Rivers are eating less methane than we hoped

On 17 July 2026, oceanographer Alberto Borges of the University of Liège published a comparative study in Limnology & Oceanography with a finding that cuts against the optimistic side of the methane ledger. Microbes living in rivers do consume methane, the potent greenhouse gas, but at nowhere near the rate required to offset rising emissions from the rivers themselves. The work, conducted in Belgium and in the Congo Basin, reframes a process once treated as a quiet natural subsidy as a partial, leaky filter.

The question matters because rivers carry roughly half a gigatonne of carbon to the oceans every year, much of it as methane produced in anoxic sediments and floodplains. If a meaningful share of that methane were re-oxidised before reaching the atmosphere, the global budget would soften. Borges's data suggests it does not soften nearly enough.

The size of the leak

River systems are not passive pipes. Methanotrophic bacteria, a class of microbes that consume methane as their primary fuel, colonise oxygenated stretches of streambed and metabolise the gas as it rises from sediments. The intuition has long been that this biological filter mops up a substantial fraction of what the river produces.

Borges and his team measured methane oxidation rates across Belgian catchments and contrasted them with measurements taken in large African tributaries. The Belgian rivers showed moderate consumption. The African rivers, draining tropical wetlands with far higher methane production, did oxidise larger absolute volumes, but the fraction consumed fell well short of what global models typically assume. The gap between modelled and measured sinks is the new headline.

"Borges found that microbial oxidation in rivers is real but insufficient," according to a 17 July 2026 write-up of the work. The finding implies that climate models which credit rivers with a large methane sink are likely overstating the offset, leaving atmospheric budgets tighter than assumed.

Why the tropics matter most

The geographic split in the dataset is doing analytical work. Belgian rivers are temperate, well-oxygenated, and small. The Congo Basin catchments are equatorial, warm, and carry orders of magnitude more dissolved carbon. Methane production rises with temperature and with the volume of organic matter delivered to floodplains during seasonal inundation. Oxidation rises too, but more slowly. The ratio tilts the wrong way as the climate warms.

This is where the structural worry sits. Tropical rivers, especially the Congo, the Amazon headwaters, and the major South Asian systems, are precisely the places where warming-driven methane production is accelerating fastest. If their microbial filters were robust, the budget would tolerate that acceleration. The Borges study suggests the filters are partial, and that the partiality is most damaging where the source term is largest.

The implication is uncomfortable for the standard model architecture used by the IPCC and by major atmospheric reanalyses. Methane budgets tend to be closed with a residual terrestrial sink. If rivers are not delivering as much of that sink as the residual assumes, the residual has to come from somewhere else, or the atmospheric growth rate has to be revised upward.

A counterpoint from the modelling side

Not every reading of the literature agrees. Some process-based models, calibrated against eddy covariance towers and atmospheric inversions, continue to credit rivers and other inland waters with a methane sink in the range of 100 to 150 megatonnes per year. Those numbers are not trivial. They sit inside a global methane budget of roughly 600 megatonnes of annual emissions and help explain why atmospheric concentrations have not risen as fast as bottom-up inventories of sources would predict.

The Borges study does not so much contradict that work as narrow the mechanism. It is consistent with the view that a real river sink exists, and it points out that the sink is not primarily the one modellers have been crediting. The microbial oxidation Borges measures is a local, biological process. The aggregate atmospheric residual is a global, unexplained gap. Linking the two requires more cross-site work, and Borges is careful to frame his contribution as one piece of that puzzle.

What it changes for policy

Methane has driven roughly 0.5 degrees Celsius of warming to date and is the second-largest contributor to current anthropogenic forcing after carbon dioxide. The Global Methane Pledge, signed at COP26 in Glasgow and now covering more than 150 countries, targets a 30 percent reduction in methane emissions by 2030 from 2020 levels. The pledge leans heavily on source-side action: plugging leaks in oil and gas systems, capturing gas from coal mines, modifying livestock diets, and limiting landfill emissions.

The Borges finding does not weaken that agenda. It reinforces it. If rivers are doing less to clean up after methane sources than the budget assumes, then reducing the sources themselves becomes more important, not less. The flip side is that natural-system offsets, the kind that have grown fashionable in voluntary carbon markets and in nature-based solutions rhetoric, look thinner when measured at the microbial scale.

There is also a measurement-programme consequence. The Congo Basin has historically been undersampled relative to the Amazon, both because of logistical difficulty and because of weaker national research infrastructure in the region. The Borges comparison, pairing Belgian measurements with African ones, is an argument for denser tropical monitoring. Without it, global budgets will continue to be built on temperate-heavy data, extrapolated across biomes they do not represent.

What remains uncertain

The study covers two regions, not the global river network. Borges's comparative design is well suited to scaling up, but the paper does not, on its own, rewrite the global budget. The Mekong, the Ganges-Brahmaputra, the Paraná, and the Ob-Irtysh system, each a major methane source, are not in this dataset. Generalising requires additional field campaigns.

There is also an open question about how microbial oxidation will respond to continued warming. Warmer water holds less dissolved oxygen, which tends to suppress methanotrophs even as methane production rises. The Borges study measures the present ratio; the future ratio is a different question, and one the data do not yet answer.

What the paper does establish, firmly, is that the optimistic framing of rivers as a self-cleaning methane system is too generous. The rivers do eat some of the gas. They do not eat enough. The gap between what they eat and what they emit is one of the more uncomfortable margins in the contemporary carbon cycle.

This article was framed by Monexus as a measurement and budget story rather than a climate-panic story. The wire coverage treated Borges's finding as a corrective to overoptimistic river-sink assumptions; Monexus carries that framing forward and adds the policy and tropical-monitoring context the press release does not.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://en.wikipedia.org/wiki/Methane_emissions
• https://en.wikipedia.org/wiki/Global_Methane_Pledge
• https://en.wikipedia.org/wiki/Methanotroph