Data from long-term ecological studies further understanding of ecosystem dynamics and can guide evidence-based management. In a quasi-natural experiment we examined long-term monitoring data on different components of the Serengeti-Mara Ecosystem to trace the effects of disturbances and thus to elucidate cause-and-effect connections between them. The long-term data illustrated the role of food limitation in population regulation in mammals, particularly in migratory wildebeest and nonmigratory buffalo. Predation limited populations of smaller resident ungulates and small carnivores. Abiotic events, such as droughts and floods, created disturbances that affected survivorship of ungulates and birds. Such disturbances showed feedbacks between biotic and abiotic realms. Interactions between elephants and their food allowed savanna and grassland communities to co-occur. With increased woodland vegetation, predators' capture of prey increased. Anthropogenic disturbances had direct (hunting) and indirect (transfer of disease to wildlife) effects. Slow and rapid changes and multiple ecosystem states became apparent only over several decades and involved events at different spatial scales. Conservation efforts should accommodate both infrequent and unpredictable events and long-term trends. Management should plan on the time scale of those events and should not aim to maintain the status quo. Systems can be self-regulating through food availability and predator-prey interactions; thus, culling may not be required. Ecosystems can occur in multiple states; thus, there may be no a priori need to maintain one natural state. Finally, conservation efforts outside protected areas must distinguish between natural change and direct human-induced change. Protected areas can act as ecological baselines in which human-induced change is kept to a minimum.