New Limbs from Old Fins

On October 20, 2011


Picture an animal – any animal, maybe your favorite animal. Then ask a nearby kid to name her or his favorite animal. I think it’s a pretty safe bet that neither of you chose a sponge or a sea squirt, or a planarian or a sea pen, or a moth or a mosquito. And let’s hope that neither of you chose a tapeworm or a trombiculid mite. Those unlikely choices are all animals. But it’s more likely that you both chose a vertebrate, and I think it’s highly likely that you both chose a tetrapod vertebrate – an animal with legs and/or wings, a skull and a backbone. Maybe we prefer these creatures because they’re a lot like us, or because they make good pets (or food), or because they’re big enough to make an impression, or because they were the animal representatives pictured on the ark in board books. (Or maybe you chose a butterfly, and now you feel a little left out.) What matters is that there is something extra interesting about tetrapod vertebrates.

As you might have guessed, tetrapods are vertebrate animals that have four limbs. The group includes reptiles, amphibians, birds, mammals… you know, the usual suspects. (Snakes and whales, which don’t have those limbs, are nonetheless classified as tetrapods, and we’ll come back to that.) At first, this might look like a wildly diverse crowd of animals with almost nothing in common: tiny hummingbirds in the air, gigantic whales in the ocean, frogs that come from tadpoles, salamanders that can regrow severed limbs, cats that eat only other vertebrates, misnamed “bears” that eat only eucalyptus leaves. But on closer inspection, some extraordinary patterns emerge. These animals, in all of their magnificent variety, seem to be built in very similar ways. It’s as though some kind of master plan has been tweaked over and over, to make a huge collection of variations on a theme.

This master plan for building tetrapods includes numerous components: plans for building backbones, for making skin, for growing a brain. Some of those components are unique to tetrapods; some are more widely employed in animals. Our focus will be the one that is most clearly associated with the tetrapods. We will explore the building of limbs – arms, legs, wings and flippers; and hands, feet, paws and paddles.

Consider, then, the human forelimb, better known as the arm. You may already be familiar with its skeletal structure, nicely illustrated in the 17th-century chalk drawing below.

bones of the arm and scapula

The upper arm contains a single large bone, the humerus, which is attached to the shoulder and to the elbow. The lower arm sports two parallel bones: the radius and the ulna. Those two bones link the elbow to the wrist. The wrist is composed of a group of small bones called the carpals (made famous by Carpal Tunnel Syndrome, which is reportedly exacerbated by the typing of blog posts). Attached to the carpals are the metacarpals, which are the bones of the fingers. So, the skeletal components of the human arm are as follows: one bone (humerus) attached to two bones (radius and ulna) attached to a set of small blocky bones (the carpals), which anchor finger bones. It’s an interesting pattern, but all by itself it’s not necessarily remarkable.

Now let’s look at the human hindlimb, or leg. The bones have a different set of names, which you may know all too well. Have a look at the 19-century illustration below.

leg bones femur tibia and fibula

The upper leg contains a single large bone, the femur, which is attached to the hip and to the knee. The lower leg sports two parallel bones: the tibia and the fibula. Those two bones link the knee to the ankle and foot. The ankle and foot are composed of a group of bones that includes a set of small bones called the tarsals. Attached to the tarsals are the metatarsals, which are the bones of the toes. So, the skeletal components of the human leg are as follows: one bone (femur) attached to two bones (tibia and fibula) attached to a set of smaller blocky bones (tarsals and others), which anchor toe bones. It’s an interesting pattern, but all by itself it’s not necessarily remarkable.

But wait. The leg pattern is essentially identical to the arm pattern. Why just one pattern? Why that pattern? Is there something special, maybe even somehow universal, about the pattern?

Questions like those were the domain of the great Richard Owen, the British naturalist and contemporary of Darwin. Owen’s detailed study of limb structure led him to write one of the more influential works in the history of biology: On the Nature of Limbs, first published in 1849 and most recently reprinted in 2007. In that book, Owen argued that all vertebrate limbs were modifications of a basic pattern or plan, called an archetype.

To see why Owen reached this conclusion, consider the wonderful lithograph below, created in Owen’s time (1860) by Benjamin Waterhouse Hawkins, who also contributed illustrations to Darwin’s Zoology of the Voyage of the HMS Beagle. The limbs of the horse are constructed in an interesting pattern, depicted in the upper left. One large bone is attached to two parallel bones that have fused over most of their length. Those two bones attach to a collection of bones which then attach to some longer bones that form the ends of the limbs.


Zoology of the voyage lithograph

The pattern is much more striking when the limbs of diverse vertebrates are compared. Have a look at these two illustrations from On the Nature of Limbs. One is a dugong, a large aquatic mammal, and the other is a mole, a tiny mammal known for burrowing and defacing lawns. Do you see the pattern? One bone attaches to two bones which attach to blocky bones that support digits.

dugong and mole exoskeleton

That pattern applies to bat wings and whale flippers and frog legs and chicken feet. It applies to dinosaurs and to newts. It’s a universal feature of tetrapod limbs, front and back. Neil Shubin, in his brilliant book Your Inner Fish, summarizes the pattern as a simple chant: one bone, two bones, little blobs, digits. Owen’s great insight was this: limbs are built according to a common pattern. One bone, two bones, blobs, digits.

Now, that’s a remarkable fact about the animal world, and we curious hominids are itching for an explanation. Why are all tetrapod limbs based on the same underlying pattern?

We can use Owen and Darwin to sketch the two main competing explanations: design and descent. In the simple version of the story, Owen the anti-evolutionist, the design theorist of his day, concluded that the archetype was a design, a basic idea in the mind of the Creator. Darwin, of course, proposed a radically different explanation: the “archetype” is a common ancestor, and the variations on that “theme” are exemplars of descent with modification. There’s no design, no Creator, just a lot of gradual tinkering with a setup that worked well enough at some time in the distant past.

That outline is hopelessly simplistic. Owen’s views on evolution were complex and malleable; indeed, he got in some trouble for suggesting that tetrapods (even humans) were descended from fish through “slow and stately steps, guided by the archetypal light.” Later in life, in the midst of various nasty disputes with contemporaries (most notably with T.H. Huxley, known affectionately as “Darwin’s Bulldog”), Owen did seem to oppose evolutionary ideas. But his writing in 1849 shows that he could see no reason to reject common ancestry while exploring the nature of the archetype. In other words, Owen was, at least earlier in his career, comfortable with common ancestry alongside strong conceptions of design.

#Comparative Anatomy

The idea that tetrapods arose from fish is not new; E.D. Cope proposed in 1892 that tetrapods descended from lobe-finned fish. (Modern lobe-finned fishes include coelacanths and lungfish, and comprise one of two divisions of the bony fishes. The other division, the ray-finned fishes, includes most familiar kinds of fish.) In the early days, biologists inferred ancestral relationships between species largely through comparative anatomy and embryology: they would carefully classify organisms according to their structure (including their structure during development) and look for relationships that generated nested hierarchies. A simple nested hierarchy goes something like this: 1) animals with backbones; 2) animals with backbones and limbs; and 3) animals with backbones, limbs, and hair. Animals in group 3 also belong to groups 1 and 2, while animals in group 2 also belong to group 1, and so the groups together define a nested hierarchy. Such studies alone could have led some scientists to infer an ancestral relationship between fish and tetrapods, and perhaps those studies did convince them. But then there were fossils of various types of fish and other vertebrates, many long extinct. That fossil record was relatively sketchy in 1892, but it nevertheless led Cope and others to conclude that certain fish had given rise to tetrapods at a particular time in natural history.

The fossil record shows a fish-to-tetrapod transition

Here are some basic findings from the fossil record that suggest a fish-to-tetrapod transition and that have been known for decades:

  • Fish, including fish with bones, lived on the earth before tetrapods appeared. Specifically, fossils of bony fish first appear in rocks from about 420 million years ago.
  • Tetrapods appear in the fossil record at a particular point in history and then persist and diversify in subsequent eons. Their arrival was long thought to have occurred about 365 million years ago, although some recent findings have challenged that hypothesis.
  • Tetrapods that still had some fishy features were prowling the planet 365 million years ago.
  • Lobe-finned fish that were starting to look more like tetrapods were eating other fish about 385 million years ago.

Even many decades ago, there were hints that something interesting happened between 400 million years ago and 365 million years ago. Let’s take a close look at the ancient animals that suggest the fish-to-tetrapod transition.

Ancient animals

The most primitive known tetrapods for which we have skeletal remains lived 365 million years ago. They were undeniably tetrapods, but there was definitely something fishy about them. (Heh heh.) One of the most famous of these creatures is Acanthostega, discovered in 1987 by British paleontologist Jennifer Clack and pictured below. Acanthostega is a card-carrying tetrapod, with fingers and toes. But it has a fish tail, with fin rays. Another well-known primitive tetrapod is Ichthyostega, which lived around the same time as Acanthostega. Like Acanthostega, it is a true tetrapod, but has several odd fish-like structural features. For example, its skull is more fish-like than that of Acanthostega. In summary, both Acanthostega and Ichthyostega already used the limb blueprint, even though both also had some fish-like anatomical characteristics. Their presence 365 million years ago shows that tetrapods must be at least that old, and their mixture of anatomical features suggests that the transition happened not long before that.

Acanthostega gunnari

And what about the lobe-finned fish that looked a bit tetrapod-ish? That animal is Panderichthys, described as “vaguely crocodile-shaped” with skeletal features that were tetrapod-like. Specifically, this ancient fish had tetrapod-like “shoulders,” and recent analysis found some finger-like bones at the ends of the fins. The creature also had a breathing hole on the top of its head. These fish lived around 385 million years ago.

The hunt for the earliest tetrapods

Taken together, these observations suggested that the fish-to-tetrapod transition occurred between 385 and 365 million years ago. Eager to see what that transition looked like, scientists began to look for 375 million-year-old rocks in which they might find animals at the beginning of tetrapod-hood. They wanted to catch evolution in the act.

Let’s stop and think about this, because it’s cool and because it’s important to note the extent to which evolutionary biology is hypothesis-driven. Critics of evolution sometimes portray the theory as an untestable historical conjecture, depicting it as fundamentally different from experimental science in the lab. But the hunt for the earliest tetrapods was an effort to test a hypothesis that had generated a prediction. Based on the hypothesis that lobe-finned fish were ancestors of tetrapods, scientists predicted that intermediate animals, “fishapods,” would be found in the gap between Panderichthys and Acanthostega. To evaluate the prediction, all they needed to do was find some suitable 375 million-year-old rocks.

Neil Shubin describes that search in the first chapter of Your Inner Fish. He and his colleagues found suitable rocks in the islands of the Arctic: the right age, nicely exposed (by erosion), and representative of the kind of environment that their quarry would frequent – freshwater streams. They made their biggest discovery on their last trip (“a do-or-die situation”) in 2004. That discovery was Tiktaalik roseae, the “fishapod.”

The “fishapod”

Tiktaalik roseae is one of the most extraordinary fossil intermediates ever described, and its public debut in 2006 was front-page news. An artist’s conception of the animal is pictured below.

There are several aspects of the anatomy of Tiktaalik that earn it the title “fishapod.” Like a good fish, it had scales and webbed fins. Like a tetrapod (more specifically, like a crocodile), it had a flat head, with eyes on the top of the head, and it had a neck. But what about those fins? Or are they limbs? Remarkably, the answer seems to be, “both.”

The fins of Tiktaalik were part fish fin, part tetrapod limb. On the outside, they looked like fins, with webbing. On the inside, though, these fins were clearly tetrapod-like. Amazingly, the fins of Tiktaalik were built using a primitive version of the limb blueprint: one bone, two bones, blobs, digits. As Shubin writes in Your Inner Fish, “We had a fish with a wrist.” (The Tiktaalik roseae web site at the University of Chicago is a great source for images and more information.)

Let’s address three questions about Tiktaalik that might have occurred to you. First, why might animals like Tiktaalikhave developed tetrapod-like fins? Shubin and his colleagues suggest that these limb-like fins may have been useful for doing “push-ups” in the shallow water. (Like PanderichthysTiktaalik had a breathing hole on top of its head and was clearly adapted for living and moving in shallow water.) Second, is Tiktaalik an ancestor of all tetrapods? No, not necessarily. What Tiktaalik shows us is that animals were developing tetrapod features in the context of fish bodies, and Tiktaalik shows us the context (shallow water) in which this likely occurred. But that doesn’t mean that our lineage arose from Tiktaalik itself. Finally, is Tiktaalik now the oldest tetrapod? No, apparently not. For one thing, Tiktaalik is truly transitional, and probably therefore not worthy of full tetrapod membership. But more notably, data published in 2010 show that tetrapods are a lot older than was thought at the time of Tiktaalik‘s discovery. The new findings show footprints that are unmistakably those of a tetrapod, in rocks about 395 million years old. Surprisingly, then, tetrapods were already on their way long before Neil Shubin’s specimen lived. Tiktaalik is truly a fish/tetrapod intermediate, which was living at the same time as animals that were fully tetrapods. A simple story of succession, in which intermediates disappear and are replaced by less intermediate types, seems to be an oversimplification.

In conclusion, the fossil record provides evidence that the fins of fish and the limbs of tetrapods are related by ancestry: limbs seem to be modified versions of fins. What other evidence supports this proposal? In the next post, we will turn to developmental biology, and explore the meaning of the term ‘homology.’


The fossil data paint a picture of common descent: animal limbs are descended from fish because tetrapods are descended from fish. This means that, in some sense, a limb is a fin or, more specifically, a modified fin. It’s not enough to say that fins and limbs are similar, or that they are constructed using similar principles of design. They are linked by descent, similar to the way an adult human is linked to the amorphous embryo that attached to her mother’s womb decades before.

In biology, there is a specific term for this kind of relationship. It’s called homology. Two structures are said to be homologous if they are both descended from the same ancestral structure. The key criterion is not structural or even functional similarity, although both may apply to a pair of structures of interest. The key criterion is historical continuity. When a biologist says that a bat wing is homologous to a mouse limb, she is saying that both are modified descendants from the same common ancestor.

So, for example, the head of a bat is considered to be homologous to the head of a rat. More interestingly, the wing of a chicken is considered to be homologous to the arm of a monkey or the front flipper of a walrus. Wings and arms and flippers are obviously diverse in their functions and appearance, but all are considered homologues.

Looking at embryonic development of fins and limbs

By contrast, consider the famous panda’s thumb (marked “rs” in the image below): it looks like a thumb and acts like a thumb, but it’s not homologous to the human digit that makes our hands so useful. The panda’s thumb is an enlarged carpal bone, and not a true digit at all. It’s a false thumb.

panda thumb

The example of the panda’s thumb raises an interesting question. How might we determine whether two biological features are homologous? We can look at fossils, to see if we can trace the development of each feature back to a common ancestor. And that might help a lot. Or we can look carefully at the structure of each contestant, to see if they are as closely related as a more superficial examination might have suggested. But here’s another strategy: we can trace the development of each feature during the life history of the animal. In other words, we can see if the two features have similar – or identical – embryological origins. Such a strategy can be decisively informative. In the case of the panda’s thumb, a quick glance at the anatomy of the thing shows it to be not very thumb-like, and the clincher comes when we watch (or infer) how it grew. That process is utterly un-thumb-like.

A look at the structure of a chicken wing suggests that it’s homologous to a walrus flipper – they both display the telltale blueprint despite their striking differences in outward appearance and function. But the hypothesis would be significantly strengthened if we found that both structures developed in the same way.

When we compare the embryonic development of fish fins and tetrapod limbs, we see remarkable evidence of homology. In this post, we’ll look at how the cells that make a limb or fin signal to each other. In the next post, we’ll look “under the hood” at the genes that control the process.

Embryonic development of limbs – in the chick

The chick has long been a favored system in which to study embryonic development. The process is accessible and easy to manipulate, and almost two centuries of ongoing work has generated a vast storehouse of basic knowledge. Until relatively recently, when the mouse became a dominant experimental system, almost everything we knew about the development of vertebrates came from studies of chicks. Extensive studies of chick embryos yielded a basic outline of the development of limbs that goes as follows.

  1. At a particular position in the early embryo, at a stage at which the embryo looks vaguely like a worm, a chemical signal is sparked and begins to tell the cells in the vicinity to divide. The chemical is a protein called Fibroblast Growth Factor-10, or FGF-10.
  2. As a result of that signal and the cell division it triggers, a bulge appears on the side of the embryo. It’s called a limb bud. The edge of the limb bud starts to make another FGF (FGF-8), and that signal tells the cells inside the bud to keep dividing. The bud continues to grow.
  3. The FGF-8 signal also tells the back part of the limb bud, but not the front part, to start making a special chemical messenger. That messenger is a protein called Sonic Hedgehog (SHH). This region of the limb bud is so important that it has its own name: the zone of polarizing activity, or ZPA. As you might guess, the ZPA functions to polarize the limb. Specifically, the ZPA causes the limb to have a front and a back, and SHH is the chemical message that accomplishes this.
  4. The SHH protein from the ZPA region stimulates the edge of the limb bud to become a special region called the apical epidermal ridge (the AER). The AER pumps out the proteins FGF-4 and FGF-8, which further stimulate the ZPA to ooze out SHH. It’s an auto-activating loop: SHH protein causes the ridge (the AER) to make more FGF proteins, which cause the ZPA region to make more SHH protein. In the chick embryo pictured on the right, the dark purple arc-shaped strips are the sites of FGF-8 production in the AER, which is the edge of the paddle-shaped limb buds.

mouce primordial limb bud

If that was a bit too much jargon for you, here’s a simplified version. A chemical signal, A, starts a the growth of a simple primordial limb bud. That signal also turns on a secondary signal, B, which causes the bud to grow more and which activates a localized signal, C. That localized signal will help the bud to have a front and a back, and that signal also turns on a self-reinforcing loop between C and another signal, D. That loop makes the limb grow.

Embryonic development of limbs – in the mouse

That’s how it works in the chick. When mouse genetics got big in the last two decades, biologists began analyzing the processes that control mouse development, using tools for genetic manipulation that chick developmental biologists could only dream about. You can probably guess what they found. The process is the same. The same FGFs act to induce the same SHH signal in a ZPA with the same characteristics. That ZPA induces the same FGF/SHH signaling loop.

What do we mean when we say that these systems are the same? Well, for one thing, the protein signals are highly similar to each other, suggesting that they are homologous. Furthermore, the signals are functionally interchangeable between different organisms. Amazingly, the SHH from a chicken can support development in a fly, substituting for a protein called Hedgehog that is thought to be a distant homologue of SHH. And finally, an extraordinary experiment in 1976 showed that the ZPA from a mouse can induce limb development in a chick. Think about it: the core developmental processes that make wings and arms are the same.

Embryonic development of fins

So what about fish fins? If chick wings and fish fins are homologous, we should expect to see substantial overlap in these developmental systems. And, remarkably, that’s what we see. FGF-10 gets things started, then FGF-8, SHH, and FGF-4 interact to keep things going. When FGF-10 is deleted from a mouse, you get a limbless mouse. When FGF-10 is damaged in a zebrafish, you get a finless zebrafish. (Both are pictured below.) What does SHH do in a zebrafish fin bud? Same thing it does in a mouse limb bud. The commonalities in the basic system are strong enough that one recent research report on the effects of FGFs in zebrafish discussed limb buds and fin development interchangeably.

mouse limb bud

zebrafish knockout

Developmental systems are the same in chick, mouse, and fish

These striking facts about vertebrate development are elegantly explained by common ancestry. The basic systems, established hundreds of millions of years ago, can be tweaked to get slightly (or significantly) different outcomes. The tetrapod blueprint – one bone, two bones, blobs, digits – is largely the same in diverse tetrapods because the developmental systems that make the limb are largely the same. And the roots of the system go all the way to ancient fish, whose fins contained similar structures that seem to have been built by core developmental systems maintained in both fish and tetrapods to this day.

How do we explain all this?

It seems to me that design is much less effective as an explanation for the apparent homology between fin development mechanisms and limb development mechanisms. Take the FGFs, for example. There are several other families of protein growth factors in animals, and there may be zillions of other ways to make a protein signal besides the small number of ways that we see in nature. And yet we see the same particular proteins used for these signaling purposes, over and over, in organisms as distinct as fish, chickens, and mice. Why is it that FGFs must be used for the purpose of directing the formation of a limb bud? Why are specific FGFs used for these purposes, when experiments show that the different FGFs (e.g., FGF-10 and FGF-4) are often interchangeable? Are these commonalities indicative of a design constraint, or do they represent the preferences of the designer? Without addressing such serious questions, we can’t establish design as a competing explanation. If limb and fin development couldn’t have been designed otherwise, then we should try to find out why. And if they could have been designed in myriad different ways (as I believe they could), then we should try to find out why things went the way they did. Modification from a common ancestor provides a coherent explanation, even when the developmental facts are considered alone. Add the fossil record, and I hope you can see that the explanatory power of common descent is impressive indeed.

#Gene Expression

You probably know that the complete collection of genes in an organism is its genome, and that every cell in the body of the organism has a complete copy of that genome. (Two copies, in fact.) So, your brain cells and your skin cells, though very different in form and function, both contain your entire genetic endowment. The difference between brain cells and skin cells lies largely in which various genes are turned on and off. Brain cells have brain-specific genes turned on, and just about everything else turned off. Genes that are turned on are said to be expressed, and the overall process is called gene expression.

The control of gene expression is critical in the patterning of a developing body, determining where the parts will be made, and directing those parts to take on their specific characteristics. Gene expression is under the control of proteins called transcription factors that specialize in turning genes on and off. Transcription factors exert this control by docking with sites in the genome that are near to the genes they seek to control. These sites are pretty specific, so that only particular transcription factors can control a particular gene. The transcription factor proteins are often members of large families of closely-related proteins, distinguished by the way they stick to sites in the genome.

Biologists have learned a lot about the network of gene expression that leads to the patterning of animals during development. One remarkable discovery is this: the transcription factor networks that direct basic development display extraordinary evidence of homology. Animals as different as fish and mice are sketched out in form by the expression of the same genes. And the patterning of tetrapod limbs has become a classic example.

Gene expression during limb development

Recall from the last post that biologists first learned about limb development through extensive studies of the chick. The process begins with a tiny bulge called a limb bud. That bud grows and takes shape under the influence of chemical signals released by cells in the limb bud or its vicinity. But how does the limb bud know where to start growing? And how do the different parts of the limb know what they should become? In both cases, the patterning decisions are controlled to a large extent by members of one family of transcription factors. That family is the Hox family.

Hox proteins are named after one of their defining characteristics: they contain a so-called homeobox, which is the part of the protein that enables it to dock with specific sites in the genome. (The homeobox is one type of genome-binding part; there are several other types that are employed by other families of transcription factors.) The Hox proteins are famous for controlling something called “segment identity,” meaning that they often help to tell a part of the body what it is, and specifically where that body part lies on the front-to-back scheme of things. So, for example, Hox proteins tell one particular part of the embryo that it is the tail.

As you might have guessed, Hox proteins tell the chick embryo where to grow its limbs. We will focus on the forelimb (which is a wing in the chick). The limb bud that will form a wing emerges from a boundary region in the embryo: the boundary between the cervical and thoracic levels. That boundary is delineated precisely by the expression of two Hox family members: B8 and C6. The limb bud grows just ahead of the boundary, meaning just ahead of the region that expresses Hox protein B8, in a region expressing Hox protein C6.

Patterning proteins in the limb: initial outgrowth

That’s in the chick. Are these systems homologous to those in other animals? Strikingly, they are. In a well-known comparative study, scientists in Cliff Tabin’s lab at Harvard Medical School showed that the Hox-based patterning system is the same throughout the vertebrates. Most notably, the B8/C6 boundary marks the position of the forelimb in animals with widely varying structures. In the diagram below, taken from that paper, the B8 expression range is marked by the shaded circles and the position of the limb bud is indicated by a bud-shaped arc. The limb bud’s position is determined by the same Hox proteins in the chick and in the mouse. And in the fish and in the frog. And even in the goose, which has a ridiculous number of added vertebrae to hold up its ridiculously long neck. The key simple point here is this: in all of these animals, the Hox expression pattern underlying the point of emergence of the limb bud is the same. Note that there are several other transcription factors known to be involved in limb emergence, including two that are specific for a forelimb vs. a hindlimb, and those are the same across the board as well. It’s a remarkable example of homology in developmental genetics.


Patterning proteins in the limb: specialization

But once the limb starts to grow, how do the different regions of the limb become specialized? Is there a Hox expression pattern behind that, too? Indeed there is, and again we’ll start with the chick and focus on the forelimb.

The development of the limb bud seems to proceed in three phases, each generating one of the three basic zones of the limb. Those zones are the upper arm/wing, the lower arm/wing, and the digits. In the diagram below, you’ll see those anatomical regions (e.g. the humerus, which is the upper arm/wing) and you’ll see some technical terms for the region (e.g. stylopod). What matters to us is the pattern of Hox protein expression. Phase I involves two Hox proteins, D9 and D10. This is followed by Phase II, which is significantly more complex. Now there are a few more proteins involved (D11, D12, and D13), and they are expressed in a more interesting pattern. Finally, Phase III introduces a new pattern, generated by some additional players. To recap: Phase I generates the upper limb, Phase II generates the lower limb, and Phase III generates the stuff at the end. Each phase is characterized by a particular pattern of expression of Hox proteins.

matheson hox proteins

The patterns of Hox expression provide clear evidence of homology. Phases I and II are the same in chicks, mice, and fish. (Even sharks.) And again, those observations are coherently explained by common descent. Moreover, the confluence of the various observations we have discussed in this series – shared structural features, fossil intermediates, identical signaling systems, and now identical genetic control systems – very strongly suggests common ancestry. Common descent provides a singular explanation for a large collection of related phenomena.

But what about Phase III? Is that phase conserved? It was long thought that the answer was “no.” For one thing, the structures at the ends of animal limbs are wildly different: fingers, hooves, flippers, wings. And more importantly, early observations from various disciplines (anatomy, genetics, development) suggested that the Phase III structures in tetrapods were something completely new. Specifically, those early findings led most experts to conclude that while Phases I and II were modified from a fish ancestor, Phase III – the part that seems to generate all the diversity – was a new invention.

#Hands and Feet, Fingers and Toes

Did hands and feet evolve from fins, too?

Until quite recently, there were fairly compelling reasons to think that the development of hands and feet involved completely new processes. Hands and feet are highly specialized for various tetrapod lifestyles, suggesting evolutionary novelty (at the time). More importantly, today’s fish – even those thought to be most closely related to tetrapods – seem not to have structures that resemble digits, nor were such structures observed in potential ancient ancestors. And finally, the genetic systems that control digit development seemed to differ in important ways from those that control fin development. Beginning just a few years ago, however, new data began to point to a different conclusion. Let’s review those recent findings, concluding with a detailed look at some fresh new results from some fascinating experiments.

Look again: proto-fingers found in an ancient fish fin

proto-fingers in ancient fish fin

Recall that the fossil record includes some intriguing evidence for a fins-to-limbs transition. Tiktaalik appears to be a tetrapod with fishy characteristics, and Tiktaalik had tetrapod-like “hands.” But then there’s the ancient fish Panderichthys, the transitional ancestor that appears to be closer to fish than to tetrapods. Panderichthys seemed not to have anything resembling hands, strengthening the conclusion that hands and digits were invented in the tetrapod lineage. But in 2008, as genetic evidence to the contrary was mounting (see below), Per Ahlberg and his colleagues took a closer look at Panderichthys using a CT scan. (Also known as CAT scanning, this is a technique that uses X-rays to build a 3-dimensional image of a solid structure. Its most familiar use is in medical imaging.) They found that previous analysis of the fossil had missed key features of its skeletal anatomy. Specifically, they found proto-fingers at the end of the creature’s preserved fin. Have a look at the color-coded image below, from their paper. The green section at the top is the humerus. The yellow and the turquoise are radius and ulna. The rust-colored shapes at the bottom are the proto-fingers. There they are: one bone, two bones, blobs, digits.

Ancient gene expression patterns in hands and fins

The genetic story is somewhat more complicated. Early molecular genetic studies had led biologists to suspect that the ends of limbs were distinct from the ends of fins, although there was never a strong consensus around that view. But about 5 years ago, around the time that Tiktaalik was revealed, several different labs reported data that began to tip the scales. Looking at the development of fish of various types chosen based on their evolutionary relationships, all of these labs found striking similarities between the gene expression patterns in fins and the patterns in limbs. In each case, biologists discovered Hox gene expression patterns that strongly resembled Phase III patterns in mice and chicks. It was unmistakable: whether in standard fish like zebrafish, or distant relatives like paddlefish or even sharks, the three-phase pattern of Hox gene expression lays the groundwork for the development of a fin, just like it does in the development of an arm – and a hand.

Taken together, the fossil evidence from an ancient fish and the surprising homology of Hox gene expression patterns strongly suggest that even the blobs and the digits are modified versions of very ancient structures in fish. But a question remained: just how much functional similarity exists between those genetic patterning systems? In other words, to what extent is that Phase III expression pattern truly homologous between fish and tetrapods? One possibility is that the Phase III Hox expression patterns are similar only superficially; perhaps somehow the pattern emerges from other conserved processes, and the similarities don’t mean much. And the other possibility is that the pattern is set by an ancient genetic switch, inherited by both fish and tetrapods from a common ancestor.

Recent experiments on interchangeable genes: chick genes and mouse paws

Biologists in Neil Shubin’s lab addressed this question in some elegant experiments published in August 2011. They already knew that the Phase III pattern was controlled in mice by a genetic switch called an enhancer, and they knew roughly where the enhancer was located in the mouse genome. Because they knew the DNA sequence of the enhancer, they were able to find regions of other genomes that are similar in sequence and located near the Hox genes of interest. They looked in mouse, chicken, frog, zebrafish, and skate, animals chosen based on likely evolutionary relationships. And they found the switch in each of those animals. Their question, then, was this: does that switch do the same thing in fish as it does in mice? In other words, will the switch turn on Phase III Hox gene expression patterns in each animal?

They answered this question in somewhat dramatic fashion by doing what we’ll call a genetic transplant experiment. They took the switch from one animal and put it into another. That alone is awfully cool, but they did the transplants in such a way that the activity of the transplanted genetic switch could be easily visualized. So, whenever the transplanted switch turned on gene expression, the pattern would be visually apparent.

In one set of experiments, the mouse switch was transplanted into fish embryos. The picture below shows the result. The developing fin is outlined, and the arrowhead points to gene expression that was turned on by the mouse switch. The genes are being expressed at the fringe, in a pattern similar (but not identical) to the Phase III pattern normally seen in developing fish fins. Think about this, because it’s really remarkable: a mouse genetic switch turned on a very specific genetic system in a fish, and in the same place and time as it would in the mouse. This is very strong evidence that the two switches (mouse and fish) are the same. They are homologous.

mouse csb in zebra fish

Then, in another set of experiments, switches from chick, fish, and skate were transplanted into developing mouse embryos. In each panel below, you’re looking at the developing limb of a mouse; notice the paddle-like shape of the soon-to-be paw. The blue color represents gene expression that was turned on by the transplanted switch. When the chick switch is put into the mouse, we see blue right at the end of the limb, at the base of the paw. That’s the same pattern as the one generated by the mouse’s own switch. And that result is notable by itself: it means that the tetrapod switches are interchangeable to a large extent. But more remarkable are the effects of the switches from the fish and the skate. Both are also capable of turning on a Phase III-like pattern of gene expression, as indicated by the appearance of blue at the base of the paw. To learn why the skate switch seems “better” than the zebrafish switch, try reading the article, which is freely available at the journal’s web site.

chick fish skate csb in mouse

A deep shared history

In conclusion, we see that studies of the molecular genetics of animal development strongly indicate common ancestry of fish fins and tetrapod limbs. We’ve looked at shared anatomical structure, curious fossil intermediates, oddly conserved signaling systems, and, now, remarkable genetic similarities that extend to the end of the fin or limb and deep into their shared history. It’s a large collection of varied observations, a collection that we have merely sampled. Descent from common ancestors offers a cohesive and compelling explanation that no competing explanation can match.


Not just homology. Deep homology.

We saw that genetic control systems involved in the development of fish fins and animal limbs are strikingly similar, so much so that genetic switches can be transplanted among organisms as different as fish, birds, and mammals. Observations like those strengthen the conclusion that chick wings, mouse paws, and (amazingly) fish fins are homologous, meaning that they are altered versions of a common ancestor. Chick wings and mouse paws are classic examples of structures that display homology.

There are, however, zillions of other animals with legs. It’s hard to think of an animal that doesn’t have appendages of some kind. What about all those other legs? Are they all homologous to each other? Specifically, are the legs of insects somehow homologous to the limbs of mice? (I’ll focus on insects here, but the same conclusions apply to all arthropods.)

The answer is: almost certainly not. For one thing, there are few structural similarities. More importantly, a look at the animal family tree shows that there are animals “between” insects and tetrapods that don’t have structures comparable to either limbs or insect legs. In other words, fins and insect legs each developed in different organisms, who originally had no appendages at all. The two branches arose independently, as near as we can tell, and so fly legs and mouse paws are not homologous.

But, shockingly, the genetic systems that control their growth are strikingly similar. Fly leg development is controlled by Hox genes. More amazingly, fly leg development is controlled by many of the same Hox genes that perform similar tasks in mice. Fly legs are not homologous to mouse paws, but it looks like the genetic systems that control their development are, to a large extent, homologous. Neil Shubin, Cliff Tabin, and Sean Carroll coined the term ‘deep homology’ to describe this phenomenon in the influential 1997 article “Fossils, genes and the evolution of animals limbs”. Deep homology, then, refers to historical continuity of genetic control systems that underlie patterns or forms that are not so evidently homologous—like fly legs and mouse paws.

Deep homology has been seen in other developmental contexts. One of the more famous examples is the development of the eye or, more accurately, the development of eyes. Eyes are wildly different in different kinds of animals (consider the camera-like human eye compared to the compound eye of a fly), and they develop in very different ways in different kinds of animals. They seem not to be homologous; on the contrary, they seem to have evolved independently dozens of times. But a look under the hood shows remarkable deep homology – the genetic control systems are homologous. In one of the most dramatic experiments in developmental genetics, scientists in Walter Gehring’s lab in Germany were able to induce the growth of extra eyes in fruit flies, by flipping a genetic switch, and it didn’t matter whether the switch came from a fly or from a mouse. The mouse switch could turn on eye development in a fly. The picture below shows a fly with eyes growing on all six of its legs! (The eyes are the reddish-brown blobs on the legs, which are on the left side of the photo.)

fly eyes on legs

Deep homology has important implications for our understanding of the development of life on earth, and its discovery helped trigger the birth of a new branch of biology, a sort of hybrid of evolutionary biology and developmental biology: evolutionary developmental biology, or evo-devo. The fins-to-limbs story is an exemplar of evo-devo thought, which emphasizes the role of conserved genetic control systems in the development of the magnificent diversity of animal forms. The basic idea is that these ancient genetic control systems make excellent “toolkits” that can be modified extensively to generate those seemingly endless “variations on a theme” that fins and limbs exemplify so beautifully. Fly legs and human arms may not be structural homologs, but the deep homology in their genetic underpinnings suggests that they are variations on a theme – perhaps we can call it the “outgrowths” theme – that was established in some of the most ancient animals of all.

Tetrapods without limbs?

We saw in the first post that snakes and whales are classified as tetrapods, even though neither has four limbs. What’s up with that?

The simplest explanation is this: snakes and whales are classified as tetrapods because they are descended from tetrapods. To a biologist interested in classification, the term ‘tetrapod’ refers to a group of organisms related by descent, and not necessarily to a group of organisms that have a particular set of features. This is more common than you might think. For example, there are flies that have neither wings nor legs, but are still classified as flies (and as insects). Darwin himself, in the Origin of Species, wrote that “all true classification is genealogical,”meaning that life is best understood as a “community of descent.” So, snakes and whales are animals with less than four limbs that have descended from animals with four limbs.

Why do biologists think that snakes are descended from card-carrying tetrapods? The evidence they rely on is the same kind of evidence that we looked at throughout this series: shared anatomical features, fossil intermediates, conserved developmental and genetic toolkits. Consider, for example, the pelvic girdle (the part of the skeleton to which the legs attach) of a tetrapod. Some snakes (like pythons) still have that pelvic girdle, and those snakes grow tiny limb buds. Fossil snakes with tiny (and shrinking) hindlimbs have been clearly described. What about the genetic toolkit? It’s conserved, with some changes that inactivate the limb-growing signal at just the right time. We know the signal is there because a limb bud from a python can cause limb growth in a chicken! Snakes, it seems, are elongated tetrapods with reduced (or erased) leg development.

The natural history of whales is better known thanks to some truly spectacular fossil intermediates discovered in the last 20 years or so. That story has been nicely covered previously on this blog, and the outline is familiar: common anatomical themes, striking fossil intermediates, and genetic data all point to a land-dwelling tetrapod mammal as the common ancestor of all whales (including dolphins). We can “watch” the hindlimbs disappear in a historical sketch based on fossils, and we can see what happens when the hindlimb-making program is accidentally turned back on. As you might have guessed, that rare event results in a dolphin with vestigial hindlimbs, arising from seemingly normal limb buds that occur in every dolphin embryo. Dolphins, it seems, are streamlined tetrapods with reduced (or erased) hindlimb development.

Descent versus design

The natural history of tetrapods is a superb example of the explanatory power of common ancestry. Each line of evidence – anatomy, fossils, developmental mechanisms, shared genetic toolkits – is interesting and suggestive. But taken together, the data form a distinctive body of observations that are convincingly explained by common descent. No competing explanation comes close.

Design, of course, is one proposed explanation for some of the data we’ve examined. As a competing explanation, design is currently a failure. Whether we look at structural themes or underlying genetic systems, we can identify no clearly-defined principles of functional optimization. In other words, we have no good reason to suppose that the tetrapod limb blueprint (one bone, two bones, blobs, digits) represents the only way – or even the best way – to build a limb (or a wing or a flipper). We have no good reason to suppose that Hox genes are the only way – or the best way – to turn on gene expression systems underlying limb development. We have no good reason to suppose that the Sonic Hedgehog protein is the only way – or even the best way – to send growth signals through a limb bud (or a fin bud). In each of those cases, we have no good reason to suppose that it could not have been otherwise, and in some cases, we know that it can work in other ways.

Design does not help us make sense of the patterns in the fossil record as exemplified by animals like Tiktaalik. Design does not help us make sense of the temporary hindlimbs of embryonic dolphins, nor does it account for deep homology. Design, to whatever extent it stands for functional optimization or even excellence, is unhelpful as an explanation when it comes to limbs and fins and their shared characteristics.

This could change, of course. Scientists may learn that there are yet-unknown design constraints that dictate one or a very few optimal configurations for limb and fin construction, such that things really could not have been otherwise. (Perhaps those scientists who prefer design-based explanation will take the lead on this one.) In fact, I argue that the only way to establish design as a competing explanation, one that can rival the explanatory power of common descent, is to convincingly demonstrate something akin to a set of profound biological design constraints. To unseat common ancestry, or at least to rival it, design theory has to show that life could not have been otherwise.

Descent with design

But design need not compete with common ancestry. Some design theorists do in fact accept common descent, and seek to identify marks of design in the processes by which the tree of life came to be. These attempts vary in quality, and the better-known examples have been discussed on this blog in detail. The point is that design thought, whether it posits supernatural “intervention” or mulls “front-loading,” need not be inherently dismissive of the explanatory power of common ancestry.

In fact, I think some emphasis on design is important to retain while considering evolution. For Christians, one reason for this is simply the fact that we confess the biological world to be the creation of the Creator God. There’s also another more subtle reason: design in some form may be a necessary explanatory principle. One point at which design and descent seem confluent is in the ideas and emphases of paleontologist Simon Conway Morris. Now in a complete web site, Conway Morris’ theme is evolutionary convergence – the striking tendency of evolution to repeatedly “discover” design themes. Perhaps ideas like his are pointing us to principles of design and purpose that run deep in the living world, or perhaps they simply tell us a bit more about how God’s world really works.

Dear BioLogos reader ...

In the escalating vitriol in our culture, “science” and “faith” have found each other on opposite sides of a polarized divide. Truth and community are under attack.

If there is one thing the pandemic has shown us, it is what science can and cannot do. Scientists and doctors have done amazing things during the pandemic—identified the virus, treated the disease, and developed safe vaccines that work.

But in these polarized times, science can’t reduce anger, forgive sins, build mutual respect, or fill us with compassion for others.

Science alone can’t give us hope. Faith can. Join BioLogos today in reaching a world desperate for hope. Your tax-deductible donation will be the difference between someone encountering misinformation, or a thoughtful, truthful, and hopeful Christian perspective that shows faith and science working hand in hand.

Give Now

Table of Contents