Abstract
Neotropical forests shelter an astonishing and underrated biodiversity. The integrity and stability of these natural ecosystems are threatened by human-induced environmental changes. Therefore, highlighting the determinants of ecosystem functioning, be they natural or anthropogenic, is a daunting but paramount scientific challenge. The intrinsic complexity of highly diversified ecosystems arouses both conceptual and logistical difficulties, which we handle, by manipulating tiny ecosystems, the tank bromeliads. The leaves of tank bromeliads form wells that hold rainwater and intercept leaf litter, allowing for a simple invertebrate aquatic community to thrive. These plants can be exhaustively sampled, are naturally replicated, and widely distributed.From local to biogeographic scales, this thesis aims at understanding how these communities respond to the natural heterogeneity of Amazonian forests, and at predicting the impacts of human-induced disturbances on the structure and functioning of these communities, using in situ and lab experiments. Particular focus is attributed to the influence of some aspects of climate changes (e.g., warming, and various precipitations scenarios) on individual physiology, behavior, and trophic interactions.