Adsa Logo White Adsa Title White

Precision and accuracy of mid-infrared spectroscopy for milk urea nitrogen analysis.

E. M. Wood


Precision and accuracy of mid-infrared spectroscopy for milk urea nitrogen analysis.
E. M. Wood*, M. Portnoy, D. M. Barbano, K. F. Reed. Cornell University Ithaca, NY.

High levels of urea in blood, milk, and urine have been linked to poor nitrogen efficiency, increased feed costs, poor reproductive performance and increased environmental impact of dairy farming. Bulk tank average milk urea nitrogen (MUN) is often used to manage herd nitrogen efficiency. Current recommendations suggest MUN should be between 8 and 14 mg/dL to maintain milk production and reduce nitrogen losses, but a previous study found that commercial analysis of MUN ranged from 6.5 to 14.9 mg/dL for the same sample set. The objective of this study was to evaluate the precision and accuracy of milk testing lab MUN measurements. Milk samples were collected from multiparous Holstein cows (n = 16) 3 times daily (06:00, 14:00, and 22:00) over 7 consecutive days during early lactation (average DIM of 40 d). Samples were sent for analysis at 2 labs (A and B). Both labs tested samples using mid-infrared spectroscopy (MIR). Lab B also employed an enzymatic spectrophotometric method for measuring MUN on de-fatted and de-proteinated milk, here considered to be the gold standard protocol. The mean (sd) of MIR MUN for Lab A and B are 8.05 (1.33) and 6.41 (1.78) mg/dL respectively. The differences between the MIR MUN values and the enzymatic assay MUN for each lab were calculated and regressed on mean-centered enzymatic MUN values in a mixed model with random effect for day. Results of the linear regression are presented below. Labs A and B had significant (P < 0.01) negative slope coefficients, indicating MIR methods over predict MUN at low values and underpredict at higher values. The intercept estimates suggest Lab A significantly (P < 0.01) over predicts MUN by 2 mg/dL while Lab B tends (P < 0.10) to overpredict MUN by less than 0.5 mg/dL. Estimates of the residual variance and random effect of date indicate similar precision between labs. In spite of existing bias, results show the MIR accuracy has improved, however, sampled data is below industry average MUN and further work on samples with a wider range of MUN is required.Table 1. Parameter estimates of mixed model regression of MUN differences

A2.02 (0.173)−0.555 (0.0639)0.3171.44
B0.481 (0.218)−0.512 (0.0694)0.4221.56

Keywords: MUN, spectroscopy.

Biography: Emma Wood is a senior studying Animal Science and Agribusiness Management at Cornell University. Her love of animals and experience as the CT FFA State Secretary initiated her interest in animal agriculture and research. During her time at Cornell, she has cultivated that interest through multiple research projects, culminating in an honors research thesis with Dr. Kristan Reed. She also plays on the women's varsity polo team and serves as the Executive Mentoring Program Coordinator for the Meinig Family Cornell National Scholars. After graduation, she hopes to pursue an MBA and work in ag business.