Patterns of seed input into our study sites

Our work strongly indicates that the most important limitation on regeneration of forest in our abandoned pastures is the availability of seeds of the forest species. We conducted a seed rain study during 1996-1998, using an extensive array of seed traps in each of our five study sites. Each trap was monitored for 14 months. The traps were arranged in pairs, in two parallel lines extending into forest, and out into the pasture. In the figure to the right, the bright green line indicates the forest boundary -- pasture is to the right, and forest to the left. In each site, the parallel lines (transects) were positioned approximately on a ridge, or on a slope. Our findings showed that seed input did not differ between the ridge and slope transects of seed traps. Our findings are summarized in a manuscript currently in press at Journal of Tropical Ecology, on which our colleague Jerald Dosch is the lead author.

The most dramatic finding of our seed rain study was the severe decrease in seed input to seed traps >5 m from the forest edge. The seed rain 15 or 30 m from forest was typically <1% of the density of seed input in the forest or at the forest/pasture boundary. This graph shows seed input to Site 3. The vertical height of the symbols is an indication of total seed input over 14 months; note that it is on a log scale. In the forest, the vast majority of seed input was from a few species with tiny seeds, in the family Melastomataceae. However, not only we seed density drastically reduced in the pasture compared to forest, but the species richness of the seeds was lower in pasture as well.

This figure shows the major shifts in contribution of different types of dispersal to the seed rain, as a function of distance from forest. Data from three of the sites is presented. The black part of the bars shows the proportion of seeds that were of bird dispersed species. The white part of the bars shows the proportion that were from wind dispersed species. Clearly, in the forest, the overwhelming majority of seeds were animal dispersed, reflecting the tremendous abundance of seeds from the family Melastomataceae, which is mostly bird-dispersed. At greater distances into pasture, the importance of wind dispersal increases, although less so at Site 2 than the other sites.