Evaluation and application of the CPM Dairy Nutrition model

Authors

L. O, Tedeschi, Texas A&M University
W. Chalupa, University of Pennsylvania
E. Janczewski, University of Pennsylvania
D. G. Fox, Cornell University
C. Sniffen, Fencrest LLC
R. Munson, University of Pennsylvania
Paul J. Kononoff, University of Nebraska-LincolnFollow
R. Boston, University of Pennsylvania

Date of this Version

2008

Comments

Published in Journal of Agricultural Science (2008), 146, 171–182. Copyright 2007 Cambridge University Press.

Abstract

The Cornell-Penn-Miner (CPM) Dairy is an applied mathematical nutrition model that computes dairy cattle requirements and the supply of energy and nutrients based on characteristics of the animal, the environment and the physicochemical composition of the feeds under diverse production scenarios. The CPM Dairy was designed as a steady-state model to use rates of degradation of feed carbohydrate and protein and the rate of passage to estimate the extent of ruminal fermentation, microbial growth, and intestinal digestibility of carbohydrate and protein fractions in computing energy and protein post-rumen absorption, and the supply of metabolizable energy and protein to the animal. The CPM Dairy version 3.0 (CPM Dairy 3.0) includes an expanded carbohydrate fractionation scheme to facilitate the characterization of individual feeds and a sub-model to predict ruminal metabolism and intestinal absorption of long chain fatty acids. The CPM Dairy includes a non-linear optimization algorithm that allows for least-cost formulation of diets while meeting animal performance, feed availability and environmental restrictions of modern dairy cattle production. When the CPM Dairy 3.0 was evaluated with data of 228 individual lactating dairy cows containing appropriate information including observed dry matter intake, the linear regression between observed and model-predicted milk production values indicated the model was able to account for 79.8% of the variation. The concordance correlation coefficient (CCC) was high (r_c=0.89) without a significant mean bias (0.52 kg/d; P=0.12). The accuracy estimated by the CCC was 0.997. The root of mean square error of prediction (MSEP) was 5.14 kg/d (0.16 of the observed mean) and 87.3% of the MSEP was due to random errors, suggesting little systematic bias in predicting milk production of high-producing dairy cattle. Based upon these evaluations, it was concluded the CPM Dairy 3.0 model adequately predicts milk production at the farm level when appropriate animal characterization, feed composition and feed intake are provided; however, further improvements are needed to account for individual animal variation.