Earth’s oceans are probably hotter than at any point in at least 1,000 years and are warming faster than they have for millennia. A 2026 review adds another warning: dissolved oxygen is rapidly declining across oceans, lakes and rivers, and although no formal planetary boundary has yet been defined, researchers argue that current losses may already exceed what can reasonably be considered safe.

Water does not need to boil, turn acidic or fill with plastic before it becomes hostile to life. It can simply become harder to breathe.

That is happening across parts of the ocean and in lakes, rivers, estuaries and streams. A 2026 review in Limnology and Oceanography argues that the decline of dissolved oxygen now belongs in the conversation about Earth’s planetary boundaries: the large processes that help keep the planet within a relatively stable operating range.

The paper does not define a tenth boundary, and it does not demonstrate that a formally agreed oxygen threshold has been crossed. No such global threshold exists. Instead, Erica Ferrer and eight colleagues assemble evidence that aquatic deoxygenation is interacting with all nine established boundary processes and moving toward what they call an “unsafe space.”

That distinction matters. “Unsafe” is the authors’ synthesis of risk, not a reading from a planetary oxygen gauge. Their case rests on the speed and breadth of observed losses, the long memory of oceans and lakes, and the limited ways oxygen can be restored once it is gone.

It arrives while the ocean is already carrying a record quantity of heat. The two trends are connected, but heat content and dissolved oxygen are not the same measurement. Keeping them separate at first makes their eventual collision easier to understand.

What it means to say the oceans are hotter

There are at least two commonly discussed ocean temperatures. Sea-surface temperature describes a thin upper layer and can swing with El Niño, La Niña, winds and other short-term patterns. Ocean heat content estimates the energy stored through a much deeper volume, commonly the upper 2,000 metres. It is the steadier measure of the heat accumulating in the climate system.

Modern instruments do not provide a thousand-year global thermometer record. Before the mid-twentieth century, and especially before the widespread Argo float network, scientists have to reconstruct ocean conditions from corals, sediments and other indirect archives. That is why “probably” belongs in claims about a millennium of ocean heat.

The longer context is nevertheless stark. The Intergovernmental Panel on Climate Change concluded with medium confidence that the global ocean warmed faster over the past century than at any time since the end of the last deglacial transition, roughly 11,000 years ago. Proxy-based summaries have also placed present ocean temperatures above those of at least the past millennium.

Instrumental data remove any doubt about the recent direction. As ScienceBlog reported in its account of the 23 zettajoules added to the upper ocean in 2025, ocean heat content set another modern record even though the global annual average surface temperature was slightly lower than in 2024. A cooler-looking surface year did not mean the deeper ocean had stopped gaining energy.

The ocean takes up more than 90 percent of the excess heat trapped in the climate system. That buffering spares the atmosphere from even faster warming, but it does not make the heat disappear. It changes circulation, raises sea level through thermal expansion, intensifies marine heat stress and contributes to oxygen loss.

Dissolved oxygen is not the oxygen in the air

Aquatic deoxygenation means a decline in oxygen molecules dissolved in water. Fish pass water over their gills to extract those molecules. Crabs, worms, zooplankton and many microbes depend on the same dissolved supply.

This does not mean the atmosphere is about to run out of breathable oxygen. Earth’s atmospheric oxygen reservoir is enormous, and most oxygen produced by photosynthesis is eventually consumed again by respiration and decomposition. The immediate risk described by the review is loss of breathable aquatic habitat and disruption of the chemistry performed by oxygenated waters.

Concentration is usually reported in milligrams of oxygen per litre or micromoles per kilogram. Researchers may also use percentage saturation, which compares the measured concentration with the maximum that water could hold at its temperature, salinity and pressure.

That adjustment matters. Five milligrams per litre does not carry exactly the same biological meaning in cold fresh water and warm salty water. Temperature changes both how much oxygen water can hold and how much oxygen an animal needs.

Why warmer water becomes harder to breathe

The first mechanism is basic physics: gases are less soluble in warmer water. Heat a glass of water and dissolved gas escapes more readily. On the scale of the upper ocean, the review says roughly half of contemporary oxygen loss above 1,000 metres can be attributed to temperature-driven changes in solubility.

The second mechanism is circulation. Warm surface water is lighter than the colder water beneath it. Stronger layering, known as stratification, makes vertical mixing more difficult. Oxygen enters primarily through contact with the atmosphere and photosynthesis near the sunlit surface, so weaker mixing can reduce the supply reaching deeper layers.

The third mechanism is biological demand. Warmer conditions often speed metabolism. Animals need more oxygen, and microbes consume it faster as they break down organic matter. Supply can fall just as demand rises.

Those mechanisms do not operate evenly. Winds, currents, freshwater inputs, salinity and the arrival of deep water can strengthen or oppose them. Some oxygen-minimum zones occur naturally and have persisted for geological timescales. As NOAA Ocean Exploration explains, these layers sit between better-oxygenated surface and deep water, and some jellyfish and microbes have evolved to live within them.

Climate-driven deoxygenation is the modern trend superimposed on that natural geography. It can expand or intensify low-oxygen zones, move their boundaries and expose ecosystems that did not evolve under persistent oxygen stress.

Nutrients create a second route to oxygen loss

In lakes, rivers and coastal waters, warming is often joined by excessive nitrogen and phosphorus from fertiliser, sewage and runoff. These nutrients stimulate the growth of algae and other primary producers. That growth can briefly increase oxygen in sunlit water during the day.

The bill arrives when the organic material dies, sinks or is eaten. Microbes use oxygen while decomposing it. If the water column is strongly layered or poorly flushed, oxygen can be consumed faster than it is replenished. Hypoxia develops; if oxygen falls to zero, the water becomes anoxic.

This is why warming and nutrient pollution multiply rather than merely add. Warmer water starts with a lower oxygen capacity, may mix less, and can accelerate respiration. A nutrient load that an ecosystem once tolerated may then produce a longer or wider low-oxygen event.

ScienceBlog’s earlier report that rivers were warming and losing oxygen found the same overlap in a different dataset: among nearly 800 rivers, 87 percent warmed and 70 percent lost oxygen. The 2026 review places such river trends within the larger ocean–freshwater system.

Dams, altered river flow, wetland loss, sediment disturbance and the destruction of seagrass or other oxygen-producing habitat can alter the balance again. “Deoxygenation” is therefore not a synonym for warming. Climate is a major driver, but land use and water quality can determine where the damage becomes acute.

How large is the decline?

The 2026 review draws together studies made at different scales and over different periods. It reports that the global ocean’s average dissolved-oxygen inventory fell by about 2 percent over the past 50 years. More than 200 coastal sites now experience persistent hypoxia, commonly defined as concentrations below 2 milligrams per litre.

Rivers in the underlying literature were estimated to be losing oxygen at roughly 1 to 1.5 percent per decade. Temperate lakes, depending on depth and category, have shown declines on the order of 1.5 to 5 percent per decade.

A separate 2026 review of inland freshwaters gives a more recent concentration-based view. Across the datasets it assessed, summer surface oxygen in lakes fell by an average of 0.034 milligrams per litre per decade from 2003 to 2023. River oxygen fell by 0.043 milligrams per litre per decade across 1980–2023.

Again, a global average is not a universal rule. That freshwater review found oxygen rising in 25.4 percent of rivers, particularly in western Africa and Southeast Asia. Asian lakes showed the fastest regional lake decline in its analysis, while the Amazon basin had the fastest river decline. Local pollution controls, flow changes, restoration and measurement coverage all matter.

The unevenness does not weaken the global concern. It explains why a single boundary number is so difficult. Oxygen changes hour by hour, season by season, from surface to bottom and between an upland stream, a eutrophic lake and a naturally oxygen-poor stretch of deep ocean.

Low oxygen is sometimes natural

The authors make a careful distinction between an ecosystem that is naturally low in oxygen and one that is losing oxygen because of modern human pressure. Some deep basins, caves, ponds and ocean layers have been hypoxic or anoxic for millennia. Their chemistry and inhabitants reflect that history.

A naturally anoxic basin is not evidence that oxygen does not matter. It is evidence that context matters. Organisms adapted to almost no oxygen can live where a tuna, crab or trout cannot. Moving the boundary of that habitat can still harm both communities.

The familiar threshold of 2 milligrams per litre is useful but imperfect. It marks oxygen stress for many animals and provides a consistent way to map hypoxia. Yet a large analysis cited by the review found lethal effects in many coastal bottom-dwelling species above that concentration and proposed 4.6 milligrams per litre as a more protective cutoff in those environments.

At the other extreme, animals that evolved inside oxygen-minimum zones can live far below 2 milligrams. A global line drawn at one concentration would protect some species poorly and label some naturally functioning ecosystems as damaged.

Why the authors want a planetary boundary

The planetary-boundary framework currently describes nine processes: climate change, ocean acidification, biosphere integrity, land-system change, freshwater change, biogeochemical flows, atmospheric aerosol loading, stratospheric ozone depletion and novel entities such as synthetic pollution.

Aquatic oxygen is affected by all of them and feeds back into several. Climate warming reduces oxygen supply. Excess nitrogen and phosphorus increase oxygen demand. Acidification and low oxygen often occur together because respiration consumes oxygen while releasing carbon dioxide.

When oxygen becomes scarce, microbes switch chemical pathways. Denitrification can remove biologically available nitrogen; phosphorus can be released from sediments and stimulate more production; nitrous oxide, a powerful greenhouse gas that also affects stratospheric ozone, can increase under some low-oxygen conditions.

Habitat also contracts in three dimensions. Large fish and predators with high metabolic demands may be pushed toward the surface, shore or cooler water. Food webs change. If midwater animals decline or move, the biological pump that carries carbon into the deep sea may change as well.

This systems argument builds on the concern covered in ScienceBlog’s 2024 article, “Loss of Oxygen in Lakes and Oceans a Major Threat to Ecosystems, Society, and Planet”. That earlier work proposed aquatic deoxygenation as a planetary process. The new review tries to show in detail how it connects to each existing boundary and what might eventually be measured.

No formal boundary has been calculated

A planetary boundary needs more than evidence of harm. It requires one or more “control variables”: measurements that can describe global movement from safer conditions into a zone of rising risk. Atmospheric carbon-dioxide concentration is one control variable for climate change. Aquatic oxygen has no agreed equivalent.

The authors discuss four candidates. The first is dissolved-oxygen concentration, perhaps expressed as the global area or volume of hypoxic and anoxic water. It is widely measured and clearly connected to biology and chemistry, but monitoring is geographically and vertically uneven.

The second is percentage oxygen saturation, which accounts for temperature, salinity and pressure. It better represents how breathable water is, but it requires more measurements and still runs into species-specific tolerances.

The third is the presence and abundance of indicator species. The disappearance of large oxygen-demanding predators, unusual surface breathing or a rise in low-oxygen specialists can reveal biologically meaningful change. The difficulty is choosing indicators that work across ecosystems.

The fourth is the Metabolic Index, a temperature-dependent ratio between oxygen supply and an organism’s demand. It ties water chemistry directly to the capacity for aerobic life. It is promising but data-hungry, and scientists lack metabolic measurements for most species.

None is ready to serve as a universally accepted planetary gauge. The review explicitly says that a precise quantitative boundary does not exist and that defining one will require more observations, physiological data, palaeo-records and modelling. Its statement that present losses may exceed what most people would consider safe is a reasoned warning, not a formally demonstrated boundary crossing.

Why oxygen loss can outlast the emissions that caused it

The ocean interior is ventilated slowly. Cold surface water takes up oxygen, sinks and moves through the global circulation. In parts of the deep ocean, replacement can take around a thousand years.

That slow conveyor gives the system memory. The review cites modelling indicating that ocean deoxygenation could continue for centuries even if carbon-dioxide emissions stopped, because committed warming persists and deep circulation cannot rapidly renew the lost oxygen.

Lakes can retain memory too, though their timescales and mechanics differ. Nutrients stored in sediments may continue feeding blooms after external inputs fall. Strong stratification can keep bottom water isolated. Restoration is possible in some systems, but a reduction in pollution does not always produce an immediate oxygen rebound.

Artificial aeration, bubbling and engineered downwelling can help particular lakes, reservoirs or coastal sites. They are not a plausible pump for the global ocean, and their effectiveness depends on scale, energy, circulation and the original cause of the oxygen deficit.

A serious warning that still has boundaries of its own

This is a review, not a new global oxygen dataset. It synthesises published observations and mechanisms, proposes indicators and makes an argument about Earth-system risk. The authors did not calculate how much oxygen every lake, river and ocean basin will lose next, and they did not identify one tipping point that applies everywhere.

Some trends remain uncertain. Observations are sparse in parts of the deep ocean, tropics and high latitudes. Models disagree about the future of some tropical oxygen-minimum zones. Species can acclimatise or adapt to a degree that simplified metabolic indices may not capture.

But uncertainty cuts both ways. The common 2-milligram threshold can miss damage to sensitive animals at higher concentrations. A 2-percent decline in the total ocean inventory sounds small while concealing much larger regional losses and the movement of critical habitat boundaries.

The useful message is not that a tenth boundary has suddenly appeared. It is that warming, fertiliser runoff, sewage, altered circulation, acidification and biodiversity loss converge on the same molecule. Oxygen is where several environmental pressures become a direct constraint on life.

Greenhouse-gas cuts address the global physical driver. Nutrient controls, wastewater treatment, wetland and seagrass restoration, and better management of river flow can improve local and regional oxygen conditions. Expanded sensors and repeated measurements are essential because a planetary average cannot tell a fishing community when bottom water in its bay is becoming uninhabitable.

The ocean’s rising heat content is already well measured. The oxygen response is patchier, more biologically variable and harder to compress into one number. That does not make it less important. It means the responsible conclusion is both serious and modest: the evidence points toward unsafe change, while the work of defining exactly where “unsafe” begins is not finished.