The Gap: Silence After Extinction

I The World After Extinction

The last essay followed the end-Permian crisis: roughly sixty thousand years for the main marine extinction at Meishan, the loss of about four-fifths of marine species, devastated reefs and the end of trilobites. Glossopteris forests also collapsed on land, but not everywhere within that same window. This essay follows the next roughly five million years: the Early Triassic.

It was a world difficult to imagine. Almost all the continents had joined into a single mass, Pangaea, stretching from near the North Pole to near the South Pole. On its eastern side it enclosed a broad gulf, the Tethys Ocean.

South China was a small piece of land drifting near the equator on the eastern side of that gulf. In many of the mountains of Guizhou, Guangxi and Sichuan today, a hammer will bring out rocks from this age: grey limestone filled with clumps of microbialite and layer upon layer of thin-shelled bivalves, with very little else. Near the equator, the surface of the sea may have reached 36 to 40 degrees Celsius, about the temperature of a bath.

As we saw in Essay Fourteen, that figure comes from Yadong Sun of the China University of Geosciences and his colleagues. They read the water temperature from conodonts, and put the words “lethally hot temperatures” in the title of their paper.

The principle behind this thermometer is not hard to grasp. Conodont elements are made of calcium phosphate. The oxygen they contain comes in light and heavy forms; the warmer the water, the less heavy oxygen enters the growing mineral.

Measure the ratio of light to heavy oxygen in one conodont element, and you can work backwards to the temperature of the water in which it formed. Sustained heat like that, together with low oxygen, would make survival difficult for many modern marine animals; tolerance cannot be reduced to one number of days for every species.

Excessive heat distorts proteins and stops them working. The water holds less oxygen just when an animal's body needs more of it. Caught in that squeeze, it cannot breathe. Much the same applies on land: in many plants, photosynthesis begins to decline above 35 degrees Celsius.

The seabed was empty. No reefs. The microbialites from Essay Fourteen spread across the shallows like fields of greyish-white cauliflowers and steamed buns. A few kinds of little snail crawled over the mud. In some samples from the western United States, 99 per cent of the snails were less than a centimetre long; even the largest were only about two centimetres, like millet grains scattered over the ground. Thin-shelled Claraia lay in layers across the bottom, and tongue-shaped brachiopods reached long stalks out of the mud.

Much of the land was red. Across many parts of Pangaea's interior, layer after layer of red sandstone and mudstone accumulated. The famous Buntsandstein of central Germany was left by the rivers and deserts of this age.

The name Triassic comes from Germany too. In 1834, the German geologist Alberti noticed that the rocks of this age in the region consistently formed three great divisions: red Buntsandstein below, marine Muschelkalk, or shell limestone, in the middle, and variegated marls above. He called the interval “threefold.”

On land, along the rivers of South Africa's Karoo Basin, there seemed to be almost only one animal: Lystrosaurus. It was a plant-eater the size of a pig, with a horny beak like a turtle's and a pair of little tusks projecting from its upper jaw. Essay Fourteen described the bones it left in Antarctica.

China had Lystrosaurus too. Around 1928, the geologist Yuan Fuli travelled to Xinjiang with the joint Chinese–Swedish expedition to northwestern China. Near Jimsar, at the northern foot of the Tianshan Mountains, he unearthed a collection of Triassic reptile fossils. Among them were the first Lystrosaurus ever found in China.

One tally found that about 73 per cent of the recorded vertebrate specimens from the earliest Triassic of South Africa belonged to this single genus. Roughly seven specimens in ten—not a direct head count of living animals in that ancient valley.

For a time, fish and reptile fossils are almost impossible to find on land or in the sea near the equator. These animals remained in the cooler middle and high latitudes. Svalbard, now inside the Arctic Circle, and eastern Greenland have both yielded large numbers of Early Triassic fish fossils. Back then, these were places to escape the heatwave. Forests were sparse. Nowhere in the world can we find a securely identified coal seam from the interval.

How long did this last? The usual answer is “five million years.” The Early Triassic extended from about 252 million to about 247 million years ago. Life slowly recovered through this interval, and only in the Middle Triassic did the world grow busy again. The Early Triassic itself has two ages, named after the Indus River in Pakistan and the Olenyok River in Siberia, both places with classic rock sections. Their lengths are very different: the first lasted less than a million years, the second more than four million.

Geologists further divide the Early Triassic into four intervals, named after places and scholars who studied these rocks. The last two, which will recur often in this essay, are the Smithian and the Spathian. The Smithian commemorates the American palaeontologist Smith, the Spathian the British ammonite specialist Spath. Both studied ammonites from this stretch of the rock record.

The questions for this essay are these: what actually happened during those five million years, and when we say “recovery took five million years,” what exactly was recovering?

II The Age Without Coal

One of the easiest gaps to see is coal.

In 1996, Retallack of the University of Oregon in the United States and two Australian colleagues assembled a worldwide compilation. They worded their conclusion carefully: no coal was known from the Early Triassic. Coal did not reappear until the Middle Triassic, and even then the seams were thin and scarce. On the timescale then in use, they estimated a gap of about seven million years between the last Permian coal seam and the next one. Retallack appeared in Essay Nine: he is also the geologist who argues that the Ediacaran organisms lived on land.

Later timescales and ways of counting shifted the figure again. About six million years is now a common formulation; recent work on global vegetation summarises it as about five million. Three things have changed: the absolute timescale has been revised; researchers may count either “the whole Early Triassic” or “the interval from the last thick coal to the first thin coal”; and they may set the endpoint at either “the first local patch of coal” or “the return of widespread thick peat.”

Peat is coal's precursor. The Zoige wetlands of northwestern Sichuan are one of China's largest peatlands. Beneath the meadows lies black peat, soft underfoot, several metres thick in places. It grows by only about a millimetre a year.

Several metres of peat take thousands of years to accumulate. Turning them into a coal seam a few tens of centimetres thick then requires millions of years under compression. What this gap measures is how often recognisable coal seams appear in the rocks, and how thick they are.

Essay Thirteen distinguished three different quantities: coal seams, the amount of organic carbon buried, and the coal we can mine today. The claim here is “we have found no Early Triassic coal,” not “nowhere in the Early Triassic was any peat accumulating.”

That claim stands up to testing. Coal is hard to overlook: black layers of substantial thickness, searched for over and over across the world by people looking for coal for a hundred or two hundred years.

Today, the age of those Triassic coals can be checked bed by bed using zircon in volcanic ash. Zircon is a tiny crystal that takes in uranium as it grows but almost no lead. Uranium decays into lead at a fixed rate; measuring their ratio gives the date when the ash fell. The thick coal-bearing succession of Tasmania, for example, dates from a little over 220 million to about 218 million years ago, placing it in the Late Triassic.

No coal does not mean no forests. Early Triassic rocks in Australia's Sydney Basin preserve impressions of tree trunks, and an earliest Triassic riverside flora has also been reported in recent years. There were trees. What did not accumulate was peat thick enough, for long enough, to become coal in the end.

The pattern is particularly clear in the coal mines of southern China. Much of South China's coal began as Late Permian swamp forest. Miners following the Permian coal seams upwards find that once they enter the Triassic, there is no longer a single substantial seam to be found.

There are three explanations for the absence of coal. All three can be true at once.

The first says that the plants which made the coal died. As Essay Fourteen described, Glossopteris dominated the Late Permian peat swamps of the southern continents. Beneath the Late Permian coal seams of the Sydney Basin, its roots are everywhere; the beds between the seams are packed with its tongue-shaped leaves.

Near the uppermost coal seam, Glossopteris disappears abruptly. High-precision dating places this collapse before about 252.3 million years ago, roughly 370,000 years before the main extinction in the sea. No simultaneous change in the local depositional environment was large enough to erase the coal seams.

A different set of plants grew after the extinction. Along earliest Triassic riverbanks, a common plant was the lycopsid Pleuromeia: an unbranched stem one or two metres tall, with a tuft of long, narrow leaves at the top, like a bottle brush stuck in the ground. Only about three million years later did a group of seed ferns called Dicroidium spread across the southern continents.

Dicroidium leaves split into two branches at their base, like an opened fork. The plant went on to dominate the Triassic forests of the southern continents. It has been found in South America, South Africa, India, Australia and Antarctica. Like Glossopteris, it is evidence that these pieces of land were once joined.

The second explanation says that the conditions for making coal had gone. Coal formation needs three things to coincide: plants growing rapidly, ground staying waterlogged long enough to prevent the plants from rotting away, and the crust sinking slowly so that peat can be buried layer upon layer.

Essay Thirteen described this narrow window. Recent vegetation reconstructions suggest extensive arid, sparsely vegetated open country and woodland in low-latitude continental interiors. This was not modern grassland: the grass family lay far in the future. Productive vegetation was largely confined to coasts and wetter places, under persistent heat and sharply divided wet and dry seasons.

Aridity and strong seasonality are different things. A place may receive plenty of rain over a year, but if it all falls in a few months, decomposition in the dry season can consume the plant remains accumulated during the rains. A forest can shed its leaves in the dry season and grow them back in the wet, year after year, leaving nothing behind.

The third explanation says the record is incomplete. Early Triassic terrestrial plant macrofossils are indeed sparse, and the ages assigned to many old fossil localities are imprecise. But an incomplete record cannot explain the entire gap. The latest global plant database does more than count fossil species: it also examines rock types, depositional environments and fossil coverage, and tests sampling effects statistically. The results still support a sharp decline in biomass and peat ecosystems on Early Triassic low-latitude land.

All three were at work. The available material cannot tell us what share belongs to each. That may sound evasive, but it says something definite: it rules out every version with only one cause.

Pollen and spores give us another picture. Hochuli of the University of Zurich in Switzerland and his colleagues studied a high-resolution pollen-and-spore record from eastern Greenland. At the very base of the earliest Triassic, they found assemblages already dominated by gymnosperm pollen. Gymnosperm dominance may have been re-established very soon after the extinction.

About half a million years later, another large-scale replacement of the flora occurred, and lycopsid spores took over. Early Triassic land looks more like a succession of crises and replacements than a blank left by one extinction. A rapid return of gymnosperms in the pollen record does not mean a rapid return of peat swamps. Wind can carry pollen a long way, and even a few surviving plants can leave some behind. Coal requires enough plants to remain piled up in one place for long enough.

The end of the gap has two endpoints as well. The coal that reappeared in the Middle Triassic was thin and scarce. Several forests of horsetail-like plants preserved where they grew have been reported from Triassic rocks in North China. Stems still stand upright in dense patches, showing that by the Middle Triassic, luxuriant vegetation had returned to riverbanks and lake shores. Its composition, however, was already entirely different from that of the Permian peat swamps centred on Glossopteris.

III Reefs

The length of the reef gap depends on how you define a reef.

The broadest definition is a carbonate structure built by living organisms and rising above the seabed. Under that definition, there is almost no gap after the extinction: microbialites appeared across wide expanses of shallow sea and made conspicuous relief on the bottom.

The narrowest definition is a three-dimensional framework built up in place, layer by layer, by organisms with skeletons, strong enough to withstand wind and waves. Under that definition, the gap is much longer.

The same outcrop, the same rocks, two answers. Both ways of counting have a case. Microbialites really were built by life, and really did raise the seabed. Their ecological role, however, was quite different from that of a coral reef. A coral reef provides countless crevices and holes, sustaining an entire community that depends on it. Today's coral reefs occupy only about one-thousandth of the seafloor, yet house roughly a quarter of marine species. A stretch of microbialite cannot provide that kind of three-dimensional space.

You can still see what living microbialites look like. In Shark Bay, Western Australia, the water is far saltier than the open sea, making life difficult for the snails and worms that eat microbes. Across the shallows grow fields of dome-shaped stromatolites. Layer upon layer of cyanobacterial mats traps sand and mud, slowly building upwards by only a few tenths of a millimetre a year.

Many Early Triassic shallows looked much like this. With fewer animals grazing on the mats, the mats spread across the bottom. In many South China sections, a microbialite bed sits directly above the last Permian limestone. Some are only tens of centimetres thick; others exceed ten metres.

The old account described a reef gap of about eight to ten million years, from the extinction until large, complex reefs flourished again in the Middle Triassic. In 2011, French, American and Swiss researchers found small reefs built jointly by sponges and microbes in Smithian and Spathian rocks of the western United States. The earliest were only about 1.5 million years younger than the extinction. Small, geographically restricted and short-lived, they look more like early, fleeting attempts.

In southern Guizhou lies a huge carbonate platform once isolated in deep water, known to geologists as the Great Bank of Guizhou. Since the 1990s, Chinese and American researchers have been tracing its shallow-marine record bed by bed, from the end of the Permian through the Triassic. Limestones from the platform's rim crop out in stretches along the valleys around Luodian and Pingtang. Walking along the mountainsides, researchers can travel from what was once the deep-water basin all the way to the platform top, reading several million years of shallow seas one layer at a time.

In 2023, researchers identified a whole succession of platform-margin reefs there, extending continuously from the end of the Early Triassic into the Middle Triassic.

Calcareous sponges, calcareous algae and early stony corals did not all arrive together. They joined one by one. Truly diverse reefs were not established until late in the first half of the Middle Triassic, another two to four million years after other marine ecological indicators had recovered and the carbon cycle had stabilised.

So “when did reefs recover?” has three answers: the first small animal-built reef, the onset of continuous reef growth, and the establishment of a highly diverse reef ecosystem. Three answers to three questions.

Today's coral reefs also show how delicate a reef can be. If seawater stays just one or two degrees Celsius warmer than usual for several weeks, corals expel the symbiotic algae living inside them and turn a deathly white. This is bleaching; if it lasts too long, they die. Early Triassic seawater was several degrees hotter even than today's tropical seas.

There is one very straightforward reason reefs took so long to recover. A great reef is not made by a single generation of organisms. Countless generations build it up, layer upon layer, in the same place.

In its thicker parts, Australia's Great Barrier Reef is tens of metres to more than a hundred metres thick, built layer by layer over hundreds of thousands of years. Something like that cannot withstand a fresh disturbance every few hundred thousand years, and Early Triassic disturbances came even more often. A coral that can survive in a troubled sea may still be unable to build a reef there: reef-building needs continuity, not merely survival.

Travel to northern Italy today and you can see what these reefs later became. Many of the steep grey-white peaks of the Dolomites are Middle to Late Triassic reefs and carbonate platforms, subsequently lifted two or three thousand metres high by the building of the Alps. They were inscribed on the World Heritage List in 2009. The mineral making up these mountains is dolomite, named after the French geologist Dolomieu, who described it in the late eighteenth century. The mountains took their name from the mineral.

From the Middle Triassic onwards, new reef builders became important, especially the scleractinian corals that build tropical reefs today. This was not the Palaeozoic rugose corals returning unchanged. Their precise ancestry remains debated; belonging to a different group is not the same as having no evolutionary relationship.

Today's tropical reef-building corals house photosynthesising symbiotic algae and depend on the nutrients they provide. In 2016, Polish researchers analysed chemical signals in Late Triassic coral skeletons and inferred that some stony corals already had such algal partners then.

IV A Dozen or So Clocks

“How long did recovery take?” assumes that one thing was recovering. In fact, a dozen or so things were recovering separately, each keeping its own time.

The first clock counts species and genera. The ammonites swimming in the sea recovered astonishingly fast. An ammonite's shell coils into a spiral, its interior divided by partitions into many small chambers. Where each partition meets the outer shell, it leaves a winding suture line on the fossil. In Triassic ammonites, these lines often look like rows of saw teeth; palaeontologists can distinguish kinds of ammonite by their sutures. In 2009, Brayard of France’s University of Burgundy and colleagues in France and Switzerland found that within about a million years of the extinction, the number of ammonite genera had returned to, or even exceeded, its end-Permian level.

During the Triassic, they diversified into hundreds of genera and filled the oceans again.

Some groups, though, were reduced to a single slender thread. Sea urchins are one example. Only a scant one or two lineages survived the end of the Permian. Every sea urchin in the ocean today, from the purple urchins of the shallows to the many kinds in the deep sea, descends from those one or two branches.

For many animals that crawled over or remained fixed to the seabed, however, genus numbers did not rise substantially until the Spathian, or even the Middle Triassic. Taking the whole ocean together, traditional estimates often put recovery at five to eight million years.

The second clock measures ecological structure: how many feeding levels there are; how many organisms live on the bottom, in the mud or in the water; and whether they make their living by filtering, grazing or hunting. A community can contain quite a few genera yet have them crowded into the same few stress-tolerant ways of life, living close to the seabed surface, attached in place and filter-feeding. A 2018 study found that in Early Triassic seas, recovery in species numbers was decoupled from recovery in ecological functions. The two clocks did not keep time together.

The third clock measures evenness: do several kinds of organism share the community equally, or does one dominate? Among South Africa's earliest Triassic tetrapods, the overwhelming dominance of Lystrosaurus brought evenness close to its minimum.

In Middle Triassic rivers and lakes, some enormous amphibians appeared. They belonged to the temnospondyls. Their great flat heads took up a third of their body length; some were several metres long, like giant salamanders on four legs, lying in wait on the bottom for fish to swim past. Cynodonts gradually became more numerous too. One of their branches would later become the mammals.

According to one tally, on entering the Middle Triassic the share of all tetrapod specimens belonging to the broader group that included Lystrosaurus fell from about three-quarters to about one-fifth. Temnospondyl amphibians and cynodonts together accounted for more than 60 per cent. Evenness can recover before species numbers do.

The fourth clock measures how deeply animals churn the seabed. Many Early Triassic mudstones preserve fine, intact layers, stacked millimetre upon millimetre like a book nobody has opened. They tell us that almost no animals were burrowing through the mud.

Most earliest Triassic burrows are narrow and shallow, as if someone had poked the mud with a toothpick. Some places show little more than shallow disturbance over long stretches of time; complex burrows elsewhere show that burrowing animals had not vanished worldwide. This clock runs very differently from region to region.

The fifth clock measures body size. The Polish researcher Urbanek coined “the Lilliput effect” in 1993. Animals tend to be small after an extinction, but there are two different ways of “getting smaller.” In one, more of the large-bodied kinds die, naturally leaving an assemblage with a higher proportion of small ones.

In the other, the same lineage, or even the same species, becomes progressively smaller as we move upwards through the beds. Both occur in the Lilliput effect discussed in Essay Fourteen.

Heat, low oxygen and scarce food can all keep animals from growing large. Yet not every animal shrank: some rock sections containing ostracods show none of the expected reduction in size.

The sixth clock is the carbon cycle. The Early Triassic carbon-isotope curve swung sharply and repeatedly, only becoming markedly steadier early in the Middle Triassic. This clock ran for about five million years.

The carbon cycle can stabilise before biodiversity has recovered, and rising biodiversity does not guarantee that the carbon cycle will settle down. The two are often plotted on the same graph, against the same time axis. It is easy to mistake them for one thing.

The seventh clock measures distribution by latitude. In today's oceans, species numbers fall from the equator towards the poles, with the tropics richest of all. In 2020, Haijun Song of the China University of Geosciences and his colleagues assembled more than 52,000 fossil records covering over 4,300 genera. After standardising for sampling, they found that throughout the Early Triassic this “most species at the equator” pattern had been almost flattened away. It did not return until the Middle Triassic.

Set the clocks side by side. Some pollen-and-spore assemblages became diverse again in less than half a million years; the first small animal-built reefs appeared after about 1.5 million; ammonites recovered in about a million; genus counts for seabed animals and land-dwelling tetrapods took about five to eight million; coal took about five to seven million, or longer; highly diverse reefs had to wait until the Middle Triassic. Compress all of that into “recovery took five million years,” and you erase the very things that need explaining.

Some places recovered astonishingly quickly. In 2017, researchers from the University of Burgundy in France described a fauna from Spathian rocks near the town of Paris, Idaho, in the United States. Named the Paris Biota, it contains at least seven phyla, including sponges, brachiopods, molluscs, arthropods, echinoderms and chordates, and more than twenty orders. It also includes groups thought to have disappeared long before, in the Palaeozoic. Together they make up a fairly complex marine ecosystem. On recently revised dates, it lay about three million years after the mass extinction. The original paper's estimate was shorter still: about 1.3 million.

China offers several windows too. Late Triassic rocks around Guanling, Guizhou, preserve whole expanses of crinoids. Their stalks were several metres long, their bases attached to floating logs; entire clusters hung beneath the timber and drifted across the sea with it. Rocks around Nanzhang and Yuan'an in Hubei, about 248 million years old, have yielded Hupehsuchus and an assortment of early marine reptiles.

Keichousaurus from Xingyi, Guizhou, is probably China's best-known Triassic fossil. It was a small marine reptile, usually only twenty or thirty centimetres long, with a long neck and paddle-like limbs. Often the entire animal lies intact on a grey slab of rock.

The palaeontologist Yang Zhongjian described it in 1958, giving it a species name that honoured its discoverer, Hu Chengzhi. Xingyi now has a museum devoted to it. Sometimes several individuals, large and small, crowd together on a single slab.

Chaohu in Anhui, which appeared in Essay Fourteen, has one of the earliest ichthyosaurs. At Luoping in Yunnan, excavations beginning in 2007 uncovered the extraordinarily rich Luoping Biota in Middle Triassic rocks about 244 million years old. Fish, marine reptiles, arthropods and plants together amount to more than ten thousand fossils. A complete marine food web, from tiny shrimps to marine reptiles several metres long, is preserved through layers of limestone. By then, the sea was busy again.

V Silence Is Not Calm

“The gap” easily suggests an interval of quiet waiting. The record shows precisely the opposite.

The principle of the carbon-isotope pen can be stated in one sentence: carbon comes in light and heavy forms, living organisms prefer the lighter one, and when marine life becomes more or less abundant, or volcanoes release large amounts of light carbon, the ratio recorded in limestone shifts with it.

The Early Triassic carbon-isotope record is no line gradually returning to normal. The great downward plunge at the Permian–Triassic boundary was only the beginning of a succession of disturbances. Through the Early Triassic, the curve underwent at least four episodes of large swings up and down, often by four to eight parts per thousand. Look more closely and there are further, smaller fluctuations. Only early in the Middle Triassic did it abruptly settle down.

The weather was extreme as well. India's monsoon is fierce enough today: each summer, moisture blown in from the Indian Ocean can deliver a year's rain in a few months.

Pangaea straddled both hemispheres, with an interior immensely far from the sea. In summer the land grew scorching hot and drew in ocean moisture with great force; in winter the circulation reversed. The result was a “supermonsoon” operating on a scale far beyond that of present-day Asia. Some places spent most of the year in drought, then flooded as soon as the rains arrived.

Temperature did not follow a straight line either. Late in the Smithian, low-latitude sea surfaces were driven into the extreme range of 36 to 40 degrees Celsius. Then, at the Smithian–Spathian transition, a pronounced cooling came with another biological crisis. The plants on land changed shifts too: lycopsid-dominated vegetation rapidly gave way to vegetation dominated by conifers. Within a very short stretch of rock, the pollen-and-spore assemblages wore a different face.

Ammonites and conodonts took another heavy blow in this crisis, about 249 million years ago. Dozens of ammonite genera that had only just returned disappeared en masse once more. Only a few branches survived, and almost all Spathian ammonites diversified anew from those survivors.

Repeated changes in temperature are harder to cope with than constant heat. If a high temperature holds, organisms can at least adapt to it slowly. Here, each time they had just settled in, the conditions changed again.

Oxygen shortages returned. Geologists often examine pyrite framboids: tiny berry-shaped aggregates of iron sulphide. Those formed in oxygen-free, sulphidic water tend to be small and fairly uniform, often only a few micrometres across. But they can also form in sediment pore water, and later alteration can change their sizes. “Smaller” is not a direct scale of “more oxygen-starved”; the grains must be read alongside other indicators.

Several kinds of redox indicator show that widespread marine anoxia after the end of the Permian did not end in a single step. The Early Triassic saw several expansions of oxygen-starved waters, associated with warming, intensified weathering on land and changes in nutrient supply. Different depths and ocean basins did not move in step. “Many sections record persistent anoxia” does not mean “the whole ocean had no oxygen for five million years.”

The seabed itself became strange. In some Early Triassic shallows, fan-shaped clusters of aragonite crystals grew directly out of the seawater, standing in bunches on the bottom, some tens of centimetres tall. Such deposits had been rare since the early Palaeozoic, because later seas held abundant shell-building animals that took up the calcium carbonate first. Once the extinction cleared those animals away, the deposits returned. Geologists call them “anachronistic sediments.”

There is direct evidence that each disturbance pushed recovery backwards. A study of seabed communities in northern Italy interpreted two subsequent Early Triassic biological crises as repeated blows to marine recovery. Functional richness and ecological tiering did not improve markedly until the Spathian. The plant record of eastern Greenland tells a similar story: initial recovery in the earliest Triassic, another drastic replacement about half a million years later, and another round of crisis after that.

There is another explanation: disturbance and slow recovery may be two expressions of the same unstable system. An ecosystem left with only a few kinds of organism, all crowded into similar positions, lacks buffers. If one group runs into trouble, nothing else can take its place. Productivity across the system then rises and falls sharply, and the carbon-isotope curve swings with it.

In this model, each swing in carbon isotopes need not be explained by a separate volcanic eruption. At present, these two explanations cannot be disentangled.

The main eruptions and intrusions of the Siberian volcanism were concentrated at the end of the Permian and in the earliest Triassic. Whether an additional volcanic pulse occurred about two million years after the extinction remains under discussion in climate research. For such later pulses, directly determined ages are far sparser than they are near the boundary.

VI Disaster Taxa

There is a term for organisms that, for a short time after an extinction, become so numerous and so widespread that they dominate whole communities: disaster taxa. In 1977, someone describing calcareous nannofossils that flourished in profusion after the end-Cretaceous mass extinction used “disaster species.” In 1992, someone else called Early Triassic stromatolites “disaster forms.” Later, “disaster taxa” became the settled term.

The classic marine examples are thin-shelled bivalves such as Claraia. In the earliest Triassic shales of many South China sections, you can count dozens of Claraia on a slab no bigger than your palm. Their shells are paper-thin, marked with fine concentric rings, and often cover entire bedding surfaces in layer after layer. Some researchers suspect they could scrape by on an oxygen-poor seabed.

Lingulid brachiopods and certain small foraminifera are just as common. A lingulid's two thin shells look like a pair of fingernails, with a long fleshy stalk trailing below and planted in the mud.

The general appearance of these brachiopods has changed very little from the Palaeozoic to today. Their close relatives can still be dug from coastal mudflats in Fujian, Guangdong and Guangxi. Known locally as “sea bean sprouts,” they can be cooked and eaten. A South China study put “disaster and opportunistic foraminifers” directly in its title, recording several small, simply built foraminifera that covered the seabed when other groups were scarce.

On land, the most famous case is Lystrosaurus. Among plants, lycopsids such as Pleuromeia gained the upper hand in several Early Triassic regions. The change in eastern Greenland is particularly clear.

The American palaeontologist Cope named Lystrosaurus in 1870. Its fossils have been found in South Africa, India, Antarctica, Russia and China's Xinjiang. It was the most widely distributed land animal of the Early Triassic.

It was once suggested that its barrel-shaped chest and short snout suited forceful breathing in oxygen-poor air, allowing it to survive in the post-extinction world. That explanation later came under considerable criticism. Burrows it dug have also been found in South African rocks, sometimes with juveniles still lying inside.

There are at least three explanations for Lystrosaurus being “everywhere.” Its own numbers really increased. Or its numbers changed little while those of other animals fell sharply, so that its share rose without any increase of its own.

Or its sturdy bones, a habitat favourable to preservation, or collectors' habit of digging in beds where it is commonly found made it easier to discover. Simply observing that “this genus occurs in many places” cannot distinguish among these three mechanisms.

In 2020, Modesto in Canada published an article called “The Disaster Taxon Lystrosaurus: A Palaeontological Myth.” His central warning was that the claim of extraordinarily high abundance worldwide rested on a geographically very uneven record, with a complicated history of classification and collecting. South Africa's 73 per cent cannot simply be promoted into “three-quarters of all land vertebrates in the world were Lystrosaurus.”

This kind of one-group dominance has followed other mass extinctions too. Essay Eighteen will describe how, after the asteroid strike at the end of the Cretaceous, a sudden spike in fern spores appeared in North America's pollen-and-spore record. The forests had been destroyed; ferns were the first to spread across the scorched ground.

The first organisms to fill the ground after a disaster are often the few kinds that grow quickly, reproduce abundantly and get by almost anywhere. The same plants are the first to grow after a wildfire today: weeds and ferns turn the burnt ground green within a year or two. Trees take decades to return.

“Disaster taxa” can also sound like a judgement, as though these organisms were speculators profiting from catastrophe. What the term actually describes is a statistical fact: in a certain stretch of strata, specimens of this group make up an unusually high share. A high share is a ratio, and a ratio's denominator can shrink all by itself.

What happened later was a gradual loosening of that monopoly. Evenness rose. Other lineages supplied a larger share of specimens. Marine communities moved from dominance by a few groups towards more ways of life and more complex tiering. In the Late Triassic, about 230 million years ago, the first dinosaurs appeared on land. That is a story for the essays ahead.

Some of the earliest dinosaur fossils come from a tract of badlands in northwestern Argentina called the “Valley of the Moon.” Wind and rain have stripped its rocks into oddly shaped mounds. In 1993, researchers described a little dinosaur from there, only about a metre long, called Eoraptor. It ran on its two hind legs and could still grasp things with its forelimbs.

The real rise in dinosaur abundance may be connected with an exceptionally rainy interval in the Late Triassic. About 233 million years ago, sediments in many parts of the world abruptly began to record wetter conditions. These lasted one or two million years. Geologists call it the Carnian Pluvial Episode.

After the rains, the plants and animals on land changed again, and dinosaurs rapidly diversified. Lystrosaurus was not defeated by “better” animals. It gradually faded from the scene after the early part of the Triassic, while other animals became more numerous.

VII How Much of the Silence Belongs to the Record?

The tools introduced in Essay Eleven come into their own here. First, we need to discuss the marine and terrestrial records separately.

The Early Triassic marine record is not poor. Conodont elements are tiny fossils, from a few tenths of a millimetre to one or two millimetres long, shaped like rows of little teeth that can only be seen clearly under a microscope. For more than a hundred years, nobody knew what animal they belonged to.

In 1983, British researchers reported specimens preserved with soft parts still attached, from Carboniferous rocks near Edinburgh in Scotland. Only then did we learn that the owner was a small, slender, eel-like animal, and that the little teeth sat in its mouth.

Conodonts changed quickly and spread widely. Researchers can use them to divide Early Triassic marine strata into a dozen or so fossil zones, each only a few hundred thousand years long.

There are abundant carbonate, shale and biostratigraphic sections that can be resolved into time bins a few hundred thousand years wide. Curves can be built for carbon isotopes, conodonts and marine animals. That Haijun Song and his colleagues could assemble more than fifty thousand fossil records for sampling standardisation is itself a sign of how much material this side offers. Each record means that someone found a particular genus once, at a particular place, in a particular bed.

The terrestrial situation is much worse. Around Shanxi, a succession of Early Triassic red sandstones and mudstones, known as the Liujiagou and Heshanggou formations, runs to hundreds of metres in thickness. Fossils are scarce, and for a long time the rocks were thought to contain almost nothing. Later, researchers found wrinkles left by microbial mats, along with a few footprints and burrows. Only then did it become clear that those riverbanks and lake shores had not been entirely lifeless. Recent studies have also reported upright stems of horsetail-like plants and Pleuromeia, preserved where they grew, in the Heshanggou Formation, only about two million years after the mass extinction.

Large plant fossils are sparse, and the ages of many old localities are poorly constrained. Studies of land plants worldwide have to use time bins at the level of geological stages. They cannot reach the marine record's resolution of a few hundred thousand years per bin. An Early or Middle Triassic stage may last less than a million years at the short end, or four or five million at the long end. For many changes in the terrestrial record, all we can say is “it happened within this stage.” We cannot say which came first.

The second task is to separate two kinds of absence: something did not exist at the time, or it did exist but left no remains we can find. As Essay Eleven explained, these can look identical in the data. To distinguish them, we must ask: if it existed, how confident should we be that we would have found it?

Coal is well placed in this respect. Coal made from a thick bed of peat is a distinctive rock, and people looking for coal have searched for it worldwide for a hundred or two hundred years.

If thick peat had been widespread in the Early Triassic, it is unlikely to have been missed altogether. “Did a particular plant still exist in the Early Triassic?” is in a much weaker position. Macrofossils are sparse, and pollen and spores repeatedly show that some plants absent from the large-fossil record were still there. The same succession of rocks can give a weighty “no” to “was there coal?” and a very slight one to “was there this particular plant?”

The third issue is Lazarus taxa. The Paris Biota offers a ready-made example: a group of slender sponges previously known only from Cambrian and Ordovician rocks. They were thought to have died out long before, yet they reappear in this Early Triassic fauna. They must have remained alive through the intervening two hundred million years or so, while leaving almost no fossils. Essay Eleven introduced the term: a group disappears from one stretch of strata, then reappears in younger rocks. It must have been alive in between; that interval simply went unrecorded.

The Americans Flessa and Jablonski proposed “Lazarus taxa” in 1983. Later, their compatriots Erwin and Droser coined a playful counterpart: “Elvis taxa.” After an extinction, a new group appears that looks very like an extinct one. At first glance the old group seems to have risen from the dead. In fact, a different branch of animals has evolved a similar appearance, just as all those Elvis impersonators in the street are not Elvis himself.

Quite a number of Lazarus taxa have been recognised between the end of the Permian and the Triassic. Their existence places a lower bound on the Early Triassic record's blanks: at least this many things were alive then without leaving fossils. But the Lazarus taxa we can recognise are precisely those recorded twice. Those that disappeared and never returned look exactly like the genuinely extinct in the data.

The fourth issue is the width of the time bins. The same gap may resolve, in marine rocks, into several episodes of rise and fall, while on land it remains one solid blank.

Essay Seven explained that a flat curve can mean the world did not change, or that the measuring tool was not sensitive enough. Here is a third possibility: the tool is sensitive enough, but the time bins are too coarse. The Early Triassic gives us a direct comparison. The marine side has already resolved five million years into a succession of rises and falls. At least some of the flatness on the terrestrial side belongs to the time bins.

The converse must be made clear too. Sampling bias cannot explain everything. Even after controlling for sample size, the number of South African tetrapod genera in the Early Triassic remains very low.

The latitudinal gradient found by Haijun Song and his colleagues remains flat after standardisation. A global pattern that survives correction cannot simply be put down to sampling again.

VIII After Recovery, a Different World

If “recovery” means returning to the world before the extinction, the answer is that it never returned.

The structure of the Mesozoic oceans genuinely differed from that of the Palaeozoic oceans. The Palaeozoic continental shelves were dominated by animals fixed to the seabed, filtering their food from the water: brachiopods, crinoids and bryozoans.

From the Mesozoic onwards, their share fell markedly. Animals that moved actively, burrowed and hunted became more numerous. A new set of hunters appeared in Triassic seas: ichthyosaurs, and marine reptiles called placodonts, whose mouths held rows of broad, flat teeth for crushing shells. Triassic rocks in Guizhou and Yunnan have yielded many of their fossils.

In 1977, the American researcher Vermeij named the changes seen from the Mesozoic onwards, as predators grew more formidable and prey developed thicker, spinier shells, the “Mesozoic marine revolution.” Ten years later, he gathered these ideas into a book called Evolution and Escalation.

Vermeij lost his sight at three. He studies shells with his hands, feeling each one for its thickness, the length of its spines, the scars where a crab crushed it and the shell grew back. A palaeontologist who could not see felt out an unseen war in the Mesozoic seas. Animals churned the seabed ever more deeply, turning a thin interface into a thick layer that was worked over again and again. The reworking of the seabed described in Essay Ten deepened and diversified in the Mesozoic; it did not reach a single final point of completion.

In 1981, Sepkoski in Chicago analysed the family-level lists of marine animals he had compiled, the sort of lists discussed in Essay Eleven. He divided the marine animals of the Phanerozoic into three “evolutionary faunas”: a Cambrian fauna represented by trilobites; a Palaeozoic fauna represented by brachiopods and crinoids; and a modern fauna represented by bivalves, gastropods and fish. Each had its own dominant groups, and they succeeded one another through time. The end of the Permian falls just at the boundary between the “Palaeozoic fauna” and the “modern fauna.”

This division is a statistical clustering. Its boundaries depend on the scope of the database and the statistical method. “The modern fauna replaced the Palaeozoic fauna” describes a statistical pattern, rather than a single ecological event we can point to.

The Triassic after recovery was an entirely new world. By the Late Triassic, the first dinosaurs, the first pterosaurs and the first mammaliaforms had appeared on land. Ichthyosaurs and the ancestors of plesiosaurs swam in the sea. Many groups in today's oceans also laid their foundations at this time.

Ichthyosaurs provide the most startling example. Those early kinds from Chaohu were less than a metre long. By the Late Triassic, some North American ichthyosaurs had reached around fifteen metres, about the length of an adult male sperm whale. Dozens of these giant ichthyosaurs were unearthed on a hillside in the US state of Nevada, which later made the animal its state fossil.

From the ground strewn with little snails after the extinction to sea monsters more than ten metres long took only a little over twenty million years. The first giant ichthyosaurs appeared earlier still: rocks in Nevada about 244 million years old have yielded an ichthyosaur around seventeen metres long, less than ten million years after the mass extinction.

“Recovery” suggests a return to where things were. This connects with the conclusion of Essay Fourteen: those that survive an extinction need not be the winners in ordinary times, and those that rule the new world afterwards need not have been the strong ones in the old.

Go to the shore to collect shells today, and almost everything you pick up will be a bivalve or a snail. Only a few hundred species of brachiopod remain, most living in deep water, cold water or cracks in rock. Few people have ever seen one alive.

Bivalves replaced brachiopods as masters of the seabed because they happened to survive and then happened to expand quickly in the new oceans, rather than because they had already been more successful in the Permian. What actually happened was the gradual establishment of a new assemblage. By several measures it equalled or even exceeded pre-extinction levels, but its composition was entirely different.

Coal gives the cleanest example. Tall forests reappeared in the Middle Triassic, but they were no longer Glossopteris peat swamps. Reefs tell the same story: stony corals had taken over as the principal builders. “Reefs recovered” and “the reefs came back” are two different statements. Essay Twenty will take up the next question: is an ecological niche an empty place already sitting there, waiting to be filled again?

What this series calls a construct comes down to drawing a line: what lies inside counts, what lies outside does not. People can draw the line, as with a set of criteria. Nature can draw it too, as with a membrane.

The construct in this essay is the line called “the timescale of recovery.” It is not clearly drawn. What is recovering: genus numbers, evenness, ecological function, the depth to which the seabed is churned, body size, the carbon cycle, reefs, peat, or distribution by latitude? These dozen or so quantities give answers ranging from less than half a million years to more than ten million, a twentyfold difference.

Each figure is quite reliable within its own method of calculation. They are different numbers, rather than different estimates of one number. When a review says “recovery took five million years,” it usually means one or two of them, and more often than not fails to say which.

The part that a construct cannot contain, yet cannot make disappear, is what this series calls the remainder. Here, the remainder is those two gaps. There really are no securely identified Early Triassic coal seams, and there really are no large animal-built reefs. Both absences have survived corrections for sampling.

What makes them remainders is not that there was nothing there. There were forests: the impressions of tree trunks in the Sydney Basin are there. There was vegetation: the pollen and spores show that gymnosperms regained dominance early.

There were small animal-built reefs: those little reefs in the western United States are there. There was a complex marine fauna with seven phyla preserved in it, at Paris in Idaho.

Two particular things are missing: peat that could accumulate over long periods and remain preserved, and large skeletal reefs able to withstand wind and waves. Both require a set of conditions to be met together, and to keep being met for a very long time.

Peat needs prolonged waterlogging, slow decomposition, plants that keep growing rapidly, and ground that sinks at the right rate. A great reef needs reef-building organisms, stable seawater chemistry, suitable water depth and currents, and a sufficiently long, uninterrupted stretch of time.

The repeated warming and cooling of the Early Triassic, its recurring oxygen shortages and swings in the carbon cycle, interrupted precisely that “sufficiently long, uninterrupted stretch of time.” A genus can survive in the space between two disturbances and be recorded there. A reef or a bed of peat cannot, because it is itself an accumulation of time.

So the silence in this essay is not empty. Throughout those five million years, organisms were living, dying and being recorded. There were forests, vegetation, little reefs, a complex ocean. In many places, what was scarce was a long enough stretch of suitable, stable conditions—not every moment of quiet for every form of life throughout those five million years.

The criteria are still sifting. The length of the gap still shifts with the definition. “Recovery” still has no definition that works across measures.