Aiming the automatization of cultivar identification by protein electrophoresis, the objective of this work was evaluate the feasibility of using band intensity values as polimorphism aditional data on cultivar discrimination and compare different similarity coefficients on the analysis of gliadins electophoregrams data. Electrophoregrams of gliadinas take from four cultivars of triticale closely related were compared by the computer-assisted analysis using six binary (presence/absence) coefficients of similarity (Jaccard, Sorensen-Dice, Nei & Li, Simple Matching, Yule and Baroni-Urbani) and five quantitative coefficients (Pearson product moment correlation, Spearman, Percent Similarity, Modified Morisita and Gower). The intensity of bands used on the quantitative analysis of electrophoregrams was distributed in different number of classes of intensity, making possible to evaluate the effect of variability on this parameter. It was used about 60 single seed samples of each cultivar. In spite of low inter-varietal polymorphism, the coefficient means within of the cultivars was always greater than between cultivars. That clearly shows the high discriminating power of the tests. With this study we demonstrate the viability of identifying triticale cultivars by the numeric comparison of gliadin electrophoresis data with a library of electrophoregrams. It allows a more objective comparison of the electrophoregrams, comparing to visual analysis. We also found a great variability on intensity (density) data of bands. The parameter intensity is therefore not much reliable on gliadins polymorphism analysis in triticale. Hence about the intensity, there is a progressive lost on reliability of the results as increase the number of intensity classes.
similarity coefficients; cultivar identification; electrophoresis
