Rivers Carry a Quarter of Land’s Carbon to the Sea, but the Numbers Remain Shrouded in Uncertainty

Every year, an enormous conveyor belt of carbon moves silently across the surface of the Earth. Rain falling on forests, grasslands, and mountains dissolves carbon dioxide, leaches organic matter from soils, and grinds down rocks, sending a continuous stream of dissolved and particulate carbon into the world’s rivers. From there, a portion of this carbon is delivered to the ocean, a portion is buried in sediments, and a surprisingly large fraction is vented straight back into the atmosphere. A new review published in Nature Reviews Earth & Environment by Robert Spencer of Florida State University, Travis Drake of ETH Zurich, Maodian Liu of Peking University, and Peter Raymond of Yale University synthesizes decades of work on this riverine land-to-ocean carbon flux, and its central message is both reassuring and unsettling: the broad outlines of the global flux are now reasonably well constrained, yet the details remain stubbornly uncertain, with model estimates for individual carbon species spanning roughly threefold differences and basin-scale export estimates diverging by as much as a factor of five.

The scale of the terrestrial handoff to inland waters is staggering. The review estimates that land landscapes deliver between 3 and 4 petagrams of carbon per year to rivers, lakes, and streams, a figure comparable to the annual carbon uptake by terrestrial ecosystems or the fossil fuel emissions of entire continents. Yet only a small slice of that carbon ever reaches the sea. The majority is outgassed from inland water surfaces as carbon dioxide and methane before it can complete the journey. This means that rivers are not merely pipes connecting land and ocean; they are reactive biogeochemical reactors, hotspots of microbial metabolism, photochemical degradation, and gas exchange that fundamentally reshape the carbon they carry. Recognizing inland waters as active processors rather than passive conduits has been one of the major conceptual shifts in carbon cycle science over the past two decades, and the new review places that shift at the center of its framework.

When it comes to the flux that actually reaches the ocean, the models converge on a total global export of roughly 0.8 to 1.0 petagrams of carbon per year. That convergence at the global scale, however, masks deep disagreements beneath the surface. The review finds that estimates for individual carbon species, such as dissolved organic carbon, dissolved inorganic carbon, particulate organic carbon, and dissolved CO2, span approximately threefold ranges across the models that were synthesized. At the scale of individual ocean basins, the picture is even murkier: estimated export of some carbon species to the Atlantic and Pacific varies by up to fivefold between models. These discrepancies arise from differences in how models represent source inputs from soils and vegetation, transport processes within river networks, biogeochemical transformations along the aquatic continuum, and the climate and hydrological drivers that govern everything from flood pulses to drought-induced contraction of river networks.

One of the most striking findings in the synthesis concerns the geography of ignorance. The Amazon and Congo rivers together contribute approximately 21 percent of global river discharge, yet these tropical giants remain far less observed than many smaller temperate watersheds in North America and Europe. The world’s 30 largest rivers by discharge account for roughly half of global cumulative river discharge, meaning that a modest number of well-instrumented stations on major rivers could dramatically improve global flux estimates. But the tropical basins that dominate the discharge statistics are precisely where long-term monitoring is weakest. Dense tropical forests, vast floodplains, political instability, and logistical challenges have all conspired to leave the planet’s most important carbon arteries under-sampled, even as small, convenient catchments in wealthy temperate countries accumulate decades of high-frequency data.

The review traces the journey of carbon through the aquatic continuum in careful technical detail. Carbon enters rivers through multiple pathways: dissolved organic carbon leached from litter and soil horizons, dissolved inorganic carbon derived from the respiration of soil microbes and the chemical weathering of carbonate and silicate rocks, and particulate organic carbon stripped from landscapes by erosion. Hydrology acts as the master control. Storm events trigger what researchers call pulse-shunt dynamics, in which floods flush accumulated organic matter from soils into streams and rapidly shunt it downstream before it can be processed. The flood pulse concept, originally developed for river-floodplain systems, captures how seasonal inundation of floodplains connects stored carbon to the river channel. Temperature, meanwhile, controls the production of dissolved organic carbon in soils, while hydrology regulates its export, a decoupling that complicates predictions under a warming climate.

Once in the river network, carbon faces a gauntlet of transformations. Sunlight, particularly ultraviolet radiation, photochemically degrades dissolved organic matter, breaking complex molecules into simpler substrates that microbes can rapidly consume, and in some cases converting organic carbon directly to carbon dioxide. Microbial respiration does the same work in the dark. Gas exchange at the air-water interface, governed by turbulence, slope, and flow regime, determines how much of the resulting CO2 escapes to the atmosphere; studies have shown that small, high-energy streams exchange gases far more rapidly than sluggish lowland rivers. Dams interrupt the journey, trapping sediment and organic carbon in reservoirs, where burial can be efficient but where greenhouse gas emissions from reservoir surfaces can also be substantial. Floodplains act as transient storage zones where deposited organic carbon may be mineralized, buried, or remobilized during subsequent floods.

The fate of riverine carbon in the ocean is itself a scientific puzzle. For decades, oceanographers have debated what happens to terrestrial organic matter once it enters the sea. Radiocarbon measurements of dissolved organic carbon in the Atlantic and Pacific have revealed that terrigenous material is far less abundant in the deep ocean than riverine fluxes alone would predict, implying efficient removal near the coast. Recent work suggests that oxidation can camouflage terrestrial organic matter to appear marine-like, and that photochemical processes in coastal waters accelerate its degradation. Yet other studies indicate that terrestrial dissolved organic carbon inputs to the ocean may have been underestimated. Particulate organic carbon, by contrast, can be efficiently buried in marine sediments, particularly in systems like the Bengal fan, where the Himalayan erosional system delivers and buries organic carbon at remarkable rates. The balance between these fates, outgassing, burial, and oceanic export, determines whether rivers represent a net carbon sink or a net source to the atmosphere.

Human activities are actively rewiring this flux. Anthropogenic perturbations have enhanced water and carbon fluxes from land to ocean through land-use change, agricultural liming, and hydrological alteration. Dissolved inorganic carbon and alkalinity exports from large rivers have been increasing, partly in response to acidification recovery and agricultural practices. Domestic wastewater has been identified as an overlooked source of riverine dissolved carbon. Fires prime terrestrial organic carbon for riverine export. In the Arctic, thawing permafrost is mobilizing old, previously frozen carbon into rivers, and recent work has shown that old carbon is being routed from land to the atmosphere by global river systems, a feedback with potentially significant climate implications. Rising dissolved organic carbon concentrations in many northern rivers signal widespread environmental change, while dam construction and eutrophication are shifting the sources and fates of carbon in river-connected coastal wetlands.

The path forward, the authors argue, requires a concerted expansion of observations in precisely the places where data are scarcest. Sensor networks capable of high-frequency, in situ measurements of dissolved organic matter, carbon dioxide, and water chemistry are already transforming temperate catchment science, and deploying them across tropical and Arctic basins would close critical gaps. Remote sensing offers complementary reach: satellites can now retrieve dissolved organic carbon concentrations, particulate organic carbon fluxes, and even the optical color of rivers across the globe, and machine learning approaches trained on these datasets are beginning to fill spatial and temporal holes in the observational record. Integrating these large datasets with mechanistic studies of the biogeochemical processes that modulate carbon export, and embedding both within process-based and knowledge-guided machine learning models, would sharpen predictions of this globally important flux. As the climate warms and human pressures on rivers intensify, the review makes clear that understanding how much carbon rivers deliver to the ocean, and how much they release to the sky, is no longer a niche question for aquatic chemists. It is a first-order problem for the global carbon budget, and one whose answer depends on finally giving the world’s great rivers the scientific attention their size demands.

Subject of Research: The global riverine flux of carbon from land to ocean and its uncertainties

Article Title: The riverine land-to-ocean carbon flux

Article References: Spencer, R. G. M., Drake, T. W., Liu, M., & Raymond, P. A. (2026). The riverine land-to-ocean carbon flux. Nature Reviews Earth & Environment. https://doi.org/10.1038/s43017-026-00834-7

Image Credits: AI Generated

DOI: 10.1038/s43017-026-00834-7

Keywords: carbon cycle, rivers, land-to-ocean flux, dissolved organic carbon, dissolved inorganic carbon, CO2 outgassing, Amazon River, Congo River, permafrost, remote sensing, machine learning, global carbon budget