Experimental error in AFLP markers: consequence and estimative

Abstract

The objective of this study was to estimate and to compare the Dice coefficients of dissimilarity through experimental error estimation, from the use of replications composed by plant bulks. The estimate of this coefficient can become more reliable and robust with the use of replications in the molecular analyses. This procedure favors either the tests of hypotheses or the evaluation of the contrasts of interest, according to the objective of the experiment. Also, to compare these estimates with each other and with the genetic dissimilarity estimate based on agronomical traits measured in the field. Among the evaluated variables (genetic and morphological), the distances estimated with base in the use of molecular versus morphological data evidenced 80% of agreement in the estimated values, compared to 5% (P < 0.05) when the replications were used, in the molecular analysis. This way, it is evident that the use of replications in molecular analyses, besides supplying an estimate of the experimental error, enables an immediate gain through the tests of hypotheses, increasing both the reliability of the estimates of the genetic parameters (i.e., repeatability), and the agreement between laboratory and field data.