A Comparison of Methods to Predict Weed Seedling Populations from the Soil Seedbank

Abstract

Accurate prediction of potential weed seedling density would allow growers to implement control measures more effectively and could help avoid inappropriate and over application of preemergence herbicides. We compared three methods for handling soil samples to predict potential weed seedling emergence in plow-disk and no-tillage corn: seedling emergence from greenhouse trays, emergence from intact cores, and seed extraction following sieving. Seedbank numbers were highest for common lambsquarters followed by annual grasses and redroot pigweed, and seed numbers were higher in no-tillage than plow-disk plots. Coefficients of variation typically exceeded 60%. Density of seedling emergence from cores was more similar to field density than was emergence from trays. The percent of seeds in the seedbank that emerged was commonly more than 90% for annual grasses and usually less than 20% for common lambsquarters. All methods gave equivalent and relatively poor predictions of field population density. Spearman rank correlation between predicted and actual populations ranged from low negative values for common lambsquarters in no-tillage to 0.60 for annual grass emergence from trays in 1991. No method consistently gave highest correlations in both years and both tillage systems. Seedling emergence from intact cores, evaluated for 4 yr in plow-disk and no-tillage soybean fields, was significantly correlated (r s = 0.15 to 0.68) with emergence in the field. Pooling data from three to five neighboring sample sites increased the correlation between core and field emergence densities.