Watch the slideshow

Ana Garrido and D. Pérez-Marín
UCO, Córdoba, Spain

Recent food crises have directly affected the animal feed manufacturing sector, highlighting the need to strengthen controls in this area. Accordingly, the UE has implemented new measures relating to the circulation of feed materials and compound feed. Particularly, the Directive 2002/2/EC of the European Parliament and of the Council imposed a compulsory declaration for all the feed materials as well as their amount in compound feedingstuffs for production animals, introducing the so-called “open-declaration”. Arising from these regulations, but also to facilitate process control, the animal feed industry has a need for analytical methods to determine the ingredient composition in the final product.

With regard to the declaration of the percentages of ingredients in compound feedingstuffs, there is considerable doubt as to whether any practical method exists that might be applied by all interested parties (industry, consumers, importers, inspection bodies). Light microscopy, currently the official method for identifying ingredients, has certain well-documented limitations when applied to quantitative analysis; moreover, it is a time-consuming technique and therefore unsuited to the routine quality-controls required of compound feedingstuff manufacturers. Therefore, the feed sector demands analytical methods that are fast, accurate and economical, and that enable real traceability in an ingredient matrix as complex as a feedingstuff.

Previous works, developed at the University of Córdoba (Spain), has proved that NIRS could predict an important number of ingredients and that could be an essential and useful analytical technology in the Safety Programs of feedcompounder plants.

Nevertheless, the prediction of ingredient composition is a complex task, where the calibration set characteristics, as set size, the presence of the different ingredient matrices and the variability of ingredients used to produce a given formula, appear to be determining factors for increasing the robustness of the equations developed to predict percentage of ingredients. Moreover, our research works have shown that the chemometric data treatment applied could be considered a key factor in enhancing the accuracy and precision of the calibrations developed, being necessary the use of advanced non-linear regression techniques to achieve adequate results.