Modeling Herbage and Plant Part Accumulation of Five Cultivars of Brachiaria Spp. Using Daylength and Air Temperature.

Pasture production models can help plan forage-based livestock systems. In Brazil, the variations in forage production are markedly affected by the seasonal pattern of climate. The objective of this research was to develop models that describe herbage and leaf accumulation as a  function of temperature and daylength combined in a single climatic variable, the photothermal unit (PU). Five commercial cultivars Brachiaria spp. were studied (Basilisk, Marandu, Xaraés, Arapoty and Capiporã) in plots that were harvested weekly at 15 cm within a regrowth of 28 days in summer and 42 in winter. For predicting forage production, a logistic model Y = a / (1 + exp (- (xb) / c)) was used, with 'Y' being the quantity of forage accumulated in kg DM ha^-1, 'x', the accumulated PU, 'a', the asymptote (maximum value of y), 'x0' the point of inflection of the model and 'b' the inverse of the rate of acceleration of the model.  Capiporã was the most productive genotype (18 Mg ha^-1 year^-1), followed by Basilisk and Xaraés (15 Mg ha^-1 year^-1) and Arapoty and Marandu (13 Mg ha^-1 year^-1). The models fit well to the data points, showing predicted yields close to observed yields, both for total forage and for leaves. Prediction errors averaged 6 and 7% for summer and winter, respectively. The use of the logistic model proved to be feasible for use in the prediction of yield characteristics. There was a difference, however, between the yield of summer and winter under the same amount of PUs accumulated, suggesting that annual production cannot be predicted only by the parameters that make up the PU. For the models to become practical tools for planning of production systems, there is a need for further study, using larger datasets and a wider range of both environmental conditions and harvest management.