Uso do NIRS como ferramenta de diagnóstico nutricional de animais mantidos em pastagem natural

Abstract

The prediction of voluntary feed intake and digestibility of heterogeneous forages in rangeland is a challenge for researchers and technicians. The objective of this study was to develop models to predict the organic matter intake (OMI), organic matter digestibility (OMD) and crude protein (CP) of sheep diet grazing rangeland Pampa biome in southern Brazil through the faeces using the near infrared reflectance spectroscopy (fecal NIRS). Six sheep were used in a natural pasture area in Santa Maria, southern Brazil. The last five day of each month (from June of 2014 to June of 2014), individual feces from animal bags, and individual forage samples was collected per animal through hard plucking and extrusas, collected at the beginning of the biggest daily meal of animals, and at the end the biggest meal. The meals were identified through behavioral assessment for 24 hours prior collections. The diet samples and feces were dried (55 ° C), ground (1 mm), and determined the contents of dry matter, organic matter (OM), total nitrogen (N). In the diet samples was also determined content of digestible organic matter (DOM) through real-in situ digestibility by 48 hours multiplied by the MO content. In fecal samples were collected absorbance spectra in the near infrared range. The OMI was estimated by daily fecal excretion of N (DFEN), using the model, OMI, g OM/animal/day = 1.1 x 101.2 + DFEN, (r2 = 0.83), proposed by KOZLOSKI (unpublished) to sheep fed the same diet of the animals in this study. The OMD was obtained by equation, OMD, % DM = (1 ((daily fecal excretion of OM) / OMI) x OM medium of simulation samples. The reference values of the diet composition and OMI were individually related to fecal spectra through the partial least squares regression method. For developing the models, 80% of the total sample was used, and the remainder was used for external validation. The performance of fecal NIRS models was dependete the accuracy of the reference methods. The best model was the OMD variable when estimated by fecal N, followed by variables from the samples extrusa independete hour of collection, and then the sample average hard plucking and daily per animal. The OMI model present excellent result front difficulty is predicting this variable, the large number of envlvidos factors besides diet chemistry. The technology of fecal NIRS can be used as a tool for predicting the OMI, CP and OMD of sheep diet on natural pasture Pampa biome in southern Brazil.