Proposta de modelagem não-linear do desempenho germinativo de sementes de milho híbrido

Abstract

This research was developed with the goal to adjust a nonlinear regression model to estimate seed germination performance of three commercial seed lots of hybrid maize OC 705, subjected to accelerated aging test at a temperature of 43°C. The logistic model y(t)=C/(1+exp (B (t-M))) was used to fit the data. Measures of bias Box and parametric intrinsic bends were used for the diagnosis of the model. The proposed model fits properly to seed germination percentage data of hybrid corn seeds for OC 705, for all seed lots. Based on estimates of the parameters and fit quality, the seed lot three was identified as being the best for marketing, by presenting less reduction in seed germination power over time.