Modelos não lineares na descrição do acúmulo de nutrientes e matéria seca em híbrido de milho

Abstract

In many situations, accurately describing the phenomena under analysis can be a challenge. The application of non-linear models represents a significant advance in data analysis, moving away from traditional linear methods and carefully considering the inherent complexities present in various phenomena. Therefore, the use of non-linear models in data analysis aims to effectively capture this complexity, providing more accurate and comprehensive information about the phenomenon under study. In this thesis, three studies were carried out with the aim of studying the growth curves of maize hybrids, based on the accumulation of dry matter and nutrients in the various phenological stages obtained in an experiment with two cultivars with different characteristics. The following non-linear models were used: Brody, Gompertz, logistic, Meloum I, Meloun II, Michaelis Mentem, modified Michaleis Mentem, Mitscherlich, Richards, Schnute, von Bertalanffy and Weibull. In the first study, the analysis was carried out using twelve non-linear models to describe the accumulation of dry matter in the corn hybrids GNZ2004 and P30F33. In the second study, data on the accumulation of macronutrients (nitrogen, phosphorus, potassium, calcium, magnesium and sulphur) in the two maize hybrids was considered. The third study presents the use of nonlinear models applied to micronutrient accumulation data (boron, copper, manganese and zinc). The analysis was conducted using the least squares method and the Gauss-Newton convergence algorithm. The selection of the most suitable models was based on the following quality of fit evaluators: adjusted coefficient of determination (), residual standard deviation (RSD), mean square error of prediction (MEP), asymptotic index (AI), Akaike information criterion (AIC), Bayesian information criterion (BIC). The results of this thesis enable nonlinear regression analysis to explore alternatives beyond the commonly used nonlinear models, and can be extended to other corn hybrids and crops.