In aquaculture operations, feed may represent more than 50% of the total cost of operations. As the industry is expanding rapidly, the average prices for aquafeeds have increased by 20-90%, depending on region and target species. Accordingly, there is a steady demand to reduce feed production price by finding novel and alternative feed ingredients and formulations, while maintaining feed efficiency and growth performance at the same time.
Currently, it is common practice to test new ingredients and formulations in laboratory trials, mostly on juvenile fish. However, there is evidence, that results obtained in such trails cannot be transferred across different life stages, and therefore contain little information regarding rearing time and harvest size.
In the current study, we have successfully tested the possibility, to model the long-term effects of different feed formulations on cultured rainbow trout. This approach can, with little effort, be adopted to other species as well. The method also allows to predict and evaluate the goals of trait-specific breeding programs.
The experiment was performed at a commercial farm. Fish (n=900) were fed with three different extruded pellets of different experimental formulations in triplicate groups. Several models of different complexity were tested against the data. The most suitable model was evaluated by a variety of statistical tests combined in a multi-criteria analysis.
Despite the massive influence of the feed formulation on fish performance, the statistical analysis revealed the same mathematical relationships related to the growth process. This means, that different growth trajectories, triggered by different feed formulations and/or feed ingredients, can be described in the same manner and by the same function within the same species. This allows evaluation of a specific model and accordingly the prediction of the influence of specific feed formulations by a mathematical equation.