Weed seedbank and community shifts in a long-term cropping systems experiment

Characterizing the long-term effect of agricultural management systems on weed communities will aid in developing sustainable weed management practices. Weed seedbanks and aboveground biomass were measured within a corn– soybean–wheat crop sequence from 1990 through 2002 at Hickory Corners, MI. Four management systems were compared: conventional (CONV; full rates of N fertilizer and herbicides, moldboard tillage), no till (NT; same as CONV with no primary tillage), reduced input (RI; reduced rates of N fertilizer and herbicides, moldboard tillage, mechanical weed control, wheat underseeded with red clover), and organic (ORG; same as RI but no synthetic inputs). Multivariate ordinations of weed seedbanks showed a divergence of the CONV and NT systems from the RI and ORG systems. The CONV and NT seedbanks were dominated by grass species (mainly fall panicum and large crabgrass), whereas the RI and ORG systems were dominated by common lambsquarters and common chickweed. Within a single growing season, weed seedbanks in the RI and ORG systems were positively correlated with weed biomass whereas seedbanks in the CONV and NT system had little predictive value. Weed biomass from 1990 through 2002 showed a strong association of grass weed species with the corn phase of the CONV and NT system and common lambsquarters and redroot pigweed with the corn and soybean phases of the RI and ORG systems. Weed biomass diversity measures were negatively correlated with soybean yields in RI and ORG and wheat yields in NT, RI, and ORG. It is not clear whether crops were less competitive in the NT, RI, and ORG treatments, allowing new weed species to enter the plots, or whether less effective weed management in the NT, RI, and ORG treatments resulted in increased species richness, causing reduced crop yields. Mechanistic studies are needed to elucidate the relationship between weed community diversity and crop performance.