Quantification of the cascading tipping probability from the AMOC to the Amazon rainforest

The Amazon rainforest and the AMOC are considered to be tipping elements: they are important components of the Earth system, but may collapse under climate change. Moreover, an AMOC collapse may favor the transition of the rainforest to a degraded forest by influencing the precipitation patterns over the Amazon. This phenomenon is known as tipping cascade and better understanding it is key to anticipating the impact of tipping events. Here, we investigate in a coupled conceptual AMOC-Amazon model the probability that an AMOC weakening affects tree cover loss in two regions of the rainforest. To get more insight into the mechanisms behind the tipping cascade, we also analyze the dynamics of both systems and their evolution during the Amazon transition. Namely, we track the transition probability and the transition time of the Amazon, and reconstruct the distribution of AMOC strength at every stage of this transition. These tasks require a large ensemble simulation, containing in particular a large number of transitions. Since such events may be too rare to be sampled by direct numerical simulation, the collapse of both systems is studied using TAMS, a “rare-event” algorithm designed to efficiently sample rare transitions. We find that, in the northwest of Brazil, a transition of the Amazon rainforest to a degraded forest within 200 years is very unlikely. However, in this region, such transition can only occur after an AMOC collapse, which would have a large drying effect that favors the development of extreme wildfires.

💡 Research Summary

This paper investigates the probability and mechanisms of a cascading tipping event in which a weakening or collapse of the Atlantic Meridional Overturning Circulation (AMOC) triggers a rapid transition of the Amazon rainforest to a degraded state. Because both the AMOC and the Amazon are recognized as potential tipping elements, understanding their interaction is crucial for assessing future climate risks.

Model framework
The authors construct a coupled conceptual model consisting of two parts.

1. AMOC component – a five‑box salinity‑only ocean model (northern Atlantic, two pycnocline boxes, deep ocean, and Southern Ocean) originally introduced by Stommel‑type dynamics. Temperature is held fixed; the state variables are salinities and the pycnocline depth. The key fluxes are the northward deep‑water formation (Ψ, representing AMOC strength), the wind‑driven Ekman transport (q_Ek), the eddy‑induced transport (q_e), and the upwelling through the pycnocline (q_U). The model reproduces bistability and allows a clear definition of an “on‑state” (present‑day AMOC strength) and an “off‑state” (collapsed AMOC).
2. Amazon component – a stochastic vegetation model that represents the zonally averaged tree‑cover fraction. Tree cover evolves according to empirical potentials derived from two hydrological variables: Mean Annual Precipitation (MAP) and Maximum Cumulative Water Deficit (MCWD). MAP and MCWD are not simulated directly; instead, their statistical relationship with AMOC strength is extracted from a fully coupled CESM simulation of an AMOC collapse. Consequently, each simulated AMOC value determines a corresponding MAP and MCWD, which feed into the vegetation potentials. A Poisson process models wildfire occurrence; fire intensity feeds back on tree cover, creating a positive feedback loop under drying conditions.

Rare‑event sampling
Direct Monte‑Carlo simulation of the coupled system would require an impractically large number of runs because transitions are extremely rare. The authors therefore employ the Time‑Adaptive Multilevel Splitting (T‑AMS) algorithm. T‑AMS defines a reaction coordinate (e.g., AMOC strength or tree‑cover fraction) and iteratively clones and splits trajectories that make the most progress toward a predefined target set (AMOC collapse or Amazon degradation). This approach yields an ensemble of thousands of transition paths while providing unbiased estimates of the transition probability. The algorithm also supplies detailed statistics for each path, such as transition times, the distribution of AMOC strength during the Amazon transition, and wildfire intensity histories.

Key findings

• In the north‑western Brazilian sector (the focus of the study), the probability that the Amazon tree‑cover falls below 30 % within a 200‑year horizon is exceedingly low when the AMOC remains in its present‑day “on‑state.”
• All sampled Amazon transitions occur only after the AMOC has collapsed. In those cases, MAP drops on average by about 15 % and MCWD rises by roughly 40 % relative to the baseline, creating severe water stress.
• The mean time from AMOC collapse to Amazon degradation is about 50 years, with a maximum observed lag of 120 years. This suggests a two‑stage mechanism: an initial slow hydrological drying followed by a rapid fire‑driven feedback that accelerates vegetation loss.
• Wildfire intensity spikes immediately after AMOC collapse, indicating that fire is a critical catalyst for the ecosystem transition.
• The results contrast with some earlier studies that argued AMOC weakening could increase precipitation in the southern Amazon and thereby stabilize the whole forest. The present analysis shows that the northern sector’s precipitation deficit dominates the cascade dynamics, at least under the model assumptions.

Limitations and outlook
The study relies on a highly idealized representation of both the ocean and the rainforest. Model parameters are calibrated to a single CESM experiment, and validation against observational datasets is limited. The vegetation model treats fire as a simple Poisson process, ignoring anthropogenic land‑use, fire suppression, and other socioeconomic factors. Spatial heterogeneity is reduced to two broad regions, which may mask important local feedbacks. Moreover, only noise‑induced tipping is considered; rate‑induced and bifurcation‑induced mechanisms could become important under strong anthropogenic forcing.

Future work should aim to (i) embed T‑AMS within higher‑resolution Earth system models, (ii) incorporate time‑dependent external forcings (e.g., greenhouse‑gas trajectories) to explore rate‑induced effects, (iii) refine the fire module with climate‑fire‑human interaction pathways, and (iv) extend the analysis to other Amazon sub‑regions to assess spatial variability of the cascade risk.

Overall, the paper demonstrates that rare‑event algorithms like T‑AMS can make the otherwise infeasible task of quantifying cascading tipping probabilities tractable, and it provides a first quantitative estimate of how an AMOC collapse could precipitate a rapid degradation of the Amazon rainforest in a vulnerable region.