A biometrical traits-to-feed model integrating fish nutrition and genetic improvement of feed utilisation traits: Predicting changes in optimal feed phosphorus levels for selectively bred rainbow trout

Feed whose composition matches nutritional requirements of farmed animals is fundamental for sustainable food production. Selective breeding of fish changes growth, body composition and feed utilisation of fish and hence also nutritional requirements. We derived equations that directly estimate the required changes in feed formulation from measured trait changes, bypassing the traditional estimation by experimental dose-response trials. We applied the model to determine changes in optimal feed phosphorus (P) levels in rainbow trout that has been improved across decades by a breeding programme. The equations show that the effect of changes in body composition (body P%, fillet%, viscera%) on digestible phosphorus inclusion level is modest and linear. The effects of FCR (feed conversion ratio) and phosphorus retention efficiency are strong and highly non-linear, implying that the available phosphorus levels for rainbow trout need to be increased due to their improved FCR, and less so due to their increased fillet% and reduced viscera%. Despite the complex non-linear relationships on the observed raw scale (i.e., additive effects), the multiplicative effects of the traits on a log-log scale are actually simple and intuitive. During the last decades, dietary phosphorus has been reduced to limit nutritional loading to environment. Forecasting showed that the required feed phosphorus levels for rainbow trout under selective breeding are expected to be further increased, and a care must be taken not to reduce the dietary levels too much. The equations integrate quantitative genetics and fish nutrition under one united predictive framework, and they are applicable to other fish species.