Figure 6

Principal Component Analysis (PCA) biplot of 67 farm samples for four variables used for the discriminant analysis (3maxcen, 2subrat2, 6sumrat2, 4maxrat3). The samples are coloured by the initial classification. The “drift” farms are clustered around (0, 0) while “application” farms are spread on left of the plot, and the “unclear” cases are distributed throughout.