RESUMO Objetivou-se, com o presente trabalho, selecionar equações de regressão com base na correlação entre o índice de clorofila e os teores foliares de nutrientes para predição em tempo real do estado nutricional de bananeiras tipo Prata. Utilizaram-se, como tratamentos, seis cultivares de bananeira tipo Prata com cinco repetições e quatro plantas úteis por parcela, dispostas em delineamento experimental inteiramente casualizado. Avaliaram-se os teores de nutrientes com base em análises laboratoriais e os índices de clorofila com um medidor portátil na terceira folha do ápice para a base. Os dados foram submetidos à análise de variância; calcularam-se as correlações entre teores foliares de nutrientes e índices de clorofila e se ajustaram equações de regressões para as associações significativas e com maior magnitude. Os modelos selecionados estimam os teores foliares de nutrientes e permitem predizer com segurança em tempo real e a baixo custo, o estado nutricional de bananeiras tipo Prata. Constitui uma técnica auxiliar que, aliada à diagnose foliar e à análise de solos, contribui para uma avaliação nutricional mais criteriosa e segura.

ABSTRACT This study aimed to select regression equations based on the correlation between chlorophyll index and leaf nutrient contents, for real-time prediction of the nutritional status of 'Prata' banana. Six cultivars of 'Prata' banana were used as treatments, with five replicates and four plants per plot, arranged in a completely randomized design. Nutrient levels were evaluated based on laboratorial analysis and chlorophyll indices using a portable chlorophyll meter, in the third leaf from the apex to the base. Data were subjected to analysis of variance, calculating the correlations between leaf nutrient contents and chlorophyll indices, and the regression equations were adjusted to the associations that were significant and with greater magnitude. The selected models estimate leaf nutrient contents and allow a real-time, low-cost reliable prediction of the nutritional status of 'Prata' banana. It is an auxiliary technique that, combined with leaf and soil analysis, contributes to a more precise and secure nutritional evaluation.

The aim of this work was to evaluate the carotenoid content and genetic variability of banana accessions from the Musa germplasm collection held at Embrapa Cassava and Tropical Fruits, Brazil. Forty-two samples were analyzed, including 21 diploids, 19 triploids and two tetraploids. The carotenoid content was analyzed spectrophotometrically and genetic variability was estimated using 653 DArT markers. The average carotenoid content was 4.73 µg.g-1, and ranged from 1.06 µg.g-1 for the triploid Nanica (Cavendish group) to 19.24 µg.g-1 for the triploid Saney. The diploids Modok Gier and NBA-14 and the triploid Saney had a carotenoid content that was, respectively, 7-fold, 6-fold and 9-fold greater than that of cultivars from the Cavendish group (2.19 µg.g-1). The mean similarity among the 42 accessions was 0.63 (range: 0.24 to 1.00). DArT analysis revealed extensive genetic variability in accessions from the Embrapa Musa germplasm bank.