Trait distribution patterns in savanna and forest plant assemblages and their relationship with soil features

Community ecologists seek to understand the processes acting on community assembly and the importance of species ecological differences to the co-existence of organisms. Here we investigated trait distribution patterns in two contrasting vegetation types, how the inclusion of intraspecific variability improves our ability to understand trait-based assembly, and if soil features predict the occurrence of non-random trait distribution patterns. We conducted our study at Emas National Park, Goiás, Brazil. We used a null model approach to investigate trait distribution patterns and a model selection approach to quantify soil features important in structuring assemblages. In savanna, we detected trait convergence at individual and species level (evidence of environmental filters) favoring plants with resource conservation strategies. In forests, however, trait dispersion at individual level equal to a random expectation corroborated a premise of neutral theory (individuals ecological equivalence). Nevertheless, at species level, we found convergence (seed mass) and divergence (height), reflecting distinct strategies related to light capture and resource use. Therefore, including intraspecific trait variability did not improve the detection of non-random trait distribution patterns. Furthermore, the influence of soil features on trait patterns was different between savanna and forest. There was a tendency for niche differentiation toward more fertile and clayey soils, but the relative importance of soil factors in assembling communities could not be generalized for savannas and dry forests.