An authentication procedure for differentiating between organic and non-organic cattle production on the basis of analysis of serum samples has been developed. For this purpose, the concentrations of fourteen mineral elements (As, Cd, Co, Cr, Cu, Fe, Hg, I, Mn, Mo, Ni, Pb, Se and Zn) in 522 serum samples from cows (341 from organic farms and 181 from non-organic farms) determined by inductively coupled plasma spectrometry were used. The chemical information provided by serum analysis was employed to construct different pattern recognition classification models for predicting the origin of each sample: organic or non-organic class. Among all classification procedures considered, the best results were obtained with the decision tree C5.0, Random Forest and AdaBoost neural networks, with hit levels close to 90% for both production types. The proposed method involving analysis of serum samples provided rapid, accurate in vivo classification of cattle according to organic and non-organic production type.
Keywords: authentication, mineral elements/metals, organic cattle, pattern recognition analysis