Recent work on complex networks has provided a theoretical framework to unravel the structure of the Web of Life. Yet, we still know little on the implications of network architecture for the robustness of beneficial network services such as pollination of crop plants. In this proposal, I intend to move beyond current studies of global environmental change that have mainly focused on its consequences for the abundance, phenology, and geographic distributions of independent species, to embrace effects for the network of interactions among species. My strategy to tackle this overall goal is through a synthetic and interdisciplinary approach that combines theory, meta-analysis, and experiment. First, I will integrate a large dataset of ecological networks with phylogenies and life-history traits to simulate the rate and shape of loss of functional groups. Second, I will set up an experimental design to address how two important ecosystem services, pollination and control of insect pests, decline with network disassembly. Third, I will explore early-warning signals of network collapse that may predict the proximity of a critical threshold in the driving forces of global change. My approach can provide a starting point for assessing the community-wide consequences of the current biodiversity crisis.