Coastal tropical forests are remarkably understudied and are well-suited for investigating how complex environmental gradients and differential dispersal can influence forest structure and composition.  In a seasonal secondary (38 to 40 years old) coastal forest in Costa Rica, I assessed the potential effects of a coastal gradient and seed dispersal on tree distribution, density, and species richness at three life stages (seedling, sapling, and adult).  To quantify the environmental factors, I established 18 permanent 100 m belt transects running perpendicular from the coastline to 350 m inland.  Within every 25 m in a stratified-random design, I placed a permanent circular plot to quantify tree densities and species richness for seedlings (stem height < 0.5 m; plot radius = 2m), saplings (2.5 m ≥ stem height ≥ 0.50 m; plot radius = 3 m), and adults (dbh ≥ 2.5 cm; plot radius = 5 m).  Seed rain was sampled every four weeks over a two-year period in 50 0.28 m² seed traps located randomly within every 20 m of the transects. Transects, plots, and seed traps were grouped into two zones, coastal and inland, because a limestone outcropping running parallel to the coastline (90-150 m from coast) naturally divided the study area

Results/Conclusions

I found differences in the forest structure and composition between the coastal and inland zones.  There were higher adult and sapling densities, canopy height, and species richness in the inland area relative to the coastal area.  Seedling densities did not differ between the two zones, but were low throughout the gradient.  Several environmental variables changed with distance from shore: altitude (+), soil texture (sand to clay), soil phosphorus (-), and soil pH (-).  Conversely soil salinity, slope, and canopy cover did not vary with respect to the gradient.  A NMS ordination showed segregation of adult plots by topographical and soil categories, but distance to shore was the only variable that explained at least 20% of the compositional differences among adult plots.  Seed rain composition reflected the shifts in tree composition along the coastal gradient, with some species' seed rain distributions extending beyond their adult distributions.  Although I found evidence of long-distance dispersal events (> 50 m), only three tree species were found in greater than 50% of the traps. My findings suggest that differences in secondary forest composition along a coastal gradient are not solely associated with differential seed dispersal, but also changes in the underlying environmental conditions.