How islands make strange animals
Elephants the size of a large dog. Pigeons that weighed as much as a turkey and couldn't fly. Eagles heavy enough to knock over something four times their weight. Tortoises you could sit on. A human species about a metre tall.
All real, all on islands, and none of them anywhere else.
Islands do something to animals. They do it repeatedly, in the same directions, to unrelated lineages on opposite sides of the world, which is the sort of pattern that means there's a mechanism underneath rather than a run of coincidences. And the mechanism isn't complicated. Cut a population off, shrink it, remove most of what used to eat it, cap the food supply, and wait.
Getting there is the first filter
Before an island can make anything odd, something has to arrive. That's harder than it sounds, and it's selective in a way that shapes everything afterwards.
Continental islands — Britain, Java, Tasmania — were joined to the mainland when sea levels were lower, so they started with a full mainland fauna and lost bits of it afterwards. Oceanic islands are the interesting ones. Hawaii, the Galápagos, the Canaries and Iceland rose out of deep water and were sterile rock. Everything living on them got there across open sea.
The palaeontologist George Gaylord Simpson called this sweepstakes dispersal, and it's a fair description. The odds against any individual crossing are enormous, but the ticket gets bought millions of times over millions of years, so occasionally somebody wins. Birds fly or get blown. Seeds ride in guts, on feet and in feathers. Insects and spiders drift in on wind at surprising altitudes. Lizards and rodents raft across on floating mats of vegetation washed out of a river mouth, which sounds like a story until you look at the distribution of iguanas and have to explain it somehow.
The filter is brutal and it isn't random. Reptiles do well on rafts because they tolerate weeks without food or fresh water. Amphibians almost never make it, because salt water kills them. Land mammals other than bats are poor at all of it. So the classic remote-island lineup is birds, bats, reptiles, insects and plants, with a conspicuous hole where the mammals should be.
Before people arrived, New Zealand had no land mammals at all apart from bats. Hawaii had one bat and one seal, and no ants, no reptiles and no amphibians of its own.
That hole is what everything else grows into.
The island rule: things converge on medium
In 1964 J. Bristol Foster published a short paper in Nature comparing island mammals with their mainland relatives, and found something suspiciously consistent. The big ones had got smaller. The small ones had got bigger.
It's since been argued about endlessly — the strength of the effect varies by group and some analyses find it weaker than advertised — but the two halves have decent explanations, and they're different explanations, which is part of why the pattern isn't perfectly tidy.
Shrinking is about the budget. An island has a hard ceiling on how much food exists, and a large body is expensive. On the mainland, being big buys you protection from predators and an edge in fights; take the predators away and the bill stays while the benefit shrinks. Smaller animals need less, breed sooner, and survive lean years better. So the population drifts downwards in size, sometimes dramatically.
Growing is about safety. Small mainland animals are small partly because it lets them hide, and hiding matters when everything eats you. On an island with no ground predators the pressure to stay small comes off, and the advantages of size — better at fighting rivals, better at surviving cold nights, able to eat tougher food, harder for the remaining predators to swallow — start to win.
Same island, opposite directions, one underlying cause: the ecology has been simplified, and the trade-offs that set body size on the mainland no longer apply.
The dwarfs
Mediterranean islands were full of miniature elephants. Palaeoloxodon falconeri from Sicily and Malta stood around a metre at the shoulder — an adult elephant roughly the height of a large dog, descended from a straight-tusked elephant several times that size. Cyprus and Crete had their own dwarf elephants and dwarf hippos. The pattern repeated on island after island from separate colonisations, which is the clearest possible sign that the island was doing the work.
Mammoths did it twice over. The Channel Islands off California had a pygmy mammoth about half the height of its mainland ancestor. Wrangel Island in the Arctic Ocean held a reduced population of woolly mammoths that survived until around four thousand years ago — meaning there were mammoths alive while the Egyptian pyramids stood.
And then there's Flores, in Indonesia. In 2003 excavators in Liang Bua cave found the remains of a small hominin, published the following year as Homo floresiensis. It stood a little over a metre tall with a brain around a third the size of ours. The dating has been revised since the first announcement and the material now looks to be roughly fifty thousand years old rather than eighteen thousand. Whether it descends from Homo erectus that shrank on the island, or from something smaller that arrived that way, is still contested. Flores also had a dwarf elephant relative, a giant rat and a giant stork, so the island was running the full programme.
The giants, and what they replaced
Islands don't just shrink big animals. They promote small ones into the jobs the missing large animals would have done.
New Zealand is the extreme case, because with no land mammals every terrestrial role went to something else. Birds became the browsing herbivores: nine or so species of moa, the largest standing well over two metres with the neck up and weighing a couple of hundred kilograms. The top predator was also a bird — Haast's eagle, the largest eagle known, at maybe thirteen to fifteen kilograms, which hunted moa. Insects took the small-mammal niches, which is how you end up with giant wētā heavier than a mouse.
Madagascar had elephant birds, up to three metres tall and over half a tonne, and they were flightless too. The Galápagos and Aldabra both produced giant tortoises independently. The coconut crab on Indian and Pacific Ocean islands is the largest land arthropod alive, with a leg span approaching a metre.
The dodo is the famous one and it's routinely misdescribed. It wasn't a failed bird or a degenerate anything. It was a pigeon — genetic work places its closest living relative as the Nicobar pigeon of south-east Asia — that arrived on Mauritius, found no predators, and over a long stretch of time got large, heavy and flightless because there was no longer any reason not to. It was well adapted to Mauritius. Mauritius simply stopped being the place it was adapted to.
Why flying gets abandoned so often
Flight is superb and it's expensive. The muscles are heavy, the keel bone that anchors them is heavy, and maintaining all of it costs energy every day whether you use it or not.
On a mainland that cost is worth paying, because flight is how you escape. On an island with nothing hunting on the ground, natural selection stops subsidising the equipment, and lineages lose it — not through disuse in any individual, but because birds that put less into flight muscle and more into breeding leave more offspring, generation after generation.
Rails have done this more than any other family. They're moderately good long-distance flyers, they colonise readily, and once they land they lose flight fast. Flightless rails evolved independently on island after island across the Pacific, over and over, from separate colonising flocks. It's about as close to a repeatable experiment as evolution offers.
New Zealand alone produced the kiwi, the kākāpō — a flightless nocturnal parrot, and the heaviest parrot in the world — the takahē, and the moa. Same island, four separate abandonments of flight.
Founder effects and the genetics of a small start
The other thing an island does is genetic, and it starts the moment the first pregnant female washes ashore.
A colonising group carries only a sample of the variation present in the population it came from. If two dozen birds arrive, whatever genetic diversity existed back home is mostly left behind, and whatever those particular two dozen happened to be carrying — including rare variants, including harmful ones — becomes the entire basis for everything that follows. That's the founder effect, and it means island populations often start out unrepresentative purely by accident.
Small populations then keep drifting. In a population of millions, chance barely shifts gene frequencies; in a population of a few hundred, chance dominates, and variants can go to fixation or vanish for no adaptive reason at all. Add a strong new selective environment — different food, different climate, no predators — and change comes quickly.
Then there's the part that makes islands famous: adaptive radiation. A single colonising species finds a range of empty roles and splits to fill them. The Hawaiian honeycreepers descend from one finch-like ancestor and produced more than fifty species with beaks for nectar, seeds, insects and prising bark. Hawaii's silversword alliance — a set of plants ranging from vines to shrubs to a spectacular rosette that flowers once and dies — all came from a Californian tarweed. Cichlid fish did the same thing in the lakes of East Africa, which behave as islands even though they're full of water.
Darwin's finches, told properly
The story everyone half-remembers is that Darwin visited the Galápagos, examined the finches, noticed their beaks varied island by island, and worked out natural selection on the spot.
That's not what happened, and the real version is more instructive.
Darwin spent five weeks in the Galápagos in 1835 and did not think the finches were important. He collected them casually, didn't consistently record which island each specimen came from, and misidentified several — he took some for blackbirds, some for grosbeaks, some for wrens. He didn't realise they were all closely related. The birds that did catch his attention on the voyage were the mockingbirds, which he noticed differed clearly from island to island, and he wrote a cautious note about it that hints at where his thinking would go.
The insight came afterwards, in London, from someone else. The ornithologist John Gould examined the specimens and reported in 1837 that the odd assortment was in fact a group of twelve closely allied species of ground finch, found nowhere else on Earth. That was the surprise. Darwin then had to reconstruct which island his birds came from by begging locality data off FitzRoy and other crew members who had labelled theirs properly.
The finches barely appear in On the Origin of Species. The name "Darwin's finches" was popularised by the ornithologist David Lack in a 1947 book, more than a century after the voyage.
The real payoff came later still. From 1973 onwards Peter and Rosemary Grant spent decades measuring every finch on Daphne Major, a small Galápagos island, individual by individual, year after year. During a severe drought in 1977 the small soft seeds ran out first, birds with deeper, stronger beaks survived better, and the average beak size in the next generation measurably shifted. Then wetter years pushed it back. They watched natural selection happen and wrote down the numbers, which is a good deal more impressive than a flash of insight in a rowing boat.
Why island species collapse when anything new arrives
Everything that makes island animals interesting also makes them defenceless.
They're often flightless, or slow, or ground-nesting, or all three. They tend to breed slowly and lay few eggs, because in a stable environment with no predation there's no advantage in pumping out young. They frequently have no fear response at all — Darwin remarked that Galápagos birds were tame enough to be knocked off a branch with a cap, because nothing had ever hunted them and no instinct had been retained. Island plants have lost thorns and toxins for the same reason.
Then a ship arrives. Rats come ashore and eat eggs and chicks that have never been defended. Cats take adults that don't run. Goats and pigs strip the vegetation and destroy the nest sites. Mongooses were introduced deliberately in several places to control rats and mostly ate the native wildlife instead. And there's no anywhere else to go — an island population is the whole species, so a local extinction is a permanent one.
The Stephens Island wren is the standard illustration. A flightless songbird known only from one small island in New Zealand's Cook Strait, it was effectively wiped out by cats around the lighthouse in 1894, within a couple of years of its discovery. The tidy version blames a single cat named Tibbles; the evidence points to a feral cat population rather than one animal, which doesn't improve the outcome.
Guam is the modern case. The brown tree snake reached the island in cargo around the middle of the twentieth century and spread. By the 1980s most of Guam's native forest birds were gone from the wild — several species extinct, others surviving only in captivity. The forest is still standing and it's largely silent, and because the birds were doing the pollinating and seed dispersal, the trees are failing to regenerate.
Islands are somewhere around five per cent of the world's land area. A commonly cited figure holds that roughly three-quarters of recorded animal extinctions since 1500 have been island species. Estimates of how many bird species disappeared from Pacific islands after human settlement run to the high hundreds, with one widely quoted analysis putting it near two thousand.
The one thing that reliably works
There's a genuinely encouraging footnote, and it's a rare one in conservation.
Because islands have hard boundaries, you can actually remove an introduced predator and have it stay removed. Eradication programmes have cleared rats, cats, goats and rabbits from hundreds of islands worldwide, and where they succeed, native populations frequently rebound on their own. Macquarie Island was declared free of rabbits, rats and mice in 2014. South Georgia, a much larger and harder target, was declared rodent-free in 2018 after a multi-year baiting operation. Seabirds returned to breed on ground they hadn't used in a century.
It's expensive, it's logistically miserable, and it has to be complete — leave one pregnant rat and the whole thing is undone. But it's one of the few conservation interventions where the result is unambiguous.
The underlying logic hasn't changed since Robert MacArthur and E. O. Wilson set it out in 1967: an island's species count settles at a balance between arrivals and losses, and that balance depends on how big the island is and how far it sits from a source of colonists. Bigger and closer means more species. Smaller and more remote means fewer, and each one is stranger, and each one is holding on by a thread that only has to be cut once.
