Tropical forests have long been a major terrestrial carbon sink. That function is weakening in both the Amazon and Congo basins, but the two systems are not at the same stage and do not face the same kind of tipping risk.
Amazon: sink weakening, parts already a source
Intact Amazon forest plots show the aboveground carbon sink peaked in the 1990s and has been declining since. Statistical extrapolations put the intact-forest sink near zero around the mid-2030s, with a very wide uncertainty range.
That picture is worse once degradation, fire, and drought are included. Aircraft and satellite work show southeastern Amazonia has already flipped to a net carbon source, driven by deforestation, fire, and a hotter, drier climate. The biome as a whole now oscillates: a narrow sink in quieter years, a source in extreme drought/fire years (2015–16, 2023, 2024). Carbon residence time in trees has also shortened (from ~70 years toward ~60), partly because storms are killing trees faster.
Deforestation of the original forest is around 17%. Recent modeling finds that if loss reaches ~22%, large parts of the basin become vulnerable to dieback even near 1.5°C of warming. Combined with current warming (1.4°C), that threshold could be approached in the 2030s if deforestation rebounds. A 2022–2026 literature synthesis gives medium confidence in a tipping risk for the drier south and east (40% of the forest), and low confidence in a basin-wide collapse this century.
Other recent work stresses that there is no single basin-wide switch. The southeast is most at risk; vast wetter areas remain more resilient if deforestation and fire are kept down. Direct human pressure (clearing, logging, fire) is still the dominant near-term threat, not a distant climate-only feedback.
Congo: later saturation, more a land-use problem
Intact African tropical forests held a stable sink into the 2000s and only later showed rising mortality—asynchronous saturation relative to the Amazon. Plot networks still project a declining but positive intact-forest sink through the 2030s.
Continent-scale biomass estimates are gloomier: some satellite analyses find African forests (Congo plus West Africa and Madagascar) net losing biomass after ~2010–2017, equivalent to 200 Mt CO₂ yr⁻¹, mainly from DRC deforestation and degradation. Other basin-focused studies still count the Congo as a weak net sink and the largest remaining tropical forest sink (0.3–0.6 Gt CO₂ yr⁻¹ depending on method and whether peatlands are included). The discrepancy is mostly about how much unmanaged clearing offsets growth in protected areas and concessions.
Climate-driven “savannization” is considered less likely than in the Amazon. Models generally project wetting over much of the Congo, and the forest–rainfall feedback is weaker or differently configured. The binding risks are logging, mining, slash-and-burn agriculture, and peatland disturbance. Peat stores a large, partly ancient carbon pool; drying or drainage could release it. Policy assessments in 2025 described a roughly decade-scale window to keep the sink from unraveling.
What “tipping point” means here
Amazon: real risk of self-reinforcing drying and dieback in the south/east if warming and deforestation continue together. Not a guaranteed basin-wide collapse by 2030, but the safe operating space is shrinking.
Congo: more a gradual sink-to-source transition from land use than a climate tipping element. Local bistability exists; a Congo-wide climate tip is not the central scenario.
Both basins still store enormous carbon (Amazon on the order of 150–200 Pg C; Congo including peat also tens of Pg C). Losing sink function plus part of the stock would add tenths of a degree of extra warming and remove a buffer that climate models still partly assume.
The practical implication is the same in both regions: stop the “hammer blows” of clearing and fire now. Intact and well-managed forest still sequesters carbon; degraded and edge forest does not. Protection and reduced deforestation change the odds more than waiting for a single dramatic threshold.