Selecting the best set of gas chromatography-derived fatty acids to discriminate between two finishing diets using linear discriminant analysis

Linear discriminant analysis (LDA) was investigated as a method for identifying the type of finishing diet fed to bulls (n = 169) based on gas chromatography fatty acid (FA) analysis. The bulls were fed ad libitum a high concentrate diet comprised of a cereal–soybean meal based concentrate plus straw offered separately (HC) or a total mixed ration made of cereal, soybean meal, maize silage and straw (TMR). Eleven variables (10 FA and one FA ratio) were selected as statistically significant predictors out of 41 variables tested. The Mahalanobis squared distance between the HC and TMR groups was 3.386 and F-test of the distance was highly significant (P > 0.001). In cross-validated classification matrices, 18 cases were misclassified in the HC group and 16 cases were misclassified in the TMR group. As a result, 79.9% of original grouped cases were classified correctly. We concluded that it was possible to classify beef samples according to their finishing diets using LDA.