Milk Composition Analysis

Proportional milk samples were analyzed in the Barbano Laboratory at the Department of Food Science at Cornell University (Ithaca, NY). On average, milk samples were analyzed within 0.6 ± 0.7 d of collection. An FTIR spectrophotometer (Lactoscope model CombiScone FT600, Delta Instruments) was used to analyze the contents of milk fat, true protein, and anhydrous lactose. As previously described by Kaylegian et al. (2009), prediction models utilized were the optimized basic model filter wavelengths, which were then validated by AOAC official validated chemical reference methods as described in Wojciechowski et al. (2016).

Woolpert et al. (2016) describes the partial least squares (PLS) prediction models and version numbers that were used to measure the concentration (g/100 g of milk) for de novo (C4 to C14), mixed (C16, C16:1, and C17), and preformed (≥C18) FA. Relative percentages (rel%) of the 3 FA groups were calculated by taking the concentration of each FA group and dividing it by the total milk FA concentration (both of which are reported in g/100 g of milk). To calibrate the milk FA parameters, gas-liquid chromatography reference chemistry was used as described previously (Wojciechowski and Barbano, 2016). The 14-sample set used to calibrate the main milk constituents was also used as the milk FA calibration set. Calibration concentration ranges, estimated yields, and PLS modeling parameters and version numbers for C16:0, C18:0, and cis-9 C18:1 were as stated in Seely et al. (2022). A new 14-sample milk calibration set was produced and used to adjust the slope and intercept of each component once every 4 wk. The set was run on the instrument each week on Monday morning to check mean differences between the instrument and reference values for each parameter to ensure the mean difference in g/100 g of milk was <0.01%. The typical standard deviation of the differences among the reference chemistry and instrument-predicted values for the 14-sample calibration set was <0.015 g/100 g of milk for each parameter. Milk urea nitrogen (mg/100 g of milk) and milk-predicted blood nonesterified fatty acids (mpbNEFA), mBHB, and mACE (expressed as mmol/L) were measured by FTIR using a PLS model developed by Delta Instruments using parameter version numbers #502, #1603, #1601, #1602, respectively (Bach et al., 2019, 2021) and calibrated using reference chemistry values from an enzymatic MUN assay (Portnoy et al., 2021).