1 A Seafloor Five Hundred Million Years Ago
About 518 million years ago, Chengjiang, near what is now Kunming in Yunnan, lay beneath a warm, shallow sea. Most of Earth's land was then joined together in the Southern Hemisphere in the supercontinent Gondwana. The land was still bare: no plants, no animals.
South China was a small landmass drifting near the equator, surrounded by warm, shallow seas. A day lasted only about twenty-one hours. The Moon was closer, and the tides were stronger.
A few trilobites crawl slowly across the mud, their bodies divided lengthwise into three lobes, left, middle, and right, the jointed plates on their backs glinting underwater. A worm with pairs of soft, fleshy legs clings to a clump of sponges, two rows of sharp spines standing up along its back. Emerging from the mud is an arthropod with a pair of long antennae on its head and a sturdy outer shell: much like a shrimp today, and yet not quite like any of them.
Through the water swims the largest hunter in this sea: Anomalocaris. It is about half a meter long, sometimes longer. Rows of finlike flaps undulate down either side of its body. A pair of large, spiny claws projects from the front of its head like two pincers, and its mouth is a ring of tooth plates that opens like a slice of pineapple.
It has a pair of large compound eyes on short stalks. Whether it could crack a trilobite's hard shell is still disputed. In 2023, researchers used computer models to work out the forces acting on its claws and concluded that they were better suited to catching soft-bodied prey.
Trilobite eyes are remarkable too. Every tiny lens in a compound eye is a transparent crystal of calcite, its optical axis aligned with the direction of incoming light.
These were animals that saw the world through stone. Their stone eyes often survive intact in fossils. In 2011, researchers found beautifully preserved compound eyes from an anomalocaridid in slightly younger rocks on Australia's Kangaroo Island. A single eye had more than sixteen thousand individual visual units, more than most insects have today.
Another animal, a few centimeters long and shaped like a little fish, swims by flexing its body. In 1999, Shu Degan and colleagues at Northwest University identified two such animals among the Chengjiang fossils. Each had a notochord running through its body and rows of muscle blocks folded into double chevrons. They are among the earliest known vertebrates. Your distant ancestors and mine were already here, in this sea.
Chengjiang fossils preserve detail fine enough to reveal nerves. In 2012, researchers from Yunnan University, Britain, and the United States identified a preserved brain in a fossil arthropod called Fuxianhuia. Its organization was already quite close to that of insect and crustacean brains today. A little head from more than five hundred million years ago left a dark shadow in the mudstone.
Some Chengjiang animals still have no settled place. One group, the vetulicolians, had bodies divided into two parts: a front half like a box with a row of gill openings, and a rear half like a segmented tail.
In 2001, Shu Degan and colleagues established a new phylum for them. Later researchers interpreted them as an early offshoot of the deuterostomes; others placed them elsewhere. A little animal only a few centimeters long has been passed from one taxonomic address to another for more than twenty years.
Buried in the mud is another kind of worm, a priapulid. Its sausage-shaped body has a proboscis ringed with spines that turns outward at the front, extending and retracting as the animal burrows through the sediment. Priapulids are common at both Chengjiang and the Burgess Shale, and a few descendants still live in seafloor mud today. Some researchers think it very likely that animals much like their ancestors dug the probing burrows we will meet in the next section, the ones used to define the beginning of the Cambrian.
These animals lie preserved in layers of yellow-green mudstone, eyes, guts, and even the fine hairs on their legs included. The Chengjiang fossils originally consisted of black carbon films and pyrite. Weathering at the surface often turns them rusty brown and pale red against the yellow-green rock, like a Chinese fine-brush painting on stone. On July 1, 1984, Hou Xianguang of the Chinese Academy of Sciences' Nanjing Institute of Geology and Palaeontology split open a piece of mudstone on Maotianshan at Chengjiang. Inside was a complete soft-bodied arthropod, later identified as Naraoia.
He said that for a moment it seemed alive, about to swim across the damp rock face. That was the discovery of the Chengjiang biota. More than two hundred species have now been described, and the site became a World Heritage Site in 2012.
China is one of the best places to see the Cambrian. Qingjiang in Hubei is roughly the same age as Chengjiang in Yunnan; Guanshan, also in Yunnan, is a little younger; Kaili in Guizhou is younger again. These fossil deposits line up in time like photographs on successive pages of an album, showing how the same sea changed over tens of millions of years.
For billions of years before this, Earth had been home mostly to microbes, along with those flat, hard-to-classify Ediacaran organisms we met in Episode Nine. Then, within a few tens of millions of years, the seas were suddenly full of animals with eyes, legs, and shells, animals that chased and fled. Early members of most animal phyla alive today can be found in the rocks of this interval.
This is the Cambrian explosion. The name “Cambrian” comes from an ancient name for Wales. In 1835, the British geologist Adam Sedgwick used it to name the oldest fossil-bearing strata in that part of the country.
Darwin worried about it. In On the Origin of Species, he acknowledged the difficulty it posed for his theory: complex animals appeared all at once in the Cambrian, yet there seemed to be nothing beneath them.
More than a hundred and sixty years later, the explosion has become one of the most famous events in evolutionary history, and one readily disposed of in a single sentence. This episode asks what in that explosion was real, what our ways of observing have produced, and what, exactly, we explain by calling it a “phase transition.”
2 The Spike in a Worm’s Burrow
On what day did the Cambrian begin? The answer is a place.
Many boundaries on the geological timescale are defined by a point on an actual rock face somewhere in the world. Above this bed, a new interval begins; below it lies the old one. Such a point is popularly known as a golden spike. We will meet it again in Episode Fourteen and devote a whole discussion to it in Episode Twenty-Three.
The Cambrian's spike is driven into a sea cliff at Fortune Head, on the Burin Peninsula in southeastern Newfoundland, Canada. Its precise position is in a sequence of strata called the Chapel Island Formation, 2.4 meters above the base of what was originally designated Member 2. This point simultaneously defines the beginnings of the Cambrian, the Paleozoic, and the Phanerozoic. “Phanerozoic” means “visible life,” a name coined in 1930 by the American geologist Chadwick to distinguish it from the long preceding age in which life was all but invisible.
The Cambrian was later subdivided into a number of stages, each of which also needs a spike to mark its beginning. Four are in China: at Paibi in Huayuan, Hunan; Luoyixi in Guzhang, Hunan; Wuliu–Zengjiayan in Jianhe, Guizhou; and Duibian in Jiangshan, Zhejiang.
Choosing the Cambrian spike took twenty years. In 1972, the International Union of Geological Sciences set up a dedicated working group. Its members examined candidate sections around the world, comparing and arguing through round after round, until a vote finally settled the matter in 1992.
One of the contenders in that 1992 vote was in China: Meishucun, in Jinning, Yunnan. Its rocks held abundant small shelly fossils from the very earliest Cambrian, and Chinese geologists had devoted years of work to it. Fortune Head won.
At Fortune Head, the spike marks a trace of behavior rather than an animal's body. The trace fossil consists of a nearly horizontal main tunnel with small, slanting exploratory burrows branching off at regular intervals, upward, downward, and to either side. Seen on the underside of a bed, it looks like a string of almond-shaped bumps.
The usual interpretation is that a wormlike, bilaterally symmetrical animal moved through the mud, probing in different directions for food as it went. No one has ever seen the animal itself. The beginning of the Phanerozoic is defined by a feeding maneuver performed by an unidentified creature.
There were good reasons to choose it. Trilobites appear distinctly later, and the earliest trilobites on different continents are strongly regional. Small shelly fossils occur earlier, but they are often scattered fragments of uncertain affinity. Complex burrows can survive in ordinary rocks without the animals that made them being preserved, while still recording the activity of complex animals.
Then something awkward happened. In 2001, Gehling, whom we met in Episode Nine as the proponent of the death-mask model, and his colleagues reexamined Fortune Head. They found this diagnostic trace more than three meters below the spike and again more than four meters below it. Between seven and fifteen meters below, they found similar burrows belonging to the same ichnogenus. The spike did not move: it was, after all, a definition. But the claim that it marked the first appearance of this trace no longer held.
The age has shifted too. In 2019, similar traces were reported from the top of a rock sequence in Namibia, between two volcanic ash beds dated to about 539 million and 538.6 million years ago.
The figure of about 538.8 million years for the beginning of the Cambrian was based partly on that material. In 2026, a reexamination suggested that those older traces belonged to a different kind of burrow, and that the oldest securely dated examples of the true diagnostic trace might be only about 533 million years old.
The numerical age rests chiefly on volcanic ash near the boundary in sections around the world. Both Oman and Namibia have ash beds close to the boundary. Again and again, dates from their zircon crystals have pulled the figure toward a younger age.
Over several decades, the numerical age of the Cambrian's beginning has been revised repeatedly: about 570 million years ago in the later twentieth century; about 544 million by the early 1990s; 542 million in 2004; 541 million in 2012; and 538.8 million in 2022. The physical point on the rock face at Fortune Head has never moved. The age of the Cambrian has changed. Its boundary has stayed put.
3 What Happened, and in What Order
Lay the fossils from this interval out in order, and what emerges is a process unfolding in stages, not a single bang.
Before the Cambrian began, Cloudina, which we met in Episode Nine, already had a hard shell, with holes drilled into it by predators. In the first few million years of the Cambrian, the seafloor was a “world of tubes”: assorted narrow tubes, primitive little toothlike structures, sponge spicules. The fossil inventory was short. Among its contents were hook-shaped spines only one or two millimeters long, occurring in groups of several. They closely resemble the hooks around the mouths of living arrow worms, used to seize prey, and may be among the oldest known predatory “teeth.”
Next come the small shelly fossils: millimeter-sized mineralized fragments, cap-shaped shells, scales, plates, little tubes. Dissolve the limestone from Meishucun or rocks of this age elsewhere in acid, and thousands of them can be recovered. Researchers often have to soak tens or hundreds of kilograms of limestone in acetic acid, piece by piece, for weeks, then pick the fossils one at a time from the residue under a microscope.
“Small shelly fossils” names a collecting basket, not an animal. When an animal died, its body often fell apart into dozens or hundreds of differently shaped plates, like a disintegrating suit of mail. What you find is a heap of little pieces that do not look alike.
Some of those suits of mail were eventually put back together. The fossil site at Sirius Passet was discovered in 1984 by researchers with the Geological Survey of Greenland, the same year as Chengjiang.
In 1990, British and Danish researchers at Sirius Passet in northern Greenland found complete fossils of an animal called Halkieria. A few centimeters long, it had a back densely covered in several kinds of differently shaped plates, with a little shell perched at either end. Fragments previously assigned to separate groups turned out to belong to the same body.
Along both banks of the Lena River and its tributary the Aldan in Siberia, Russia, stretches a long succession of Cambrian cliffs, their beds stacked like the pages of a book. Several early Cambrian stages take their names from small places in this region, Tommot among them. In the 1950s and 1960s, Soviet paleontologists systematically collected small shelly fossils there and, for the first time, put the earliest Cambrian animals into sequence.
By about 525 million years ago, animals with skeletons were more diverse: brachiopods, hyoliths, and a group of sponges called archaeocyaths. Together with microbes, the archaeocyaths built the Cambrian's first large-scale animal reefs in shallow seas. They looked like double-walled cups perforated with tiny holes, most only a few centimeters tall, standing in clusters on the shallow seafloor.
They flourished for more than ten million years, then almost all vanished by the middle Cambrian. Earth's reefs were left largely vacant for a long time, until other animals eventually took over. About 521 million years ago, the first trilobites appeared. One of the commonest early Cambrian trilobites in China is Redlichia, with a large head, a small tail, and long spines projecting from its cheeks. It is found everywhere in the shales of Yunnan, Guizhou, and Shaanxi, and is the first trilobite many geology students learn to recognize.
More recent studies suggest that the earliest trilobites entered the record on several continents almost simultaneously, about 521 to 520 million years ago. The difference is roughly a million years, much the same as the uncertainty involved in correlating the continents. The earliest echinoderms appeared around 520 million years ago. By about 518 million years ago, Chengjiang's soft-bodied fauna already included highly differentiated arthropods, chordates, and many other groups.
Tiny animals were preserved too. Around Kuanchuanpu in southern Shaanxi, rocks about 535 million years old contain abundant millimeter-scale animals preserved in phosphate.
In 2017, Han Jian and colleagues at Northwest University described one of them: a wrinkled little creature just a millimeter across, with its mouth gaping open, called Saccorhytus. It was initially regarded as one of the earliest members of the deuterostomes, the great group to which humans belong. In 2022, more complete scans showed that it was actually an early member of another major group, a distant relative of today's insects and roundworms. A creature one millimeter across had its family tree changed within five years.
The Cambrian's internal stages are still not all settled. Several beginnings remain without a spike, because no fossil has been found that can be matched across continents and that appears abruptly enough. How long the “explosion” lasted depends on which stretch of this sequence we choose to call the explosion.
Take 530 to 520 million years ago, and it lasted about ten million years. Count from the base of the Cambrian to the arrival of trilobites, and it lasted about eighteen million. Include the groundwork before the Cambrian, and it becomes about thirty million.
One more thing to remember: many claims about “the first appearance of a phylum” mean the earliest recognizable member of its total group. Episode Nine introduced total groups and crown groups. A Cambrian animal can genuinely belong to a branch of a particular phylum without looking anything like that phylum's familiar living members.
Many small animal phyla alive today, including flatworms and ribbon worms, also have no undisputed Cambrian fossils. An animal a millimeter or less in length, without any hard structures, has almost no chance of leaving anything in the Cambrian record. Its absence from that record does not tell us it was absent from the world.
4 The Record Brightens
Part of the Cambrian's apparent richness comes from how we are able to observe it.
The first route is a window for preserving soft bodies. In 1909, Charles Walcott of America's Smithsonian Institution discovered black shale containing soft-bodied animals near Burgess Pass in the Canadian Rockies.
Walcott was then Secretary of the Smithsonian. Every summer he took his family into the mountains to collect fossils. The story goes that his wife's horse slipped on the trail and kicked over a slab of rock.
The story may not be reliable, but the Burgess Shale became famous all the same. Its animals lived at the foot of an underwater cliff. Mudflows swept down the slope again and again, burying them rapidly where they lay, in oxygen-poor mud. Before scavengers and bacteria could get to work, the clay pressed their bodies into thin black films.
Walcott collected more than sixty thousand specimens over his lifetime and squeezed most of the animals into existing phyla. The commonest animal there was Marrella, an arthropod only two centimeters long, with two pairs of long spines extending from its head. More than twenty thousand specimens were collected.
Sixty years later, Harry Whittington at Britain's University of Cambridge and his students Simon Conway Morris and Derek Briggs reexamined the specimens. Many of the animals, they found, would not fit into any known phylum. One, Opabinia, had five eyes on its head and a long proboscis projecting from the front like a fire hose, with a pincer at the end. When Whittington showed his reconstruction at a scientific meeting in 1972, the audience burst out laughing. They thought he was joking.
The story of Anomalocaris took even more turns. Its pair of large claws had been described as the bodies of a kind of shrimp as early as 1892. Its ring-shaped mouth had been taken for a jellyfish.
Its body was first interpreted as a sea cucumber, and later as a jellyfish fossil lying on top of a sponge. Each of the three parts had its own name. Not until 1985 did Whittington and Briggs recognize that they belonged to a single animal.
Another animal had an odder fate still. It bore two rows of spines on its back and a row of soft, fleshy legs below, but its 1977 reconstruction was upside down: the spines became legs, and the legs became tentacles on its back. It was named Hallucigenia. In 1991, comparisons with a relative from Chengjiang finally turned it the right way up. In 2015, researchers worked out which end held its head.
The Burgess story is still unfolding. In 2014, Jean-Bernard Caron of Canada's Royal Ontario Museum and colleagues reported the discovery of a new Burgess-type fossil site at Marble Canyon, about forty kilometers from the original locality. A few weeks of collecting yielded thousands of specimens, including many new species. The Kaili biota in Guizhou, China, is close in age to the Burgess Shale and also preserves abundant soft-bodied animals, forming a bridge across the interval between Chengjiang and Burgess.
The Burgess Shale became a World Heritage Site in 1980 and was later incorporated into the Canadian Rocky Mountain Parks. Today, visitors can see it only with a guide, after several hours on a mountain trail.
More than fifty fossil deposits like Burgess and Chengjiang are known, scattered across several ancient continents. In 2019, Fu Dongjing and colleagues at Northwest University reported another: the Qingjiang biota, from Changyang in Hubei, about 518 million years old. More than half the taxa represented among its four thousand-plus specimens were new. It contains abundant jellyfish, comb jellies, and delicate soft-bodied animals, and its composition differs greatly from that of Chengjiang at the same time.
No single fossil deposit stands for the entire ocean. Burgess and Chengjiang are two sites within a worldwide window when the chances of preserving soft bodies were unusually high. Such windows became markedly rarer in the middle and late Cambrian. Move to a different fossil deposit, and the “animal diversity” on view can change severalfold. One quantitative comparison found that Chengjiang recorded about 300 percent more taxa than the similarly aged Sirius Passet, and about 27 percent more than the classic Burgess quarry.
There is a smaller window too. Sweden's “stinkstones,” which we will visit in Episode Eleven, are limestone nodules from the late Cambrian. Dissolve them in acid and out come three-dimensional arthropods around a millimeter long, with even the bristles on their legs intact. They show us a level of life that neither Chengjiang nor Burgess can bring sharply into focus: the smallest animals in the crevices of seafloor mud.
The second route is the acquisition of skeletons. Within a very short span of geological time, many lineages independently developed mineralized hard parts, using different materials: calcium carbonate, calcium phosphate, silica.
Calcium carbonate skeletons alone are estimated to have arisen independently in animals at least twenty times or so. Once shells, plates, tubes, and exoskeletons existed, even ordinary shallow-marine limestone and shale could preserve recognizable fragments. What changed was not just how many animals there were, but how many could be seen. Even if the actual number of species rose steadily, the acquisition of skeletons could put a steep upward bend into the diversity curve.
Skeletons also introduced another way to be preserved. A trilobite molted several times during its life, leaving an empty shell on the seafloor each time. Many trilobite fossils in the rocks are really cast-off clothes. One animal could leave several “fossils.” Hard-shelled animals were not only more likely to be preserved; they were more likely to be preserved repeatedly.
The third route is the rock itself. In 2024, researchers compared North American stratigraphic and paleontological databases. From the end of the Ediacaran to the late Cambrian, the amount of marine sedimentary rock available for sampling increased more than fivefold. The correlation coefficient between fossil abundance and the amount of marine sedimentary rock exceeded 0.8. Episode Eleven will explore two ways to read that correlation.
We cannot yet disentangle how much these three routes contribute between them. There is no generally accepted correction factor that can divide a peak in diversity into “40 percent real radiation, 60 percent preservation.”
5 Four Curves
The set phrase “Cambrian explosion” entered the scientific literature only around the 1960s. It has never consistently referred to the same quantity.
At least four different curves are involved: when a group first appears, the numbers of species and genera, the extent of differences in body structure, and the complexity of ecology. Those four curves need not move in step.
Ecological complexity can be drawn quite directly. In 2008, American researchers used fossil gut contents, fossilized dung, bite marks, and the shapes of mouthparts to connect the food webs of Chengjiang and Burgess, link by link: who ate whom, who was eaten by whom, and how many levels of predators there were.
The networks that emerged were already quite close in structure to marine food webs today. More than five hundred million years ago, the seas had a thoroughly developed web of eating and being eaten.
Differences in body structure, known as morphological disparity, are the least intuitive of the four. Imagine one clothing shop selling a thousand kinds of shirt, and another selling only ten things: one or two kinds each of shirts, trousers, shoes, hats, and umbrellas.
The first shop has more kinds of merchandise. The second has greater differences among them. Suppose we score every kind of animal for a series of bodily features: whether it has segments, how many pairs of appendages it has, whether it is bilaterally symmetrical.
Each kind of animal then becomes a point in a multidimensional space. Disparity measures how widely those points spread. This is different from the number of species. Ten species that look completely unlike one another can occupy a vast space; ten thousand similar species may fill only a small patch.
In 1989, the American paleontologist Stephen Jay Gould published a book about the Burgess Shale, Wonderful Life. Its title was borrowed from the 1946 American film in which the protagonist sees what the world would have been like had he never been born. Gould argued that the early Cambrian produced an explosion of body structures so varied that their range was never equaled afterward. Those strange animals represented separate phyla, and all subsequent history had done was prune them away.
He also imagined rewinding the tape of life and playing it again: a different set of animals would survive. The book was enormously influential. Conway Morris later wrote a book specifically to challenge it, The Crucible of Creation. Most of those animals, he argued, could be accommodated within known phyla, and evolution, even on a replay, would repeatedly arrive at similar solutions.
A man who had studied the Burgess specimens with his own hands and a man who had told the whole world about them were arguing openly in print. One of Conway Morris's favorite examples was the eye. Image-forming eyes have evolved independently many times in animals. An octopus's eyes and a human's are similar in structure, yet each evolved separately.
In 1992, Briggs, Fortey, and Wills put the claim to a quantitative test. They subjected Cambrian and modern arthropods to the same morphological analysis. The range of disparity among modern arthropods turned out to be broadly comparable to that of their Cambrian counterparts. Morphology had evolved rapidly in the Cambrian, but it had not reached a maximum that could never be approached again. Debate then shifted to how “morphospace” should be calculated: which characters to choose and how to measure distances both affect the result.
Gould ended his book with an inconspicuous little animal from the Burgess Shale: Pikaia, a few centimeters long, with a flattened body and a structure resembling a notochord along its back. At the time, it was considered one of the earliest representatives of the chordates. Gould wrote that if Pikaia failed to survive when the tape was replayed, there would be no vertebrates afterward, and no us. Myllokunmingia and Haikouichthys, later discovered at Chengjiang, are older than Pikaia and more fishlike. The protagonist of that sentence can be replaced; its point remains.
At bottom, the disagreement between the two books concerns contingency and necessity. Gould stressed contingency: replay the tape and the outcome will be entirely different. Conway Morris stressed necessity: life has fewer routes available than we imagine, and on a replay it will reach similar destinations. Both sides can find examples in the Burgess fossils. The argument is still unfinished.
One by one, those “bizarre” animals found their places. Anomalocaris belongs to a branch of the arthropod stem group. Hallucigenia is a distant relative of velvet worms, which still live among leaf litter and rotting wood in tropical forests. Soft-bodied, with pairs of short, plump legs, velvet worms hunt by firing a stream of slime from each side of the head to glue small prey in place. Opabinia also lies along the arthropod stem. They became less strange because we found places to put them.
Another bias comes from the window labeled “still alive today.” Researchers call it the “push of the past.” In 2018, Britain's Budd and Mann pointed out that lineages surviving to the present are not a random sample of all past lineages. Those that expanded quickly early on and built up greater diversity had a better chance of lasting until today.
When we look back at the Cambrian through “all living phyla,” we naturally concentrate on the lineages that expanded most successfully at the time. Living phyla therefore seem to have appeared early and fast. No special biological mechanism is required for this. We need only notice that we are looking backward from the outcome.
The evidence pointing the other way is just as solid. Burrows and trails become increasingly complex, regardless of whether the animals had hard shells. Within tens of millions of years, a series of independent mineralizing systems appeared on several continents.
Trilobites appeared on several continents almost at once. Multiple deposits preserving soft bodies from about 518 million years ago already contain many deeply differentiated groups. These four findings come from four different kinds of record.
Burrows, mineralized hard parts, and carbonaceous soft bodies follow different routes into the record. They can still share biases in which rocks survive, become exposed, and get sampled; these are not wholly independent witnesses. But mineralization alone cannot account for the complexity seen across them. Mineralization can amplify a “first appearance”; it cannot, on its own, explain the rise in ecological complexity.
In 2019, researchers used trilobite morphological characters to estimate evolutionary rates across the entire Cambrian and found them broadly stable. This suggests that by the time trilobites appeared in abundance, the burst of rapid change responsible for the major differences in body plans had largely happened already. The “fast” part that needs explaining is pushed back into an earlier, more poorly recorded interval.
6 Why Then?
Why more than five hundred million years ago? Several sets of explanations each have hold of part of the answer.
First, the timing. When the Cambrian began, fewer than a hundred million years had passed since the end of the last Snowball Earth glaciation. After the frozen planet thawed, ocean and atmospheric chemistry swung wildly. The Ediacaran organisms we met in Episode Nine appeared in this post-thaw interval. Animals rose hard on the heels of a succession of global upheavals.
The first explanation is oxygen. As Episode Four discussed, claims about oxygen's role come in different strengths. The weakest says that it “made things possible.” In oxygen-poor waters today, the proportion of carnivorous animals is markedly lower. Large, active animals need enough oxygen.
The middle claim is that “a threshold was crossed around the beginning of the Cambrian.” A 2015 study inferred that the area of oxygenated seafloor expanded during the early Cambrian. The strongest claim is that “an oxygen pulse directly triggered the explosion.” This has the weakest evidence. The latest combined estimates put Cambrian oxygen at perhaps 10 to 40 percent of today's level, a very wide range. Animals themselves were also changing the oxidation state of the oceans. Causation can run the other way.
The evidence for predation is written on the fossils. The American paleontologist Loren Babcock examined large numbers of Cambrian trilobites and found many with healed bite injuries. Where a piece of the edge had been missing, it had subsequently grown back.
Among trilobites with scars, about 70 percent were injured on the right side. Some researchers have suggested that predators favored an attack from the right, or that trilobites tended to flee to the left. Fossilized dung at Chengjiang and Burgess also often contains fragments of trilobite shells.
Oxygen has another, less conspicuous use. Animal bodies are held together by a protein called collagen, found in skin, ligaments, and bone. A crucial step in making it requires molecular oxygen. As early as 1970, researchers proposed that large animal bodies need collagen scaffolding: without enough oxygen, the scaffolding cannot be built.
The second explanation is predation. Prey were finding ways to cope too. Some Cambrian trilobites could already curl their bodies toward the belly, rolling into a ball like a pill bug and hiding their soft undersides and legs inside their hard shells. By the Ordovician, this ability was very widespread.
Once predators existed, movement, senses, defenses, and skeletons all became useful, setting off an arms race. The drill holes in Cloudina shells lie below the boundary. Predation had begun long before; the Cambrian was when it escalated. The British researcher Andrew Parker proposed a “light-switch” hypothesis: eyes allowed hunters to see prey at a distance for the first time, and obliged prey, for the first time, to hide and put on armor. He made it the subject of a book, In the Blink of an Eye.
The third explanation is the changing seafloor. As Episode Nine described, large stretches of the Ediacaran bottom were covered by tough microbial mats. More and more animals burrowed through the mud, gradually turning a stable mat-covered surface into a soft, water-rich, mixed sediment bottom. Researchers call this the “agronomic revolution.”
The revolution unfolded in stages, and estimates still disagree. Some models suggest that the average depth of seafloor mixing in the early and middle Cambrian was only a fraction of a centimeter; others find a substantial mixed layer already present in the early Cambrian. Either way, it was a long way from the five to ten centimeters of mixing on the seafloor today. Animals were both cause and consequence: they burrowed and changed the bottom; the altered bottom then changed the animals.
This explains how a process can accelerate. Its beginning has to be sought elsewhere. It also had winners and losers. Animals that made their living lying on microbial mats, the Ediacaran way of life, were pushed into a few remaining corners. As Episode Seven noted, stromatolite diversity declined markedly in the late Precambrian, especially around 700–600 million years ago—not in one continuous, worldwide fall from a billion years ago. In many Cambrian seafloor settings, their territory shrank further.
The fourth explanation is the developmental toolkit. In 1978, Edward Lewis at the California Institute of Technology in the United States published work on fruit-fly mutations he had studied for decades. One mutation gave the fly two pairs of wings: the segment that normally bore balancing organs became another wing-bearing thoracic segment.
He received the Nobel Prize in 1995 for this work. A class of genes called homeotic genes controls the arrangement of an animal's body from head to tail. A mutation in a fruit fly can make a leg grow where an antenna should be.
Comparative genomics shows that these genes and the signaling pathways between cells were already present in the common ancestor of animals long before the Cambrian. Some are older than animals themselves. Eyes tell a similar story.
In 1995, Walter Gehring at the University of Basel in Switzerland performed an astonishing experiment on fruit flies. He switched on an eye-development gene in their legs and wings, and extra compound eyes grew on the legs and wings. More astonishing still, when he substituted the equivalent mouse gene, the flies still grew fruit-fly compound eyes.
Insect and mammal eyes look utterly different, yet the same ancient switch sets them in motion. Episodes Five and Eight encountered this same pattern: the parts come first, the uses change later. The genome alone cannot tell us when the change of use took place.
There was also an unusual abundance of phosphorus in the seas at the beginning of the Cambrian. Around Kunyang, near Kunming in Yunnan, a bed of earliest Cambrian phosphorite forms one of China's largest phosphate deposits. Meishucun is right beside it, and many of those small shelly fossils were preserved in the phosphorite. More nutrients meant more life the sea could support.
The fifth explanation is seawater chemistry. Episode Eleven will discuss the Great Unconformity in the Grand Canyon of the United States, where hundreds of millions of years of rock have left not a single page behind. In 2012, the American researchers Peters and Gaines proposed that, at the beginning of the Cambrian, seas swept far inland across continents weathered over a very long time. They washed huge quantities of ions, including calcium, phosphate, and potassium, into the oceans, suddenly making shells cheap to build. Chemistry can explain why shell-building became easier. It cannot explain eyes and legs.
Finally comes the synthesis: “there was no single trigger.” This is the position most likely to be right, and the one most easily turned into an after-the-fact checklist. Any observation that fails to fit can be patched with another factor. To be useful, it must specify in advance which factor should precede which event, in what setting its effect should be strongest, and what we should fail to see in its absence.
7 Molecular Clocks and Fossils
Darwin had already predicted this. In On the Origin of Species, he wrote that if his theory was right, then long ages must have passed before the lowest Cambrian beds were laid down, with the world already full of living things. We simply had not found their traces. More than a century later, molecular clocks put a number on that prediction.
Molecular clocks have to be calibrated with fossils. The oldest reliable fossil supplies a minimum age. The earliest trilobite, for instance, tells us that arthropods and their closest relatives had already parted company before then.
Place many such minimum ages on a tree and combine them with rates of genetic change, and we can estimate when branches with no fossil record diverged. Molecular clocks generally date animal origins earlier than fossils do. A study in 2004 placed the common ancestor of bilaterally symmetrical animals between about 650 million and 570 million years ago.
Even the first branching at the base of the animal tree remains disputed. For a long time, sponges were thought to be the earliest branch to split away: they have no nerves, no muscles, and the simplest organization.
Some genomic analyses, however, put comb jellies at the base. Comb jellies have nerves and muscles, and swim through the water on rows of cilia that flash with rainbow colors. In 2023, a study comparing the order of genes on chromosomes again supported comb jellies as the first branch. If that is right, nerves and muscles may have arisen more than once, or sponges may once have had them and subsequently lost them.
Definite bilaterally symmetrical animals already existed before the Cambrian. Kimberella and Yilingia, which we met in Episode Nine, were scraping food from the seafloor and crawling across it more than 550 million years ago.
Some molecular-clock studies place their central estimate for the common ancestor of the animal crown group more than seven hundred million years ago. A recalibrated study in 2024 brought it forward to the early Ediacaran instead. Most estimates predate the Cambrian fossils by tens of millions to more than a hundred million years, while their confidence intervals often span a hundred million years.
The clocks themselves may also run too far back. When animal lineages first split, small bodies and a rapid succession of generations could have made genetic change much faster than it was later.
Extrapolate backward using the later, slower rate, and the inferred split becomes too old. Working out how to allow rates to vary across the tree is one of the most demanding parts of molecular-clock research.
The size of the gap depends on which two endpoints are being compared. Set divergences dated to 650–600 million years ago against complex burrows at 538 million years ago, and the difference is sixty to a hundred and ten million years.
Compare them with unmistakable animal fossils at 521 million years ago, and the gap is eighty to a hundred and thirty million years. In the statement “molecular clocks put animals a hundred million years before their fossils,” the hundred million has been measured between two particular endpoints.
As Episode Nine mentioned, a 2024 study of an Ediacaran sponge-like fossil from South China placed it within the sponge crown group. Other clues also point to animals before the Cambrian. Beginning in 1998, abundant spherical fossils only a fraction of a millimeter across were found in phosphorite about six hundred million years old at Weng'an in Guizhou. Inside were dividing cells: two, four, eight, resembling the cleavage stages of animal embryos.
Whether they really are animal embryos has been debated for more than twenty years. Some researchers think they look more like the dormant stages of other single-celled organisms. Others have found a molecular fossil in rocks more than six hundred million years old that was thought to be made only by sponges, though that molecule too may have had another source. Animals probably existed before the Cambrian, but they were small and soft, leaving traces that were scarce and hard to recognize.
Small animals have a hard time leaving traces in any case. Scoop up a handful of mud from the seafloor today, and it will often contain hundreds or thousands of animals a millimeter or less in length: roundworms, copepods, tiny annelids. They make up most of the animals on the seabed, yet almost never become fossils. If most pre-Cambrian animals were little creatures like these, their absence from the record tells us little.
It is a little like developing an old-fashioned photograph. The image is there the moment the film is exposed, but remains invisible until it is immersed in developer and begins to emerge. Animal lineages may have been “exposed” long before. The Cambrian was when their image developed in the rocks.
A common reconciliation goes like this. The lineages branched early, when the animals were small, soft-bodied, few in number, and ecologically inconspicuous. Tens of millions of years later, body size, skeletons, mobility, and abundance rose together, and the lineages entered the ordinary fossil record. Molecular clocks constrain when lineages separated. The Cambrian record chiefly captures when those lineages became visible.
One fact is particularly interesting. If a group has hard parts that preserve well, fossils are more sensitive to its origin. In such cases, molecular estimates and the earliest fossils tend to agree more closely.
The largest disagreements cluster in the deep stem-group stages, before hard parts existed. It is not that both methods become more accurate when applied to groups with hard parts. The fossil side of the comparison has become more sensitive there.
8 Phase Transition, and Concealment
“Phase transition” is borrowed from physics. Water freezes at zero degrees Celsius. A tiny difference in temperature, and the arrangement of its molecules suddenly changes from disordered to orderly. Physicists call this abrupt change of state after a critical point has been crossed a phase transition. The analogy is tempting: the Cambrian, too, looks as though life on Earth crossed some critical point.
This series has used that term several times to describe particular moments: phase transition. “The Cambrian was a phase transition” sounds as though it explains something. Let us give the term as narrow a definition as we can, then ask whether the Cambrian qualifies.
In this series, a phase transition means a change in the unit of selection. Units that could previously replicate separately and bear fitness independently become integrated into a new, higher-level entity that reproduces as a whole. The lower-level units systematically lose their independence. There is no “large,” “fast,” or “important” in the definition. It asks just one question: what is being selected?
The mitochondrion in Episode Five is one example. A free-living bacterium was incorporated into another cell and lost its ability to live independently; thereafter, the combined entity was copied and selected as a whole. Multicellularity in Episode Eight is another. Cells ceased to go their separate ways and formed a body that reproduced as a whole, something that happened forty-five times on the broadest count. This criterion, then, does not pick out “spectacular events.” It picks out a particular structural change, one that may be quite common.
We will meet another candidate later in the series: eusociality in Episode Sixteen. In ant and bee societies, most individuals forgo reproduction, and the colony as a whole is selected like a body.
The concept comes from a book coauthored in 1995 by Britain's John Maynard Smith and Hungary's Eörs Szathmáry, The Major Transitions in Evolution. Maynard Smith is the biologist we met in Episode Six who began by designing aircraft. They listed several major transitions in life's history: molecules enclosed by membranes within a common replicating system; prokaryotic cells to eukaryotic cells; asexuality to sexuality; single cells to multicellular bodies; solitary organisms to social groups with a division of labor; and human language.
Their list is broader than the criterion used here, since it also includes changes in how information is transmitted. The Cambrian radiation of animals is not on it.
We can break the criterion into three questions that can be checked. Did formerly independent reproducing units lose the ability to reproduce on their own? Did a new reproducing entity emerge, bearing fitness as a whole? Were conflicts between lower-level units systematically suppressed by some mechanism? Only if all three conditions hold can we call it a shift in the level of selection.
Episodes Twelve and Fourteen will use the same yardstick to measure the move onto land and the end-Permian mass extinction. Apply it to the Cambrian, and none of the three conditions holds.
Multicellular animals already existed before the Cambrian. The changes the Cambrian records all took place within the existing level of individual animals: skeletons, appendages, eyes, larger bodies, predation, burrows. Food webs and ecosystems grew more complex; the fates of different animals became increasingly intertwined. But the ecosystem as a whole did not become something that reproduced as one.
Another substantial radiation followed the Cambrian. During the Ordovician, about 480 to 450 million years ago, the number of marine animal genera multiplied severalfold. Brachiopods, bryozoans, corals, nautiloid cephalopods, and others diversified. Researchers call it the Great Ordovician Biodiversification Event.
It added more diversity than the Cambrian did, yet few people call it an “explosion.” Where a label lands depends partly on which interval was noticed first, and which was first made into a story.
Look the other way, and the moments when selection really changed levels were often quiet. As Episode Eight described, cells gathered into bodies on roughly forty-five separate occasions. Most left only inconspicuous filaments and clusters that are hard to recognize in fossils. How spectacular an event is and whether it changes the level of selection are separate questions.
Three other things changed: existing animal individuals came to be organized into more varied body structures; ecological relationships among animals grew more complex; and visibility in the fossil record rose abruptly. Use a different criterion, ecological complexity, for instance, and the Cambrian ranks very high indeed.
The term has two uses, and they are not interchangeable. If “phase transition” simply means that many interacting variables cross thresholds and the overall state changes rapidly, the Cambrian can perfectly well be described that way. If it requires a change in the unit of selection, we currently have no such evidence for the Cambrian.
This is a little like saying that a patient's “constitution has changed.” It sounds like a diagnosis, but merely restates the symptoms. Which organ is actually at fault remains unanswered.
Quietly exchange the first meaning for the second, and the word turns from explanation into concealment. Once “this was a phase transition” has been said, the question seems to have been answered. But which question? Did the unit of selection change? Did several variables speed up together? Or did our camera suddenly begin admitting more light during this interval?
Other grand terms conceal the same risk. Call something a “revolution,” an “explosion,” a “leap,” and it is easy to feel that something has been explained. The more a grand term can encompass, the more we need to ask exactly which variable it names.
What this series calls a construct comes down to drawing a line: what falls inside counts; what lies outside does not. The line may be drawn by people, as with a set of criteria, or by nature, as with a membrane. The construct in this episode is the line drawn by the label “Cambrian explosion.” Inside it lies the explosion we can see.
What a construct cannot contain and cannot eliminate is what this series calls the remainder. Here, the remainder is the part we cannot cleanly separate out. Of the diversity visible in the Cambrian, how much comes from genuine radiation? How much from skeletons allowing ordinary rocks to start taking in fossils? How much from a worldwide window for preserving soft bodies? How much from a fivefold increase in marine sedimentary rock? The different modes of preservation act entirely differently on soft bodies, hard parts, microscopic organisms, and burrows. There is no common denominator.
Put these together, and the Cambrian looks roughly like this. Animals had quietly separated into several branches at an earlier time, small and soft-bodied. Over tens of millions of years, multiple lineages independently acquired shells and plates, eyes and appendages that could grasp and run. They dug ever more complex tunnels through the mud, while predators chased and prey fled.
Meanwhile, the sea spread over the continents. Rocks capable of holding fossils multiplied severalfold, and there was a rare interval when even soft bodies could be pressed intact into mud. Real change and visible change overlapped in the same span of time, together making that “explosion.”
The spike is still driven into Fortune Head. More than three meters below it lies the worm's burrow it was meant to mark. According to recent reexaminations, the burrow's own oldest reliable age may be five or six million years younger than the boundary it marks. The criteria are still sifting. The spike has not been driven in anew.