Estimation of overlapping seed shadows in a northern mixed forest

Abstract

It is difficult to make a direct measure of seed dispersal patterns from individual trees in forests that are poor in species richness due to overlapping dispersal distances. We estimated the `seed shadows' of three species in a mixed forest in northern Japan by measuring seed densities at various points during three dispersal seasons. We also measured the distance from each sampling point to surrounding mother trees, diameter at breast height of the mother trees, and the direction from each point to the mother trees. We assumed that the relationship between seed number, at a distance in all directions around a mother tree, and the distance followed a log-normal distribution in the case that wind blew equally in all directions. We estimated seed density from multiple mother trees at a given point using the above relationship and an allometric relationships between seed production and diameter at breast height of the mother trees. We then predicted the seed shadow as the log-normal curve that best explained the spatial variance in observed seed densities. The estimate was improved when wind direction was included for wind dispersed species such as Betula ermanii and Abies sachalinensis because of the existence of a prevailing wind direction. However, the estimate became worse for seeds not adapted to wind dispersal such as Quercus mongolica var. grosseserrata. The median of dispersal distance was negatively correlated to the seed weight among the three species. We conclude that the estimation of seed shadow from multi-located mother trees is possible by using the distance to mother trees, diameter at breast height of the trees, and wind direction data. This method may aid in studies of forest regeneration processes and regeneration control for forest management.