Two ways to look alike
Two living things can resemble each other for two very different reasons.
The first reason is inheritance. They share a trait because an ancestor had it and passed it down. Biologists call this homology.
The second reason is a shared problem. Two unrelated lineages faced the same challenge and hit on similar answers. Biologists call this analogy, and the process that produces it is convergent evolution.
Telling these apart is one of the core skills in biology. Get it wrong and you build the wrong family tree. Get it right and both kinds of similarity become evidence.
Homology: the same bones, again and again
Hold up your arm. Feel the single long bone from shoulder to elbow. That is the humerus. Below the elbow there are two bones side by side, the radius and the ulna. Then comes a cluster of small wrist bones, and then five digits.
Now open a bat’s wing. Humerus, then radius and ulna, then wrist bones, then digits. The digits are enormously long and skin is stretched between them, but the parts and their order are the same.
A whale’s flipper. Humerus, radius and ulna, wrist bones, digits. Wrapped in a smooth paddle of flesh, but there they are.
A horse’s leg. Same plan again. The horse walks on the tip of one enlarged digit, and much of the rest is reduced, but the sequence is unchanged.
Four animals, four completely different lives, one shared underlying structure. This is not what you would expect if each limb were designed separately for its job. A paddle, a wing, and a running leg have almost nothing in common as engineering problems. There is no reason a good paddle should contain a wrist.
There is an excellent reason if all four are modified versions of the same limb, inherited from a four-limbed ancestor. Biologists call four-limbed vertebrates tetrapods. They trace back to fish that were moving toward land roughly 370 million years ago. One of those fish, the 375-million-year-old Tiktaalik roseae, already had the beginnings of this arrangement inside its fin. Fossil evidence covers that find in full.
Homology is not only about bones. The same logic applies to genes, to embryo stages, and to whole body plans.
Convergence: the same answer, found twice
Now the other kind of similarity.
Eyes
Your eye has a lens, an iris that opens and closes, and a retina. So does an octopus eye. The resemblance is remarkable. Yet the last common ancestor of humans and octopuses had nothing like either. It had, at most, simple light-sensitive patches.
Look closer and the difference shows. The two eyes are wired in opposite directions, and only the vertebrate version ends up with a blind spot.
Two lineages built a camera eye. They wired it differently, because they built it separately. How did the eye evolve takes that comparison apart in detail.
Wings
Birds, bats, and insects all fly with wings. The wings are built three different ways.
A bird’s wing is feathers extending along the arm. A bat’s wing is a membrane of skin stretched between hugely elongated finger bones. An insect wing has no bones at all. It is an outgrowth of the body wall with a network of veins running through it.
Here is a nuance worth holding onto, because it shows how careful the reasoning has to be. A bird wing and a bat wing are analogous as wings, because flight evolved separately in the two groups. But they are homologous as forelimbs, because birds and bats both inherited that humerus, radius and ulna from a shared four-limbed ancestor. The same pair of structures can be homologous at one level and analogous at another. You have to ask which trait you are comparing.
Fast swimmers
A shark, a dolphin, and an ichthyosaur have nearly the same silhouette. Torpedo body, pointed snout, a dorsal fin on the back, paired fins at the front, a tail fin at the rear.
They are not close relatives at all. A shark is a fish with a skeleton of cartilage. A dolphin is a mammal with bones, lungs, and warm blood. An ichthyosaur was a marine reptile whose ancestors walked on land; the group died out around 90 million years ago.
Water is unforgiving. A body that pushes water aside efficiently at speed has a fairly narrow range of possible shapes. Three separate lineages were pushed into that shape by the same physics.
The giveaways are internal. A shark breathes through gills. A dolphin comes to the surface for air. A shark’s tail beats side to side; a dolphin’s tail flukes beat up and down, because its ancestors ran on land with a spine that bent that way. The outline converged. The insides did not.
Echolocation, right down to the genes
Some bats and some dolphins both hunt by sending out high-pitched sounds and listening for echoes. The two groups are separated by tens of millions of years, and the shared ancestor did not echolocate.
The surprise came from the DNA. A protein called prestin sits inside the ear and helps animals hear very high sounds. In 2010, Yang Liu and colleagues reported something odd. Echolocating bats and dolphins had picked up matching changes in the prestin gene, separately.
If you build a family tree using prestin alone, dolphins land next to bats. That answer is wrong, and we know it is wrong because every other kind of evidence says so.
In 2013, a team led by Joe Parker published a genome-wide study in Nature and found this pattern was not limited to one gene. Many independent genes, especially hearing genes, showed convergent changes across echolocating mammals.
This is convergence at the deepest level we can look. It is also a warning: a single gene can mislead you, which is exactly why biologists test relationships across whole genomes and many traits at once.
Plants that copied each other across an ocean
Cacti grow in the Americas. Succulent euphorbias grow mainly in Africa and Madagascar. Many species in the two groups look almost interchangeable: thick green water-storing stems, ribs, no real leaves, sharp spines.
The two families are not close relatives at all. They sit far apart on the flowering plant family tree, and they live on separate continents. They faced the same problem, which is staying alive in hot, dry places, and they reached much the same shape.
You can tell them apart with two checks. Cactus spines grow from small cushion-like structures called areoles, which euphorbias do not have. And a broken euphorbia stem leaks milky white latex, which cacti do not produce. Their flowers are built on entirely different plans.
Two mammal groups running the same experiment
Australia’s mammals are mostly marsupials, which raise young in a pouch. Most mammals elsewhere are placentals. The two groups have been mostly separate for tens of millions of years, and each has produced its own version of similar lifestyles.
The thylacine, Thylacinus cynocephalus, hunted in Tasmania and had a skull strikingly like a wolf’s. The last known one died at Hobart Zoo in 1936. The marsupial mole burrows with the same shovel-like build as a placental mole. The sugar glider glides between trees on a skin flap, just as the flying squirrel does on the other side of the world.
None of these pairs are close relatives. Each is a marsupial and a placental arriving at the same answer, twice.
How biologists actually tell the difference
Surface appearance is the least reliable clue. Here is what researchers use instead.
Deep structure. Are the internal parts arranged the same way, in the same order, connected to the same neighbours? The humerus is always the single upper bone, always attached at the shoulder. That kind of correspondence survives even when shapes change wildly.
Developmental origin. Where does the structure come from in the embryo, and out of which tissue? A bat’s wing membrane grows between digits. An insect’s wing grows out of the body wall. They can never be the same structure, however similar the finished shape.
Fit with the wider tree. No single trait decides anything. Biologists build family trees from many characters, from fossils, and from whole genomes. A similarity that shows up in a group that everything else places together is probably inherited. A similarity that would force you to break a well-supported tree is probably convergent.
Genetic detail. Convergent traits usually get there by different genetic routes, or by different changes to the same gene. Even the striking prestin case shows measurable differences alongside the matches.
Why both kinds of similarity are evidence
Homology gives you the branching pattern of life. Whale flippers and horse legs carrying the same bones is the sort of thing common ancestry predicts and separate creation does not.
Convergence gives you something else. It shows that natural selection is a real, repeatable process. Run a similar environment twice, in unrelated lineages, and you get similar answers twice. Sharks, dolphins, and ichthyosaurs are three independent runs of the same experiment, and all three came out looking alike.
Together they explain both the pattern and the process. For more on that process, read natural selection. For the other lines of evidence, read evidence for evolution.