Terrestrial ecosystems take up approximately a third of anthropogenically emitted carbon and are a key component of climate mitigation strategies. However, recent evidence indicates constraints on land-based carbon uptake and mitigation potential.
Whether terrestrial systems will continue to absorb anthropogenic CO2 — and at what rate — has been an active research question, important for mitigation plans and carbon budget estimations, as well as for understanding potential carbon cycle feedback under mitigation scenarios that lower atmospheric CO2. Research has pointed to constraints in continued growth of the land sink, for example, those set by the availability of nutrients essential to growth, such as nitrogen and phosphorus, as well as decreasing benefits to biomass from elevated CO2 and increasing intensity of disturbances. Models generally indicate a decreasing sink over time with future climate change.
Credit: Michele D’Ottavio
However, rather than a gradual decline, 2023 saw an unexpected collapse of natural carbon sinks, exemplified by very high CO2 growth rate. In this year, the terrestrial biosphere took up almost no net carbon, due in part to fire emissions from Canadian wildfires and drought in the Amazon1. The impacts were not limited to land; Jens Müller and colleagues show in an Article in this issue of Nature Climate Change that the non-polar marine carbon sink also decreased in 2023, mainly on account of high surface temperatures that subsequently impacted CO2 solubility.
The year 2023 was remarkable, as El Niño conditions brought warmer temperatures and drier conditions to parts of the world, which negatively impacted biomass and facilitated the heightened emissions from wildfires. However, the weakening of El Niño in 2024 has not brought the expected relief that would allow the biosphere to recover. A global increase in hot, wet conditions in 2024 also led to large terrestrial carbon losses, and 2025 has seen sustained hot conditions as well; for example, June 2025 was the third warmest June globally, just behind 2023 and 20242.
While the carbon cycle dynamics for a few years do not necessarily indicate a long-term trend or breakdown, the abrupt collapse of the land sink was not expected. These developments in the response of the land carbon sink, as well as the failure of models to predict them, are thus concerning and have implications for current mitigation strategies that rely heavily on the land sector for many countries. Notwithstanding the competition with agriculture and other land uses, which reduce the land available for reforestation efforts and their potential carbon impact, high temperatures, wildfire risk and drought all erode the effectivity of restoration efforts meant to draw down further CO2 (ref. 3). The impacts are not only in the future; for example, the carbon sink of European forests has declined more than expected, putting removal targets for 2030 for the transition to net neutrality at risk4.
Carbon uptake by the terrestrial biosphere is a balance of many different processes spanning ecosystems and scales, which are difficult to fully account for in models. These processes include feedback that lowers sink capacity but also include processes that might contribute to a sustained sink. In particular, nutrient availability has been cited as a potential constraint on vegetation biomass that could strongly impact the direction of the future land sink. Yet, plants have strategies for alleviating nutrient limitation that are not currently well integrated into models. Writing in a Review Article in this issue, Trevor Cambron and colleagues discuss these strategies and their incorporation into Earth system models. Preliminary estimates indicate that better accounting for nutrient acquisition could increase estimates of land carbon uptake.
In addition to natural feedback, deforestation is also still a major contributor to declines in the land carbon sink, despite recent initiatives such as the Glasgow Declaration and indications of falling deforestation rates in many countries5. In addition to the impacts on the carbon sink, deforestation also has consequences for local warming and heat-related mortality, as discussed by Reddington and colleagues in another Article in this issue. While forest regrowth offsets some of the emissions, curbing deforestation and conserving existing forests are critical, as deforestation and degradation undermine the resilience of remaining forest cover and carbon stores.
These unexpected developments in the ability of the natural world to mitigate some of the impacts of climate change, evidenced by the unprecedented failure of the land sink in 2023, highlight the importance of observations of carbon dynamics from long-term records such as Mauna Loa, as well as satellite infrastructure and flux tower measurements. These data are urgently needed to monitor carbon cycle responses and benchmark mitigation efforts in an era where models increasingly struggle to predict and simulate events and carbon impacts.