The 8th Revised Edition of the Nutrient Requirements of Beef Cattle: Development and evaluation of the mathematical model

Abstract Text:

The beef cattle nutrient requirements model (BCNRM) is a spreadsheet-based computer software program compatible with Microsoft Excel 2007 or earlier versions. The BCNRM contains two levels of solutions (empirical = ELS and mechanistic = MLS) to compute the supply of energy and nutrients to the animal. The calculation of animal requirements for energy and nutrients is the same for ELS and MLS. In the ELS, users can choose to either use tabular values for ME and NE or compute NE, ME, and DE from tabular TDN. Methane (CH₄) is computed based on empirical equations that combine animal and dietary chemical information. In contrast, the MLS computes TDN based on: 1) rumen digestibility of five carbohydrate pools (CA = sugars, CB1 = starch, CB2 = pectin, CB3 = available NDF, and CC = unavailable carbohydrate) and three protein pools (PA = NPN, PB = soluble CP, and PC = ADIN), assuming their fractional degradation rates (kd, %/h) and a fractional passage rate (kp, %/h) for each feed; 2) intestinal digestibilities for CB1, CB2, and PB for each feed; and 3) endogenous matter production for each feed. Then, similar to ELS, NE, ME, and DE are computed from TDN. In the MLS, CH₄ is computed based on the stoichiometric relationship of VFA produced in the rumen. The BCNRM includes an optimizer to assist with diet formulation and balancing, an ability to perform stochastic modeling, and a table generator that allows the user to create tables of nutrient requirements through an optimization procedure. The BCNRM was compared with the NRC (1996, 2000) levels of solution 1 (L1) and 2 (L2) using data from 20 experiments (n = 2,539 pen-fed animals). For ME-allowable gain, ELS and L1 predictions were nearly identical (r² of 0.999, root of mean square error (RMSE) of 0.018 kg/d, and accuracy (Cb) of 0.998). The MLS predictions tended to be greater than L2 predictions by approximately 0.158 kg/d, though there was a strong correlation between them (r² of 0.999 and Cb of 0.9). The opposite was observed for MP-allowable gain, MLS and L2 predictions were nearly identical (r² of 0.999, RMSE of 0.023 kg/d, and Cb of 0.998) while ELS and L1 predictions differed by 0.234 kg/d (r² of 0.975 and Cb of 0.9). A stochastic simulation (n = 5,000) predicted 122 and 97 g CH₄/d for ELS and MLS, respectively, with a 67% prediction overlap.

Keywords: Computer, Modeling, Simulation, Spreadsheet