The use of linear regression to predict digestible protein and available amino acid contents of feed ingredients and diets for fish

An aquatic environment and unique biological characteristics complicate the determination of apparent digestible protein (ADP) and apparent available amino acid (AAAA) contents of feed ingredients and diets for fish. This urges for alternative methods of calculation, such as prediction from dietary chemical composition. Analysis of 617 means from 69 studies, which evaluated the protein digestibility of 73 feed ingredients with 35 fish species, revealed a linear relationship (ADP = − 10.0731 + 0.8942 CP, R2 = 0.9462) between dietary crude protein (CP; g/kg dry weight) and ADP (g/kg dry weight) contents. In compound diets (n = 831, 170 studies, 49 fish species) this relation was described by ADP = − 51.4001 + 0.9872 CP (R2 = 0.8632). Validation of the above with values from independent studies presented a high degree of agreement (R2 > 0.8000; mean prediction error < 0.1100) between predicted and determined values. AAAA profiles in feed ingredients and compound diets, expressed as g/kg dry weight, are predictable (R2 mostly > 0.9500) from its dietary amino acid contents through linear regression. The above equations provide a useful tool that could be use in practical feed formulation for fish. However, further work is needed on the validation and development of the prediction of AAA contents.