily feed intake from 5245 young boars were analysed in this study in order to improve feed efficiency by changing the feed intake curve. Individual logistic and linear‐segmented functions were fitted, based on daily data of feed intake per boar, recorded during a test period of 9 weeks. Heterogeneous variance was considered by means of the power‐of‐the‐mean variance model. The mean estimated day on test corresponding to maximum increment in feed intake varied from 71 to 81 days using logistic function, and the mean estimated day on test at which the feed intake plateau was reached varied from 45 to 48 days on test, using a linear‐segmented function. High differences in the individual variation pattern of feed intake were obtained. A mixed model was fitted to individual estimates of function parameters. The uncertainty in individual estimation was taken into account by incorporating the estimated asymptotic covariance matrices. Estimates of heritability for parameters characterizing feed intake at the beginning of the test period ranged from 0.03 to 0.30. Selection based on parameters of a logistic function, using absolute residual estimation of variance heterogeneity, reduced feed intake at the end of the test period by 0.2 kg/day, while feed intake at the beginning of the test was slightly modified. With increasing individual information on daily feed intake, the two‐stage method was recommended. If individual observations on daily feed intake are sparse, random regression models were preferred in order to change the feed intake curve for breeding purposes.