Gradients within gradients: the mesoscale distribution patterns of palms in a central amazonian forest

Abstract

Questions: What are the relative contributions of environmental factors and geographic distance to palm community structure at the mesoscale, and how do they depend on the length of the environmental gradient covered? How do soil and topography affect variation of the canopy and understory palm community structure at the mesoscale? How does fine-scale variation within the broad edaphic/topographic classes affect palm community composition? Location: Reserva Ducke, terra-firme forest, Manaus, Brazil. Methods: Palms were sampled in 72 plots 250 m × 4 m, systematically distributed over an area of 100 km 2. Soil, topography and distance to watercourses were measured for all plots. The relationship between community structure axes, summarized by NMDS ordinations, and environmental predictors, was analysed with multivariate regressions. Matrix regressions were used to determine the proportions of variance explained by environmental and geographic predictors. Results: Floristic variation at the mesoscale was mostly related to environmental variation, and the proportion of variance explained depended on the amplitude of the environmental gradient. Soil was the main predictor of floristic change, but its effects differed between life forms, with the understory palm community structured within one of the edaphic/topographic classes, in association with distance to watercourses. Conclusions: Dispersal limitation does not explain palm composition at the mesoscale, and the amplitude of environmental gradients covered by the analysis can be as important as the scale of analysis, in determining the relative contributions of environmental and geographical components to community structure. Soil and topography can predict a large proportion of palm composition, but gradients differ in scale, with some environmental gradients being nested within others. Therefore, although all environmental gradients are nested within distance, they do not necessarily coincide.