Gibberellic acid on postharvest quality of banana 'Maçã' evaluated by artificial neural networks

Abstract:

The objective of this work was to evaluate the effect of gibberellic acid doses on the extent of banana 'Maçã' postharvest life and the applicability of the use of artificial neural networks. A completely randomized design was carried out in a split-plot arrangement, with five doses of GA₃ in the plots and evaluation times in the subplots, with four replicates, with a three-fruit bunch by sample unit. Fruit were harvested with totally green peel and were immersed for 10 min, at room temperature, in the following treatments: 0 (control), 25, 50, 75, and 100 mg L^-1 GA₃. The bunches were stored in a cold chamber (20±1°C, with 91% air relative humidity). Artificial neural networks are effective, providing estimates of the coefficient of determination higher than those obtained by multiple linear regression models. After 19 days of storage, the dose of 100 mg L^-1 GA₃ is more favorable to banana 'Maçã' postharvest conservation, in all the evaluated traits during storage.

Index terms:
Musa; computational intelligence; fruit quality; growth regulator.