Effect of long-term agricultural management systems on occurrence and composition of weed species

This study aims to assess the composition of weed communities as a function of distinct selection factors, at neighboring areas submitted to distinct soil management and diverse use for sixteen years. Four areas submitted to distinct managements (conventional tillage system; no-till system; integration crop/livestock and continuous livestock) were sampled in relation to the occurrence and severity of weed species by the beginning of the planting season, being estimated the relative abundance, relative frequency and relative dominance of each weed species under each area, as well as the Importance Value Index for each species. Areas were also compared by the Sørensen's similarity coefficient. Areas where pasture and grazing were never present, exhibited a number of seedlings of weed species 250% higher than areas periodically or continuously under grazing, while the area of soil covered by weeds was 87% superior at the conventional tillage system in relation to the average of the other treatments. Grass weeds were the most important at the conventional tillage area while broadleaved weeds where more important at the no-till area, probably due also to herbicide selection factors. Under crop/livestock integration there may be the need to care about controlling seedlings of the forage species inside grain crops in succession.