A magnitude 9.2 earthquake makes an unlikely landlord.

At 5:36pm on Good Friday, 27 March 1964, the crust beneath Prince William Sound let go and kept going for about four and a half minutes. The US Geological Survey still ranks the Great Alaska Earthquake as the most powerful ever recorded in the United States, and the second largest recorded anywhere, behind Chile in 1960. Around Montague Island, the seabed did something stranger than shake. It stood up.

Geologists working on the island’s drowned lagoons, in a USGS study of the Patton Bay splay fault system, put the local jump at 3.7 to 4.3 metres, with some shorelines lifted by as much as 11. Salt water that had been ocean hours earlier was sitting above the tideline by the next morning, slowly going fresh on rain and snowmelt.

A few of those puddles had fish in them.

What the ponds trapped

They were threespine stickleback: silver, finger-length, and armoured from gill to tail with a row of bony side plates. Marine stickleback wear that armour because the ocean is full of things that want to eat them. The new ponds had fewer predators, no salt, and a winter that seals the surface under ice for months.

Half a century later, Emily Lescak, then at the University of Alaska, led a team out to those awkward, barely reachable islands to see what had become of the castaways. Working with William Cresko’s lab in Oregon, the group sampled more than a thousand animals and read over 130,000 individual DNA differences across them. The method compares the same short stretches of genome, letter by letter, in every specimen. Their results, reported in the Proceedings of the National Academy of Sciences, landed two findings side by side.

First, the genetics confirmed the obvious-sounding part: the pond fish really were descended from local ocean stickleback stranded within living memory, and the same thing had happened separately on more than one island. Second, and less obvious, those young populations had already drifted almost as far from their ocean parents as freshwater stickleback that got started thousands of years ago, when the glaciers pulled back. Eyes, body shape, colour, bone size, armour. All of it shifted.

Cresko, quoted in a University of Oregon release, suggested the clock might run even faster, since in some populations “we found evidence of changes in fewer than even 10 years.”

Why the fish were ready

How does a fish rebuild itself in fifty generations? Not by waiting around for lucky mutations, which takes far longer than that. The useful variants were already swimming in the ocean.

Most of the armour difference traces back to a single gene called Ectodysplasin, or Eda, which tells the developing fish how many plates to grow. Natasha O’Brown and colleagues at Stanford, reporting in eLife, traced the difference to a regulatory switch: a small tweak to the DNA that tells the gene where to turn on, not a change in the gene itself. Low-plate versions of that switch turn up at low frequency in marine fish worldwide. Drop a few hundred ocean stickleback into a pond and selection has something to work with from day one.

Susan Bassham, a senior research associate in Cresko’s lab, made the point that this is a genuine genetic rearrangement rather than a flexible response like skin tanning in summer. The follow-up work she and Julian Catchen published in the journal Genetics found the same broad blocks of genome reshuffled in the earthquake ponds as in far older lakes. Up to a quarter of the genome had been rearranged in just a few dozen generations.

A lake that ran the same experiment

In 1982 the Alaska Department of Fish and Game poisoned Loberg Lake, a small body of water north of Anchorage, to make room for trout and salmon, wiping out its resident stickleback. Sea-run fish found their way in between 1983 and 1988. Nobody is quite sure how.

Michael Bell at Stony Brook started sampling them in 1990 and kept at it. Writing in Evolution, he and his co-authors recorded a population that was almost entirely fully plated in 1990. By 2001 the fully armoured form was down to 11 per cent and the stripped-down form was up to 75. Twelve years, one small lake, armour mostly gone.

What this does and doesn’t prove

Three islands and one unusually well-prepared species don’t add up to a general law of nature. Stickleback carry an enormous reservoir of freshwater-ready variation because the species has been shuttling between salt and fresh for millions of years, and plenty of animals have nothing comparable in the tank. A coral or a specialist alpine insect handed the same shock would simply be in trouble.

What the Alaskan work does unsettle is the assumption that visible evolutionary change needs geological patience to show up. The ponds are still there, still being sampled, and still slowly filling in. The USGS geologists reading Montague’s sediment layers found silt giving way to peat, which is the signature of a lagoon becoming a bog. Those earthquake ponds are on the same path. Whatever the fish are going to do next, they will need to get on with it.