Microhabitat associations and seedling bank dynamics in a neotropical forest.

We conducted a rigorous test of tropical tree seedling microhabitat differentiation by examining microhabitat associations, survival and growth of established seedlings of ten tropical tree species representing a four-factor gradient in seed size. Eight microhabitat variables describing soil and light conditions were measured directly adjacent to each of 588 seedlings within twelve 10x100 m belt transects at Paracou, French Guiana, and at 264 reference points along the transects. From these measurements, we defined three principal components describing soil richness, soil softness and canopy openness. Six of ten species (in 9 of 30 total cases) were distributed non-randomly with respect to microhabitat along at least one principal component. However, few species demonstrated clear microhabitat specialization. All shifts in distribution relative to reference points were in the same direction (richer, softer soil). Furthermore, of 135 pairwise comparisons among the species, only 7 were significantly different. More than three-fourths of all seedlings (75.3%) survived over the 2-year monitoring period, but survival rates varied widely among species. In no case was the probability of survival influenced by any microhabitat parameter. Relative height growth rates for the seedlings over 2 years varied from -0.031 cm cm(-1) year(-1) (Dicorynia guianensis, Caesalpiniaceae) to 0.088 cm cm(-1) year(-1) (Virola michelii, Myristicaceae). In only 4 of 30 cases was height growth significantly associated with one of the three principal components. Because the conditions in this study were designed to maximize the chance of finding microhabitat differentiation among a group of species differing greatly in life history traits, the lack of microhabitat specialization it uncovered suggests that microhabitat partitioning among tropical tree species at the established seedling stage is unlikely to contribute greatly to coexistence among these species.