Marine reserves are rapidly being adopted as a conservation and management tool around the world. Much of this interest in using marine reserves has developed in response to the many studies documenting positive changes within reserves. However, while the reasons that different species respond differently to protection are now relatively well understood, differences in community-level responses remain difficult to predict. To be most effective, reserves need to be designed with an understanding of how species and communities will respond to protection. A major impediment to filling this gap in our understanding is the lack of ecological data at a scale large enough and over long enough a timeframe to allow differentiation of ephemeral and local changes in biological communities from persistent and general patterns. Data from the kelp forest monitoring program around the Channel Islands provide a unique opportunity to analyze the effect of multiple factors on community dynamics at large temporal and spatial scales. Entire communities have been monitored at sites spanning hundreds of kilometers across 20 years, documenting that community structure and stability vary significantly between sites. Using a variety of time-series and multi-factor statistical models, I will first evaluate which factors affect community variables such as stability, trophic and community structure, and productivity (measured as biomass). Second, I will test predictions of community response to marine reserves developed from these analyses in datasets of community change that span smaller spatial and temporal scales. Finally, I will experimentally test in the field how factors predicted to be particularly important interact to control community dynamics. Results from this work should help ecologists, conservationists, and resource managers develop appropriate expectations for how biological communities respond to reserve protection, and these expectations will in turn allow for optimal site selection and monitoring programs to be pursued.