Análise exploratória de constituintes inorgânicos em tecido, líquido ruminal e fezes de bovinos da raça Nelore (Bos Taurus Indicus)

Abstract

The inorganic constituents supplied in the diet are essential for the development of cattle. This dissertation aimed to propose a procedure for the multi-element determination of Ca, Mg, P, K, S, Na, Cu, Fe, Mn, Zn, Co, Cr, Mo, Se, and V in ruminal liquid samples, muscle, and feces of cattle submitted to two different diets, as well as to evaluate possible alterations according to the diets, in the contents of these analytes in the different samples evaluated. The experiment developed out with 52 male Nellore steers to evaluate possible changes depending on the diets in the analytes' levels in the different samples evaluated. The animals were confined and divided into 2 groups. The first was submitted to a conventional diet (with silage and soybean meal) and the second to a diet with by-products (citrus pulp and peanut meal). The samples were lyophilized, ground, and microwave-assisted digested in dilute nitric acid and hydrogen peroxide media. The analytes were determined by inductively coupled plasma optical emission spectrometry (ICP OES) and inductively coupled plasma mass spectrometry (ICP-MS). The accuracy was determined using certified reference materials (CRMs), NIST 8414, NIST 1577a, NIST 1573, and NIST 1515. Satisfactory results were obtained for the evaluated CRMs, and adequate detection and quantification limits for the intended use. Due to the large volume of results generated, the Student's t-test and the Mann-Whitney U test were performed for each analyte to verify if there was a statistical difference between the diets. The results indicated significant differences for Ca, P, Fe, Mn, Mg, Cr, and V in the feces and ruminal liquid samples. Feces samples also showed significant differences for Na, Zn, S, Cu, and Co. In muscle samples, only K showed a significant difference between diets. Exploratory data analysis was performed using linear discriminant analysis (LDA), principal component analysis (PCA), and random forest algorithm (RF). The results obtained by the LDA indicated that the groups (ruminal liquid, muscle, and feces) are very well defined and distinct and that the levels of nutrients are specific to each group. Through the PCA, it was possible to observe that only the feces samples had a clear separation between the groups of animals submitted to the conventional diet and the diet with by-products. The RF algorithm was used to verify whether the analyte's concentration could be used as a predictor of treatment groups and to verify which analytes would be the most important for this classification. The results obtained by the algorithm showed that it is only possible to classify the samples using the concentration of analytes present in the feces sample, confirming the results obtained by the PCA. The elements that contributed most to this classification were P, Zn, Mg, Ca, and K.