The Mammal Line

The Question

How did the descendants of a small Carboniferous amniote come to include both an albatross and a human being?

The previous entry traced life from the deep freeze of the Cryogenian to the appearance of vertebrates that could lay their eggs on dry ground. From here, the arc narrows. Two lineages, both descended from those early amniotes, will spend the next two hundred and fifty million years dividing the world's land vertebrates between them. One will rise spectacularly, dominate the planet for more than a hundred and thirty million years, suffer the worst ecological catastrophe since multicellular life began, and re-establish itself in radically different form. The other will live in its shadow for nearly the whole of that time, then inherit the world. A small subset of the second lineage will, much later, climb into trees and look out from them with forward-facing eyes — and from that vantage, eventually, write these words.

This is the story of that division and the events that decided it.

The Two Branches

The split appears to have occurred in the late Carboniferous, around 318 to 315 million years ago, in the equatorial coal swamps of what was then a single supercontinent assembling toward Pangaea. The two daughter lineages are now distinguished by a pair of holes in the skull behind the eye socket. The arrangement, position, and number of these temporal fenestrae are a key diagnostic feature of the early amniotes that have left fossils, and they remain the marker by which the two halves of the modern tetrapod world are most often distinguished.

The lineage with one pair of openings, low on each side of the skull, became the synapsids. The lineage with two pairs, or with the openings positioned higher up, became the sauropsids. Sauropsids are what readers today would loosely call reptiles, plus the birds — turtles, lizards, snakes, crocodiles, dinosaurs, and the surviving descendants of the dinosaurs that still nest in hedgerows. Synapsids are the mammals, plus an enormous extinct radiation that filled out most of the Permian.

The earliest known animals firmly on the synapsid side of the split are Archaeothyris florensis and the more poorly preserved Echinerpeton intermedium, both described from late-Carboniferous rocks at Florence, in Cape Breton, Nova Scotia, not far from the locality where Hylonomus itself was recovered. They lived around 310 million years ago, in the late part of the Moscovian stage, on current stratigraphic calibration of the host rocks ^[1]. They are not impressive creatures by modern standards. They are unmistakably the kind of animal from which the rest of this entry's cast descends.

For the first stretch of their history, both branches were sidelined. Through the late Carboniferous, the dominant tetrapods of the equatorial swamps remained the large amphibian-grade vertebrates of the previous entry: Eryops-like predators, hippo-sized waterside hunters, and a variety of forms with no modern analogue. The early amniotes were small and ecologically marginal. What changed that, in the Permian, was the climate.

The Sails And The Teeth

The Carboniferous coal forests collapsed at around 305 million years ago in an episode now known as the Carboniferous Rainforest Collapse, driven by a shift to drier and more seasonal climates as Pangaea continued to assemble. The amphibian-grade vertebrates, dependent on water for reproduction, suffered disproportionately. The amniotes, which did not need standing water to breed, did not. The early Permian — running from roughly 299 to 252 million years ago — opens with the synapsids expanding into ecological roles that the disappearing amphibian-grade animals had occupied.

The first wave of synapsids is the group informally called pelycosaurs — a paraphyletic collection of basal synapsid lineages that includes the most famous early-Permian animal of all, Dimetrodon. Pelycosaurs were not yet mammals in any meaningful sense. They were sprawling, lizard-shaped quadrupeds with limbs splayed sideways from the body, and their reproduction, metabolism, and skin were all closer to those of a modern reptile than to those of any modern mammal. What they brought to the world was teeth. For the first time among amniotes, the teeth in a single jaw were differentiated — small incisor-like teeth at the front for nipping, larger canine-like teeth for puncturing, and smaller cheek teeth behind for processing. Dimetrodon itself, with its enormous dorsal sail supported by elongated vertebral spines and its powerful canines, was the apex predator of its early-Permian ecosystem and probably the first vertebrate to occupy that role on dry continental ground.

The pelycosaurs did not last. Sometime in the late early Permian, around 280 million years ago, a more derived group of synapsids — the therapsids — appeared, evolved from one of the pelycosaur lineages, and progressively replaced their ancestors over the following several million years ^[2]. Therapsids were better adapted to active life on land. Their limbs were drawn in beneath the body rather than splayed out, their gait was more upright, their teeth more elaborately differentiated, and the bones of the lower jaw began the long simplification that would, much later, leave the mammalian jaw as a single bone on each side. By the late Permian, therapsids had radiated into a great variety of forms: tusked herbivores, sabre-toothed carnivores, dog-sized burrowers, and the small, slender, mammal-like cynodonts from which the mammals themselves would eventually derive. They were not mammals — fur, milk, and warm-bloodedness in the modern sense almost certainly came later, in scattered features, across multiple cynodont lineages — but they had the unmistakable look of animals on the way to becoming something else.

For roughly forty million years, the synapsids were the dominant land vertebrates on Earth. The branch that would eventually produce the dinosaurs — the sauropsids — remained relatively quiet through this stretch, diversifying in the background but not yet challenging the synapsid hold on terrestrial ecosystems. Then, at the very end of the Permian, the world ended.

The Great Dying

The boundary between the Permian and the Triassic is the most catastrophic moment recorded anywhere in the fossil history of complex life. The formal boundary is now placed at 251.902 ± 0.024 million years ago, with the principal pulse of the extinction itself constrained, by high-precision uranium-lead dating of volcanic ash beds in the Chinese sections at Meishan that bracket the horizon, to an interval of about sixty thousand years between roughly 251.941 and 251.880 million years ago ^[3]. Around eighty to ninety per cent of marine species and the majority of terrestrial vertebrate species disappear from the rock record across this interval. The event is now informally referred to as the Great Dying. Geologically, the pulse was a flash; biologically, it was an epoch.

The principal cause is now broadly settled. Coincident with the extinction, a vast outpouring of basaltic lava and underlying intrusive sills was emplaced across what is now western Siberia. The Siberian Traps large igneous province covered millions of square kilometres at its peak and represents one of the largest volcanic events in the planet's history. Recent work has shown that the most lethal phase of the eruption was not the surface lava flows but the intrusion of magma sheets into the surrounding sedimentary rocks, including thick deposits of coal and evaporites, which were cooked by contact metamorphism and released vast quantities of carbon dioxide, methane, and halogenated compounds into the atmosphere ^[4]. The atmospheric loading drove rapid global warming, ocean acidification, and the spread of oxygen-poor and sulphide-rich conditions through much of the marine water column. Coral reefs collapsed. Marine invertebrate diversity reached its lowest point in the entire Phanerozoic. On land, the diverse late-Permian therapsid faunas that had ruled for forty million years were almost entirely wiped out.

A handful of synapsid lineages came through. One of them was a small, burrow-dwelling cynodont group — animals roughly the size of a modern weasel, with the early hints of mammalian dentition and posture. They are not the ancestors of any specific later lineage that can be uniquely identified, but they are members of the kind of group from which the next chapter of the synapsid story would unfold. The sauropsids fared somewhat better, in part because some of their lineages had already moved into ecological niches — semi-aquatic, small and burrowing, partly arboreal — that buffered them against the worst of the surface catastrophe.

The post-extinction world of the Early Triassic was bleak and depopulated. The reefs were gone. Coal swamps would not return for nearly ten million years — a stretch sometimes called the coal gap, during which the planet's vegetation was so impoverished that no thick organic deposits formed anywhere. The recovery of land ecosystems took roughly five to ten million years to begin in earnest, depending on the metric. When it did, the dominant terrestrial vertebrates were no longer the synapsids.

The Triassic Recovery

The Triassic — running from 251.9 to 201.4 million years ago — opens with the surviving sauropsids in a position to expand. Within the sauropsid branch, a particular subgroup, the archosaurs, took the lead. Archosaurs split early in the Triassic into two daughter lineages whose modern survivors are very different: the pseudosuchians, the lineage that today contains the crocodiles and alligators, and the avemetatarsalians, the lineage that produced the pterosaurs and the dinosaurs.

For most of the Triassic, the pseudosuchian lineage was the more successful. It produced large, often heavily armoured, often bipedal predators and herbivores that dominated terrestrial ecosystems through the middle and late Triassic. The early dinosaurs, by contrast, were small, lightly built, bipedal carnivores and omnivores at the margins of these ecosystems. Eoraptor lunensis and Herrerasaurus ischigualastensis, both from late-Triassic rocks of the Ischigualasto Formation in Argentina, are among the oldest unambiguous dinosaurs in the rock record, dated to roughly 231 to 228 million years ago ^[5]. Slightly older Brazilian dinosaurs from the Santa Maria Formation, including Buriolestes schultzi, push the well-supported record back to around 233 million years ago. There are older candidate fossils still — Nyasasaurus parringtoni from Tanzania, dated to roughly 243 million years ago, has been argued by some workers to be the earliest dinosaur, but the available material is fragmentary and the animal is now more often placed as a basal dinosauromorph just outside Dinosauria proper. The placement remains contested.

What changed the balance was another extinction. At the end of the Triassic, around 201.4 million years ago, a second mass extinction — driven by the volcanic eruptions of the Central Atlantic Magmatic Province, associated with the early break-up of Pangaea — eliminated about three-quarters of marine and terrestrial species. The pseudosuchians were almost entirely wiped out, except for the lineage that would become the modern crocodilians. The dinosaurs, for reasons that are still actively debated, came through largely intact. From the Early Jurassic onward, they radiated into the ecological space the pseudosuchians had vacated, and for the next hundred and thirty-five million years, they were the dominant land animals on Earth.

While this was happening on the sauropsid side, the synapsid side was producing a much smaller revolution. Within the cynodonts, a series of innovations — increasingly differentiated teeth, a secondary palate, an expanded brain, refinement of the jaw bones — accumulated across the Late Triassic and Early Jurassic in animals progressively closer to the modern mammalian form. By the end of the Triassic, the Mammaliaformes — the group containing the true mammals and their nearest extinct relatives — were established. Morganucodon, a mouse-sized insectivore from the latest Triassic and earliest Jurassic of Wales, China, and elsewhere, dated to around 205 million years ago, is the canonical benchmark mammaliaform ^[6]. A more contested claim has been made for an older animal, Brasilodon quadrangularis, from the Late Triassic of Brazil at around 225 million years ago: a 2022 study of Brasilodon's tooth-replacement pattern argued that it shows the diphyodont (two-set) dentition characteristic of mammals, and on that basis claimed it as the earliest known mammal ^[7]. The interpretation is not universally accepted, and as of 2026 most workers continue to treat Brasilodon as a non-mammalian eucynodont near the boundary, with Morganucodon-grade animals as the more secure benchmark for the early mammaliaforms. As with all such "earliest" claims, the benchmark is a moving target.

The Long Shadow

For the next hundred and forty million years — almost exactly half the entire stretch since the synapsid–sauropsid split — mammals lived in the shadow of the dinosaurs.

This is a long time. To put it in perspective: the whole span of recognisable human evolution, from the earliest hominins to today, is about seven million years. The Mesozoic shadow under which the early mammals lived is more than twenty times that long. Through the Jurassic and Cretaceous, dinosaurs filled almost every large terrestrial vertebrate niche. The largest were the long-necked sauropods such as Argentinosaurus and Patagotitan, which probably exceeded seventy tonnes — among the largest land animals ever to have existed. The fastest were the ostrich-like ornithomimids. The most heavily armoured were the ankylosaurs and stegosaurs. The dominant predators were the theropods — the bipedal, mostly carnivorous group that included Allosaurus, Tyrannosaurus, and the small, agile dromaeosaurs.

In this world, mammals were small. The largest known Mesozoic mammals reached the size of a modern badger; the great majority were mouse-sized to rat-sized. They were, on the available evidence, mostly nocturnal — a long stretch of obligate night activity that has been argued, on the basis of the genetic and anatomical features of their modern descendants, to have left lasting marks on mammal vision, hearing, and olfaction. This nocturnal bottleneck, as it is sometimes called, is a plausible explanation for why most modern mammals have relatively poor colour vision compared with most reptiles and birds, and unusually keen hearing and smell. It is also a plausible explanation for the development of fur as insulation in the absence of solar warming, and of an elaborate inner ear capable of detecting insect movement at night.

The Mesozoic mammals were nevertheless evolutionarily busy. The three modern groups — monotremes (egg-laying mammals such as the platypus and echidnas), marsupials (pouched mammals such as kangaroos and opossums), and placentals (the much larger group containing every other modern mammal) — all originated during this stretch. The earliest fossil widely interpreted as a eutherian — a member of the lineage that contains the placentals and their nearest relatives — is Juramaia sinensis, a small shrew-like animal from north-east China, dated to about 160 million years ago and described in 2011 ^[8]. The phylogenetic placement of Juramaia has been revisited in subsequent cladistic analyses, and a few workers now argue it sits just outside the eutherian crown, but the 160-million-year age and the broad interpretation remain the standing benchmark in 2026. Genomic analyses calibrated against the fossil record, including the comprehensive 2023 re-analysis of 241 placental mammal genomes by Foley and colleagues, place the divergence between marsupials and placentals around 160 million years ago and the deepest divergences among the modern placental orders in the Late Cretaceous, before the K–Pg boundary ^[9]. The intraordinal radiations — the splitting of placentals into the rodents, primates, carnivorans, ungulates, bats, and so on — happened almost entirely after the Cretaceous ended.

Birds Within Dinosaurs

The most consequential development on the dinosaur side of the divide, viewed from a 2026 perspective, was the appearance of birds.

Birds are not the surviving cousins of the dinosaurs; they are dinosaurs. The lineage that today wakes up at dawn outside any reader's window is a subset of the theropod dinosaurs — specifically of the small, lightly built, often feathered theropods that included the dromaeosaurs and their close relatives. The transition is one of the best-documented in the entire fossil record.

The first specimen recognised as transitional was Archaeopteryx lithographica, recovered from the Late Jurassic Solnhofen Limestone of Bavaria, Germany, beginning in 1861 ^[10]. Archaeopteryx dates to approximately 150 million years ago and looks, to a modern eye, like a magpie-sized animal with a long bony tail, toothed jaws, clawed forelimbs, and unmistakable asymmetric flight feathers preserved as impressions in the rock. It was difficult, at the time of its discovery only two years after the publication of On the Origin of Species, to interpret as anything other than the kind of intermediate form Darwin's theory predicted but had not yet been able to point to.

The much fuller picture came more than a century later, from a region of north-east China called Liaoning. From the mid-1990s onward, a series of fossil deposits in the Yixian and Jiufotang Formations — Early Cretaceous sediments dated to between roughly 130 and 120 million years ago — produced an extraordinary census of feathered theropod dinosaurs, preserved with skin, feathers, and even traces of pigment cells. Sinosauropteryx prima, described in 1996, was the first non-avian dinosaur preserved with unambiguous filamentous body coverings. It was followed by Caudipteryx, Microraptor (a four-winged glider), Anchiornis, Yutyrannus (a tyrannosauroid covered in down-like feathers), and many more. Within roughly a decade, the Liaoning material had transformed the relationship between birds and dinosaurs from a respectable scientific hypothesis into one of the best-supported evolutionary continuities known. Feathers, it turned out, were not a bird invention. They were a theropod feature that had been around for tens of millions of years before any of the lineages that bear them today learned to fly. By the time Archaeopteryx arrived in the Late Jurassic, feathered dinosaurs had already existed for a long time. Birds, in the modern sense, are simply the one feathered theropod lineage that survived what came next.

The Flowers

A second major Mesozoic transition unfolded on the plant side of the world's surface. Through the Triassic and most of the Jurassic, the dominant land plants were the gymnosperms — the conifers, cycads, ginkgoes, and their relatives, all of which reproduce via exposed (literally "naked") seeds. They built the great forests under which the dinosaurs lived for most of the Mesozoic.

Sometime in the Early Cretaceous, a new lineage of plants became visible in the rock record: the angiosperms, or flowering plants ^[11]. Pollen of probable angiosperm affinity appears around 130 million years ago. The first widely accepted body fossils — Archaefructus from the Yixian Formation of Liaoning, dated to about 125 million years ago, and Montsechia vidalii from Spain, of comparable age — are small herbaceous aquatic or semi-aquatic plants. Within roughly forty million years, however, the angiosperms had exploded into a position of ecological dominance: they had colonised most terrestrial habitats, evolved trees with broad deciduous leaves, formed elaborate co-evolutionary relationships with insects, birds, and mammals, and produced the fruits, nuts, and grasses on which most modern terrestrial vertebrate ecosystems depend. By the close of the Cretaceous, flowering plants were everywhere; by the Cenozoic, they would be the substrate on which the entire mammalian radiation took place. The grasses on which modern hoofed mammals graze, the fruit on which modern primates feed, the seeds on which modern rodents and birds depend — all of them were assembled, ecologically, in the second half of the Mesozoic.

The Impact

The Mesozoic ended with the largest, sharpest, best-documented mass extinction in the fossil record. The boundary is now formally placed at 66.043 ± 0.043 million years ago at the two-sigma level. At that boundary, an asteroid roughly ten to fifteen kilometres across struck the shallow tropical sea covering what is now the northern Yucatán peninsula in Mexico. The impact crater — Chicxulub, named after the modern village near its centre — is about a hundred and eighty kilometres in diameter and is buried beneath younger sediments, but has been mapped in detail by gravity and seismic surveys ^[12].

The energy released by the impact was on the order of one hundred million megatons of TNT — more than six billion times the energy of the Hiroshima bomb. The immediate consequences in the surrounding region were catastrophic: a fireball, a megatsunami crossing the Gulf of Mexico, the ejection of vaporised rock and impact glass into ballistic trajectories on a continental scale. The longer-term consequences were planetary. Pulverised rock and soot from continent-scale wildfires, lofted into the upper atmosphere, blocked sunlight worldwide for months to years. Surface temperatures plunged. Photosynthesis on land and in the surface ocean was severely curtailed. The food chain collapsed from the bottom up.

A subsidiary contributor to the end-Cretaceous extinction has been argued in the recent literature: the Deccan Traps eruptions in what is now western India, which were already underway at the time of the impact, may have stressed the global ecosystem in the run-up to the Chicxulub event and affected the pace of recovery afterward. The relative contributions of the asteroid and the volcanism are still actively debated, but the timing of the principal extinction pulse is unambiguously synchronous with the impact, to within about thirty thousand years.

The effect on the world's vertebrates was selective and severe. All non-avian dinosaurs disappeared. So did the pterosaurs, the marine reptiles called mosasaurs and plesiosaurs, the ammonites, and a substantial fraction of marine plankton. Among the survivors were the crocodiles, the turtles, the lizards and snakes, some birds (the modern lineage descended from small ground-dwelling avian dinosaurs that came through), some fish, and — importantly — the mammals. Of the mammal lineages alive at the time, only a fraction made it across, but the fraction that did included representatives of all three modern groups: monotremes, marsupials, and placentals.

The world entered the Cenozoic — the era running from 66 million years ago to the present — almost empty of terrestrial vertebrates above about twenty-five kilograms in body mass, with the exception of a handful of ectothermic and semi-aquatic survivors such as the crocodiles and the freshwater turtles.

The Radiation

What followed is the clearest large-scale example of an adaptive radiation anywhere in the fossil record.

For more than a hundred and forty million years, the mammals had occupied the small, fast, nocturnal, often arboreal niches at the edges of dinosaur-dominated ecosystems. With the dinosaurs gone, every large-bodied terrestrial niche on the planet was vacant. Within roughly ten million years of the impact, mammals had begun to fill them. Within twenty million years, mammals included rhinoceros-sized herbivores, horse-like running animals, hoofed grazers, large burrowing diggers, semi-aquatic forms, fully aquatic forms (the ancestors of modern whales and dolphins moved into the seas during the Eocene), arboreal forms, and predators ranging from weasel-sized to bear-sized. The radiation drew on lineages that were already established by the end of the Cretaceous — the placental mammal interordinal divergences, as the genomic analysis of Foley and colleagues showed in 2023, had largely happened before the impact — but the intraordinal radiations, the splitting of orders into the diversity of modern families and genera, took place almost entirely in the Paleocene and Eocene aftermath of the K–Pg event ^[9].

The mammals were not unique in radiating. Birds — the surviving avian dinosaurs — also radiated rapidly into vacated niches, producing the modern groups of ducks, songbirds, hawks, and the rest within roughly the same window. Flowering plants continued their Cretaceous expansion into a world without dinosaurs. Insects, which had also come through the impact in reduced but representative form, recovered with the plants. The reorganised terrestrial ecosystem of the early Cenozoic was, in its broad shape, recognisably the world of 2026: forests of broadleaved trees, grass-eating mammals on open plains, fruit-eating mammals and birds in the canopy, predators of various sizes, and underneath it all the slow biochemical recycling of soil microorganisms and fungi.

Among the mammalian lineages that began to fill the vacated niches, one is of particular interest to anyone reading this entry.

Into The Trees

The mammals that would eventually become the primates originated at the very base of the Cenozoic, perhaps even before it.

The earliest fossils that can plausibly be called primate relatives belong to the genus Purgatorius, recovered from earliest-Paleocene sediments in Montana and described in detail by Wilson Mantilla, Chester, and colleagues in 2021 ^[13]. Purgatorius fossils from that study are dated to within about 105,000 to 139,000 years after the Cretaceous–Paleogene boundary — making Purgatorius one of the very earliest small mammals to fill an arboreal niche after the impact. The animals were the size of a modern shrew and ate insects and fruit. They were members of the Plesiadapiformes, an extinct group of small, often squirrel-like, omnivorous mammals usually placed at or just below the base of the primate stem in modern phylogenies. Whether plesiadapiforms count as primates depends on how one draws the line between "primate" in the strict modern sense and the broader cluster of arboreal placental mammals from which the modern primates derive. By the looser conventions, Purgatorius and its relatives are the earliest known primates. By the stricter conventions, they are stem-primate forms — the kind of animal from which the primates emerged, but not yet themselves on the main primate trunk.

The first animals that almost everyone agrees to call primates appear about ten million years later. Teilhardina, a tiny mouse-lemur-sized primate known from earliest-Eocene rocks on three continents (North America, Europe, and Asia), dates to approximately 56 to 55 million years ago and seems to have spread rapidly during the Paleocene–Eocene Thermal Maximum — a brief, intense, globally warm episode at the boundary between the two epochs ^[14]. Teilhardina and its near-contemporaries had the diagnostic primate features: large forward-facing eyes capable of stereoscopic vision, grasping hands and feet with nails rather than claws on at least the first digit, and a lifestyle dependent on moving accurately through the small branches of the canopy of the broadleaved Eocene forests then spreading across the Northern Hemisphere. Within the early Eocene, the primates split into two main branches: the strepsirrhines (the lineage leading to the modern lemurs and lorises) and the haplorhines (the lineage that would later produce the tarsiers, the monkeys, the apes, and ourselves).

The Eocene primates were not yet our direct ancestors in any close sense. The lineage that leads to humans is one branch within the haplorhines, which itself splits later into the New World monkeys, the Old World monkeys, the apes, and, much later still, the bipedal hominins. But the basic primate body plan — small, tree-dwelling, social, fruit- and insect-eating, dependent on accurate vision and dexterous hands — was now in place.

What Follows

The two and a half hundred million years covered in this entry took life on Earth from the small lizard-like amniotes of the late-Carboniferous swamp forests to a Cenozoic world of grass plains, broadleaved forests, songbirds, primates in the canopy, and whales offshore. The most consequential events of this stretch were the Permian–Triassic extinction that destroyed the synapsid-dominated terrestrial world; the rise of the dinosaurs in the Triassic and their hundred-and-thirty-five-million-year dominance of the Mesozoic; the appearance of birds within the theropod dinosaurs; the rise of the flowering plants; the Chicxulub impact that ended the dinosaur world; and the explosive Cenozoic radiation of the mammals into the ecological space the dinosaurs had vacated.

Within that radiation, a small group of arboreal placentals — the early primates — established a body plan and a way of life in the trees of the Eocene forests. The forward-facing eyes, the grasping hands, the social groups, the long childhoods: these are features of life in the canopy, refined over tens of millions of years before any primate ever climbed back down.

The next entry follows what happens when one branch of those primates does climb down. It will cover the earliest hominins, the move to bipedal walking, the appearance and expansion of the genus Homo, the emergence of language and complex tool use, the spread of Homo sapiens out of Africa, and the genetic record of recent common ancestry preserved in the mitochondria and Y chromosomes of every person alive today. From that point on, the story will be one we can recognise as our own.

The world had been remade for the mammals. The question of what one particular kind of mammal would do with that world is the subject of what comes next.