The non-linear model y = ae(−hW−c/W) was fitted to weekly calculated daily gain (DG), daily food intake (FI) and food efficiency (FE) of 653 barrows and gilts fed ad libitum or restrictedly during a growing period from 27 to 108 kg live weight. Where y was DG, FI or FE, W was live weight and a, b and c were parameters. The model fitted well to the expected course of the traits, with an accuracy similar to that of quadratic polynomials. Parameters for one trait could simply be derived from the parameters for the other two traits. For each trait, four basic patterns were distinguished, depending on the signs of b and c. Curves with a maximum (b > 0 and c > 0) occurred most frequently. In cases of curves with a maximum, the model could be reparameterized to a model with parameters having a simple biological meaning.
Coefficients of determination in barrows averaged 0·29 for DG, 0·88 for FI and 0·45 for FE, whereas these values were somewhat lower in gilts. With ad libitum feeding, a DG curve with a maximum was fitted in proportionately 0·83 of the barrows and 0·61 of the gilts. The maximum DG was on average at live weights of 64 kg for barrows and 77 kg for gilts. A maximum in the FI curve was predicted in proportionately 0·60 of the barrows and 0·39 of the gilts. Curves for DG and FI in gilts were less curvilinear than in barrows. FE curves, with ad libitum feeding, had a maximum in proportionately 0·59 of the barrows and 0·52 of the gilts. This predicted maximum FE was, on average, before the start of the growing period. Gilts had a higher FE than barrows from 35 kg body weight onwards, and the difference increased with increasing live weight. Differences in FE between ad libitum and restricted feeding were small, with a tendency for animals fed at a restricted feeding level to be more efficient at the end of the growing period.
Average FE curves and individual FI or DG curves were used for indirect prediction of individual DG or FI curves, respectively. The correlation between directly and indirectly predicted values of DG and FI at live weights of 30, 65 or 100 kg was about 0·7 in ad libitum fed barrows and gilts, and greater than 0·8 in pigs fed at a restricted level. This indicates that the model is suitable to predict and control the course of individual daily gain by influencing the course of food intake.