The uncertain impacts of an 1.5 °C overshoot-and-decline scenario on carbon stocks and productivity in terrestrial ecosystems

Overshoot scenarios, in which global temperatures peak and then decline through emission reductions and CO 2 removal, are increasingly central to climate policy as we are already approaching 1.5 °C warming. Yet the ecological reversibility of such pathways remains poorly understood, even though the trajectories of productivity and carbon stocks in terrestrial ecosystems have major implications for land-based carbon sink projections. Here we draw on four complementary lines of research to study the land carbon sink under overshoot: (i) CMIP6 models project declines in net ecosystem productivity (NEP) after peak warming. Nevertheless, NEP remains positive, leading to a stabilisation of the carbon stocks in terrestrial vegetation (cVeg). The CO 2 ramp up-down model experiment by CDRMIP underscores that these models predict near-reversibility of the vegetation carbon stock. (ii) A systematic review on global change experiments identified 256 papers (warming, drought, irrigation or elevated CO 2 , all elements of an overshoot scenario). However, only 15 included a recovery phase following treatment. This reveals a gap in experimental design for understanding ecosystem responses under overshoot conditions. Where recovery was assessed, responses were mixed, with persistent legacy effects documented after drought. (iii) Modern-day overshoot analogues, including two El Niño events and the 2018–2020 European droughts, show that compound heat and drought can cause carbon losses that recover slowly, with repeated extremes progressively reducing ecosystem resilience. (iv) Paleoclimate records from the Holocene Climatic Optimum and the Paleocene–Eocene Thermal Maximum demonstrate that exceeding ecological thresholds can trigger vegetation changes and carbon cycle feedbacks that long outlast the climatic perturbation itself. Together, these findings show that the trajectory of the terrestrial carbon sink during climate overshoot is uncertain, because ecosystem responses are often delayed, nonlinear, and threshold dependent.