Walloon Agricultural Research Center Walloon Agricultural Research Center, Quality Department Chaussée de Namur, 24 – 5030 GEMBLOUX - Tél :++ 32 (0) Fax : (0) Potential of Mid-Infrared Spectrometry for Prediction of Fatty Acid Contents in Cow Milk H. Soyeurt 1,7, F. Dehareng 2, P. Dardenne 2, D. Veselko 3, G. Lognay 4, C. Bertozzi 5, P. Mayeres 1,5, V. Baeten 2 & N. Gengler 1,6 1 Gembloux Agricultural University, Animal Science Unit, B-5030 Gembloux, Belgium 2 Walloon Agricultural Research Center, Quality Department, B-5030 Gembloux, Belgium 3 Milk Committee, B-4651 Battice, Belgium 4 Gembloux Agricultural University, Analytical Chemistry Unit, B-5030 Gembloux, Belgium 5 Walloon Breeders Association, B-5530 Ciney, Belgium 6 National Fund for Scientific Research, B-1000 Brussels, Belgium 7 F.R.I.A., B-1000 Brussels, Belgium 1. Aim and Objectives Since many years, gas chromatography is the most widely used method to determine the fatty acids profile. Even if this method is efficient, it involves a time consuming procedure, some expensive and pollutant reagents and qualified staff. Mid-Infrared (MIR) Spectrometry should be considered as a good alternative to assess fatty acids contents in milk. The aim of this study was to explore the possibilities offered by MIR Spectrometry for the calibration of the fatty acid concentrations in milk. 2. Material and methods Sampling and recording spectra files Between April to June 2005 in Wallonia, 600 milk samples were taken from 275 cows among 6 breeds in 7 reference herds chosen using different criteria (e.g. the percentage of milk fat or the type and number of breeds). Then the milk samples were analyzed by a MIR Spectrometer (FOSS MilkoScan TM FT6000) and the spectra files were recorded in a database. Figure 1 : Example of milk spectra Reference values Among the 600 spectra, 49 samples covering the whole variability in the space determined by a Principal Component Analysis (PCA) were selected. Then, the milk fat was extracted according to ISO 14156:2001 and analyzed by gas chromatography based on a methodology derived from Collomb et al. (2000). Calibration equations Using chromatographic and spectral data, multivariate calibration equations were built by using Partial Least Squares regression method (PLS). Chromatographic data were expressed in two ways : fatty acid concentrations in milk (g/dl) and fatty acid concentrations in milk fat (g/100g of fat). References Collomb, M. and T. Bühler Analyse de la composition en acides gras de la graisse de lait. Mitt. Lebensm. Hyg. 91: Soyeurt, H., Dardenne, P., Dehareng, F., Lognay, G., Veselko, D., Marlier, M., Bertozzi, C., Mayeres, P. and Gengler, N Estimating Fatty Acid Content in Cow Milk Using Mid-Infrared Spectrometry. J. Dairy Sci : Acknowledgement : Hélène Soyeurt acknowledges the support of the FRIA through a grant scholarschip, Danny Trisman for his laboratory work, the Walloon Breeding Association (AWE), the Walloon Milk Committee, Walloon Agricultural Research centre and the Walloon Regional Ministry of Agriculture for his partial financial support. 3. Results and discussion Table 1 : Estimated statistical parameters for each calibration equation that characterize concentrations of fatty acid in milk (g/dl of milk) 1 SD = Standard deviation ; 2 SEC = Standard error of calibration ; 3 R² C = Calibration coefficient of determination ; 4 SECV = Standard error of cross-validation ; 5 R² CV = Cross-validation coefficient of determination ; 6 RPD = Ratio of standard error of cross validation to standard deviation ; 7 SAT = Saturated fatty acids ; 8 UNSAT = Unsaturated fatty acids ; 9 MONO = Monounsaturated fatty acids ; 10 POLY = Polyunsaturated fatty acids The estimated concentrations of fatty acids obtained with PLS were more reliable for milk than for milk fat. In order to have good results of prediction, we need to have a good repeatability of the reference values and the statistic of the equations (coefficient of determination and RPD) must be as high as possible. The correlations between reference GC fatty acids and the percentage of milk fat were lower than the correlations between the fatty acids and the predicted values from the spectrum. This is the prove that the MIR spectra contain information for the different fatty acids independently and not only for the total fat content. 4. Conclusion This experiment showed that the MIR spectrometry technique associated to the multivariate calibration methods is a promisingly combination for quantifying C12:0, C14:0, C16:0, C16:1 9-cis, C18:1, C18:2 9-cis,12-cis, total saturated and monounsaturated fatty acids in milk. All these components represent a majority (>70%) of fatty acids present in milk. This combination could be used in many fields, such as, nutrition and dietetics, genetic and animal selection, animal feeding, … Figure 2 : Relation between total saturated fatty acids predicted and measured