Associating phenotypic to molecular data can be a powerful tool for the selection of parental genotypes for breeding and mapping purposes. Thus, the objectives of the study were: i) to estimate the genetic dissimilarity among 30 maize inbred lines (15 tolerant and 15 sensitive to flooding); ii) to select potential parents for mapping and breeding; iii) to compare the efficiency of different graphical models in displaying the calculated distances. A total of 21 RAPD primers were used for the estimation of genetic dissimilarity. The genetic dissimilarity was obtained according to the complement of Dice similarity coefficient, clustering procedure was performed by the average linkage method and the cophenetic coefficient was obtained. The complement of Dice similarity coefficient was subjected to principal components and multidimensional scale analyses, and the output efficiency was tested by the correlation between the original distances and those presented in the graphs. The clustering techniques did not reveal a perfect agreement with the original matrix, with correlations of 0.70, 0.53 and 0.75 for the dendrogram, principal components and multidimensional scale analyses, respectively. Among the tested techniques employed, multidimensional scale analyses gave more precise outputs, since this technique showed higher agreement with the original distance matrix, and preserved distances between all genotype pairs. Besides, this technique is the most indicated when the objective is to plan crosses, since it displays the distances between genotype pairs.
Zea mays; flooding tolerance; clustering techniques
