Loading the record…
A Geohistory of Life on Earth · sub-page
Biosphere Disruptors
Everything that has ever changed life on Earth at global scale sorts into two kinds. One kind arrives from outside the living world, does its damage quickly, and leaves a poorer planet locked in place for millions of years. The other arises from inside the living world, takes tens of millions of years, and in every case the record preserves has left a planet more capable of supporting life.
We came from the inside. We are behaving like the outside.
episodes · published measurements · sources. Every point on every chart traces to one of them.
One
Two kinds of disruptor
The difference between them is not how hard they hit. It is how long they last, and where they come from.
A transient disruptor is geologically short-lived and almost always non-biological: an asteroid, a continent-sized volcanic outpouring, an ocean gateway opening as the plates move. It arrives from outside the biosphere and eventually dissipates out of the Earth system. What it leaves behind is the surprising part. The damaged biosphere does not simply recover; it settles into a state that is stable precisely because it is simplified, and stays there for millions of years. A depleted world is not necessarily a fragile one, and that is the problem.
A persistent disruptor is long-lived, and in this record it is alive. Oxygen-producing photosynthesis. The first complex cells. The first animals to burrow into the seafloor. Plants walking onto land. Flowering plants, and everything they made possible. It arises from inside the biosphere and never leaves. Several of them poisoned the world that produced them before improving it — to the microbial biosphere that released it, oxygen was lethal. Every one of them, in the end, raised the planet's capacity to support life. That is what the record shows; it is not a guarantee. A runaway climate would be persistent too, and would do the opposite.
The panels below are a schematic — the shapes of the argument, not measurements. Everything after this act is data.
Two
How long they lasted
Here is the same distinction, measured. Every episode in the record, youngest at the top, running down into deep time. Each bar spans the shortest and longest duration anyone has published for that episode; the dot is the best estimate. For the five human rows the ends are the authors' stated assumptions rather than published measurements of an episode, since ours has not finished happening.
The gap is the point. Transient episodes cluster between ten thousand and a million years. Persistent ones sit a hundred to a thousand times further right — a mean of across of them, against across transient ones. (Five of the six green rows drawn here, not six — the notes explain which is left out of the statistics, and why.) And the human rows sit far to the left of everything, in centuries. Only one thing in four billion years moved faster — and it is the row directly below.
One bar breaks the pattern. K-Pg, the asteroid that ended the age of dinosaurs, stretches across the entire plot, because its published duration runs from to . It is not clipped here. That bar is the honest picture of what we know, and it is what act five is about.
Read this chart as a table
Three
What they did
Now put duration on one axis and the size of the change on the other. Above the middle line, the biosphere gained; below it, the biosphere lost. Four things are counted: how many species there were, how many genera, how much living matter by weight, and how much new growth the planet produced each year.
The picture sorts itself. Green sits above the line and far to the right — big gains, taken slowly. Red sits below the line and in the middle — losses, over hundreds of thousands of years. The human points sit below the line and far to the left — losses, over centuries. That single arrangement is the paper's whole thesis, and you can read it before you read a single number.
The published version of this figure plots all measurements at once and is close to unreadable. It opens here as one point per episode. Expand it when you want to see what it is made of.
Read every measurement as a table
Four
The ranking
Divide each magnitude by its duration and you get a rate: how fast the biosphere changed. Rank every episode by its worst rate, fastest loss at the bottom, fastest gain at the top, and this is the order that comes out.
Humanity lands from the bottom of — the fastest driver of biosphere loss in Earth's history, between the asteroid that ended the Cretaceous just below us and, just above, a projection of the rest of this century if nothing changes course. Not the largest. The fastest but one.
Read down the list and the point starts to bite. The Great Dying — the Permian catastrophe that killed roughly nine species in ten, the worst thing life has ever survived — sits only fifth. Not because it was small. Because it took sixty thousand years, and a rate is a magnitude divided by a duration. Divide a catastrophe by long enough and it stops looking fast.
The two business-as-usual rows make the same point against each other. The run to 3100 destroys more than the run to 2100 — and ranks four places slower. Same world, same trend, longer window, lower place.
One number here is not what it seems. A rate of hundreds of thousands of per cent per million years is an extrapolation, not a forecast: it is what the present pace would come to if it ran for a million years, which it cannot. It is a unit for comparing speeds across deep time, nothing more.
Before you carry that ranking anywhere, read the next act. It is what the number rests on.
Read this ranking as a table
Five
What the ranking rests on
A rate is a magnitude divided by a duration. The magnitudes in this dataset are reasonably well constrained — they come from decades of counting fossils. The durations are not. They are the least certain quantities in the entire analysis, and the two most quotable numbers in the paper are the two that depend on them most.
The asteroid holds first place on a chosen duration of , drawn from a published range running from to . Humanity holds second place on a chosen , from a range of to .
So move them. The control below sweeps every episode across its own published range — nothing invented, nothing outside the literature. Watch the ranking re-sort as you drag.
It answers two different questions, and you can switch between them. Move every episode and something counter-intuitive happens at the far end: the asteroid's published range is far wider than ours, so it slows down more than we do, and humanity ends up in first place. Move humanity's duration alone, holding the geological record at the paper's own estimates, and the opposite happens — we fall back through the ranking toward the end-Permian, the worst extinction in the record. Neither is a trick. Both are what the published ranges allow.
None of this makes the ranking wrong. It makes it a claim with a stated assumption rather than a fact, and the assumption is now yours to inspect.
Six
The ones nobody can count
Act five showed what the ranking rests on. This is what it leaves out. The charts above hold only what can be measured, and the record describes more persistent disruptors that cannot be put on any axis. Two are older than any surviving percentage; two leave a record too diffuse to pin to a single duration. They are not smaller for being unmeasured. One of them is the enzyme that makes photosynthesis possible — which is to say, one of them is most of the carbon in your body.
Seven
Where that leaves us
The paper's title leads with the hopeful half: humanity could become the greatest driver of biosphere gain in Earth history. That claim deserves exactly the scepticism we just applied to the other one, and for the same reason.
It is a claim about rate, not amount. The Great Oxygenation Event — when photosynthesis first filled the air with oxygen — raised the planet's primary productivity by somewhere between 460 and 188,000 per cent — one source, one episode, two model calculations, and the width of that range is itself a warning about how well any of this is known. The stewardship scenario in this dataset imagines 5 per cent more species by the year 3100. Humanity wins that comparison only because 3,300 years is roughly a hundred and fifty thousand times shorter than 500 million. And 5 per cent more species than today is measured against a baseline we have already lowered — not against the world of ten thousand years ago.
Two things nonetheless remain true, and neither depends on any contested number. Every persistent disruptor this record identifies arose from inside the biosphere, and every one increased the planet's capacity to support life. The paper is careful to add that this is a pattern, not a law: a runaway climate state would also be persistent, and would lock in the opposite. And we are the only disruptor in the whole record that can know which of the two it is becoming. Oxygenic photosynthesis could not choose.
The record is also not the only evidence. Pathways toward the persistent class are already documented, at small scale: landscapes where sustained human presence has raised local biodiversity rather than lowered it, land and sea both spared and shared, and a shift in the law of the ocean away from freedom to take and toward a common responsibility for the Earth system. None of it is planetary yet. All of it is the shape a persistent disruptor would have, if that is what we turned out to be.
That is not consolation. The categories are descriptive — drawn from rock, not from prophecy — and nothing in the record says which one we will turn out to be. What the record does say is narrower and more useful: the biosphere has been remade from within before, repeatedly, and always slowly. Nothing here says it cannot be remade well. Nothing here says it can be done fast.
Wong Hearing, T. W., Williams, M., Zalasiewicz, J., et al. (2026). Humans could become the greatest driver of biosphere net gain in Earth history, but we are currently the second fastest driver of biosphere loss. bioRxiv preprint, doi: — not yet peer reviewed. Supplementary data CC BY 4.0.