Linear models to predict the digestible lipid content of fish diets
2009
SALES, J.
Values for the digestible contents of nutrients in diets and feed ingredients are of utmost importance in nutritional strategies for fish. Prediction from dietary composition would eliminate lengthy, tedious and demanding digestibility experiments with fish. Apparent digestible lipid (DL) content [range 7.6-353.4 g kg⁻¹ dry matter (DM)] in compound diets can be predicted with high accuracy (n = 610; studies =127; fish species = 34; R² = 0.9515; RMSE = 16.9504) from dietary crude lipid (CL) content (range 12.0-388.7 g kg⁻¹ DM) by the linear regression equation DL =-2.7303 + 0.9123 CL. Validation of this equation against 65 values from 15 independent studies presented R² and mean prediction error (MPE) values of 0.9947 and 0.0671, respectively. The corresponding equation for 37 individual feed ingredients evaluated in 24 studies with 18 fish species (n = 180) was found to be DL = -1.5824 + 0.8654 CL (R² = 0.9717; RMSE = 8.3765). However, validation of the latter is currently hampered by a lack of independent values.
显示更多 [+] 显示较少 [-]