Future tipping of key parts of the Earth system, driven by human-induced climate change, poses serious risks to society. These concerns have spurred the development of Early Warning Systems (EWSs) that use data to show how close we are to tipping. However, existing EWSs are often generic statistical forms. We first review current EWS methods, then present a broad set of hypotheses that may help improve them. We argue strongly for process knowledge to guide more refined EWS forms.

Human societies and the natural environment constantly shape one another, yet researchers usually study and model them separately. We reviewed 314 studies on global, two-way coupled social-environmental models and propose a typology to structure the field. While we find clear differences between the groups, models overall could be improved by better representing social-cultural dynamics, agent heterogeneity, diverse feedback loops, endogenous tipping dynamics, and cross-scale interactions.

A century ago, scientists suggested that life is more than just a passenger on Earth. It benefits, they argued, from influencing the conditions on Earth, co-shaping its own habitat. This idea is now widely accepted, but our review shows we still lack a comprehensive theory of the biosphere’s role in the Earth system. Science-based Earth stewardship calls for collaboration of fields such as evolutionary theory, ecological thermodynamics, ecosystems modelling, biogeochemistry and land use science.