Evaluation of nitrogen excretion equations for ryegrass pasture-fed dairy cows

• Published in: Animal (Cambridge, England) England), 2021-09, Vol.15 (9), p.100311-100311, Article 100311
• Main Authors: Christodoulou, C., Moorby, J.M., Tsiplakou, E., Kantas, D., Foskolos, A.
• Format: Article
• Language: English
• Subjects: Cattle; Environmental impact; Management; Zero-grazing
• ISSN: 1751-7311; 1751-732X
• DOI: 10.1016/j.animal.2021.100311

Accurate and precise estimates of nitrogen (N) excretion in faeces and urine of dairy cattle may provide direct tools to improve N management and thus, to mitigate environmental pollution from dairy production. Empirical equations of N excretion have been evaluated for indoor dairy cattle but there is no evaluation for cows fed high proportions of fresh forage. Therefore, the objective of the current study was to evaluate N excretion equations with a unique data set of zero-grazing experiments. Through literature searches, 89 predictive equations were identified from 13 studies. An independent data set was developed from seven zero-grazing experiments with, in total, 55 dairy Holstein-Friesian cows. Models’ performance was evaluated with statistics derived from a mixed-effect model and a simple regression analysis model. Squared sample correlation coefficients were used as indicators of precision and based on either the best linear unbiased predictions (R2BLUP) or model-predicted estimates (R2MDP) derived from the mixed model and simple regression analysis, respectively. The slope (β0), the intercept (β1) and the root mean square prediction error (RMSPEm%) were calculated with the mixed-effect model and used to assess accuracy. The root mean square prediction error (RMSPEsr%) and the decomposition of the mean square prediction error were calculated with the simple regression analysis and were used to estimate the error due to central tendency (mean bias), regression (systematic bias), and random variation. Concordance correlation coefficient (CCC) were also calculated with the simple regression analysis model and were used to simultaneously assess accuracy and precision. Considering both analysis models, results suggested that urinary N excretion (UN; R2MDP = 0.76, R2BLUP = 0.89, RMSPEm% = 17.2, CCC = 0.82), total manure N excretion (ManN; R2MDP = 0.83, R2BLUP = 0.90, RMSPEm% = 11.0, CCC = 0.84) and N apparently digested (NAD; R2MDP = 0.97, R2BLUP = 0.97, RMSPEm% = 5.3, CCC = 0.95) were closely related to N intake. Milk N secretion was better predicted using milk yield as a single independent variable (MilkN; R2MDP = 0.77, R2BLUP = 0.97, RMSPEm% = 6.0, CCC = 0.74). Additionally, DM intake was a good predictor of UN and ManN and dietary CP concentration of UN and ManN. Consequently, results suggest that several evaluated empirical equations can be used to make accurate and precise predictions concerning N excretion from dairy cows being fed on fresh forage.