1799
Evaluation of the updated version of CNCPS (v6.5)
The first version of The Cornell Net Carbohydrate and Protein System (CNCPS) was released in 1991, and since then it has been continuously under evolution. Our objectives are to describe the latest updates of the model resulting in version 6.5 and to evaluate model predictions against both literature and on-farm data. Degradation rates of protein and carbohydrate fractions were modified to meet new fractionation schemes, updated amino acid (AA) profiles on a whole feed basis were made and a combined efficiency of essential AA use was adopted representing an improved understanding of AA metabolism. Three different datasets were developed to evaluate lysine (Lys) and methionine (Met) requirements (AA dataset), rumen N balance (rumen dataset) and metabolizable energy (ME) and protein (MP) allowable (lactation dataset). In total 96 peer-reviewed studies with 367 treatments and 15 regional farms with 50 different diets were included. The AA dataset was used to estimate the concentration of Lys and Met that maximizes milk protein yield and content according to the broken line model with plateau: results suggested concentrations of 7.00 and 2.60 %MP for Lys and Met, respectively for maximal protein yield and 6.77 and 2.85 %MP for Lys and Met, respectively for maximal protein content. Proposed concentrations are slightly higher for Lys and 11-18% higher for Met compared with CNCPSv6.0 which can be attributed to changes in the AA profile of feeds. The ability of the model to predict post-ruminal flows of N and milk yield was assessed using the correlation coefficient based on the BLUP (R2BLUP) or model predictions using a mean study effect (R2MP) and the concordance correlation coefficients (CCC) to simultaneously account for accuracy and precision. The model predicted accurate and precise post-ruminal flows of rumen degraded and undegraded N (RDN and RUN, respectively; R²BLUP = 0.98 and 0.92 and CCC = 0.88 and 0.80 for RDN and RUN, respectively), bacterial N (R²BLUP = 0.97; CCC = 0.84) and provided a uniform offset of non-ammonia N that is robust with little bias (R2BLUP = 0.98; CCC = 0.92). For the lactation dataset, the model predicted accurate and precise milk yield according to the first limiting nutrient (MP or ME) with a R²BLUP = 0.95, R2MP= 0.78 and CCC = 0.83. Results suggest a significant improvement of the model due to current updates.
Keywords: CNCPS, evaluation, dairy cattle