Prediction of the nutritional value of grass species in the semiarid region by repeatability analysis

Abstract

The objective of this work was to estimate the repeatability (r) and the number of samples required to measure the nutritional value of four warm-season forage grasses growing in a semiarid region. The grasses evaluated were Urochloa mosambicensis, Cenchrus ciliaris, Digitaria pentzii, and Megathyrsus maximus. Evaluations occurred under two forage management conditions: stockpiling and grazing. Hand‑plucked forage samples were analyzed for dry matter (DM), crude protein (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF), lignin, and in vitro DM digestibility (IVDMD). Four methods were used to estimate r and the number of samples required: analysis of variance, method of principal components based on the covariance (PCCOV) and the correlation (PCCOR) matrices, and structural analysis (EVCOR). Species were compared by the probability of the difference using the t-test. The method PCCOV presents the highest coefficient of repeatability and, therefore, a lower number of samples required. Lignin is the trait that have the highest number of samples required. In terms of qualitative traits, D. pentzii and M. maximus show the best forage qualities among the species evaluated.