Avaliação da curva de crescimento de variedades de cana-de-açúcar utilizando modelos não lineares

Abstract

Sugarcane (Saccharum spp.) is a very important crop in the Brazilian agribusiness scenario. Evaluating the growth patterns of stalks height of the main varieties of this crop is of great importance to better understand the duration of the phenological phases for better management. This study aimed describe the growth curve of the stalks height of four different sugarcane varieties: RB92579, RB93509, RB931530 and SP79-1011 in plant cane and ratoon cane irrigated, using non-linear Logistic and Gompertz models and considering any assumptions violations. We used data from Almeida (2008), who conducted an experiment at the agro-meteorology experimentation area of the Agrarian Sciences Center of the Federal University of Alagoas (UFAL), Delza Gitaí Campus, Rio Largo, Alagoas. The parameters of the models were estimated by the method of least squares using the Gauss-Newton convergence algorithm. The fit quality of the models was compared using the adjusted coefficient of determination (R 2 ad j ), the residual standard deviation (RSD) and the corrected Akaike information criterion (AICc). The stalks growth rates (SGR) were determined by the first derivative of the best model for each variety. All statistical analyzes were developed in the R statistical software (version 3.4.3). The results indicated that the varieties showed sigmoidal growth pattern, adjusting well in both studied models. Overall, the models without addition of the autoregressive parameter AR(1) better described the stalks growth. The Logistic model was the most suitable for plant cane, while the Gompertz model best described most of the varieties in ratoon cane. Plant growth was faster for RB92579, which presented the higher SGR at the inflection point, followed by SP79-1011 in pant cane and RB93509 in ratoon cane.