The take-home message of this essay is a simple one: The transition of animals from water to land in the Devonian period, 370 million years ago, was profoundly important in one sense and entirely trivial in another. It had a major impact on our world, but it did not involve any unusual or extraordinary biological processes. The effects of the transition are all around us. We see them in the rocks. We see them in ponds and seas around the world. We even see them when we shake hands. Let me explain.
When we look back after 370 million years of evolution, the invasion of land by fish appears special. However, if we could transport ourselves by time machine to this early period, it isn't clear whether we would notice anything extraordinary. We would see a lot of fish, some of them big and some of them small, all of them struggling to survive and reproduce. Only now, 370 million years later, do we see that one of those fish sat at the base of a huge branch of the tree of life—a branch that includes everything from salamanders to humans. It would have taken an uncanny sixth sense for us to have predicted this outcome when our time machine deposited us in the middle of the Devonian.
To get a glimpse of the water-to-land transition, we need to see the creatures that lived on Earth at that time, then we need to look at our world today. When we do this, we see something sublime: The ancient world was transformed by ordinary mechanisms of evolution, with genes and biological processes that are still at work, both around us and inside our bodies.
The gulf between water and land looks like an unbridgeable divide. The challenges of life on land are vastly different from those in water. It would seem that completely different animals must live in these distinct habitats. Animals that walk on land need to cope with gravity; unlike water, air does not support animals as they move about. Animals also can dry out on land; this is particularly dangerous, because water is needed for many basic metabolic processes. And, of course, breathing is different in water than on land. Animals that breathe air need a more efficient way than gills to take in air and extract oxygen.
Because of all these factors, there are a daunting number of features that distinguish land- living animals from their fish ancestors: limbs with fingers and toes, necks, backbones with bony connections between vertebrae, a bony inner ear, a large scapula, ribs, paired nostrils, and so on. Biologists have singled out one of these characteristics for special treatment: True limbs are not seen in any living fish; for this reason, everything that is descended from fish is called a tetrapod (from the Greek for "four-footed").
For a long time it was thought that the shift from fish to tetrapod was driven by a transition from life in water to life on land. For example, it was thought that fins gradually evolved into limbs as animals began to walk. This thinking was captured by a famous hypothesis originally proposed by the American geologist Joseph Barrell in 1916 and later by the great American paleontologist Alfred Sherwood Romer. Romer and Barrell speculated that fish were forced out of water when Earth's climate supposedly became drier some 370 million years ago. As the ponds dried, so the story went, the fish had to learn to survive on land and so developed features that enabled them to hop from pond to pond.
When Romer did his work, in the 1920s through the 1960s, there was only one early tetrapod known: a limbed creature recovered from 365-million-year-old-rocks in East Greenland. At present, East Greenland is a cold desert—dry, mountainous, and well within the Arctic Circle. Temperatures there rarely rise above freezing and for much of the year are colder than -20 F. But 365 million years ago East Greenland was a much warmer place, containing warm-water swamps, streams, and ponds. In the 1920s, a Swedish team led by Gunnar Save-Soderberg discovered the skeletons of the then-earliest-known tetrapods in these rocks. These animals had robust limbs, appeared to be partly land-living, and supported Barrell's and Romer's hypothesis—at least, initially. To see how our theories have changed since Romer's day, we need to follow new evidence, whose trail leads to notions completely unforeseen even twenty years ago. This change in thinking attests to the power of evidence and the way it can change our view of the world.
In 1987 my colleague, Jenny Clack, began new studies in East Greenland and found the first important piece of evidence bearing on this water-land transition in over fifty years. She discovered the skeleton of another truly extraordinary tetrapod—one even more primitive than the one discovered by Save-Soderbergh. Sure enough, this creature has limbs with fingers and toes. It also has a very tetrapod-like hip, neck, and ear. What is remarkable is that this, the most primitive known tetrapod, is aquatic. It is not remotely specialized for life on land. It has fingers and toes but they are set within a limb that looks like a flipper. The limbs are delicate structures and seem unable to have supported the weight of the animal on land. It has a pair of hind limbs, but behind that is a tail that resembles that of a fish. Most important, this tetrapod has big gills.
The inescapable conclusion is that the most primitive tetrapod was an aquatic creature. The implications are profound: The fish-to-tetrapod transition likely happened not in creatures that were adapting to land but in creatures living in water. Moreover, everything special about tetrapods—limbs, digits, ribs, neck, the lot--might well have evolved in water, not on land.
This hypothesis made a prediction that could be tested: Aquatic animals more ancient than this new find should have intermediate structures. A search for these kinds of fossils dovetailed nicely with my own expeditionary research program in the late 1980s. Back then, my colleague Ted Daeschler and I were uncovering fish and tetrapods of the same age as Jenny's Greenland fossils in the roadcuts of central Pennsylvania. Pennsylvania is dotted with rocks of the same age as those of Greenland, but they need to be uncovered by dramatic means. The good news is that the state is not a frozen desert; the bad news is that fossils and rocks are mostly covered with trees, lawns, and cities. As a consequence, Ted and I made paleontological careers out of following the Pennsylvania Department of Transportation every time it cut a new road in central Pennsylvania. We found many fossils, but all of them were too young to test the issue at hand. We needed to go to a different area.
Ted and I ultimately found inspiration in an atypical place. We began a whole new research program that sprang from a single figure in a twenty-year-old textbook. I was thumbing through my old college geology text and found a map that seemed unremarkable at first. It was a map of North America with colored patches showing where rocks between 360 million and 380 million years old are preserved. One big splotch was on the east coast of Greenland, home to Jenny's find; another patch covered the part of Pennsylvania where most of our field effort had been focused. There was still another such area, though, and this is what made the figure interesting. Large, and running east-west across the Canadian Arctic, this patch extended over 500 miles and had never been explored by vertebrate paleontologists, although it was well known to geologists, particularly the Canadian geologists and paleobotanists who had mapped it extraordinarily well. The rocks turned out to be older than those in Pennsylvania and Greenland.
Ted and I first visited this area in 1999 and found little of interest. As it happened, we were fumbling around in the wrong part of the section; the rocks we were looking at were in the middle of an ancient ocean environment. When we shifted the expedition to areas that preserved ancient streams, lakes, and ponds, we found more fossils. During the 2004 field season, in these ancient environments, we found what we were looking for. Buried within a 370-million-year-old shallow stream was a collection of whole skeletons, one on top of the other. One of these creatures is an astonishing new kind of fish.
The new fish has fins, scales, and gills. By all definitions, it is a fish. This designation seems to hold until we look at its skeleton. Inside the fin is the skeletal pattern of all tetrapod limbs, in primitive form. It has an arm bone, a forearm, even a wrist. The new fish has a neck much like that of the earliest amphibians. The skull of this fish is not cone-shaped, as fish skulls are, but flattened like a crocodile's, with a nostril on either side. This creature also has expanded ribs, something unknown in any fish. We had found, one of my colleagues mentioned in jest, a fishapod.
The fishapod underscores one important point: It is no longer easy to distinguish a fish from a tetrapod. The arctic fossils were only the tip of a paleontological iceberg; after subsequent discoveries in Latvia, Scotland, and China, the distinction is now so fuzzy that many of my colleagues do not even try to define tetrapods by ticking off a list of features. Our earlier definition of tetrapods distinguished them from fish by their possession of limbs. In what group, then, do we put our fish with wrists? What other characteristics might help us? Perhaps we could use lungs to distinguish tetrapods from fish. Then we would have to explain why lungfish use gills and lungs both, yet have fully formed fish fins. Scales? Even here, we run into the same problem, because early limbed and lunged animals also have belly scales. Indeed, the difficulty that our taxonomists have in distinguishing tetrapods from fish is the inevitable result of finding fossil intermediates.
This practical problem reflects a significant reality. One of the major transitions in the history of life is now bridged by a series of fossils dating from 380 million to 360 million years ago. The fact that we have discovered intermediates is not surprising; the surprise is that these creatures all appear to be aquatic and not specially adapted to life on land. This insight begs the question: Is there really a great divide between life in water and life on land? Answers to this question come from the study of fish alive today.
Modern fish have adapted to live in very different environments, including on the sea floor, in the shallows of lakes or streams, even partly in air. To cope with these environments, they have a remarkable set of features that enable them to walk, breathe, and even climb. For example, the various species of walking fish have evolved "armlike" bones and joints allowing them to prop up and propel their bodies along the ground. Some fish, like the mudskipper, maneuver in mudflats and spend a considerable period of their lives outside water, able to breathe air because the back of their mouth can absorb oxygen and relay it to the bloodstream. Mudskippers can hop good distances on the mudflats; some of them even climb trees by reaching up the trunk with their front fins and holding on with their hind fins.
What is important is that these various adaptations to land have evolved many times in fish. Several different kinds of fish climb trees; in addition, there are many different species of fish that breathe air, live part of their life on land, and walk about. The boundary between water and land is quite porous and bridged by modern fish from around the world. In fact, the adaptations we see in the fossils of the fish-tetrapod transition seem almost trivial in comparison to the living animals.
Mudskippers and the other walking fish are all very interesting, but are they extraordinary in an evolutionary sense? No, they are not, and the reason is instructive. Hopping, climbing, and breathing fish are just animals that have evolved to live in different kinds of aquatic and subaereal habitats. They are able to breathe air, hop, or climb because of subtle changes to their bodies; no revolutionary changes are needed. In evolutionary terms, the only way they will be notable is if their lineage is prolific and their descendants do great things. The relatives of the fish and tetrapods from Canada and Greenland were prolific; they are part of a trunk of the evolutionary tree that gave rise to every tetrapod—every bird, mammal, reptile, and amphibian. The mudskipper has a long way to go, and many hurdles to leap, before we will know whether its part of the evolutionary tree is special. If paleontologists 300 million years from now dig up the remains of a mudskipper, they will write chapters about its role in a "great" transition only if its part of the evolutionary tree has branched into many twigs. The mudskipper will get extra special treatment if one of its evolutionary branches leads to the paleontologists' own species.
Our understanding of the fish-to-tetrapod transition is not limited to long-dead fossils or obscure fish that climb trees. We have access to the DNA of every creature alive today. This is an enticing record of evolution, because DNA builds our bodies and is passed from generation to generation. By knowing how DNA works, we can dissect the molecular machinery that builds animals. This defines a whole new research program, one that was unimaginable in Romer's day. We can now compare the genetic recipe that builds a fish to the one that builds a tetrapod, in