Questions
We test the hypothesis that soil water availability operates as an ecological filter on water economy plant traits, particularly in savanna stands, with consequences for species distribution and the functional profile of plant assemblages across a savanna–forest mosaic.
Location
A savanna–forest mosaic, northeastern Atlantic forest, Paraíba state, Brazil.
Methods
We sampled all trees and shrubs in 11 forest and 12 savanna plots in the same climatic region. Soil water storage capacity was measured, as well as the following traits: specific leaf area (SLA), internode mass fraction (IMF), maximum height (Hmax), wood density (WD), presence of trichomes (LT) and presence of abiotic seed dispersal (DD). Differences in trait values and water soil content between forest and savanna plots were assessed using Wilcoxon tests. Standardized effect size of functional richness and functional dispersion (SESFRic and SESFDis, respectively) were compared between forest and savanna plots with Wilcoxon tests, and the relationships between both functional indices of each trait and soil water content assessed using simple linear regression. We also used null models to test for habitat filtering in each plot, considering species assemblage (SESFRic) and abundance (SESFDis).
Results
Vegetation types differed in terms of soil water availability, with savanna soils having, on average, half the content of forest soils. Compared to forest, savanna plants had heavier internodes (30% increment), lower SLA (about 42% lower) and more plants with trichomes, but plants were much smaller across savanna assemblages. Savanna assemblages also exhibited a higher probability of trait convergence (i.e. reduced functional diversity) than forest assemblages in terms of SLA and maximum height, but savannas were less prone to trait convergence in terms of leaf trichomes, considering species abundance. Considering both vegetation types, a variable proportion of assemblages exhibited significant values of either FRic or FDis, but soil water availability did not correlate with functional diversity measures.
Conclusions
Habitat filtering not related to soil water availability favours the occurrence of conservative resource‐use plants across savanna stands, reinforcing the notion that filtering is able to act as an assembly force by affecting the abundance/frequency of particular traits, thus contributing to the occurrence of spatially organized mosaics of vegetation.