Flatfish, in the order Pleuronectiformes, have long been an evolutionary puzzle, for all the fish in this order lie on the substrate—on their sides—with both eyes on one side of their body, like the flounder below:

Phylogenetic analysis shows that flatfish evolved from “regular” fish, fish having one eye on each side and swimming vertically, that evolved over time to lie on their sides. The bizarre thing about this evolution is that it involved genetic changes so that “normal” fish had their eyes move over the top of their head so that both eyes look upwards. Their skin changes color and texture, too, with the top half colored, as above, and the bottom half pale.
And all flatfish begin their development like “normal fish”, swimming vertically and having one eye on each side of the head. Then, as the fish gets older, one eye migrates over the top of the skull to the other side! (You can see that in the video below.)
When the eyes are both on one side, the flatfish tip onto their sides and spend the rest of their adult life lying on one side. (The side varies among species: some have 100% right-sided individuals, others 100% left-sided, and some species are random, with half of the individuals having the right eye move over (and lying on their right side), and the other half having the left eye move.
Living on the substrate like this, and often camouflaged as the flounder above, is an advantage for the fish, both protecting them from predators and, since they are predatory piscivores (fish eaters), hiding from their prey.
Here’s a video of the development of a young flatfish, showing the eye migration. Since the ancestor had both eyes on one side, like the young flatfish, this is a case of “ontogeny recapitulating phylogeny”—that is, the development of a single living fish goes through a process mimicking the evolution of their adult ancestors.
But since the weird developmental pathway is presumably an adaptation that evolved by (presumably stepwise) natural selection, two big questions immediately arise:
- What were the intermediate evolutionary stages of eye migration?
- What were the evolutionary advantages of this migration, which presumably involved a gradual evolutionary movement of the eye from the side to the top of the head, and then over the head to the other side? It’s hard to see how, for example, an eye that’s halfway around, so it’s close to the top of the skull but hasn’t moved to the other side, could leave more offspring, or survive better, than their ancestors. What would be the advantage of each small step of the migration?
It’s hard to envision a gradual Darwinian process that could produce this migration. As Carl Zimmer wrote in a new NYT article that summarizes recent flatfish findings (click below), Darwin’s critics used both questions about to cast doubt on his theory. In response, some “saltationists”, who assumed that major evolutionary changes occurred in one huge step rather than a series of gradual steps, said that a single mutation moved the eye from one side to the other. (But that would not be advantageous unless the fish had already evolved to lie on its side!)
Click below to read the Zimmer piece in the NYT here (the drawing is animated), or find it archived here.
As Carl reports, there was another weird finding that now seems doubtful: a 2001 paper by a group of Chinese researchers who, using DNA=based family trees, seemed to show that flatfish evolved twice. You can see that paper in Nature Genetics by clicking on the headline below, or read the pdf here. The discovery that flatfish seemed to be “polyphyletic”—with more than one evolutionarily independent origin—was deeply weird, because the hormone-induced eye migration, which is extraordinarily complex, would have had to evolve twice. It’s not impossible, but seemed unlikely. One of the doubters was evolutionist Matt Friedman, who got his Ph.D. here and is now a professor at the University of Michigan and director of its Museum of Paleontology.
A while back, when he was still at Chicago, Friedman published what I see as the most interesting of the three papers highlighted here. This one was in Nature, and you can read it by clicking below or seeing the pdf here
Note that this paper was a lot of work, and yet, unlike the others, Friedman was the sole author. I love to see single-person research efforts like this. That aside, what Friedman found were two fossil evolutionary intermediates between adult “normal” fishes (the presumed ancestors of flatfish) and modern flatfishes, having both eyes on one side. Friedman reanalyzed a neglected species, Amphistium paradoxum, and a described a new fossil fish, Heteronectes chaneti, both from the lower Eocene, about 50 million years ago.
Amazingly, both species (the former randomly sided and the latter lying on its left side) showed an intermediate placement of the eyes in the adult fish. Both eyes were on the same side of a vertically-oriented fish, but one eye had migrated upwards toward the top of the skull, so that the fish could presumably see both to the side and also, perhaps, a bit above them. Thus we have two evolutionary intermediates of the adult stage, likely showing that the eye movement did not occur in one big evolutionary leap.
Here’s a photo from the 2008 paper of the left and right sides of the H. chaneti skull, showing the eye sockets, which I’ve circled. The asymmetry is obvious:

And a reconstruction of the Amphistium species, showing both sides. The asymmetry is again clear, but the eyes of the adults are still on opposite sides of the head:

You’ve probably realized that this addresses question #1 above, showing that the movement was presumably gradual over evolutionary time, though we need more fossils to show that it was a continuous series of small steps. But at least the movement didn’t seem to involve one big leap.
But that leaves question #2, which I’ll address in a moment.
The reason Zimmer’s note came out now, though the papers above date from 2008 and 2021, is that a group of authors recently published another DNA based analysis in Nature Genetics showing that the Chinese group was probably wrong: flatfishes and their eye movements seem to have had a single evolutionary origin. (The Chinese group maintains that their “polyphyly” conclusion is still the best one.)
Click below to read, or find the pdf here.
Before returning to the Big Unsolved Question, I’ll show the phylogeny advanced in the 2001 paper (bottom), showing two origins of flattening and eye migration, and the newer analysis by Duarte-Ribiero et al. at the top (Friedman is the third author), showing a single origin of flatfish (I’ve circled it). This newer paper also singles out some genes that, showing signs of selection in their DNA sequence, may be involved in the evolutionary transformation, but I’ll leave that issue aside. Green silhouettes are flatfish, black are nonflat fish.

Now for the big mystery. How could there possibly be an evolutionary advantage to each step of the eye movement? Presumably the adult either laid on its side or swam “normally”, and what would be the advantage of intermediate stages when the eye gradually moved up, across the top of the skull, and settling on the other side? The movement is presumably advantageous only when the fish is already on its side, but then what would be the advantage of moving a few mm towards the top of the skull?
Well, perhaps the fish didn’t lie fully on its side. Here’s one clue in a quote from the 2008 paper:
Questions about the possible selective advantage of incomplete orbital transit arise from the discovery of stem flatfishes. Clues are given by living taxa, which often prop their bodies above the substrate by depressing their dorsal- and anal-fin rays. Similar behaviour might have permitted Amphistium and Heteronectes—both of which have long median-fin rays—the use of both eyes while on the sea floor. The unusual morphology and resting orientation of pleuronectiforms have been interpreted as adaptations for prey ambus, and it is clear that stem flatfishes, like morphologically primitive living forms, were piscivorous; one specimen of Amphistium (MCSNV V.D.91+92) contains the remains of a fish nearly half its own length.
So perhaps this happened: a normal ancestor, through behavioral evolution, adapted to hanging around the sea bottom, as they were less conspicuous and could get more prey. But they’d have a more difficult time seeing upwards with eyes on both sides of the head. Movements of the eyes toward the top of the skull could be advantageous so long as they occurred in concert with behavioral changes (first perhaps learned, then evolved) involving propping themselves up with their fins. The advantage of tilting a bit would be that the fish might become a bit less conspicuous.
This whole scenario, as I proposed it (and I’m sure others have before in some form) presumes that the eye movement is either induced by or occurs in concert with changes in the fish’s behavior, which initially could have been learned and not coded in the genes. (Ernst Mayr once said something like “all major evolutionary changes begin with a change in behavior”). I don’t know how to test the hypothesis, as even finding more fossils with intermediate stages of eye migration will tell us little about the selective pressures involved. But for sure the movement involved natural selection rather than other evolutionary forces like genetic drift, for we have a big directional change involving many genes, genes that involve both morphology and behavior.
In short, I don’t know how it happened. But seeing that modern fishes can use their fins to prop themselves up on the sea floor may give us a clue. And other scenarios may be possible; readers can entertain themselves by finding alternative ways this change could have occurred by natural selection.















































