Coming Ashore: Turning an Environment’s Remainder into Building Material

1 A Set of Footprints

In January 2010, Nature published a set of footprints. They came from an abandoned quarry at Zachełmie, in the Holy Cross Mountains of southeastern Poland. The quarry had produced dolomite, and the prints were stamped into the bedding surfaces of its successive layers. They were discovered by Niedźwiedzki, a young paleontologist at the University of Warsaw, who published the research with Ahlberg of Uppsala University in Sweden.

Tracks ran across the rock. There were digits; forefeet could be distinguished from hind feet; and the feet had moved in alternation. In the best-preserved trackway, the gait approached a diagonal sequence: left forefoot and right hind foot moving together, as in a lizard today. Between the two rows of prints, there was no furrow left by a dragging belly. This animal had held its body off the ground as it walked.

That matters. When a fish uses its fins to propel itself along the bottom, its body generally stays close to the substrate. In water, buoyancy carries most of your weight for you. On shore, the buoyancy is gone. Your weight has not changed at all.

A fish a meter long can hover in water with hardly any effort. Put it on shore, and it cannot even lift its head. Holding a body clear of the ground requires limbs, a shoulder girdle and pelvis that can bear weight, and a backbone that will not collapse.

The footprints varied in size. Estimates based on stride length put the smaller trackmakers at about forty to fifty centimeters long. The largest individual footprint was about twenty-six centimeters wide, suggesting that its maker may have exceeded two meters in length. In some prints, small triangular pits marked the tips of the digits.

The shock was their age. These footprints were about 390 million years old, more than ten million years older than the earliest tetrapod bones known at the time. Textbooks had a famous procession: Eusthenopteron, Panderichthys, Tiktaalik, Acanthostega, Ichthyostega, each more tetrapod-like than the last, each later in time. Fins became legs by degrees; water receded little by little behind them. Tiktaalik, the one that looked most like a creature halfway between fish and land animal, lived about 375 million years ago. These footprints came before the whole procession.

Footprints are made by feet. A foot cannot be younger than its own footprint. Yet in the fossil record, the footprints came first, and the feet came later.

The prints were dated this way. Their own layer contained no minerals that could be dated directly. About fifteen meters above it lay a horizon of tiny fossils belonging to a known interval of geological time, and that interval’s absolute age had been established elsewhere by radiometric dating.

On the timescale used in 2010, this worked out to about 395 million years ago. When the timescale was revised in 2015, the same interval was calibrated to about 390–391 million years ago. The rock layers had not moved. The scale, and the way it was read, had been updated together.

Challenges followed. In 2015, the American researcher Lucas reexamined Devonian tetrapod tracks from various localities. He argued that the Zachełmie impressions varied too much in size and were too irregularly arranged to qualify as reliable footprints. There were also fossil droppings among them, and some of the pits might have been made by fish rooting around on the bottom for food or excavating nests.

Fish really can “walk.” In 2011, researchers at the University of Chicago filmed an African lungfish taking alternating steps along the bottom with a pair of long, slender pelvic fins, lifting its body clear of the bottom as it advanced.

“Alternating left and right” alone does not prove the existence of legs with digits. The original authors’ side replied that lungfish fins were thin as whips and had no true digits: they could not make those deep, separate digit impressions. Nor could feeding pits explain several trackways in which forefoot and hind-foot prints repeated regularly. The dispute concerned a handful of the best trackways. Neither side treated every pit in the quarry as a genuine footprint.

One of the most interesting results came from Ichthyostega. In 2012, researchers at Britain’s Royal Veterinary College scanned its skeleton into a computer, then rotated the joints one by one to find each joint’s maximum range of motion.

Their conclusion was that this animal, one of the earliest known tetrapods, could not have taken the alternating steps of a lizard. It was more likely to have hauled itself forward on its forelimbs, dragging the rear half of its body behind, rather as a seal comes ashore. Mudskippers on the mudflats of southeastern China move this way today. After the tide goes out, they prop themselves up on a pair of stout pectoral fins and hop across the mud in pursuit of insects. They also puff out their cheeks with a mouthful of water to keep their gills moist.

There are two completely opposite ways to read this result. One says: if even the earliest tetrapods could not walk that way, the Zachełmie footprints cannot have been made by tetrapods.

The other says: the animals that made those footprints were better walkers than the earliest tetrapod skeletons in our possession, and we have not found their bones yet. The fossil record itself cannot decide between these readings. A 2025 paper on Australia’s earliest reptile tracks still used the Zachełmie trackways as a constraint on when limbed tetrapods had appeared.

However the dispute ends, it exposes a more general problem. Footprints and bones require different conditions for preservation. An animal can leave hundreds of footprints on a surface of wet mud without dying there even once.

It might die in a river channel or in the middle of a lake, where its body can be buried but footprints are rarely preserved. The footprint-bearing layers at Zachełmie contain almost no bones. The rocks there were initially interpreted as coastal intertidal mudflats. When the original team revisited them in 2018, they concluded that the setting was more likely a series of intermittent shallow lakes: water rose and fell; mud surfaces alternated between wet and dry. Whether tetrapods first ventured about beside the sea or beside lakes is itself still being reconsidered through these footprints.

Episode 11 called the fossil record a sieve. Here there are two sieves: one for footprints and one for bones, their openings facing different places. Arrange what these two sieves have caught in chronological order, and what you get is the earliest moment each of two sampling processes allows you to see. The actual order in which things happened must be established separately.

2 A Timetable for Coming Ashore

Coming ashore was not a journey completed by a single procession, either. Set the records of life on land in chronological order, and a collection of unrelated pioneers comes into view, separated by billions of years.

About 3.48 billion years ago, a set of deposits formed in the Pilbara of Western Australia that has been interpreted as a terrestrial hot-spring environment. It contains the silica deposits characteristic of hot springs, stones ejected by geysers, and layers that may have been left by microbes. The authors who published the study in 2017 regarded it as one of the oldest candidates for terrestrial life yet known. What it tells us is that something was living beside a hot spring. That is a long way from an entire terrestrial ecosystem.

About 3.22 billion years ago, extensive microbial mats spread across tidal-flat and river deposits in South Africa, bearing the marks of repeated exposure and drying cracks. Almost three billion years before true plants appeared, a thin layer of living things already occupied the edges of the land, drying out and getting soaked, getting soaked and drying out again.

About 635 million years ago, filamentous microfossils were preserved in pyrite in a set of cave deposits in South China. Rainwater had dissolved out these caves, and the fossils lay at the land surface or just below it. The original authors interpreted them as possible fungi.

About 500 million years ago, before plants had come ashore, something had already crawled across an ancient coastal dune near Kingston, Ontario, in Canada. The sandstone bedding surfaces there preserve trackways of many-legged animals: two rows of closely spaced footprints with a drag mark between them, some extending for more than two meters.

The shape of the sand grains and the bedding showed that this was a dune piled up by wind, not a place underwater. The trackmakers left no bodies behind. Researchers inferred that they were arthropods, something between shrimp and centipedes in appearance, that had crawled out of the sea onto the beach, perhaps to lay eggs, perhaps simply to escape something.

Once again, footprints preceded bodies. Later, body fossils of an animal whose leg spacing matched the trackways were found in tidal-flat strata of the same age in Quebec, Canada. It was a long-extinct arthropod with a segmented body and a long, pointed tail; some researchers considered it a distant relative of the group that includes modern millipedes. These animals were probably scouts, venturing onto the beach as the tide went out and returning to the sea, over and over again.

About 470 million years ago, spores appeared in Argentina whose walls had a structure much like those of moss-like plants today. They are among the oldest widely accepted, reliable evidence of land plants. Spores tell us that the plants producing them already existed. What those plants looked like, the spores cannot say.

From about 430 million years ago, land plants began to appear in the rocks with their overall form still recognizable: a few centimeters tall, their bare stems forking in two, then in two again, each topped by a little spore-bearing capsule. One was Cooksonia, commemorating the Australian botanist Cookson by taking her name.

The vegetation on land at the time was probably something like a layer of green velvet, a few centimeters high, beside the water. Rocks more than 400 million years old around Qujing in Yunnan, China, also contain a rich array of these early land plants, making the region an important source for studying how plants came ashore.

At Stonehaven on Scotland’s east coast, the amateur fossil collector Newman chipped a small millipede out of coastal rock. Its age is still disputed, with estimates ranging from a little over 410 to a little over 420 million years. He worked as a bus driver, and the millipede’s species name honors him. Its body preserved spiracles, tiny openings in the surface through which it could breathe. This is some of the earliest direct evidence of an animal breathing air.

About 407 million years ago, the Rhynie chert in Scotland sealed an entire terrestrial ecosystem in place in silica: plant cells, fungal threads, fungi growing inside plants, lichen-like partnerships, and tiny arthropods. Episodes 6 and 11 both mentioned it.

At the same time, the tallest things on land were Prototaxites. Their “trunks” could reach a meter across and eight meters high, but they were not trees. Inside were bundles of interwoven tiny tubes.

Long thought to be a gigantic fungus, Prototaxites was proposed in a 2025 study to belong to an entirely extinct, independent group, distinct from both fungi and plants. Out of a layer of green velvet a few centimeters high rose columns several meters tall. We still cannot say for certain what they were.

About 390 million years ago, the Polish footprints. About 375 million years ago, Tiktaalik and the bones of some of the earliest tetrapods.

This timetable includes bacteria and archaea, various eukaryotic microbes, fungi, plants, and several separate ventures onto land within the animals. Even arthropods came ashore more than once. Molecular phylogenetics treats myriapods, which include millipedes; hexapods, which include insects; and arachnids, which include spiders, as separate colonizations of land. The ancestors of insects and the ancestors of spiders did not come ashore together. The phrase “arthropods colonized the land” compresses at least three distinct histories, each with its own time, place, and route.

The gaps between dates on this timetable also need to be read with an eye to how they were measured. The earliest spores are about 470 million years old. Plants whose whole form can be recognized appear roughly forty million years later.

Count still earlier millimeter-sized tissue fragments as plant body fossils, however, and the gap shrinks to about twenty million years. Forty million years and twenty million years differ by a factor of two. All that changed was “what counts as a plant fossil.” The gap between footprints and bones in the first section was measured in the same way: where you draw the boundary at either end determines how much time lies between.

3 The Space Called Land Was Made

Before vascular plants appeared, the continents were already more than bare rock. At least four layers need to be distinguished. There was bedrock and the debris of physical weathering. There was a mantle altered by chemical weathering.

There were ancient soils capable of developing soil horizons. And there was the surface covering of microbial mats, lichen-like communities, and moss-like growth. We should be clear here about what weathering means.

Rock changes when exposed to air and water. Physical weathering breaks it apart through freezing and thawing, temperature changes, and abrasion. Chemical weathering happens when water and the substances dissolved in it react with minerals, carrying away some components in solution and leaving others behind. Chemical weathering does not require plants. On a planet utterly devoid of life, water and rock would still react.

The difference made later by roots, fungal threads, organic acids, and carbon dioxide from respiration was to change where those reactions happened, how intensely they proceeded, and what they produced. We can still see this surface layer in deserts today.

A thin, dark-brown crust covers the ground in many deserts. Cyanobacteria, lichens, and mosses form it by binding sand grains together with slime; it is called a biological soil crust. It can hold sand in place and retain a little water. Crush it underfoot, and it often takes decades to grow back.

The most useful modern analogy for the ecology of early land plants is not a forest but a cryptogamic ground cover: that layer of mosses, lichens, and microbes spreading close against the ground.

It traps dust, retains moisture, and keeps mineral grains in prolonged contact with organic acids and fungal threads. Some researchers have imagined the pre-vascular-plant continents as a “mat world.” The central idea is that microbial mats acted as ecosystem engineers, changing weathering and nutrient cycling across the continents.

In other words, before true plants appeared, a living layer on the continental surface was already changing weathering and the destinations of nutrients. That layer was thin, low, and close to the ground.

What vascular plants later brought, then, was not a transition from nothing to something. It was a change of scale in thickness, depth, and flux. And experiments give us direct estimates of the size of that effect.

Researchers have used living liverworts to model systems of early land plants and fungi. Compared with controls without plants, plants together with fungi could increase calcium release from basalt by about three to seven times, and phosphorus release by about nine to thirteen times. The same study also produced a very important result that checks the rhetoric.

Once the grain-scale rates were integrated over the full depth affected, early plants anchored in a shallow layer only about 0.1 meters deep may have had a total weathering flux at least an order of magnitude below that of later trees, whose roots reached about 0.75 meters or deeper. Put these two sets of numbers side by side, and they teach a useful discipline.

A local rate and a global flux are two different quantities. A system can multiply weathering several times over on every square centimeter of contact surface, yet, because its influence extends only about ten centimeters down, its global total can remain far below that of later forests.

So whenever we read that “such-and-such an organism greatly accelerated weathering,” we need to ask which quantity is meant. A process that is much faster locally need not be much larger in its global total. The development that truly changed the spatial scale later was deep rooting.

Shallow rhizoids mostly altered the top few centimeters to a few decimeters. As true root systems and tree-sized forms gradually appeared in the Devonian, the reaction front moved downward. Soil thickness, pore structure, slope stability, and the supply of organic carbon below ground all changed. Phosphorus has two faces here.

Plants and mycorrhizal fungi turn poorly soluble mineral phosphorus into phosphorus that organisms can use. When runoff carries some of it into the sea, it may in turn increase marine productivity and the burial of organic carbon.

The same “mineral residue” thus becomes a biological resource on land, then an input to an aquatic ecosystem after leaving it. Phosphorus still makes that journey today. Most of the phosphate fertilizer farmers put on their fields comes from phosphate rock laid down on ancient seafloors. Crops absorb some of the phosphorus applied; rain washes some into rivers and lakes, where it causes algal blooms. In Section 7 we will see that Devonian forests may have done much the same thing.

Organic matter was a building material that only later became increasingly abundant. Before terrestrial life became widespread, resources already existed on the land surface: light, carbon dioxide, water, and elements such as phosphorus, iron, and potassium in the rocks. What was really missing was a pool of organic matter continually replenished by local producers and capable of cycling repeatedly through the surface layers.

This mechanism later left a large-scale signal in the sedimentary record. Researchers who compiled some 3.5 billion years of terrestrial deposits found a marked, irreversible increase in muddy deposits in alluvial environments after the appearance of early land plants.

Another experiment went further: there was no need to wait for deep-rooted forests. Organic matter from early plants could itself promote clay flocculation, making mud more likely to settle out of river water onto floodplains. Flocculation means that tiny particles, which would otherwise stay suspended and drift away, stick together into larger clumps and can therefore settle.

This is almost a literal instance of a remainder becoming building material. Organic matter left by plant metabolism began as an output of production. By changing the way clay particles clustered together, it created more stable, fine-grained floodplains. Those fine-grained layers then became a new substrate better able to hold water and nutrients and support roots.

Notice the direction of the loop: plants first changed how mud settled, and the altered mud then suited plants better. An output became the next round’s input. But we should not give organisms all the credit for making the continents habitable.

A global model of ancient landscapes published in 2023 proposed that tectonics, past climate, erosion, and internally drained basins controlled whether continents could accumulate a sedimentary cover and develop soils suitable for rooted plants. The study even suggested that poor soils had, for a time, limited the spread of early land plants. Rivers offer the clearest counterexample.

It used to be common to write that deep, sinuous, single-channel rivers appeared only after plants did. Yet deep channels and single-thread meandering rivers existed before vegetation. Modern unvegetated settings, and even models of Mars, show that vegetation is not a prerequisite for a meandering channel, provided mud, ice, or some other mechanism gives the banks cohesion.

What plants really changed was the likelihood of such channels forming, their stability, their rate of migration, and the preservation of mud. A two-way model is therefore more appropriate. Tectonics, climate, and the atmosphere first establish the boundary conditions for life. Once organisms enter, they modify those very conditions.

4 These Genera Are Not Rungs on a Ladder

Return to the textbook sequence in the first section. These fishes come from all over the world. Eusthenopteron comes from Miguasha in Quebec, Canada, where coastal cliffs contain so many fossils that the site became a World Heritage site.

Gogonasus comes from Gogo in Western Australia. The fish fossils there are sealed inside limestone nodules; dissolve the stone in acid, and three-dimensional skulls emerge. Panderichthys comes from Latvia, Tiktaalik from the Canadian Arctic, and Acanthostega and Ichthyostega from eastern Greenland.

In 2002, Zhu Min and colleagues at the Chinese Academy of Sciences’ Institute of Vertebrate Paleontology and Paleoanthropology found a tetrapod lower jaw in Late Devonian rocks in Ningxia. Asia’s first Devonian tetrapod fossil showed that these animals had already reached the ancient landmass on which China now lies. One lower jaw extended the map of this group halfway around the globe to the east.

Inside the pectoral fin of Eusthenopteron, the bones nearest the body already corresponded to a humerus, radius, and ulna. Farther out, however, it retained the structure and rays of a fin. Its significance is not that it was “already coming ashore.” It shows that the basic materials of the proximal limb skeleton already existed in wholly aquatic fish.

Gogonasus presents an especially striking mosaic. Some parts of its skull remained relatively primitive, while its spiracular region and pectoral-fin skeleton had features closer to those of tetrapods than expected. Traits were not being added one by one in a fixed order.

Panderichthys had a low, flattened body, eyes toward the upper side of its head, and a robust series of bones inside its pectoral fins. The bones farther out in the fin displayed some of the spatial relationships later found in the wrist and digit regions. But it was still a fin.

Tiktaalik retained scales, gills, and fin rays, while also possessing a low, flat skull, eyes on top of the head, markedly enlarged ribs, and a “neck” formed by the functional separation of the shoulder girdle from the skull.

The neck deserves a moment of its own. In fish, the skull and shoulder girdle are joined, so a fish cannot turn its head: to look another way, it has to turn its whole body. Once the shoulder girdle separates from the skull, the head can move independently. The distal part of Tiktaalik’s appendage, meanwhile, was still enclosed in a fin, with rays along the outer edge and no truly free digits.

“Wrist-like bones,” then, do not yet make a hand. A 2014 study found its pelvis to be extraordinarily large, much larger than initially recognized, showing that the capacity for propulsion driven by the hind appendages had begun to strengthen before true digit-bearing limbs appeared. Its pelvis still lacked a complete tetrapod-style connection to the sacrum, however.

In 2020, a complete specimen of Elpistostege showed a series of bony elements inside its fins that could be matched to multiple digits, while fin rays still covered them. This is crucial evidence that the skeletal pattern of digits first appeared inside a fin, making the image of “a fin suddenly becoming a foot” even less satisfactory.

The story of Acanthostega is itself a story of looking again. In 1970, a geology student from Cambridge collected some fossil bones almost in passing during fieldwork in eastern Greenland, then brought them back and put them in a museum drawer.

More than a decade later, the Cambridge paleontologist Clack read his field notes. In 1987, she led a team back to the hillside they described and excavated a collection of complete Acanthostega specimens. In 1990, she and her colleagues published its forelimb: eight fingers.

Acanthostega already had limbs with eight digits, yet retained many aquatic features, including a tail adapted for swimming and limbs relatively ill suited to prolonged weight-bearing. Here, for the first time, “having digits” and “being genuinely good at walking on land” were separated with exceptional clarity. Digits are not a synonym for terrestrial competence.

And there were eight digits, not five. Five became fixed only later. Acanthostega had eight, Ichthyostega seven, and Tulerpeton from Russia six. The counts differed. “Tetrapods have five digits” was not a rule laid down at the beginning; it was the result of a later settling on that number.

Once the count had settled at five, it became remarkably stable. Five is the ancestral condition of the vast majority of tetrapods alive today. Ichthyostega was the first member of this sequence to become known.

A Swedish paleontologist named it in 1932 from fossils collected in eastern Greenland. For decades afterward, it served as the standard image of “the earliest tetrapod”; museum reconstructions showed it standing on shore with its body held high. Its trunk and ribs were more robust than those of Acanthostega.

Yet three-dimensional reconstructions of its joints show that it still did not step along like a modern salamander. The real picture is therefore a matrix of traits, not a ladder of successive rungs.

Lay these animals out side by side, and something becomes clear: these traits did not all change together, and their order of appearance does not follow our intuitions about function. Intuitively, “able to walk on land” ought to mean that digits, a weight-bearing pelvis, a neck, and air breathing all arrived together.

The actual record says otherwise: a neck before digits, the skeletal pattern of digits before the loss of fin rays, a stronger pelvis before a sacral connection, digits before effective walking. On a phylogenetic tree, the safer description is therefore “these genera occupy different branches of the tetrapod stem,” rather than “Eusthenopteron became Panderichthys, and Panderichthys became Tiktaalik.”

Most are samples from side branches, and they can overlap in time. Tiktaalik was found by following a map.

The research team first drew up a search map using the age of the gap in the phylogeny and the sedimentary environments needed to preserve shallow-water vertebrates. Their targets were well-exposed Late Devonian river and delta deposits. Ellesmere Island happened to have extensive outcrops of just that kind.

Formal fieldwork began around 1999. The crucial bones were found in 2004 and published in 2006. Leading the work were Shubin of the University of Chicago and Daeschler of the Academy of Natural Sciences in Philadelphia. They searched the Arctic permafrost for four summers before seeing a flat skull emerge from the rock.

Local Inuit elders were asked to choose the name. In their language, it refers to a large freshwater fish living in shallow water. The discovery is a textbook case precisely because the reasoning was not “we know animals were bound to colonize land, so let us look for an animal doing that.” It was a testable prediction. And finding Tiktaalik did not establish it as the direct ancestor of later tetrapods.

What it established was that the map had been drawn correctly: animals of this kind could indeed be found in rocks of that age and that facies. Shubin later turned the experience into a bestselling book, Your Inner Fish, tracing the story from Tiktaalik to the old fish parts still present in human hands, lungs, and ears.

Now return to the first section. If those tracks are accepted, the actual transition from fish to tetrapods must have taken place before the tracks were made.

The stratigraphic sequence originally assembled from body fossils may then be an illusion. But that would not deprive Tiktaalik of its transitional significance. A transitional form in phylogenetics is a lineage preserving a mosaic of ancestral and derived traits. It need not have lived before the more derived groups appeared.

A side branch retaining a more primitive form can perfectly well coexist with more derived descendant lineages. Living lungfish provide a ready example. They retain many ancient features, and they are alive at the same time as we are.

The more precise formulation, then, is this: if those tracks are accepted, fossils such as Tiktaalik preserve several morphological stages in the transition from fish fins to limbs, yet lived after digit-bearing tetrapods already existed.

A 2025 paper on Australia’s earliest reptile tracks went further, estimating that the most recent common ancestor of all living tetrapods lived at about the same time as Elpistostege and Tiktaalik, perhaps even a little earlier. What this strengthens is the case for long ghost lineages and coexistence, not a neatly ordered procession of ancestors replacing one another.

5 The Tools for Making Limbs Are Older Than Limbs

Many of the same genes regulate both paired fins in fish and limbs in tetrapods. A 2021 study of the lungfish genome was especially important. The Australian lungfish has a genome roughly fourteen times the size of ours, the largest animal genome sequenced at the time, crammed with repeated stretches of DNA.

The study found several “limb-like” patterns of gene expression already present in the distal lungfish fin. But “the tools were already there” cannot be promoted to “fingers were already written into the fish’s genes.”

Functional experiments in 2025 even suggested that a regulatory region used in the distal tetrapod limb may have been co-opted wholesale from an older fish network originally serving structures such as the cloaca. We need to be clear about what “co-opted wholesale” means.

It means that a regulatory region already involved in development elsewhere was recruited for another part of the body and another developmental window. What shifted was its use, not a stretch of DNA physically travelling from one body part to another.

In 2016, the University of Chicago conducted an experiment with zebrafish. Researchers labeled cells in embryos and followed them through development. They found that cells becoming the slender rays at the ends of fish fins came from regions expressing the same class of genes as the regions producing fingers in tetrapod embryos. A fish’s fin rays and our fingers are two destinations for the same population of cells.

The parts are old. Their wiring has a new use. Change need not start one connection at a time: an existing regulatory landscape can be recruited together. Episodes 5 and 8 encountered a similar pattern.

Episode 5 said that the toolkit precedes the system. Episode 8 said that parts exist before their use changes. Here one mechanism becomes concrete: co-opting a whole regulatory landscape. That does not exclude smaller modifications. It means a new use need not be wired from scratch.

“Fish already had the genes needed to make limbs” sounds as though tetrapods were ready and waiting. What it actually says is that those genes were doing something else at the time. Why, then, did these abilities appear? There are several hypotheses, each with its own merits and its own difficulties.

Shallow water and complex substrates. Low water levels, plant debris, and complicated bottom surfaces favored bodies that could use appendages to prop themselves up and turn. The difficulty is that these features could work perfectly well entirely underwater.

Seasonal drying. Intermittent drying of bodies of water may have selected for air breathing and the ability to move between them. The difficulty is that tolerating low oxygen and walking while supporting one’s weight on land are two different axes of adaptation.

Low oxygen in water. Lungs and air breathing in fish can precede locomotion on land. The difficulty is that this explains respiration more directly than it explains digits and pelves. Escaping predators. Leaving the water briefly might reduce exposure to certain aquatic predators. But the land surface simultaneously introduces new problems of water loss, weight-bearing, and body temperature.

Pursuing food on land. Before tetrapods appeared, plants, fungi, and arthropods already supplied resources there. The timing makes this entirely possible. Yet the teeth and stomach contents of early tetrapods rarely provide independent proof that a particular morphological innovation was driven by hunting on shore.

Opportunities for vision. Researchers have quantified how looking from water into air can enormously increase effective visual range: from a few meters underwater to tens or hundreds of meters in air. They found that eyes had already enlarged substantially before truly terrestrial tetrapods appeared. Eye sockets approximately tripled in size, and the eyes shifted from the sides of the head to the top, like those of a crocodile lying in the water with only its eyes showing.

Larger eyes paid for themselves because air made it possible to see farther. Underwater, even very large eyes could not see that far. And this change took place while the animals were still in the water. Here is another case of “capacity before use,” and its direction was, quite literally, outward from the water.

The merit of this hypothesis is that it can be quantified. Given the water’s turbidity and the attenuation of light, effective visual range can be calculated. Its difficulty resembles those of the other hypotheses: larger eyes can arise for other reasons too, such as hunting in murky water. The greatest difficulty all these hypotheses share is that historical causation cannot be rerun.

The same body form often improves maneuvering in shallow water, breathing, lifting the head to look around, and briefly leaving the water, all at once. It is therefore hard to identify the original selective pressure from the bones that ultimately survive.

The problem is that these pressures existed at the same time, and their demands on body form overlapped. A bone cannot record “which job it was chiefly selected for in the first place.”

The safest account is that several kinds of aquatic ecological pressure first produced capacities that could be used on land. Those capacities were then recombined in different environments.

6 Every Solution Brings New Problems

The problems land poses to a body are specific, and no solution comes free. Water loss. Water-vapor pressure in the air is usually far lower than it is inside tissues, so every exposed moist surface continually loses water by evaporation. Plants answer with a cuticle and waxes; animals with a keratinized body surface or an exoskeleton.

On the same surface, however, gas exchange and the prevention of water loss work against one another. Plant stomata are a direct product of this conflict: they must open to admit carbon dioxide, and opening them loses water.

The trade-off can be calculated. For every carbon atom a plant fixes, it loses hundreds of water molecules. On a sunny summer day, a large tree can move hundreds of liters of water from its roots to its leaves, then release it through its stomata into the air. Many of the clouds above forests begin this way.

There is no way to admit only carbon dioxide while refusing water an exit: both pass through the same hole. The entire water economy of land plants therefore revolves around that hole. When should it open? How wide? For how long?

This regulation requires sensing: sensing moisture, sensing light, sensing the concentration of carbon dioxide. Behind a simple hole hangs a whole control system. And that system came later. The earliest land plants had far less control over their stomata than plants have today.

A stoma, then, is not simply an air inlet. It is an adjustable means of managing a trade-off. Gravity and support. Once water’s buoyancy is gone, a body’s weight has to be carried by its structure. Plants answered with lignification and vascular tissue, simultaneously addressing support and the long-distance transport of water.

Animals answered by reorganizing bones and muscles. As we have already seen, having digits does not mean being able to walk effectively while bearing one’s weight. Between “having limbs” and “being able to walk” lies a whole set of changes to the pelvis, backbone, and muscle attachments. Put a fish on land, and its fins do not enable it to hold itself up.

Gas exchange itself. Oxygen diffuses far faster through air than through water, so from the standpoint of oxygen supply, air is the better medium. The real problem is that the exchange surface must stay moist, and a moist surface loses water. Terrestrial respiratory organs therefore generally turn inward: lungs and tracheal systems conceal their exchange surfaces inside the body.

This calls for a ventilation mechanism, actively moving air in and out. Gills, too, need water to pass over them, and many fish actively pump it. Air ventilation likewise costs energy and requires muscles and nervous control. So “breathing on land is easier” holds only on the chemical side. Mechanically, it brings complications of its own.

Breathing can also constrain how large a body grows. Insects deliver air toward their tissues through tracheal tubes, using both diffusion and, in many cases, active ventilation. Greater size brings challenges of delivery distance and the space the tubes occupy; oxygen supply is not the only limit. The next episode turns to the oxygen-rich Carboniferous, when some distant relatives of dragonflies approached seventy centimeters in wingspan.

Two other matters deserve a mention. Temperature. Water has a high heat capacity and changes temperature slowly. Day-to-night temperature swings on land can be dozens of times greater than those in water. A lizard today has to lie on a rock in the morning sun for a while before its body warms enough to run. Early tetrapods probably also had to learn to live by the sun.

Ultraviolet. Water absorbs most short-wave ultraviolet radiation; the land surface has no such shield. Episode 4 discussed ozone’s two thresholds. Protection from ultraviolet at the surface arose only after oxygen had reached a substantial level.

Reproduction. Fertilization that depends on swimming sperm ties reproduction to a film of water. Plants later evolved seeds and pollen; animals later evolved the amniotic egg. The earliest seeds appeared at the end of the Devonian, and amniotic eggs later still. Both lie at the tail end of the period covered here, or beyond it.

In other words, during the period this episode covers, most organisms that came ashore still had to return to the water’s edge to reproduce. Sensory systems also had to be rebuilt, a point often overlooked. Light scatters and attenuates much less in air than in water, while the speed and impedance of sound in air are entirely different from those in water.

The ear is a good example. Underwater, sound can pass straight through the body to the inner ear. In air, most sound striking the body is reflected back, so a mechanism is needed to amplify the air’s faint vibrations. The tiny bone called the stapes in our middle ear was originally, in fish, a bone supporting the jaw region and gill cover. Early tetrapods repurposed it, and step by step it became an auditory ossicle.

The effective range, directional discrimination, and signal-to-noise ratio of the same set of senses all have to be recalculated for the two media. The result is not always worse. The increased visual range in the previous section is an example of improvement. What matters for this episode, however, is that these solutions themselves created new dependencies.

A cuticle reduced water loss, but it also reduced the ability to exchange substances across the entire body surface. Lignification provided support, but it also produced a material resistant to decay. Hiding exchange surfaces inside the body solved the problem of drying out, but required a whole ventilation system.

This follows the same pattern established in earlier episodes: a repair does not remove the remainder; it moves it somewhere else. Episode 3 offered the prototype. A membrane solved dilution and simultaneously created osmotic pressure. Coming ashore replays the same pattern at another scale.

Each physical constraint solved adds another set of structures that must be maintained continuously. Those structures in turn become constraints for the next round. Lignin is the clearest example here. It provided support and water transport, It also made plant tissues harder to break down. Which organisms could degrade them then, and how effectively, cannot simply be read backward from the coal that survives today.

Lignin has an irregular structure, without a fixed repeating unit, whereas most enzymes are made to work on regular structures. Paper mills today still need large quantities of chemicals to remove lignin from wood pulp.

Cheap paper turns yellow and brittle with age partly because the lignin left in it slowly oxidizes. In the next episode, this will become the heart of an entire debate. The question is whether lignin’s resistance to decay can explain why so much carbon was buried during the Carboniferous.

How did this “remainder of a solution” become a resource for other organisms? Why was some decomposed while some was buried? The next episode carries those questions forward.

7 The First Forest

The greatest change on Devonian land came in the middle and later parts of the period: trees appeared.

In 2019, on the floor of an abandoned quarry near Cairo in New York State, researchers uncovered the remains of a forest about 386 million years old. Tree root systems spread across several thousand square meters of exposed rock.

One kind of tree sent roots around ten meters in all directions, with lateral roots branching and branching again, already much like those of trees today. This is one of the oldest forests currently known.

At nearby Gilboa, a nineteenth-century flood had scoured away a riverbank and exposed another set of fossil tree stumps of almost the same age. In the 1920s, construction of a reservoir for New York City unearthed many more. During repairs to the dam in 2010, workers cleared away the overlying rock and exposed the entire floor of that ancient woodland. Before it was covered again, researchers mapped the position of every tree.

Another kind of tree in the New York forests looked stranger still. In 2007, researchers found its complete crown: a bare pole several meters high, topped by a brush-like tuft of branches. It had no leaves, and the branches dropped off one by one. It looked like a giant palm, or a broom planted upside down.

China has forests from the same period too. In 2019, researchers from Peking University and other institutions identified a roughly 365-million-year-old lycopsid forest in quarries at Xinhang, Anhui. At least 250,000 square meters were exposed, with the bases of tree trunks standing in their original positions across the rock. The tallest trees were estimated to have reached seven or eight meters. This is the oldest known forest in China.

The most successful tree of the time was Archaeopteris. It could grow more than twenty meters tall, with wood already resembling that of modern conifers, but fern-like leaves and reproduction by spores. By the end of the Devonian, it was present on nearly every continent, from North America to China.

Trees brought depth. Shallow-rooted plants could alter only the top few centimeters to a dozen or so centimeters of soil. Tree roots pushed the weathering front down a meter, then several meters.

Soils thickened and developed layers; organic carbon began accumulating underground. Atmospheric carbon dioxide declined markedly during this part of the Devonian. Many researchers think the expansion of forests accelerated rock weathering, drawing large quantities of carbon from the atmosphere into soils and marine sediments.

One hypothesis also connects forests with catastrophe in the sea. Roots accelerated weathering, washing large quantities of nutrients such as phosphorus into shallow seas. Algae bloomed, then died; their decomposition exhausted oxygen at the seabed. This may have contributed to several episodes of marine anoxia and extinction in the Late Devonian.

There is evidence for every link in this chain, and every link has been challenged. When Episode 14 reaches the end-Permian, we will see again that a mass extinction often has more than one cause.

Forests also brought fire. Wildfire requires two things: combustible vegetation and enough oxygen. The earliest charcoal has already been found in rocks more than 400 million years old in the Welsh borderlands of Britain, showing that something on land could burn by then, and that there was enough oxygen in the air for it to do so. By the later Devonian, charcoal was becoming increasingly common.

The soil was growing busy too. By the Devonian, its tiny arthropods had differentiated into fungus-eaters, consumers of decaying plants, and small arachnids that preyed on them. A soil food web had taken shape. Dead plants, fungal threads, and feces became meals for the next set of organisms.

The Devonian ended with an extinction. Nearly all the armored placoderm fishes vanished from the seas. Surviving vertebrates generally became smaller, and small-bodied fish and tetrapods gained the upper hand.

After the extinction, the tetrapod fossil record fell almost blank again for more than ten million years. Paleontologists call this interval “Romer’s Gap,” after the American researcher Romer. Around 2016, successive discoveries of tetrapods from this interval in river and lake deposits in the Scottish Borders began filling it in, little by little. Here was another sieve. Tetrapods were probably present throughout the interval; the places where they died simply left few rocks that could later be sampled.

Does coming ashore count, then, as the kind of shift in level discussed in Episode 10? The criterion there asked whether units that had previously reproduced separately became locked into a new, higher-level individual reproducing as a whole. Fish that came ashore were still individual fish reproducing separately; plants that came ashore were still individual plants. By that criterion, coming ashore did not change the level.

It changed something else: the surface of a planet. The capture of solar energy spread from water onto the continents. Organic carbon began to remain in soils and floodplains.

Mineral nutrients cycled through roots, fungal threads, and fallen leaves, rather than merely flowing straight from rock into the sea. While alive, the earlier arrivals changed the environment faced by those that came later. The changes remained after their makers died. A layer of soil can persist for thousands of years after the plants that made it are gone.

“No change of level” and “the surface of a planet rewritten” can both be true. Saying clearly which variable changed is more useful than attaching a label.

8 The Remainder Becomes Building Material

The title calls coming ashore a matter of turning an environment’s remainder into building material. Along a Devonian riverbank, that was almost literally what happened.

Section 3 described the experiment: organic substances released by early plants could make fine clay particles in river water stick together and settle onto floodplains. What plant metabolism left behind was originally a by-product of production. It held onto mud that would otherwise have drifted away.

The fine mud that stayed behind held water and nutrients better and made a good place for roots. More plants grew; more plants released more organic matter. Dead cells and fallen leaves became soil organic matter. Fungal threads etched minerals and released nutrients. Rotting roots left pores and carbon. Slime secreted by microbial mats held sand grains in place. An output became the next round’s input.

The oxygen in Episode 4 was also something released, but its loop was different. Oxygen-producing bacteria discharged oxygen as waste. It rewrote the atmosphere and oceans, yet did not make oxygen production easier. On the contrary, it caused trouble for the cells that produced it: they had to make special enzymes to counter its damaging effects.

The loop involved in coming ashore has feedback. Once started, it reinforces itself. The study compiling 3.5 billion years of terrestrial sediments measured precisely this step upward: muddy deposits beside rivers increased markedly after the appearance of early land plants and never fell back to their previous level.

Darwin’s last book was about earthworms. For decades he watched them in his own garden and nearby fields, measuring the casts they left. Year after year, he found, they brought fine soil up from below to the surface, enough to slowly bury stones lying on the surface and ancient Roman ruins beneath the soil.

An organism eats something, passes it out, and what comes out becomes the next layer of soil. Ecologists later gave such organisms a name: ecosystem engineers. The usual example is the dam-building beaver. A single dam remakes an entire stream valley.

Life did not make the land. Continental topography, rock, water, climate, and sedimentary basins were supplied by the Earth itself. The 2023 model of ancient landscapes also suggests that basins and climate may first determine whether plants have enough soil in which to root. The boundary conditions come first. Organisms move in, and the boundary conditions begin to change.

The consequences of coming ashore are still visible. According to the 2018 compilation of global biomass, land plants account for about four-fifths of all biomass on Earth. Most known species also live on land. The green velvet a few centimeters high beside the water more than 400 million years ago grew into the overwhelming majority of the living matter on this planet today.

A construct, as this series uses the word, comes down to drawing a line. What lies inside counts; what lies outside does not. The line can be drawn by people, as with a set of criteria, or by nature, as with a membrane.

In this episode, the line is the water’s edge: life possible on one side, impossible on the other. Nature drew it, but the organisms crossing it kept drawing it again. Each wave that came ashore pushed the line a little farther inland.

What a construct cannot contain and cannot get rid of is what this series calls the remainder. Here, the remainder is sequence. If the footprints in the first section are accepted, the order pieced together from bones alone is not the order in which traits were acquired. The fossils called transitional fall after an interval they cannot record. They have been lying in museum drawers all along, studied for more than a century. The problem lies with the two sieves.

Footprints need a damp surface of mud that is not too deep, quickly covered by the next layer of sediment. Bones need a place where sediment accumulates underwater, or some other setting that keeps a body from being pulled apart.

These two places are usually not the same place. An animal can be active in both, leaving footprints in one and bones in the other. Its bones fell into another sieve, whose opening faced somewhere else.

The criteria are still sifting. Those footprints are still an open case. The sequence laid out before us is still pieced together from a stretch taken from each of two sieves.