Evaluation of in vitro methods and mathematical models to estimate the dry matter and fiber digestibility of feeds for cattle

A presente dissertação foi elaborada com base em dois experimentos. No primeiro experimento comparou-se a digestibilidade in vitro da matéria seca (DIVMS) e da fibra em detergente neutro (DIVFDN) de forragens e concentrados obtidas por intermédio das incubadoras DaisyII e TE-150, combinadas com filter bags tipo F57 ou tecido não-tecido (TNT, 100 g/m2) em relação aos valores obtidos utilizando-se o método de Tilley e Terry (1963). Foram utilizadas 25 amostras de alimentos concentrados e 25 de forragens. Não houve diferenças entre filter bags (P>0,39) para a DIVMS de forragens. Maiores valores de DIVMS (P0.39) for IVDMD of forages. Higher IVDMD (P<0.01) was observed for TE-150 compared to DaisyII. Estimates of IVDMD obtained with incubators and filter bags were higher (P<0.01) than those obtained with the Tilley and Terry (1963) method. Higher IVNDFD (P<0.01) were obtained with filter bags F57 compared to NWT and TE-150 when compared to DaisyII. All values obtained with incubators and filter bags were higher (P<0.01) than those obtained with the Tilley and Terry (1963) method. Both IVDMD and IVNDFD variables were greater using TE-150 incubator than using DaisyII (P<0.04). Estimates obtained using NWT filter bags were higher (P<0.01) than F57 filter bags. All methods were positively and strongly correlated (P<0.01) to each other with regard IVDMD and IVNDFD. The repeatability and variances among samples for IVDMD and IVNDFD were similar among methods. It is concluded that all methods are similarly able to discriminate samples. In the second experiment, it was suggested model for predicting the digestible fraction of neutral detergent fiber (NDF) of forages and concentrates offered for cattle managed under tropical conditions as an alternative to the current model adopted by the BR-CORTE System. Two studies sequentially were carried out. The first one study was carried in laboratory and it aimed to build models that could allow the fractionation of NDF from forages and concentrates based on its potential of ruminal degradation by using chemical characteristics. Three hundred seventy one samples of forages and sixty five samples of concentrates were used. The models built for both types of samples based on the prediction of pdNDF. For forages, the pdNDF prediction model was given by the linear effect of NDF corrected for ashes and protein contamination (NDFap; P < 0.01), linear and quadratic effects of acid detergent fiber (ADF; P < 0.01 and P < 0.05, respectively) and linear effect of lignin (P < 0.01). For concentrates, the pdNDF prediction model was given in terms of linear effects of NDFap (P < 0.01) and ADF (P < 0.01), a "dummy" variable was also introduced (P < 0.01) in the case of feedstuffs containing fiber of lower potential degradation. The second study was carried out aiming to build empirical model for predicting the pdNDF digestibility of forages and concentrates. The models were built from information of 60 animals receiving diets exclusively composed of forages. The mathematical basis of the empirical model was built by integrating the ruminal degradation rates (kd) and ruminal passage (kp) of pdNDF and average utilization in the large intestine, called intestinal adjustment factor (IFA). The models built to estimate kd, kp, and IFA were obtained on the basis of chemical characteristics and feed intake level of the animals in the database. The comparative evaluation of the empirical model and the model adopted by BR-CORTE was performed using forages and concentrates information from the CQBAL 3.0 database, through a simplified assessment of the mean square prediction error. The estimates of digestible NDF obtained from the empirical model are more accurate and precise than the estimates obtained from current BR-CORTE model.Fundação de Amparo à Pesquisa do Estado de Minas Gerai