Organização das comunidades infestantes de plantas daninhas na cultura da cana-de-açúcar em agrupamentos-padrão

Abstract

The aim of this study was to evaluate the possibility of grouping stands of sugarcane harvested mechanically, and without previous straw burning in the region of Ribeirão Preto-SP, according to the potential of weed infestation. Clustering analysis and other multivariate analysis techniques were applied, using as variable the relative rate of infestation assigned by visual assessments, in two phases: in the first, twenty sugarcane plots with an 18-month cycle were selected. These areas were used to compare two methods to estimate the specific composition of the weeds, namely, a phytosociological analysis, described by Mueller-Dombois & Ellenberg (1974) and an analysis of the visual percentage of general coverage (GC) and specific coverage (SC). In the second phase, the specific composition of the weed community was evaluated in 189 plots in areas of ratoon sugarcane harvested in 2008, attributing to these plots only GC and SC. Based on the information obtained on the weed community, a database was built, and subjected to exploratory analysis of multivariate statistical techniques. For the main species within the stands (DIGNU, ARACH, IPOHF, MRRCI and IPOQU, followed by CYPRO, ELEIN and EPHHS), 75% of coinciding results were verified between the two methods. It was found that the visual assessments of species coverage percentage may substitute, for purposes of convenience, flexibility and applicability, the phytosociological evaluations, as these assessments provided good detection capability of the main weeds in each stand. The techniques of multivariate statistics showed that the plots can be grouped according to similarities in infestation intensity and species composition.