ABSTRACT
Fertigation management of banana plantations at a plot scale is expanding rapidly in Brazil. To guide nutrient management at such a small scale, genetic, environmental and managerial features should be well understood. Machine learning and compositional data analysis (CoDa) methods can measure the effects of feature combinations on banana yield and rank nutrients in the order of their limitation. Our objectives are to review ML and CoDa models for application at regional and local scales, and to customize nutrient diagnoses of fertigated banana at the plot scale. We documented 940 "Prata" and "Cavendish" plot units for tissue and soil tests, environmental and managerial features, and fruit yield. A Neural Network informed by soil tests, tissue tests and other features was the most proficient learner (AUC up to 0.827). Tissue nutrients were shown to have the greatest impact on model accuracy. Regional nutrient standards were elaborated as centered log ratio means and standard deviations of high-yield and nutritionally balanced specimens. Plot-scale diagnosis was customized using the closest successful factor-specific tissue compositions identified by the smallest Euclidean distance from the diagnosed composition using centered or isometric log ratios. Nutrient imbalance differed between regional and plot-scale diagnoses, indicating the profound influence of local factors on plant nutrition. However, plot-scale diagnoses require large, reliable datasets to customize nutrient management using ML and CoDa models.
ABSTRACT
Na literatura, são escassas informações a cerca da avaliação nutricional para a maioria das culturas, em especial o amendoim. Objetivou-se avaliar o estado nutricional do amendoim cultivar BR-1, através da chance matemática. O experimento foi conduzido em Fortaleza-CE, no período de agosto a dezembro de 2009. As plantas foram cultivadas em canteiros, adubados com N (3, 21, 30, 39 e 57 kg ha-1), P2O5 (7, 46, 65, 85 e 124 kg ha-1) e K2O (5, 32, 45, 59 e 86 kg ha-1), sendo composto por dezesseis tratamentos conforme a matriz Plan Puebla III e quatro repetições. Durante o período de florescimento foi coletada folhas para a determinação dos teores de NPK. Utilizou-se a ChM matemática para fazer a diagnose foliar. Para N, as maiores chances matemáticas para obtenção de altas produtividades ocorreram com teores de N na faixa de 31,11 a 40,23 g kg-1, para P, de 1,66 a 2,11 g kg-1 e para K entre 13,42 a 18,15 g kg-1. Pela ChM, estimou-se que os teores de 35,67; 1,89 e 15,78 g kg -1 de N; P e K, respectivamente, como sendo os teores ótimos para a cultura. O método mostra-se promissor para determinar a faixa de suficiência e nível crítico para a cultura do amendoim cultivar BR-1.
In the literature, there is little information about the nutritional assessment for most crops, especially peanut. Therefore, the objective of this study was to evaluate the nutritional status of the peanut cultivar BR-1, by the method of mathematical chance (ChM). The experiment was conducted in Fortaleza-CE, from August to December 2009. Plants were grown in beds, fertilized with N (3, 21, 30, 39 and 57 kg ha-1), P2O5 (7, 46, 65, 85 and 124 kg ha-1) and K2O (5, 32, 45, 59 e 86 kg ha-1), comprising sixteen treatment as the matrix Plan Puebla III, with four replications. During the flowering period leaves were collected to determine the levels of NPK. The method of mathematical chance was used to make the leaf diagnosis. The greatest mathematical chance to obtain high yields occurred at contents from 31.11 to 40.23 g kg-1 for N; 1.66 to 2.11 g kg-1 for P and 13.42 to 18.15 g kg-1 for K. By the method of ChM, it was estimated that the levels of 35.67, 1.89 and 15.78 g kg-1 of N, P and K, respectively, were the optimum levels for the crop. The method shows promise for determining the sufficiency range and critical level for the peanut crop cultivar BR-1.