Relationship between the diet fed to goats and the odd and branched chain fatty acid contents in their milk

The contents of odd and branched chain fatty acids (OBCFA) in 94 milk fat (MF) samples, determined by gas chromatography, were used to identify the diet fed to dairy goats by linear discriminant analysis. The diets consisted on a control without added oil (CON), and the same diet supplemented with high oleic sunflower oil (OSO), regular sunflower oil (RSO) or linseed oil (LO). Four variables were selected as valid predictors (C7: 0, methylltetradecanoate, C15: 0 iso and the sum of C17: 0 plus C17: 1 cis-9) out of 21 OBCFA identified in milk fat. Wilk's test was highly significant (P < 0.001), which indicated that the data were appropriate for a linear discriminant analysis. The first discriminant function accounted for 90.9% of the variance between groups. Mahalanobis squared distances between the CON treatment and the other three oil treatments were similar (5.1 to 6.6, P < 0.001). Within the oil supplemented rations, the RSO and OSO treatments showed the highest distance (0.95, P < 0.05). The distances between the LIN treatment and the OSO and RSO treatments were not significant (0.27, P = 0.58, and 0.48, P = 0.28, respectively). According to the centroids, function 1 discriminated the CON treatment from the oil treatments. Methyltetradecanoate and C7: 0 presented the greatest discriminant ability in this function in positive and negative directions, respectively. Function 2 discriminated the RSO and OSO treatments in a weakened mode, with C15: 0 iso and C17: 0+C17: 1 cis-9 showing the greatest discriminating ability in positive and negative directions, respectively. The error rate count of the Fisher's functions was lower in the CON treatment (82.1% of observations were classified correctly) than in the oil treatments (31.8, 50.0 and 40.9% of observations were classified correctly in the OSO, RSO and LIN treatments, respectively). Individual observations in the OSO treatment were mainly misclassified in the LIN treatment, whereas several individual observations of the latter were misclassified in the RSO treatment. Obtained results indicate that the OBCFA content in milk fat MF does not predict the type of plant oil fed to goats, but allows an accurate classification of the milk fat MF from goats fed a diet with or without added oil by linear discriminant analysis.