Stand structure and tree species diversity patterns were examined plot-wide and among four topographically defined habitats (plateau, cliff, low plain and depressions) in a 120-ha permanent plot in an Acacia-dominated savanna in Mpala Ranch, central Kenya. The four habitats were defined by clustering the 3000 quadrats of 20 × 20 m in the plot based on their altitude, slope and convexity. Structural and floristic differences among the four habitats were examined and species-habitat associations were tested for the 30 most abundant species using torus translation randomization tests. The plot included 113 337 trees in 62 species with diameter at knee height ≥ 2 cm (18.4 species ha−1), 41 genera and 23 families. Fabaceae with the genus Acacia were the dominant family, followed by Euphorbiaceae and Ebenaceae. Tree density and basal area were twice as high on low plain and depressions than on the plateau. Species richness was highest in the cliff and was seven times higher than in the adjacent plateau. Half of the species assessed showed significant positive associations with one habitat and 21 showed significant negative associations with at least one habitat. The variation in stand structure and tree species diversity within the Mpala plot shows that topography is among the important drivers of local species distribution and hence the maintenance of tree diversity in savannas.

In tropical plant communities with diverse species, many congeners are found to coexist. Do environment or biotic interactions structure the coexistence of congeners in tropical forest communities? In this paper, we aimed to disentangle the effect of environment (first-order effects) and species interactions (second-order effects) on the spatial distributions of tree species. We used a classification scheme and torus-translation to test the first-order interaction of 48 species from 17 genera in a fully mapped 20-ha dipterocarp tropical seasonal rain-forest plot in Xishuangbanna, south-west China. Then we used heterogeneous Poisson null models to reveal significant uni- and bivariate second-order interactions. The results demonstrated that (1) 34 of the 48 studied species showed a significant relation with at least one topographic variable. This confirmed that topographical heterogeneity is important for distribution of these congeners. Spatial segregation (36.6%) and partial overlap (34.8%) were the most common bivariate association types in Xishuangbanna plot, which indicated first-order effects (environment) were strong. (2) For small-scale associations, 51% saplings (1 to ≤ 5 cm) (68.8% for large trees with dbh > 5 cm) of the species showed non-significant associations. For large-scale associations, 61.6% saplings (81.2% for large trees) of the species showed non-significant associations. Lack of significant species interactions provides evidence for the unified neutral theory. In conclusion, both environment and biotic interactions structure congeneric species' coexistence in tropical seasonal rain forest in this region.

Whereas broad-scale Amazonian forest types have been shown to influence the structure of the communities of medium- to large-bodied vertebrates, their natural heterogeneity at smaller scale or within the terra firme forests remains poorly described and understood. Diversity indices of such communities and the relative abundance of the 21 most commonly observed species were compared from standardized line-transect data across 25 study sites distributed in undisturbed forests in French Guiana. We first assessed the relevance of a forest typology based on geomorphological landscapes to explain the observed heterogeneity. As previously found for tree beta-diversity patterns, this new typology proved to be a non-negligible factor underlying the beta diversity of the communities of medium- to large bodied vertebrates in French Guianan terra firme forests. Although the species studied are almost ubiquitous across the region, they exhibited habitat preferences through significant variation in abundance and in their association index with the different landscape types. As terra firme forests represent more than 90% of the Amazon basin, characterizing their heterogeneity – including faunal communities – is a major challenge in neotropical forest ecology.