Uso de modelos não lineares na descrição do acumulo de boro em diferentes partes do feijoeiro cultivar jalo

Abstract

The study on bean growth through accumulation over time of the micronutrient Boron is interesting because it represents an information tool that helps management of the plant development and detect factors that affect it. Modeling this type of growth produces a better understanding and an effective application of agricultural practices. In this context, the present study aims at fitting nonlinear models (Logistic, von Bertalanfy and Exponential) using original and transformed data on Boron accumulation in two planting systems and three seeding densities of the cultivated bean Jalo. The experiment was conducted at the Federal University of Lavras, Brazil, during the water harvest period (spring-summer) of the years 2006-2007 in a randomized block with three repetitions. In each planting system, the factorial scheme 3 x 7 was used. It consists of three seeding densities (75, 215 and 355 thousand plants per hectare) and seven evaluation periods (13, 23, 33, 43, 53, 63 , 73 days after emergence). The Boron accumulation was analyzed in the following plant parts: leaves, stem and leaves, stem, leaf and pod, and the complete plant. Both, the nonlinear Logistic and von Bertalanfy models are suitable for the description of Boron accumulation in the Jalo bean. Logistic model presents better fitting properties. Data transformation was used to ensure normality and homogeneity of errors but, since accumulation characteristics change over time, the use of an Exponential model to represent them was necessary. As a result, the validity of the ANOVA analysis is guaranteed.