Nonlinear models to describe the maize seed quality during the maturation stage: A bayesian approach

Abstract

The adjustment of linear and non-linear models to describe the longevity of seed was studied here. The Bayesian analysis is a robust statistical procedure with many possible applications. In this study, the Bayesian method was used to fit the seed germination data of two maize hybrids (OC705 and CD5501) as a function of the number of days after female flowering on two sowing dates (E1 and E2) to the following non-linear model: ) y(t)=A-Bexp(Ct)The accumulated dry biomass was also fit to the following nonlinear model: )y(t)=A(/1+exp((B/-t)/C). Ten consecutive assessments corresponding to intervals of four days were performed with four replicates. The sampling started 23 days after female flowering and the last sample was collected 59 days after this period. The Bayesian method enabled the study of germination curves. The results suggested the first sowing date (E1) for both hybrids as the most viable for planting. The results indicated 57 days after the female flowering as the ideal time for harvesting, when the maximum germination percentage (96 %) occurred. The procedure used in the present study also allowed an accurate comparison through the credibility intervals of the difference of the fit parameters, the hybrid types and the sowing dates. Therefore, this method was successfully applied to the seed industry to study germination percentage.