Padrões de distribuição de plantas ao longo de gradientes ambientais na Amazônia Central: uma comparação entre duas paisagens

Abstract

Although an extensive literature deals with plant distribution patterns in tropical rainforests and their relation to environmental variables, these studies usually do not take in account that their results may be, at least in part, due to the amplitude of environmental variables present in the sampled sites and that within the main predictor variable one or more subordinate gradients may also be important in defining compositional groups. We sampled legume trees and terrestrial pteridophytes in two areas, each about ~25 km2 in size and separated by ~150 km. Both sites have similar topographic characteristics. One site has only infertile soils, while the other has a broad range of soil fertility, as it is located at the contact between two distinct geological substrates. Principal Coordinates Analysis and similarity matrices were used to ordinate the plots. Multivariate Regression Trees were used to organize the environmental predictors in a hierarchy according to their importance in determining compositional groups. At the site with only infertile soils, clay content was the most important predictor of composition. At the site with a broad range of soil fertility, the sum of exchangeable bases was the most important determinant of compositional differences. This was true for both plant groups. For legume trees, within certain a range of values of the main predictive variable, terrain slope appeared as a subordinate predictor. For pteridophytes, the most important differences in composition were nested within the low end of the values of the main predictor, i.e. a nonlinear relationship. When all samples from both sites were pooled for an analysis at the regional scale, soil fertility was the main compositional predictor. Low fertility plots from the two different sites were more similar than low and high fertility plots within the same site. At this regional scale, clay content was an important subordinate predictor for pteridophytes. This study shows that, when generalizing plants community patterns, we must take into account that landscape heterogeneity can change in an unexpected way and that the importance of an environmental predictor tends to be constrained by its amplitude.