When are buttresses and stilt roots necessary for a tree in terra-firme Amazonian forests?

Determinants of the occurrence of buttress and stilt roots are still poorly understood. These may be linked to individual's properties as a way to increase stability as trees get higher, to environmental factors that increase falling risks or interactions of both. We took advantage of a large dataset (presence of buttress, stilt roots, and tree diameter in 8415 trees from 35 1-ha plots in central Amazonia) to investigate how tree and environmental properties interact to determine the occurrence of support structures. We also made detailed measurements of allometry and canopy size in 67 trees of two dominant species. The probability of occurrence of support structures was modeled with multiple logistic regressions and boosted regression trees. We established a best-fitting subset model based on AIC ranking using unsupervised model selection. At the landscape scale, support structures were more common in large trees (b_std = 0.88, p < .001) and valleys (b_std = −0.09, p < .01), due to species turnover along topography and also due to intraspecific variation in the development of buttresses within species, linked to interactions of tree size and topography. The relationship between height and diameter (Height:Diameter) was the most important determinant of buttress occurrence (b_std = −1.57, p < .001). We conclude that less stable soils select a higher frequency of trees with support structures. However, coordinated allometric relationships among stem and crown sizes also influence the need of support structures. Thus, the presence of support structures depends on the interplay of individual plant's allometry and environmentally imposed conditions of instability. Abstract in Portuguese is available with online material.