Skip to main content
ARS Home » Research » Publications at this Location » Publication #211190

Title: Comparison of Diffuse Reflectance Fourier Transform Mid-Infrared and Near-Infrared Spectroscopy with Grating-Based Near-Infrared for the Determination of Fatty Acids in Forages

Author
item Calderon, Francisco
item Reeves Iii, James
item Foster, Joyce
item Clapham, William
item Fedders, James
item Vigil, Merle
item Henry, William

Submitted to: Journal of Agricultural and Food Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/21/2007
Publication Date: 10/17/2007
Citation: Calderon, F.J., Reeves III, J.B., Foster, J.G., Clapham, W.M., Fedders, J.M., Vigil, M.F., Henry, W.B. 2007. Comparison of Diffuse Reflectance Fourier Transform Mid-Infrared and Near-Infrared Spectroscopy with Grating-Based Near-Infrared for the Determination of Fatty Acids in Forages. Journal of Agricultural and Food Chemistry 55:8302-8309.

Interpretive Summary: We compared the performance of three analysis procedures for individual fatty acids in forages using the absorbance of infrared light. A total of 182 samples from thirteen different forages belonging to 11 different species were sampled at 3 different times. Samples were analyzed for fatty acid content using conventional methods. Samples were then analyzed with three different types of infrared instruments available commercially. Our data analysis shows that the three instuments perform very well, and it is not clear whether any of the infrared methods is distinctly better than another. In summary, these results are valuable because we have shown how a new technology can be used to quantify forage fatty acids, which are important components of cattle diets, and can subsequently affect human nutrition.

Technical Abstract: Diffuse reflectance Fourier transform mid-infrared (FTMIR) and near infrared spectroscopy (FTNIR) were compared to scanning monochromator-grating-based near infrared spectroscopy (SMNIR), for their ability to quantify fatty acids (FA) in forages. Thirteen different forage cultivars belonging to 11 different species were sampled at 3 different times in a 5 replicate experiment. Samples were analyzed by gas chromatography to determine concentrations of individual FA. Samples were then scanned using SMNIR, FTMIR,and FTNIR. A total of 182 samples were used in the study. Three calibration analyses were conducted for lauric (C12:0), myristic (C14:0), palmitic (C16:0), stearic (C18:0), palmitoleic (C16:1), oleic (C18:1), linoleic (C18:2), and '-linolenic (C18:3) acids. Using all 182 samples in a one-out Partial Least Squares (PLS) calibration, statistics showed that the three spectroscopic methods performed similarly. Average R2 for the complete FA set was FTNIR (0.95)>SMNIR (0.94) >FTMIR (0.91). Excluding C12:0, a FA with low concentration and high variability, narrowed the differences for the three spectrometers to average R2 of 0.94-0.96. Constituents C18:2 and C16:0 had among the highest R2 regardless of the spectroscopic method used. The FTNIR did better for C12:0, C14:0, and C18:3. The SMNIR did better for C16:0, C16:1, C18:0, C18:1, and C18:2. A second set of calibration equations developed with half of the samples as the calibration set and the remaining samples as the validation set showed that all the methods could produce acceptable calibrations, with calibration R2 above 0.9 for most constituents. However, the SMNIR had a better average calibration relative error than the FTNIR, which was slightly better than the FTMIR. A third set of calibration equations developed using 100 random PLS runs with the 182 samples split randomly (122 for the calibration set and 60 for the validation set) also shows that the three spectral methods are satisfactory for predicting fatty acids in forages, although the SMNIR had more favorable error terms and R2 for the validation set. It is not clear whether any of the spectral methods is distinctly better than another. Calibration R2 and validation R2 were higher for most FA with the SMNIR than the FTMIR and FTNIR.