O ensaio de citogenotoxidade em Allium cepa L.: variações metodológicas e influência sobre a qualidade dos resultados

Abstract

Environmental mutagenesis and cytogenotoxicity are areas that are concerned with chemical or physical agents with the potential to induce cytological and genetic damage. Several biological models have been used for environmental monitoring studies, and Allium cepa L. is a consecrated plant in the evaluation of cytogenotoxicity. The method lacks guidelines, and a wide methodological variation exists in the articles, including variations in design and planning for statistical analysis. These are important in obtaining quality data essential to a good test of your hypotheses. The objective of this work was to perform a bibliographic review in order to understand all the variations found in the method and analyze the impact of these variations on the quality of the data obtained, and later to test two alternative forms of statistical analysis in a cytogenotoxicity experiment in A. cepa. For each treatment used, 9 slides were made and arranged in different ways in the two forms of analysis. In the form of analysis A we considered each slide as a repetition (9 repetitions) and in the form of analysis B we considered the mean analysis of each 3 slides as a repetition, where we had, then, 3 repetitions. The slides that constituted each repetition were previously defined. Our hypothesis was that the 2nd way of analysis would obtain better results and a more robust statistical analysis. Actually, the form of analysis B was more robust, with higher F calculated in the Variance Analysis, lower estimate of experimental error and greater detection of differences between means in the Tukey and Dunnett tests (all tests performed at p<0.05). In the Tukey test, the d.m.s. (significant minimum difference) was lower in the use of 3 repetitions (2.02) compared to the use of 9 repetitions (d.m.s = 2.80). In Dunnett's test, the reduction of d.m.s. was even greater (1.70 for the use of 3 repetitions compared to 2.33 in the use of 9 repetitions). Regarding the achievement of the means and standard deviations of the treatments, we also perceived the advantage of adopting the form of analysis B. This reduced by 59.25% the mean values between treatments of the standard deviation, consequently reducing the estimate of experimental error. However, the observations raised do not seem to be conclusive and do not seem to work for any number of treatments, so we continue to make efforts to better understand these methods of analysis.