Seleção de descritores e análise de agrupamento em acessos de tabaco

Abstract

This study aimed to select descriptors morphological and agricultural traits and determine their relative importance in the characterization, as well as propose a minimum number to quantify the genetic divergence between accessions of tobacco type Sumatra and apply different methods of cluster analysis based on the simultaneous analysis and individual by means of quantitative descriptors and qualitative selected. For the characterization of accessions were initially used 43 descriptors morphological and agricultural traits, being 17 quantitative and 26 qualitative. Was the identification of quantitative descriptors redundant, by two procedures: 1) direct selection, proposal by Jolliffe and 2) Selection based on the coefficient of Singh. The selection of qualitative descriptors was performed by means of the entropy level of the characters (H), proposed by, Renyi's series. 10 Were selected quantitative descriptors and 8 qualitative descriptors. The disposal of 58% of the descriptors has not lost considerable information, since the descriptors are correlated with the remaining descriptors. For the analysis of divergence were used methods UPGMA and WARD, in which the UPGMA method was the best explained the genetic divergence among the accessions into study. They were used for the individual analyzes the Mahalanobis distance and Euclidean average for obtaining the arrays from the quantitative data and the qualitative data and original quantitative data processed was used the distance of Cole-Rodgers. For the simultaneous analysis of quantitative and qualitative data were tested three different methodologies: The algorithm of Gower; algebraic Sum of individual arrays and data integration through the transformation of quantitative traits in multicategoric by two distinct strategies (Rule of Sturges and Rule the square root). The correlation values between the arrays of quantitative data processed and original quantitative were significant at 1 and 5% of probability, one of them being of high magnitude, offering support to extrapolate the results of a set of data to another. With this, the strategy of square root was the most indicated, with a correlation of 0.75 and 0.82 between the arrays of dissimilarity of encoded data and quantitative originals. As a criterion for the definition of the optimal number of groups was used the index Pseudo-t2, this was possible the formation of 3 groups by UPGMA method for all the methodologies of simultaneous analysis used. Access to14 (125 PD) has distinct behavior of others and the methodologies of simultaneous analysis, based on matrices of distances generated, captured this divergence. All arrays of joint analysis were compared, showing great correspondence between them, with high correlations ranged from 0.824 to 0.998. The results of this work showed that the methodologies of simultaneous analysis were effective in identifying the existence of genetic divergence among accessions of Nicotiana tabacum L. type Sumatra and show the importance of the combination of methods, since they have been able to leverage a significant way to the interpretation of the results for greater knowledge of germplasm in study.