The objective of this research was to determine the number of soil samples needed to estimate with reasonable precision level the amount of seeds in the soil and the emerged weeds, in a seedbank of experimental and cropland areas, in order to help management decisions. Sampling the soil seedbank was carried out by using a tubular soil sampler core of 5 cm diameter, from 0 to 10 cm soil depth. Emerged weeds were scouted by placing iron quadrats (0.5x0.5 m), randomly. The number of samples required was estimated as a function of the average seeds per sample, given a required precision (CV = 20% or 40%). For the small plots, means between 10 and 20 seeds/soil sample (500 to 1,000 seeds/m²) and CV of 20%, it would be necessary 40 to 90 samples, respectively; with CV of 40% (lower precision), the need would be 10 to 20 samples. For the same seed number per soil sample in cropland areas, estimated sample size is about tree times higher. Emerged weed scouting presents lower data correlation between mean and variance, and was less appropriated for handling weed management decisions.
soil seed banks; weed control; soil sampling; method
