Design guide · six problems, one audit
Every guide to believable creature design says the same three things: study real animals, give every part a purpose, and the creature will feel real. None of them says how to check. Biologists have a check, and they have been running it for two centuries. It is called convergent evolution. When two animals with no common ancestor worth mentioning end up with the same body, that body was the answer to a problem, not an accident of family. The standard example is a shark and a dolphin: a fish and a mammal whose lines parted more than four hundred million years ago, both torpedo-shaped, both with a fin on top.
For a designer that is better than a rule of thumb. If nature has posed your creature's problem more than once and the answers agree, the shared part is a constraint to honour, and the parts that differ are where the creature gets to be itself. Below are six problems nature has solved at least twice, laid out as a ledger: who solved it, what each came up with, where the answers meet, and the one detail that still gives each lineage away. The page ends with a five-question audit and two creatures from the hybrid animal generator, one that fails it and one that passes.
The problem
Sharks
Cartilaginous fish · in the sea for over 400 million years
Torpedo body, a stiff fin on the back, a crescent tail swept from side to side.
Ichthyosaurs
Reptiles · at sea from about 250 to 90 million years ago, from land-living ancestors
Torpedo body, a boneless fin on the back, a crescent tail with the spine bent down into the lower lobe.
Dolphins
Mammals · back in the sea for about 50 million years
Torpedo body, a boneless fin on the back, horizontal flukes beaten up and down.
Shared answer
The fusiform body, thickest a third of the way back and tapering to a narrow tail stalk, with a fin on top to stop it rolling. The convergence goes deeper than the outline. In 2018 a Stenopterygius from the Holzmaden shales turned out to preserve a layer of blubber and a dark back over a pale belly: the same insulation and the same countershading a living dolphin has, on a reptile that died 180 million years ago.
The tell
The tail. Sharks and ichthyosaurs beat theirs side to side; every whale and dolphin beats it up and down, because their ancestors ran on land with a spine that flexed vertically. Draw a torpedo with flukes and you have promised the viewer a mammal, so it also needs a blowhole.
The problem
Bats
Mammals · most of the thousand-plus species echolocate
Ultrasonic calls made in the larynx and sent out through the mouth or the nose; echoes caught by large outer ears.
Toothed whales
Mammals, on a different branch · dolphins, porpoises, sperm whales
Clicks made by a pair of “phonic lips” in the nasal passage, focused through the fatty melon on the forehead; echoes received through the fat-filled lower jaw.
Shared answer
Echolocation, and more of its hardware than anyone expected. Prestin is the protein that lets the ear's outer hair cells amplify high frequencies. In echolocating bats and in dolphins it has picked up the same amino-acid changes, so that a family tree drawn from that one gene groups bats with dolphins instead of with their real relatives. A 2013 scan of whole genomes found nearly two hundred genes carrying the same convergent signal, many of them for hearing.
The tell
Where the sound comes out. A bat's face is all ears and mouth; a dolphin has no outer ears at all and shouts from its forehead. Give a creature echolocation and you have to choose which body you are borrowing. Big ears on a dolphin-shaped head read as a cartoon.
The problem
Grey wolf
Placental mammal · order Carnivora
Long muzzle, big canines, shearing cheek teeth, a skull braced for a bite that hangs on.
Thylacine
Marsupial · Tasmania · the last known animal died in 1936
The same muzzle and the same canines, on a skull that behaved like a dingo's when a computer loaded it with the forces of a bite.
Shared answer
The dog skull, reached from opposite ends of the mammal tree; the two lineages parted about 160 million years ago. When the thylacine genome was published in 2018, the authors traced the likeness partly to changes in the regulatory DNA that shapes the head rather than to the genes for the parts themselves. A 2007 simulation loaded three-dimensional models of a thylacine and a dingo skull with biting forces and got similar performance, with the thylacine coming out weaker under the loads a large, struggling animal would apply, so it probably hunted smaller prey than its build suggests.
The tell
Count the teeth. A wolf has 42; a thylacine had 46, with four molars on each side of the upper jaw where a wolf has two, and eight upper incisors to the wolf's six. Below the skull the tell was a pouch, in both sexes, and a stiff tail that could not wag. Convergence copies the tool, not the whole animal.
The problem
Vertebrates
Fish to humans · one eye design inherited down the whole line
A single lens, an iris, a retina at the back. Focus by changing the lens's shape. Photoreceptors point away from the light, behind their own wiring, so the wiring leaves through a hole: the blind spot.
Octopus and squid
Molluscs · closer kin to snails than to anything with a spine
A single lens, an iris, a retina at the back. Focus by moving the lens forward and back, like a camera. Photoreceptors face the light with the wiring behind them. No blind spot.
Shared answer
The camera eye, invented twice in lineages whose common ancestor had at best a patch of light-sensitive cells. In 2004 a comparison of the genes switched on in an octopus eye and in a human eye found a large shared set, including many of the same genes recruited to build the eye in the embryo. The two lineages arrived at similar cameras partly by reaching for the same old parts.
The tell
The pupil. An octopus pupil is a horizontal bar, and the animal rotates its eyes to keep the bar level with the horizon however its body tilts. Draw a round pupil in a white eyeball on anything built like a mollusc and the eye will read as a mammal's, glued on.
The problem
True crabs
Brachyura · about 7,000 living species
Wide flat shell, abdomen folded under the body, four pairs of walking legs behind the claws.
King crabs
Anomura · descended from hermit-crab-like ancestors
Wide flat shell, abdomen folded under, three visible pairs of walking legs; the fourth pair is tiny and tucked inside the shell to clean the gills.
Shared answer
The crab shape: a flattened, widened carapace over a tail tucked out of sight. A 2021 review counted at least five separate occasions on which decapod crustaceans went crab-shaped, from ancestors that looked more like lobsters or hermit crabs, and a few on which crab-shaped lineages turned back. Biologists call it carcinisation, and it is the clearest case there is of one body plan being the answer to a way of life on the sea floor.
The tell
Count the walking legs from above. Four pairs and it is a true crab. Three pairs, often with a lopsided shell, and it is a hermit crab's descendant that has folded itself into a crab; the asymmetry is the leftover of a spiral shell it no longer lives in.
The problem
Mexican tetra, surface form
A river fish of north-eastern Mexico and Texas
Eyes, silver scales, a dark stripe: a standard small fish.
Mexican tetra, cave form
The same species, in about thirty caves · colonised more than once
Eyes gone, pigment gone; more taste buds, a bigger jaw, a sharper lateral line, more fat in reserve.
Shared answer
Subtraction. Convergence does not only add fins and teeth; in caves it removes the same things every time, because an eye costs energy to build and a colour costs energy to make, and neither pays off in the dark. The Mexican tetra is the textbook case because the surface fish and the cave fish are one species, and the cave populations lost their eyes separately in different caves, which is convergence running inside a single animal.
The tell
The eye that starts and stops. A cave tetra embryo begins to build an eye, then the lens dies and the eye sinks back under the skin. Loss in evolution is rarely clean; the animal keeps a trace of what it came from. A cave creature drawn with smooth skin where the eyes were is less convincing than one with a dent.
When the likeness lies
Convergence is a good check, and your own eye can still game it. Thylacosmilus atrox, a South American relative of the marsupials from roughly nine to three million years ago, carried canines as long as a sabre-toothed cat's, growing all its life, with roots that curved back over the top of the skull. For a century it was written up as the marsupial Smilodon.
In 2020 a team led by Christine Janis went through the skull tooth by tooth and argued that it was nothing of the kind. It had no upper incisors, the front teeth every cat uses to grip and to strip meat. Its canines wore in a pattern that fits pulling and slicing soft tissue, not stabbing. Its skull was built to resist forces pulling backwards, not the twisting of a struggling animal. Their reading is a beast that opened carcasses and ate the soft insides.
The lesson for a designer is the whole page in one animal: a sabre tooth is only believable if the rest of the head shows up for the same job. Ask what the tool is for, then check whether the other parts agree.
The audit
The six ledgers above collapse into one routine. Run it on a part at a time; most creatures fail on the first question and never get to the fifth.
Not what it looks like: what it does. “Big claws” is a look. “Digs burrows in hard ground” is a problem, and a problem tells you what the rest of the body has to be.
Find two unrelated animals that faced the same problem. If both went the same way, that is the shared answer, and it is the strongest evidence you will get that the design works.
Whatever the lineages converged on is a constraint: the torpedo, the fin on top, the folded abdomen, the single lens. Break it on purpose, with a reason the viewer can see, or not at all.
Every convergent trait was paid for. The cave tetra gave up its eyes; the dolphin gave up its hind legs and can never leave the water. A creature with every advantage and no bill is the fastest way to lose a viewer.
Leave one detail that gives the ancestor away: the flukes, the tooth count, the bar-shaped pupil, the missing pair of legs, the dent where an eye began. That is the detail people remember, and it is what separates a creature from a costume.
Try it
Both start from a great white shark's body. The difference is where the other parts were borrowed from. Each link loads the creature with its parts locked; reroll the empty slots and see whether the tell survives.
Fails at question 1
Shark body, wolf legs, eagle talons, an eagle's brow. The body answers one problem, moving fast through water; the legs and talons answer two others. Nothing on it tells you where it lives, so nothing on it can be checked.
Load Parts List →Passes, with a tell
The same shark body, but every borrowed part now comes from an animal that solved the same problem: a sea otter's short limbs and webbed feet, a dolphin's flukes, a penguin's flipper-wings. Questions one to three pass at once. The tell is the flukes on a fish's frame, which promise a mammal underneath.
Load Second Dolphin →This page is the biology half of a pair. The hybrid drawing guide is the drawing half: base body, borrowed parts, seams, silhouette, colour, in that order. Run the audit here before you start it, and the borrowed parts will already agree with the body. For a set of animals where evolution did the borrowing and the result still looks assembled, the guide to real animals that look fake takes nine of them apart. Every animal named above, from the thylacine to the blind cave tetra, has a card in the animal library, with its scientific name, its group and a photograph to draw from. One borrowed part gets a guide of its own: the eye, whose pupil shape and placement can be read off with two questions.
Where this comes from
Every case above rests on a published study, linked here by DOI. The definition of convergence follows the University of California's Understanding Evolution site. Photographs carry their licence in the caption and are listed again on the credits page.
The working definition used on this page: unrelated lineages shaped the same way by the same challenge, and how biologists tell a convergent trait from an inherited one.
“Soft-tissue evidence for homeothermy and crypsis in a Jurassic ichthyosaur”: the Holzmaden Stenopterygius with blubber and countershading.
“The hearing gene Prestin unites echolocating bats and whales”: the gene tree that groups dolphins with bats.
“Genome-wide signatures of convergent evolution in echolocating mammals”: the scan that found nearly two hundred convergent loci.
“Genome of the Tasmanian tiger”: the thylacine genome and the regulatory basis of its wolf-like skull.
“Computer simulation of feeding behaviour in the thylacine and dingo”: the finite-element bite test, and the smaller-prey conclusion.
“Comparative analysis of gene expression for convergent evolution of camera eye between octopus and human.”
“How to become a crab: phenotypic constraints on a recurring body plan”: the count of at least five carcinisations, and the reversals.
“Regressive evolution in Astyanax cavefish”: eye degeneration, pigment loss and the gains that came with them.
“An eye for a tooth: Thylacosmilus was not a marsupial ‘saber-tooth predator’”: the tooth-by-tooth case against the Smilodon comparison.
