ESTIMATION OF THE SEEDBANK IN CULTIVATED AND DIRECT DRILLED SYSTEMS

The seed bank in a maize monoculture was estimated in eight different situations (with two cultivation systems, two sampling frequencies and two levels of weed control) and the statistics of the distribution of the principal species, as well as the number of samples needed for a given level of precision at various mean values, were studied. A total of 56 species was found and the number of seeds m-2 varied with the method of cultivation and the time of sampling from 11,300 to 106,900. None of the 186 data sets showed a normal distribution. Species occurring infrequently (mean of less than 1 per core) often followed a Poisson distribution while those of higher frequency (4-5 seeds/core) often showed an aggregative normal distribution. The negative binomial distribution fitted 72.6% of the cases. There were more species that did not conform with any of the distributions considered in the direct drilled than in the cultivated plots. In the eight situations the seeds of species that followed one of the distributions considered varied from 17.9-98.4% of the total. Soil cultivation and time of sampling affected fundamentally the uniformity of seed distribution while the nature of the flora did not affect the precision of the sampling procedure. In these experimental conditions 200 cores were necessary to determine, with good precision, the majority of species present in a plot.