The Cornell Net Carbohydrate and Protein System (CNCPS) and NRC (1985)
models were evaluated for accuracy in predicting metabolizable protei
n (MP) and essential amino acid (EAA) allowable ADG, using chemical bo
dy and feed composition data from feeding trials with Holstein steers.
Nine Holstein steers (113 to 200 kg) were slaughtered and determined
to have the following whole-body essential amino acid composition of (
grams/100 grams of protein): arginine, 5.94; histidine, 2.07; isoleuci
ne, 2.28; leucine, 5.72; lysine, 5.81; methionine, 1.99; phenylalanine
, 3.04; threonine, 3.52; tryptophan, .57; and valine, 3.32. The NRC an
d CNCPS were then tested against data from 25 feeding periods, each re
presenting the 56-d growth of 10 Holstein steers (mean BW of 162 kg),
to determine their ability to predict the gain allowed by the supply o
f MP and the first-limiting EAA. The NRC (1985) system accounted for 4
6% of the variation in MP allowable gain, with an average bias of -30%
. The CNCPS accounted for 87 and 73% of the variation in MP and EAA al
lowable gain, with a bias of 8 and 5%, respectively. The bias was redu
ced to 3% (R2 of .82) when ADG was predicted by the factor (ME, MP, or
EAA) first-limiting ADG.