It remains a challenge to quantify mammal-plant interactions including seed predation or herbivory of different body sized mammals and its consequences on tree community structure and diversity. Here we reviewed the results from different studies to demonstrate how exclosure experiments can be powerful methods to test various hypotheses in the Amazon. First, we quantified seed removal of Astrocaryum in dense groves and under solitary palms within and outside of large mammal (peccary, Tayassudae) exclosures. One major finding was that peccaries, more frequently, preyed upon seed in groves whereas rodents foraged more often underneath solitary palms. In a 7-year study, we used semipermeable exclosures to simulate large mammal defaunation and its consequences on seedling demographics. Results showed that seedling density, survival, and recruitment were higher in exclosures. Finally, we tested how mammalian seed predation affects forest community structure. We monitored the fate of 8,000 seeds from 24 different species in exclosures that were selectively permeable to small, medium, and large mammals for up to 4.4 years. Small, medium, and large mammals reduced survival time by 10.3, 15.1, and only 2 months, respectively. Small and medium-sized mammals reduced survival of 17 and 14 species respectively and changed beta diversity. Therefore, somewhat surprising, small and medium-sized mammals had a stronger effect on community structure and may promote species coexistence compared to large mammals. Short and long-term exclosure experiments can provide unique insights that contribute to our ecological understanding and may also allow to predict what changes to expect in over-hunted or empty forests.