Ground seed density patterns under conditions of strongly overlapping seed shadows in <Emphasis Type="Italic">Abies alba</Emphasis> Mill. stands

In three subsequent years, seeds were collected from 66 to 75 seed traps in three Abies alba stands that differed in vertical structure and the number of potentially reproducing trees. The objective was to compare empirical patterns of seed density with a random (Poisson) model, which assumes that seed density would be the same in every location, and a binomial negative model that is appropriate for over-dispersed data. The seed density patterns were tested for spatio-temporal independence. The effect of some dispersal variables (number of mother trees and seed fall seasons, dispersal distance, fecundity rate) on seed density pattern was studied in a simple simulation experiment. Several local stand density measures (stem density, basal area, distance-weighted indices, canopy openness) were also tested for their ability to predict local seed density. The Poisson model was rejected in all the stands studied, and a much better fit achieved using the binomial negative model. The simulation study revealed that random seed distribution may occur only at considerable dispersal distances, concave seed shadows, and at very low fecundity rates. The seed density patterns identified exhibited both spatial and temporal correlation, which can probably be linked to the arrangement of mother trees and variation in their seed output. To a minor degree, this seed density pattern also correlated with local stand density.