Speciation
or, how one kind quietly becomes two — with no magic step anywhere in the middle
Want to split a population yourself? Jump to the valley ↓
Start at the top of California’s Central Valley, where the foothills close in, and go looking for a small, plain salamander called Ensatina. Walk south down the western rim of the valley, turning over damp logs, and you will keep finding them — blotched, then a little more orange, then plainer again, the pattern sliding by degrees as you go. Never once do you cross a line where the salamanders stop being the salamanders next door; each population interbreeds happily with the one just up the slope. Follow the ring all the way around the southern end and back up the eastern rim, and the sliding continues, mile after mile, neighbor blending into neighbor. Then, near the bottom of the valley, the two arms of the ring finally come back around and meet — and the salamanders from the western line and the salamanders from the eastern line, standing on the same hillside, will not breed. Same animal, walked in a circle. Somewhere along that loop, one species became two, and you can put your finger on the general place it happened — except there is no place, no line, no moment. Just a gradient that got long enough to matter.
I lead with those salamanders because they answer, in one hike, the single honest objection that stops more people from accepting evolution than any other. It usually comes out as: sure, things vary — finches get bigger beaks, bacteria shrug off a drug, dogs come in a hundred shapes — but that’s just variety inside a kind. I’ve never seen one kind turn into a genuinely different kind. It is a fair thing to want to see. And the reason it feels like a wall — small change yes, big change no — is that we picture the big change as needing some extra ingredient the small change doesn’t have. A spark. A threshold event. Some moment where the machinery does something new.
The step that isn’t there
Here is the whole secret of this page, and it is almost a letdown: there is no extra ingredient. Nothing new gets invented at the split. The thing that makes two species out of one is the same thing that gives you bigger beaks and drug-proof bacteria — ordinary variation and selection — plus two boring additions: isolation and time. Cut a population in two so the halves stop trading genes, let each half be worked on by its own weather and predators and luck, and wait. Each half drifts and adapts on its own account. The differences pile up, the way differences between two isolated dialects pile up into two languages. And at some unremarkable point along the way, the two halves have drifted far enough apart that their bodies and their genes no longer fit together well enough to make fertile offspring.
That is the entire event. The instant two populations can no longer interbreed, they are, by definition, two species — and notice that nothing had to happen at that instant. No new organ, no new plan, no leap. The animals on either side didn’t feel a click. They just kept doing what every generation does — being a little different from their parents — until the accumulated distance quietly crossed a line that only a biologist with a checklist even names.
What a species even is
It helps to be honest about the word. We tend to imagine a species as a fixed, essential kind — a Platonic template stamped somewhere in the nature of things, so that a dog is a dog and a wolf is a wolf and never the twain, etc. But when biologists need a working definition, the most useful one is disarmingly practical: a species is a group whose members can breed together and produce fertile offspring. Not a template — a test. Who can still make babies with whom.
Once you hold the word that way, speciation stops being mysterious and starts being almost obvious. Two varieties become two species the moment that test starts coming back negative — when the door between them closes, whether because their courtship songs no longer match, or their breeding seasons drifted apart, or their chromosomes no longer pair up cleanly, or the hybrids that do get born turn out frail or sterile like a mule. There are many ways for that door to swing shut, and none of them require a designer to install a lock. Distance does it, given long enough. The salamander at the bottom of the valley isn’t refusing its cousin out of principle. Long, separate histories simply left the two of them unable to make it work.
Speciation you can walk
The salamanders are what biologists call a ring species, and they are precious because they lay the whole process out in space where you can see all of it at once. Normally speciation happens in time, which is exactly why it’s hard to believe — the intermediate steps are dead and gone, so all you ever meet are the finished, separate species, and the connecting gradient has to be taken on faith. A ring species hands you the gradient in the flesh. Every intermediate is still alive and still breeding with its neighbors; you can literally walk the length of it, from “one species” at one end to “two” at the other, and never once cross a line where the one abruptly turned into the other.
It isn’t only the salamanders. Around the great treeless dome of the Tibetan plateau lives a small bird, the greenish warbler, in a ring of populations wrapped clear around the highlands. Go around the loop and the birds’ songs change note by note, and at every step a warbler can court and answer the warbler next door. But where the two ends of the ring overlap in Siberia, the songs have drifted so far apart that the birds treat each other as strangers and do not interbreed. Two species where the ring closes; one continuous chain of interbreeding everywhere else. The same story the salamanders tell, told in birdsong instead of blotches — which is itself the point. Nothing about this trick is special to one animal. Hand any population a long enough gradient and it will do this.
The rest of this page is just more of the same fact from other angles — but before you read it, go split a population yourself. The simulation below is the salamander valley boiled down to its bones: one population, a wall you can raise and drop, and a running count of how many species you’ve made. Raise the wall and wait, and one kind becomes two. Then — and this is the part worth staying for — drop it, let them mix, and raise it again.
Things to try
Press Begin and let the single population wander a while — one calm colour, strolling as a group. Then hit Raise the wall. West of it is a cooler world, east a warmer one; with no genes crossing, the two sides start shading apart — one drifting toward blue, the other toward gold. Watch the two colours pull apart. There’s no jump — the colour slides, and somewhere along the slide the count ticks from 1 to 2.
This is the one that settles it. Once it reads two species, press Drop the wall. The two colours pour back across the old divide and wander the whole valley, thoroughly mixed in the same dirt — and stay two. The wall is gone; the split isn’t. Two species don’t interbreed, so nothing re-blends. The wall was never what kept them apart; the banked difference is.
With two colours now spread across the valley, Raise the wall again. Each colour that straddles it feels the same pull — west toward cool, east toward warm — and starts shading apart a second time. Raising the wall divides; dropping it lets what you have share the ground. You can keep dividing the room, and the count keeps climbing.
Catch it early. Raise the wall, then Drop it while the two sides have only begun to drift apart — before the count has ticked to two. The colour, still one kind, blends right back together and the count doesn’t move. Isolation has to last long enough for the two sides to drift far enough apart that they’d stay apart even if the wall came down; interrupt it before then and the drift you built washes out. Only a finished split is permanent.
The one-way door. There is no button that folds species back into one. Raising the wall only ever makes more; dropping it only ever lets the ones you have share the ground. You can walk through this door again and again — and never back.
Now the control. Reset, tick leave a gap in the wall, then raise it. Genes keep crossing through the gap, so the two sides never finish drifting apart and the count holds at one — isolation switched off. Then play the dials: push how different the two sides are up and the colours part in fewer generations; slow the speed right down near the line and try to catch the single frame a new species is “born.” You can’t. There’s only more sliding.
Odd numbers. Tick let the west world race ahead and raise the wall — now the cool side pulls away faster than the warm one, so when several colours straddle the wall, one crosses the line before another. Watch for that moment, when one split has finished and the next hasn’t, and Drop the wall right then. The finished one stays; the unfinished one re-blends. Land it well and you’re holding three species, or five — not just the tidy two, four, eight.
The one-way door
Run it a few times before reading on, because two moves are worth catching for yourself. The first is the split itself: with the wall up and sealed, a species doesn’t leap into two — the two colours just drift apart. Generation by generation the west half and the east half shade toward their two different worlds, until the gap between them is wide enough to hold even if the wall came down, and the count ticks up by one. There is no single frame where “one species” becomes “two.” The line is ours, drawn for bookkeeping; the animals only ever drifted. That climb is the answer to “where’s the big leap?” There isn’t one. Big is just small that kept going.
The second move is the one an earlier version of this page got exactly backwards, and it’s the whole point. Raise the wall a second time and each colour that straddles it shades apart again — the count climbs — and no amount of raising and dropping ever runs it back down. Let one side race ahead of the other and the splits finish at different moments, so a well-judged drop can leave you holding an odd number, three or five, not just the tidy doublings. Drop the wall and the daughters pour across the valley and mix in the same dirt, but they don’t answer each other; they stay separate. You can keep dividing the room; you can never un-divide a species. (Before the line, yes — an interrupted drift washes right back out, reversible up until it isn’t.) Leave a real gap in the wall and the crossing genes stir the two pools back into one, so the split never finishes and the count holds — that is the control, isolation switched off. But once a split has actually crossed the line, sealing or opening the valley does nothing to undo it. That one-way quality isn’t special to salamanders. It’s the same asymmetry this site keeps running into — the way a civilization can shed its complexity far faster and more permanently than it built it (Why Civilizations Fall). Order, once it has paid for itself in accumulated difference, doesn’t offer refunds.
Speciation you can watch
The valley is a model, so it’s fair to ask for the thing itself — speciation caught happening, in the wild, on a human clock. It has been. In the eastern United States there is a fly, Rhagoletis pomonella, that for ages laid its eggs only in the fruit of native hawthorns. Then colonists planted apple orchards, and around the middle of the 1800s some flies started using apples instead. Apples ripen weeks earlier than hawthorn fruit, so the apple flies now breed on a different schedule, in a different place, from their hawthorn cousins — and the two groups have begun to pull genetically apart while living in the very same orchards. Nobody moved them; a change in taste built a soft wall out of timing, and the split started on its own. That is speciation with a founding date roughly inside American history.
There are stranger cases, and one of them happened with people standing right there taking notes. In 1981, on the small Galápagos island of Daphne Major, Peter and Rosemary Grant caught a male finch that was too big and sang the wrong song. He had blown in from Española, a hundred kilometers off, and he was never going home. He bred with a local medium ground finch, and his descendants — the Grants’ field notes call them the Big Bird lineage — came out with a beak and a song that matched nobody else on the island. Within about three generations they were breeding only with each other. Reproductive isolation, the part that is supposed to take forever, showed up inside the working life of two people who had cameras, and it is still going. In the great lakes of East Africa, cichlid fish have fountained into many hundreds of species in a span that, by evolution’s standards, is the blink of an eye — whole galleries of distinct fishes carved out of a few founders by little more than who preferred to mate with whom. And in a couple of cases botanists have watched a brand-new species appear in a single generation: when two plant species’ chromosome sets accidentally combine and double — a hiccup called polyploidy — the offspring can be instantly unable to breed back with either parent yet perfectly able to breed with each other. It happened in the American Northwest with the weedy goatsbeards (Tragopogon) in the last century, and in a greenhouse at Kew around 1900 with a primrose. New species, dated and witnessed, no deep time required.
Speciation you can point to on a map
The commonest way it happens leaves the plainest fingerprint: a physical barrier drops across a population’s range and does exactly what your barrier does in the valley. Stand on the rim of the Grand Canyon and you can almost read it off the ground. On the forested plateau on one side lives the Kaibab squirrel, dark-bellied with a flashing white tail; on the other side, its close relative the Abert’s squirrel, cut from the same recent cloth. The river carved the gorge; the gorge split the forest; the two squirrel populations have been drifting apart on their opposite rims ever since, not yet fully two species but visibly on the road. Rivers, canyons, rising seas, a new mountain range, an island cut loose — the planet is forever laying down these walls, and on either side of each one the same slow arithmetic runs. Selection, isolation, time.
I’ll add one honest note, because this site keeps its own rule that a thing recurring in nature is not a thing nature endorses. The same tidy mechanism that fills the world with warblers and cichlids also turns out crop pests, and disease vectors sorting themselves into the niches we built for them, and new lineages of things we’d rather do without. It can even run backwards — but only where the door hasn’t fully closed. When farm runoff clouded parts of Lake Victoria, the female cichlids could no longer see the colors they had been choosing mates by, so they stopped choosing — and species that were still young enough to merge, held apart by preference rather than by any settled incompatibility, flowed back together into single mixed populations. Species un-happening, in our lifetime, for a reason as stupid as bad water. Speciation is not a moral engine handing out rewards. It is a blind consequence of separation and time, as willing to make a nuisance as a marvel, and — while a split is still half-made — as willing to abandon it as to finish it. That it happens to have made the tangled, glorious tree of everything alive is not a promise it made us. It is just what falls out when populations get divided and left alone.
The wall was never the thing
Come back to the salamander at the bottom of the valley, refusing its cousin on the same hillside. Nobody drew a line through that species. No moment arrived when the machinery reached for a new tool. Two arms of one population walked two long, separate roads, and where the roads met again the walkers had simply changed too much to go home. If you granted the small step — and you do; you’ve seen beaks and bacteria and every breed of dog — then you were always, quietly, granting the big one too. They were never two different claims. The big one is the small one with the patience to keep going, and a wall to keep the two halves honest while it does.
You just made it happen yourself: raised a wall, watched a count climb, dropped the wall and found two where there had been one — then raised it again and split them further. If some part of you still wants to keep a little skepticism in reserve — good. Keep it, and keep it pointed at the evidence. Then go raise the wall one more time and watch the colors part. Nobody talked you into anything. You ran the valley and read the result.
If speciation never sat right with you — if you could nod at the finches and the drug-proof bacteria but something balked at “and that’s how you get a whole new kind” — I suspect it’s because the phrase new kind smuggled in a promise of some grand event, and then nobody ever showed you the event, because there isn’t one. There is a population, a wall, and enough time. The salamanders ringing the valley are one species and two species at once, depending only on how far apart along the ring you stand. That isn’t a paradox. It’s a photograph of the exact thing you were told to take on faith — held still, in daylight, so you can walk right up and look.
The mechanism runs on any substrate you care to name, which is the drum this whole site keeps beating. See it building adaptations with a visible, intentional hand in Unnatural Selection, and with no hand at all in Survivors Reproduce; watch it caught in the act, filmed and measured, in Evolution You Can Watch. This page is just those same three words — variation, heredity, selection — given a wall to divide the room, and enough patience to finish the job.
Survivors Reproduce — the engine right underneath this page: variation, heredity, and a sieve, with no hand on the dial.
Unnatural Selection — the same machine with a visible, intentional hand; you play the breeder.
Evolution You Can Watch — the mechanism caught in the act, filmed and measured.
- Ring species — Ensatina eschscholtzii around California’s Central Valley: the classic case, developed by Robert C. Stebbins and later by David Wake’s lab. Overview: UC Berkeley Understanding Evolution — Ring species and Ensatina (Wikipedia). On the genetic leakage along the ring and the hybrid zone where the two terminal forms meet: Pereira & Wake, “Genetic leakage after adaptive and nonadaptive divergence,” Evolution (2009), and Devitt et al. on asymmetric reproductive isolation between the terminal forms (2011). Note: whether the ring is “perfect” is debated — there are breaks and some hybridization along it — but the core observation (a continuous chain of interbreeding populations whose end-points do not interbreed where they meet) holds.
- Greenish warbler (Phylloscopus trochiloides) around the Tibetan plateau: Darren Irwin and colleagues, e.g. Irwin, Bensch & Price, “Speciation in a ring,” Nature 409:333 (2001). nature.com/articles/35053059; lab background at the Irwin lab. The honest complication — genomes show the ring’s history was lumpy, with real breaks in gene flow and later reconnections rather than one smooth spread: Alcaide, Scordato, Price & Irwin, “Genomic divergence in a ring species complex,” Nature (2014); and the skeptical read, Jerry Coyne, “There are no ring species” (2014).
- Apple maggot fly (Rhagoletis pomonella) host-race formation on apple vs hawthorn — a textbook case of incipient sympatric speciation. Feder et al., “Genetic differentiation between sympatric host races,” Nature (1988); on fruit odor as the isolating cue, Linn et al., “Fruit odor discrimination and sympatric host race formation,” PNAS (2003).
- The Big Bird lineage on Daphne Major — a reproductively isolated lineage founded by one immigrant and followed in the field for three decades: Lamichhaney et al., “Rapid hybrid speciation in Darwin’s finches,” Science (2017). Plain-language accounts: Uppsala University; Quanta Magazine.
- Speciation in reverse — Lake Victoria cichlids losing their reproductive barriers as turbidity blocked color-based mate choice: Seehausen, van Alphen & Witte, “Cichlid fish diversity threatened by eutrophication that curbs sexual selection,” Science 277:1808 (1997). science.org; on gene flow between similar species in turbid water, Konijnendijk et al.
- One case we deliberately took OFF this page: the “London Underground mosquito” (Culex pipiens f. molestus), long told as a species that evolved in the tunnels since the 1860s (Byrne & Nichols, Heredity 1999). Whole-genome work published in 2025 places its origin above ground in the Middle East / Mediterranean, on the order of a thousand years ago or more — so it is not a case of speciation we watched happen. Summary: Natural History Museum; paper: “Ancient origin of an urban underground mosquito,” Science (2025). The story was better than the evidence, so out it goes.
- African cichlid radiations (Lakes Victoria, Malawi, Tanganyika) — hundreds of species from few ancestors on a short timescale: overview at Cichlid speciation (Wikipedia).
- Instant speciation by polyploidy: the goatsbeards Tragopogon mirus and T. miscellus, allopolyploids that arose in the northwestern United States in the 20th century (Ownbey, 1950); and Primula kewensis, an allopolyploid that appeared at the Royal Botanic Gardens, Kew, around 1900. The Tragopogon allopolyploids were first collected in the Palouse in 1949 and reported by Ownbey in 1950; the three diploid parents did not co-occur there before about 1928, so the new species cannot be much more than 80–90 years old, and each has formed more than once. Soltis et al., “Synthetic polyploids of Tragopogon miscellus and T. mirus… 60 years after Ownbey’s discovery,” American Journal of Botany (2009).
- Allopatric split at the Grand Canyon: the Kaibab squirrel (Sciurus aberti kaibabensis) on the north rim vs the Abert’s squirrel (Sciurus aberti) on the south — commonly cited as divergence in progress; they are usually treated as subspecies, i.e. speciation not yet complete. Overview: Kaibab squirrel (Wikipedia).
- Background on the biological species concept and modes of speciation (allopatric, sympatric, ring): Coyne & Orr, Speciation (2004); UC Berkeley’s Understanding Evolution — Speciation. On why Mayr’s definition is a working test rather than a final answer — and why something like two dozen rival species concepts are still in play — James Mallet, “Ernst Mayr and the modern concept of species,” PNAS (2005). The simulation reproduces the shape of the process — divergence under isolation, gene flow preventing it — not the biology of any particular organism.