The last issue signed off by promising a lake that turns itself over twice a year, because fresh water is heaviest at about four degrees above freezing rather than at the freezing point — so a temperate surface must cross that density peak twice, and each crossing lets the whole column mix to the floor.
That issue already exists. It is No. 65, fifty-seven issues back, and it is the same lake. The promise had been kept in advance and nobody noticed.
The other half was never kept, and it is the better half: the fish at the bottom spend the whole summer on one breath drawn last spring. This is that part.
Two rooms and a floor
By July a temperate lake is two rooms with a floor between them. Upstairs is warm, lit and in contact with the air. Downstairs is cold, dark and, for the season, sealed. No wind reaches it. No light reaches it either, so nothing down there is making oxygen. Whatever was dissolved down there when the lid closed in spring is the entire supply — and it is being spent, because everything that dies above sinks, and decomposition breathes.
The budget only runs one way. A lake’s deep water in September is living on air that dissolved into it in April, minus five months of funerals.
Which sets the problem for anything cold-adapted swimming in it. A lake trout needs two things at once: cold water, and water with oxygen in it. The cold is downstairs and the oxygen is upstairs. Its habitat is not a layer — it is the overlap of two layers, and an overlap can go to nothing while neither layer does anything dramatic.
Limnologists measure that overlap directly. TDO3: the coldest water in the lake still carrying three milligrams of oxygen per litre. The coldest water you can breathe in. When it climbs, the room is closing.
Nothing down there got hot
In 2021 a team led by Stephen Jane pooled 45,148 oxygen-and-temperature profiles from 393 temperate lakes, some series running back to 1941. Among the 369 lakes that ever fell below three milligrams per litre anywhere in the column, TDO3 rose 0.17 °C a decade, and 68 per cent were trending upward.
The obvious explanation is that warm water holds less gas: heat the lake, lose the oxygen. For the surface that is roughly right — those lakes’ surfaces warmed 0.39 °C a decade, and solubility accounts for about two-thirds of the oxygen they lost.
It is not what is happening at the bottom. The deep water is not warming. Its trend across the whole dataset is −0.01 °C per decade: flat, and flat for as long as anyone has measured. Cold water dissolves more gas, so on solubility alone it should have gained a little — the prediction is plus 0.01 milligrams per litre. It lost 0.23. Between 1980 and 2017 the deep water gave up 18.6 per cent of its dissolved oxygen while remaining exactly as cold as it had always been.
Nothing down there got hot. The room is suffocating at an unchanged temperature.
What changed is the lid. A warmer surface means a larger density difference across the thermocline, and a larger difference is harder for wind to break. The seal sets earlier, holds harder, lasts longer. The same fixed budget is asked to cover a longer season, and that is the entire mechanism.
This has an uncomfortable shape. Stratification is not a malfunction. It is the lake’s ordinary, healthy, self-maintaining summer structure — the very thing No. 65 was about. Nobody is breaking the lake. The lake is getting better at being a lake.
The account that runs the wrong way
One subset does the opposite, and it is the most telling group in the study. Eighty-seven lakes got warmer at the surface and gained surface oxygen — the murky, fertile ones, once summer temperatures pass roughly 24 °C. Algae: a large enough bloom makes oxygen faster than warming takes it away.
Those same lakes hold the worst deep water in the entire dataset. Median deep oxygen of 0.64 milligrams per litre, against 3.42 for everything else.
It is one process seen twice. The bloom makes oxygen at the top, then sinks and is decomposed at the bottom, and decomposing it consumes oxygen. The lake gains air in the room that had plenty, by the same act that strips it from the room that has none. Two accounts, one transaction, and the transfer runs the wrong way.
What this is not
Winterkill is a different seal — there the lid is ice and snow, the lake is shallow, and what fails is photosynthesis under the cover. Different room, different door.
No. 65 is the door. That issue is about a boundary a lake assembles every spring and dismantles every autumn on its own schedule, using nothing but its own temperature, and about the collapse being the maintenance rather than the failure. This is about what is shut in behind the boundary while it holds.
The maintenance still arrives. Turnover comes in the autumn and the deep water is recharged, exactly as it always was. The fish’s problem was never that the lake stops working. It is that the interval got longer, and a fish does not breathe on an annual schedule.
≈
One loop I’m watching
Next: a hurricane is a heat engine, and engineers have run the numbers on it the way they would run them on a turbine — an intake temperature, an exhaust temperature, and a hard theoretical ceiling on how strong the storm can possibly get. The number has a name, the ceiling is rising, and almost no storm ever reaches it. Next time.