Mass extinctions

Foundations

Not the asteroid

Everyone knows the asteroid. It is the best story in earth science: a rock the size of a city, a crater under the Gulf of Mexico, and the end of the dinosaurs. It killed about seventy-six percent of species.

It is the second worst thing that has happened to life on this planet, and it is not close to the first.

About 252 million years ago, at the end of the Permian, an estimated eighty-one percent of marine species disappeared. Trilobites, which had survived for nearly three hundred million years, ended here. So did most of the reef-building organisms, most of the dominant land vertebrates, and, unusually, a large fraction of insects, which almost nothing else in the record has managed to kill in bulk.

It is sometimes called the Great Dying, and if you want to know how close complex life has come to being erased, this is the number to remember.

Five events, and what counts as one

Species go extinct all the time. Over ninety-nine percent of all species that have ever existed are gone, and most of them went quietly, outcompeted or unlucky, at what palaeontologists call the background rate.

A mass extinction is different: a short interval in which the rate rises far above background, across many unrelated groups, worldwide, at once. Raup and Sepkoski identified five such intervals in 1982 by analysing the marine fossil record statistically, and the list has held up.

The percentages come from different methods and should be read as rankings rather than measurements.

The thing that keeps doing the killing

Here is the pattern that surprises people who came for the asteroid. Four of the five coincide with an enormous volcanic outpouring. One coincides with an impact.

And in every case the lava is not the killer. Flood basalts cover large areas but not the world. What does the damage is gas.

A large igneous province releases carbon dioxide and sulphur dioxide on a scale and at a rate that the systems in lesson 4.7 cannot absorb. Sulphur produces short, sharp cooling and acid rain. Carbon dioxide produces long warming. Warm water holds less oxygen, so the oceans become anoxic. Dissolved carbon dioxide acidifies seawater, which dissolves the shells of anything that builds them. And where the magma intrudes through the right rocks, it can bake out additional carbon: the Siberian Traps erupted through coal and evaporites, which appears to have made a bad situation much worse.

That combination, rapid carbon release followed by acidification, anoxia and heat, is the recurring signature. It is worth sitting with, because it is also a description of what is happening now, at a rate that is faster.

How fast the worst one was

Until recently “the end-Permian” was an interval of uncertain length. High-precision uranium-lead dating on volcanic ash beds in South China narrowed it dramatically: the main marine extinction happened within about sixty thousand years.

On the cosmic calendar that is a few seconds. In geological terms it is close to instantaneous, and it rules out slow explanations.

The aftermath is equally stark. Oxygen isotopes in conodont fossils show equatorial sea surface temperatures reaching about forty degrees Celsius in the Early Triassic. That is hot enough to exclude most animals from the tropics entirely, and there is a corresponding gap in the fossil record at low latitudes. Recovery to pre-extinction diversity took something like five to ten million years, which is a hundred times longer than the killing.

The asteroid, properly

The impact story deserves its fame because of how it was found.

In 1980 Walter Alvarez, a geologist, and his father Luis Alvarez, a physicist, were trying to work out how long a thin clay layer at the Cretaceous boundary in Italy took to deposit. They measured iridium, a metal rare in Earth’s crust because most of it sank into the core, and steadily delivered by micrometeorites. Instead of a small background level they found a spike, tens of times above normal. Then the same spike turned up in Denmark, and New Zealand, and everywhere else anyone looked.

The inference was an asteroid roughly ten kilometres across. There was no crater to point at, and the hypothesis was contested for a decade. The crater was identified around 1990, beneath the Yucatán Peninsula, buried under a kilometre of sediment and originally found by geophysicists prospecting for oil. About 180 kilometres across, and the right age.

And the volcanoes at the same time

Except that something else was happening at that exact moment, and it happened in India.

The Deccan Traps, whose step-like landscape covers much of Maharashtra and neighbouring states, erupted across the same boundary interval. Hundreds of thousands of cubic kilometres of basalt. The obvious question is whether the dinosaurs were killed by the rock from space or by the volcanoes, and it has been argued fiercely for forty years.

The current state of play is more interesting than either answer. Precise dating shows the Deccan was erupting before the impact and continued after it, and one analysis found a marked change in eruption style within tens of thousands of years of the impact, raising the possibility that the impact’s seismic energy affected the volcanism. Meanwhile, statistical analysis of climate proxies across the boundary indicates that the extinction itself tracks the impact rather than the volcanism, and climate modelling finds that impact winter produces conditions matching the observed pattern of who died and who survived, while volcanic warming does not.

The reasonable summary: the volcanoes stressed the world before and after, and the impact did the killing. Not everyone agrees, and this is a live argument in which Indian geology sits at the centre.

What extinctions actually do

The word suggests subtraction, and that is only half of it. Mass extinctions restructure.

Before the end-Permian, the seas belonged to brachiopods, crinoids and trilobites. After it, to molluscs, and eventually to fish and marine reptiles. Before the end-Cretaceous, mammals had existed for a hundred and fifty million years without getting large. Afterwards, within ten million years, there were mammals the size of rhinoceroses.

Nothing about mammals improved at the boundary. What changed was that the competition was gone. Being marginal is a survival strategy when the dominant become a target, which is the subject of lesson 5.8 and, indirectly, the reason you exist.

The sixth

Current extinction rates are far above background by any credible estimate, driven by habitat loss, hunting, invasive species and now climate change. Whether the present episode has reached the magnitude of the five is a separate question from whether it has reached their rate, and the honest answer is that it has not yet, but that at current rates it could within centuries.

That distinction matters. The five events took tens of thousands to millions of years. The current one is being measured in decades. If it does reach that magnitude, it will have got there faster than any of them.

Deeper

What the Raup and Sepkoski analysis did

Their approach was to compile stratigraphic ranges for marine families and genera and plot extinction rate per interval through the Phanerozoic. Five intervals stood out as statistical outliers above a declining background trend. Later work has refined the compilation, corrected for sampling and preservation biases, and questioned whether the five form a distinct class or the tail of a continuous distribution. Bambach’s reassessment argues that the events differ enough in cause and selectivity that treating them as a single phenomenon obscures more than it reveals.

Why sixty thousand years matters

Constraining the end-Permian to about 60,000 years does two things. It eliminates explanations operating on million-year timescales, including gradual sea-level change and slow climate drift. And it forces the carbon release to have been extremely rapid, which is what allows comparison with anthropogenic emissions: the isotopic excursion implies a release rate that, while enormous in total, may have been slower per year than present emissions, a comparison that is widely quoted and depends sensitively on the assumed duration.

Selectivity as evidence

Who dies tells you what killed them. At the end-Permian, heavily calcified organisms with poor respiratory control fared worse than active animals with efficient gas exchange, which points at acidification and hypercapnia rather than temperature alone. At the end-Cretaceous, the pattern favours small-bodied generalists, freshwater ecosystems and detritus-based food chains, and disfavours large-bodied animals dependent on primary production. That is the signature of an abrupt collapse in photosynthesis, which is what impact winter predicts.

The Deccan debate, and what would settle it

The two lines of evidence pull in different directions. Radiometric dating establishes that Deccan volcanism spanned the boundary, and one dataset resolves it into pulses with a major one beginning shortly before the boundary. Proxy records of temperature and carbon cycling show warming several hundred thousand years before the boundary, consistent with volcanic carbon release, and then no further environmental deterioration attributable to volcanism at the boundary itself. Resolving this requires eruption chronologies precise to tens of thousands of years alongside proxy records at comparable resolution, and both are at the edge of what current methods deliver.

Frontier

Are five the right number

Some analyses identify additional intervals of elevated extinction, including in the Capitanian before the end-Permian, and question whether the traditional five are qualitatively distinct from lesser events. The answer affects how the present episode should be compared.

Rate versus magnitude in the current episode

Estimating modern extinction rates requires knowing how many species exist, which is not known to within a factor of several, and detecting extinctions that may not be confirmed for decades. Reconciling those uncertainties with the fossil record’s coarser resolution is an unsolved methodological problem, and it is why credible scientists give different answers to whether a sixth mass extinction is under way.

Transcript and shot list

Target runtime about 8 minutes 30 seconds. Built as a ranking reveal: start with the famous one, demote it, then work through what actually did the killing each time. Shot types: ANIM is rendered from code with Manim, AI is a generated shot from Higgsfield with the shared style reference, REAL is agency imagery with credit, CAM is the narrator.

[00:00] S01 · AI · An asteroid entering the atmosphere over a Cretaceous coastline, seen from the ground.

Everyone knows this one. A rock the size of a city, a crater under the Gulf of Mexico, the end of the dinosaurs. It killed about seventy-six percent of species. And it is the second worst thing that has ever happened to life on this planet. It is not close to the first.

[00:30] S02 · ANIM · Five bars of extinction severity; the end-Cretaceous bar is not the tallest; the end-Permian bar rises above it and is labelled 81 percent.

Two hundred and fifty-two million years ago, at the end of the Permian, an estimated eighty-one percent of marine species disappeared. Trilobites, which had lasted nearly three hundred million years, end here. Most reef builders. Most of the dominant land vertebrates. And, unusually, a large fraction of insects, which almost nothing else in the record has managed to kill in bulk. It is sometimes called the Great Dying. If you want to know how close complex life has come to being erased, that is the number.

[01:10] S03 · ANIM · A background extinction curve with five spikes rising above it; the definition appearing: far above background, many unrelated groups, worldwide, at once.

Species go extinct constantly. Over ninety-nine percent of everything that has ever lived is gone, and most of it went quietly. A mass extinction is different: a short interval where the rate goes far above background, across unrelated groups, worldwide. Raup and Sepkoski found five of them in 1982 by analysing the marine record statistically, and the list has held up.

[01:45] S04 · ANIM · The five events as a table with dates, severity and chief suspect; four of the five highlighted as volcanic; one as an impact.

And here is the thing that surprises people who came for the asteroid. Four of the five coincide with an enormous volcanic outpouring. One coincides with an impact.

[02:15] S05 · ANIM · A flood basalt erupting; the lava covering a region but not the world; then the gases rising: sulphur dioxide, then carbon dioxide; a chain of consequences: cooling, acid rain, warming, ocean anoxia, acidification.

And in every case, the lava is not what does it. Flood basalts cover a lot of ground, not a planet. What does the damage is gas. These eruptions release sulphur dioxide and carbon dioxide faster than the systems from lesson 4.7 can absorb them. Sulphur gives you short sharp cooling and acid rain. Carbon dioxide gives you long warming. Warm water holds less oxygen, so the oceans suffocate. And dissolved carbon dioxide acidifies seawater, which dissolves the shell of anything that builds one. Rapid carbon release, then acidification, anoxia and heat. That is the recurring signature, and it is worth sitting with, because it is also a description of right now, at a faster rate.

[03:05] S06 · ANIM · Uranium-lead dating of ash beds in South China; error bars collapsing onto a 60,000-year interval; then conodont oxygen isotopes showing 40 °C equatorial seas and a tropical gap in the fossil record.

For a long time the end-Permian was an interval of unknown length. High-precision dating on volcanic ash beds in South China pinned it down: the main marine extinction happened inside about sixty thousand years. On the cosmic calendar, a few seconds. And the aftermath is worse than the event. Oxygen isotopes show equatorial sea surface temperatures reaching about forty degrees. That is hot enough to empty the tropics of animals, and there is a matching hole in the fossil record at low latitudes. Getting diversity back took five to ten million years. A hundred times longer than the killing.

[03:55] S07 · REAL · The Gubbio boundary clay in outcrop, a thin pale band between limestones (research image; confirm credit line). S08 · ANIM · An iridium measurement expected to be flat; the spike appearing; then the same spike appearing in Denmark, New Zealand and everywhere else.

Now the asteroid, and it deserves its fame for how it was found. In 1980 Walter Alvarez, a geologist, and his father Luis, a physicist, were trying to work out how long this thin clay layer took to deposit. So they measured iridium: a metal that is rare in Earth’s crust, because most of it sank into the core, and that arrives steadily from space in micrometeorites. Instead of a small background level, they got a spike tens of times above normal. Then the same spike turned up in Denmark. And New Zealand. And everywhere anyone looked.

[04:40] S09 · ANIM · The inference: a 10 km asteroid; a decade of contested argument; then the Chicxulub crater revealed under the Yucatán by gravity anomaly data, 180 km across.

The inference was an asteroid about ten kilometres across. There was no crater to point at, and people argued about it for a decade. Then the crater turned up: buried under a kilometre of sediment beneath the Yucatán Peninsula, originally spotted by geophysicists looking for oil. A hundred and eighty kilometres across. And exactly the right age.

[05:15] S10 · REAL · The Deccan Traps step topography of the Western Ghats (photograph; confirm credit line). S11 · ANIM · An eruption timeline spanning the boundary; a state shift within tens of thousands of years of the impact.

Except something else was happening at that exact moment, and it happened here. The Deccan Traps, whose step-like landscape covers much of Maharashtra, erupted across the same boundary. Hundreds of thousands of cubic kilometres of basalt. So: killed by the rock from space, or by the volcanoes? People have argued about it for forty years.

[05:55] S12 · ANIM · Two lines of evidence: proxy records showing warming before the boundary attributed to volcanism, and extinction tracking the impact; then a climate model comparison where impact winter matches the survivor pattern and volcanic warming does not.

The current answer is more interesting than either option. Precise dating shows the Deccan was erupting before the impact and kept going after. One analysis found the eruption style changed sharply within tens of thousands of years of the impact, which raises the possibility that the impact’s shaking affected the volcanism. But statistical analysis of climate proxies across the boundary finds the extinction tracks the impact, not the volcanism. And climate modelling finds that impact winter reproduces the pattern of who died and who lived, while volcanic warming does not. Reasonable summary: the volcanoes stressed the world before and after, and the impact did the killing. Not everyone agrees, and Indian geology sits right at the centre of the argument.

[06:50] S13 · ANIM · Before and after the end-Permian: brachiopods, crinoids and trilobites replaced by molluscs and fish. Then before and after the end-Cretaceous: small mammals becoming rhinoceros-sized within ten million years.

And notice what extinctions actually do, because the word suggests subtraction and that is half of it. They restructure. Before the end-Permian the seas belonged to brachiopods, crinoids and trilobites. Afterwards, to molluscs and fish. Before the asteroid, mammals had existed for a hundred and fifty million years without getting large. Within ten million years afterwards, there were mammals the size of rhinos. Nothing about mammals improved at the boundary. The competition was simply gone.

[07:35] S14 · ANIM · Modern extinction rate plotted far above background but below the five in cumulative magnitude; two axes distinguished, rate and magnitude; a projection reaching the threshold within centuries.

Which brings up the obvious question. Current extinction rates are far above background, by any credible estimate. But rate and magnitude are different things. The honest answer is that the present episode has not yet reached the magnitude of the five, and at current rates it could within centuries. That distinction matters, because those five events took tens of thousands to millions of years. This one is being measured in decades. If it does get there, it will have got there faster than any of them.

[08:15] S15 · AI into ANIM · A quiet modern coastline, unremarkable; the calendar returns.

Five times, the biosphere came close to ending, and each time what was left rebuilt into something the previous world would not have recognised. Next lesson: the small warm-blooded animals that spent a hundred and fifty million years being unimportant, and what happened when the competition disappeared.

Shot list

Two AI shots, eleven code-rendered shots, two research photographs. Estimated Higgsfield use: 3 to 6 generations including retries.