The global environment is changing rapidly through processes that act at multiple spatial and temporal scales. It is now clear that local processes can affect broader-scale ecological dynamics, and that broad-scale drivers may at times overwhelm local patterns and processes. Understanding these cross-scale interactions requires a conceptual framework to predict continental-scale ecosystem dynamics based on connectivity in material and information flow across scales. In this paper, we (1) discuss a multi-scale conceptual framework, including the key drivers and consequences of ecological connectivity acting across spatial and temporal scales, (2) provide a series of testable hypotheses, predictions, and an approach derived from this conceptual framework, and (3) propose the development of a network of networks largely taking advantage of existing research facilities and cyberinfrastructure to address these hypotheses. This framework and associated research infrastructure based on connectivity across scales will uniquely enable our understanding and enable us to better forecast global environmental change at multiple spatial scales, from local research sites to regions and continents.