Models to estimate the growth dynamics of Pennisetum purpureum cv. Cuba CT-169

Resumen

Different regression models were used to study the growth dynamics of Pennisetum purpureum cv. Cuba CT-169 through the agronomic variables accumulated yield of dry matter, percentage of leaves, and plant height. The data of the experiments were from the Pastures Department at the Institute of Animal Science. Every 14 d, data of the pasture biomass accumulation were collected at twelve cut ages during the rainy season, and at ten ages in the dry season, without fertilization or irrigation. Three linear models (one-way, quadratic, and cubic) and three non-linear (logistic, Gompertz, and exponential) were fitted. Four statistical criteria were applied to select the models of best goodness of fit. The models accounted for most of the total variability, with coefficients of determination superior to 89 %, minimum residual variance, significant parameters and adequate values for the rest of the criteria. The models of best fit in the rainy season were the Gompertz, for the accumulated yield of dry matter and the height, and the exponential, for the leaf percentage. In the dry season, that of best fit was the logistic, for the accumulated yield of dry matter and the height; and the one-way linear, for the leaf percentage. In the conditions of this study, the forage cut at 70 and 106 d of age is recommended in the rainy and dry seasons, respectively. Deepening into the performance of the bromatological indices is suggested, as well as validating the proposed models in further researches to simulate the response of this pasture to different conditions.
Key words: regression models, growth dynamics, Cuba CT-169, goodness of fit.