Question
How much variation in plant community structure and composition can be predicted using soil, flooding and topography in white‐sand vegetation of northern Amazonia?
Location
Brazil, northern Amazon, Roraima State, Viruá National Park (01°46′34″ N, 61°02′06″ W).
Methods
Data from 17, 1‐ha permanent plots distributed across a 25 km2 landscape within the Viruá National Park, northern Amazonia, was used to assess the soil, flooding and topography effects on tree species composition, diversity and structure. Our analyses were based on 16 599 trees (DBH ≥ 1 cm) belonging to 303 species. Principal components analysis (PCA) was used to reduce the dimensionality of soil variables, whereas dimensionality of floristic composition assemblage was reduced using non‐metric multidimensional scaling (NMDS). Multiple regression analysis was used to test combined effects of soil, flooding and topography on floristic composition, diversity, basal area and tree density.
Results
Depending on the diameter class or the type of data (qualitative or quantitative), NMDS axis 1 explained 45–84% of total variation in species composition among plots. The first and second axes from PCA explained, respectively, 54.8% and 15.1% of the variation related to soil gradients. Soil texture and fertility (PCA axis 1) were the most important predictors of the variation in tree composition, diversity and structure for the three different size classes analysed (DBH ≥ 1 cm, 1 cm ≤ DBH ≤ 9.9 cm and DBH ≥ 10 cm). Overall, plots on more fertile clayey soils showed higher tree diversity and basal area than those on well‐drained sandy soils. Although flooding had little effect on the tree community, the water level was higher in high‐stature forests established on more clayey and fertile soils.
Conclusions
Even though soils in the study area are predominantly sandy and oligotrophic, our findings highlight the importance of soil in structuring plant communities of seasonally flooded white‐sand vegetation at mesoscales in the Amazon Basin.