Variáveis críticas para estimativa de produtividade do milho em função da população e espaçamento

Abstract

Corn is one of the most important cereals grown and consumed in the world. The objective of this work was to find a group of independent variables that influence and estimate maize (Zea mays L.) productivity modeled by multiple linear regression. The experimental design was in a completely randomized blocks, 2 x 2 factorial scheme, being two populations (45,000 and 65,000 plants ha-1 ) and two spacing ( 0.45 and 0.90 m), with 20 replicates. The model found consists of a linear combination of the logarithm of several factors, such as yield per hectare, number of ears per hectare, number of grains per row, number of rows and weight of grain. The assumptions of the model were analyzed as to the absence of serial autocorrelation between the residues, multicollinearity between the independent variables, residue normality, homoscedasticity of the residues, linearity of the coefficients. The next step was to verify through the model the estimation of the productivity with approximation of the real, for this we used data from the experiments carried out in the field by other authors. The results showed that the variables, in order of impact on productivity, are: EH (spikes per hectare), NGF (number of grains per row), MCG (mass of 100 grains) and NF (number of rows). The model proved to be effective, requiring calibration in all cases, due to possible changes that the variables can suffer regardless of the management and environmental factors.