Convergent evolution
A Socratic walk-through of convergent evolution — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why do unrelated animals living in similar places end up looking so alike?
Put a shark, an ichthyosaur and a dolphin in a row and they read as variations on one animal: torpedo body, dorsal fin, tapered snout, flippers. Their ancestries could hardly be further apart — a fish, a reptile, a mammal, separated by hundreds of millions of years and by ancestors that walked on land in two of the three cases.
The naive reading is that resemblance means relatedness. Here it plainly does not. So what is the resemblance actually reporting?
Reasoning it through
REASONING #Ask what the three animals have in common that is not ancestry. They all move fast through water, which is roughly eight hundred times denser than air. That is not a preference; it is a physical constraint, and it does not care what lineage you belong to. Drag rises steeply with speed and with the wrong shape, so any lineage that moves fast in water is pushed by the same physics towards the same profile.
Notice what that reframes. Selection is not free to invent; it is answering a problem posed from outside the organism. If the problem is well-posed and the number of good solutions is small, then different starting materials will be pushed towards the same answer. Similarity of form is then evidence about the problem, not about the family tree.
Does that hold up when we look closely? The test is whether the solutions are built the same way. They are not — and the differences are the giveaway. A shark's tail beats side to side; a dolphin's beats up and down, because a dolphin descends from a running mammal whose spine already flexed vertically. The outside shape converged; the machinery kept its history.
The clearest case is the eye. Vertebrates and squids both have a camera eye — a lens, an iris, a focusing mechanism, a retina at the back. But the vertebrate retina is inverted: the light-sensing cells face away from the incoming light, so light passes through a layer of nerve wiring before reaching them, and the bundled wires must exit through the retina, leaving a blind spot. The cephalopod retina faces the right way and has no blind spot. Both eyes solve the problem of forming an image; only one of them carries a manufacturing decision that looks like an error. That is exactly what independent origins should look like — and it is why the resemblance cannot be inherited, since no ancestor would have handed down one arrangement to squids and its inside-out version to us.
How often does this happen? Powered flight has been invented at least four separate times — by insects, pterosaurs, birds and bats — and no two of them build a wing the same way. A pterosaur stretched a membrane on one enormously elongated finger; a bat stretched one across four; a bird made an aerofoil out of feathers on a fused forelimb; insects grew wings that are not modified limbs at all.
The analogy
THE ANALOGY #Think of separate engineering teams, in different countries and without contact, each handed the same specification: carry a load quickly across water. They will not produce identical boats, but every one of them will produce something long, narrow and pointed at the front, because the water is writing much of the specification. What differs will be what each team already had in the workshop.
Engineers work towards a stated goal and can discard a bad start, whereas selection has no goal and cannot discard anything — it can only modify what the lineage already carries, which is why the dolphin is stuck with a vertically flexing spine and we are stuck with a blind spot.
Clarifying the model
THE MODEL #The distinction to hold on to is between characters that are similar because of shared ancestry and characters that are similar despite it. The forelimb bones of a bat and a whale correspond one for one because both lineages inherited them from a common ancestor; that is homology. A bat's wing and an insect's wing correspond in function and in nothing else; that is analogy, and the general term for a resemblance arrived at independently is homoplasy. Sorting the two is the working problem of comparative anatomy, and the reason convergence is a nuisance as well as a lesson: it plants false evidence of relatedness in every dataset, which is much of why molecular sequence data changed the field.
Two honest complications.
First, "wholly independent" is often too strong. Vertebrate and cephalopod eyes are separately built, but both lineages deploy related master control genes in eye formation, inherited from a distant common ancestor that had no camera eye. Biologists call this deep homology: the solutions are independent, the toolkit partly is not. Where to draw that line is a live argument.
Second, convergence is not universal, and its frequency is a real dispute about how constrained life is. The C4 photosynthetic pathway has evolved independently in plants on the order of sixty times, which suggests some problems have very few good answers. Others — feathers, the vertebrate jaw, the eukaryotic cell — appear to have happened once. The interesting question is not whether evolution repeats itself, but which problems force a repeat and which admit only a single accident.
A picture of it
THE PICTURE #How to readRead left to right as deep time, and read each section as a separate lineage, not a stage in one story — nothing here is descended from the entry above it. The dates are the earliest reasonably secure fossil evidence, so they mark when we can see flight, not when it began. The second line in each section is the point: the same capability, four times, and four structurally different wings. If flight had been inherited, those four descriptions would match.
What became clearer
WHAT CLEARED #Resemblance between distant lineages is not a claim about ancestry but a measurement of how tightly the environment constrains the answer. Where the physics leaves few good solutions — moving through water, forming an image, staying aloft — different starting material converges on the same outward form, while the underlying construction stays stubbornly loyal to whatever each lineage happened to bring with it. Which is why the informative part of a convergence is rarely the similarity. It is the seam where the two solutions were built differently.
Where to go next
ONWARD #- Why molecular data resolved family trees that anatomy could not, and where convergence still fools it.
- How predictable evolution is in principle — whether replaying the tape would produce the same body plans again.
Key terms
TERMS #| Term | What it means |
|---|---|
| Convergent evolution | independent lineages arriving at similar traits under similar selective pressures. |
| Homology | similarity inherited from a common ancestor, such as the shared bone plan of tetrapod forelimbs. |
| Homoplasy | similarity not inherited from a common ancestor; convergence is its main cause. |
| Deep homology | independently evolved structures built using the same ancient genetic toolkit. |
Every term the collection defines is gathered in the glossary.