Climate warming will likely disrupt the flow of matter and energy within ecosystems, threatening the global carbon balance. Microorganisms are fundamental components of carbon cycling and are thus integral to ecosystem climate responses. However, ecosystem responses to warming are uncertain due to the functional and trophic complexity of microbial food webs. Here, we expose two major black boxes hindering our ability to anticipate ecosystem climate responses: viral infection and predation by microbial predators. We review current knowledge and uncover critical gaps in knowledge about how warming will impact these important top-down controls on the global carbon cycle. Understanding and predicting ecosystem responses to climate change will require disentangling complex direct and indirect responses within microbial food webs.

Mixotrophs are ubiquitous and integral to microbial food webs, but their impacts on the dynamics and functioning of broader ecosystems are largely unresolved. Here, we show that mixotrophy produces a unique, dynamic type of food web module that exhibits unusual ecological dynamics, with surprising consequences for carbon flux under warming. We find that mixotrophs generate alternative stable carbon states across temperatures---including an autotrophy-dominant carbon sink state, a heterotrophy-dominant carbon source state, and cycling between these two. Moreover, warming always shifts this mixotrophic system from a carbon sink state to a carbon source state, but increasing nutrients erases early warning signals of this transition and expands hysteresis. This suggests that mixotrophs can generate critical carbon tipping points under warming that will be more abrupt and less reversible when combined with increased nutrient levels, having widespread implications for ecosystem functioning in the face of rapid global change.

Understanding which foodwebs thrive or collapse is a major challenge that has been mostly studied in terms of topology and interaction strength. Yet the relative importance of these properties is hotly debated due to limited research on how they interact and which forces generate them. Here, we construct a foodweb model that incorporates mass-based constraints on density dependence, maximum consumption rate, and the likelihood and strength of interactions, which in turn control overall topologies and interaction strength distributions. Our model predicts both stability and connectivity that closely match real foodwebs ranging widely in size (29-163 species) and connectivity (113-1086 interactions). Despite their absence in most research, density dependence and maximum consumption rate are not only required to accurately predict stability but have stronger impacts than the more-frequently-studied interaction strength. Our results demonstrate that predicting foodweb stability requires simultaneously considering multiple foodweb properties—all of which naturally emerge from species masses.