The error associated with the prediction of digestible protein contents of fish diets from tabulated values

Abstract

Numerous species available for culture, and lengthy, tedious and demanding digestibility experiments necessitate the use of tabulated values to calculate the apparent digestible nutrient contents of compound fish diets. The error associated with the above practice was evaluated in the present study with the use of apparent protein digestibility coefficients (APDCs) and dietary crude protein (CP) contents obtained in independent studies on fish nutrition. Prediction of APDCs of 21 compound diets evaluated in 18 studies with rainbow trout (<I>Oncorhynchus mykiss</I>), by using APDCs of 27 individual feed ingredients determined for this fish species, presented a mean prediction error (MPE) of 0.0489 between observed and predicted values. However, the above APDCs resulted in an overestimation of 7.86% (MPE = 0.0899) for APDCs of compound diets (<I>n</I> = 15) tested with 7 species other than rainbow trout. The CP contents of the above compound diets were overestimated with MPE’s of 0.3029 and 0.3200 for rainbow trout and other species, respectively, when using CP contents of individual feed ingredients as predictors. This eliminates the use of tabulated values to calculate apparent digestible protein contents of compound fish diets, except if databases are regularly updated with CP values for feed ingredients used.