ABSTRACT. The conservation of okra landraces [Abelmoschus esculentus (L.) Moench] in gene banks is essential for the success of their use in breeding programmes. This study evaluated the genetic diversity among okra landraces in Brazil based on morphoagronomic descriptors and AFLP markers. We studied 30 accessions of the vegetable gene bank of the Universidade Federal de Viçosa. To this end, 17 morphoagronomic descriptors and five combinations of AFLP primers were used. Genetic parameters were estimated for the quantitative traits and the accessions were grouped by Ward’s method, using the Gower’s and Jaccard’s distance measures, respectively, for the morphoagronomic and molecular data. Polymorphisms were observed for all qualitative traits, while the quantitative traits were significant by deviance analysis. The genetic parameters confirmed the existence of variability among accessions, and high accuracy and heritability indices were found for the traits related to fruit and plant height. Ward’s grouping showed no relationship between the clusters formed with the morphoagronomic and molecular data and the geographical origin of the accessions. No association between morphoagronomic descriptors and AFLP markers was observed. The lack of correlation suggests that both approaches of characterization are important to understand and differentiate the okra accessions.

RESUMO. O presente trabalho teve como objetivo estimar e caracterizar a divergência genética entre acessos de feijão-de-vagem determinado e indeterminado do banco de germoplasma da UEL com base em marcadores AFLP. Para tanto foram caracterizados 40 e 32 acessos de hábito determinado e indeterminado, respectivamente. Para análise do AFLP, sete combinações de primers EcoR1 e Mse1 foram testadas, sendo selecionadas as combinações E-AAG/M-CTC, E-ACT/M-CTT, E-ACC/M-CTT. Essas combinações de iniciador/enzima revelaram um total de 485 locos informativos, sendo a combinação E-ACC/M-CTT a que detectou o maior número de locos informativos (49%). A análise da distribuição das frequências da dissimilaridade revelou que a distribuição foi uniforme variando de 0.1285 a 0.7310 com média de 0.4801, com os acessos de hábito indeterminado apresentando uma maior variabilidade quando comparado com os de hábito determinado. Pelas análises de agrupamento UPGMA, PCoA e Bayesiana verificou-se a formação de dois grandes grupos, para os quais sugere-se uma possível predominância do hábito de crescimento do feijão-de-vagem quanto ao pool gênico. Os determinados podem estar mais associados ao pool gênico Andino, enquanto o indeterminado ao pool gênico Mesoamericano. Pela análise Bayesiana, verificou-se acessos intermediário aos dois grupos sugerindo-se uma introgressão entre os pool gênicos Andino e Mesoamericano.

ABSTRACT. The present study aimed to estimate and characterize the genetic divergence between determinate and indeterminate snap bean accessions from the Universidade Estadual de Londrina (UEL) germplasm bank based on amplified fragment length polymorphism (AFLP) markers. A total of 40 and 32 accessions with determinate and indeterminate growth habits, respectively, were characterized for this purpose. Seven combinations of primers corresponding to EcoR1 and Mse1 were tested for the AFLP analysis, and the combinations E-AAG/M-CTC, E-ACT/M-CTT and E-ACC/M-CTT were selected. These selective AFLP combinations revealed 485 informative loci in total, and the combination E-ACC/M-CTT detected the greatest number of informative loci (49%). The analysis of dissimilarity frequency distribution showed that the distribution was uniform, ranging from 0.1285 to 0.7310 with a mean of 0.4801, and the accessions with indeterminate growth habits exhibited greater variability than the accessions with determinate growth habits. The clustering UPGMA, PCoA and Bayesian analyses showed the formation of two large clusters, wherein there is a possible association between snap bean growth habit and gene pool. The determinate accessions may be more closely associated with the Andean gene pool, while the indeterminate ones may be associated with the Mesoamerican gene pool. The Bayesian analysis showed accessions intermediate to both groups, suggesting introgression between the Andean and Mesoamerican gene pools.