The fantastic eye of the scallop revealed in a new paper

December 4, 2017 • 9:00 am

Perhaps you didn’t realize, like reader Gregory (who sent me the Science paper), that scallops have eyes. But they do indeed—up to 200 tiny eyes lining the mantle, each a millimeter across: about the size of an “o” on a printed page.

Here’s what the array looks like in the scallop Pecten:


And a close up of the miniscule baby blue eyes:

A close-up view of a scallop’s eyes. Photo Dan-Eric Nilsson/Lund University, source: New York Times

Why do they need them? Because scallops aren’t sedentary molluscs: they swim actively by “jet propulsion,” flapping their shells to get away from predators or to find new resting sites. To wit:

It’s been known for a while that these eyes probably involve mirror reflection of incident light onto a retina, but how that reflection was achieved wasn’t clear, except that the mirror probably involved guanine crystals (guanine is one of the four nucleotide basis that make up the “code” of DNA). But a new paper in Science by Benjamin Palmer et al. (free access, reference below), elucidates how the eye works, and it’s amazing. The mirror, formed of overlying sheets of guanine crystals, reflects light back on retinal tissue that sits in front of the reflector, and there is not one but two retinas, each giving information about different parts of the scallop’s environment. And the mirror, besides functioning very efficiently, is a thing of beauty: a marvel of natural selection.

First, though, another picture (from the paper) of the eyes lining the mantle (captions of all figures come from the paper):

The scallop Pecten maximus with numerous eyes lining the mantle (the white arrow points to an individual eye)

So here’s how the eye works. Figure “A” below is an image produced by the technique that enabled this research to be done: cryogenic scanning electron microscopy (cryo-SEM), in which a frozen sample is scanned. (A Nobel Prize in Chemistry was awarded this year to the researchers who developed the method.) This enabled the researchers to visualize not only the entire structure of the eye, as in “A” below, but also sections of it, so they could look at the fine structure of the guanine “mirror” as well as make computer models of how light would travel after entering the eye.

Image “A” is analyzed in “B”, with different colors represent the parts of the eye and the directions of light.  Incoming light (red lines) hits the mirror (green) after passing through the cornea (black), the  iris (navy blue), the lens (light blue), and the transparent retinas (gray cloud). After hitting the mirror, light rays (now yellow) are reflected through the guanine layers, eventually striking the two retinas. One retina is proximal (closer to the scallop’s body) and the other distal (closer to the front of the eye):


A) Volume rendering of an x-ray micro-CT scan of a whole scallop eye, showing the eye anatomy. (B) Segmentation of the micro-CT in (A). Black, cornea; navy, “iris;” blue, lens; gray, gross retinal volume; green, mirror. Rays traced through the eye from a point source aligned with the axis of the lens (red) are reflected (yellow) and focused on the retina. The border of the best-focused region encompassing all reflected rays denotes a 3D circle of least confusion (COLC; black line). The inset is a side view of the mirror showing the optical axes of the lens (blue), central mirror (green), and center of the visual field (cyan). The lens and mirror axes are offset by 7.3°.

Here’s a cross section of the tiny eye with the elements labeled. You can see the two retinas (iii and iv), with the bowl-shaped mirror (v) right below the retinas. As I said, the retinas are transparent so they don’t block incoming light. Yellow arrows show the direction of light entering the eye:

Fluorescence microscopy image of an eye cross section, showing the cell nuclei stained with DAPI (4′,6-diamidino-2-phenylindole). The (i) cornea, (ii) lens, (iii) distal retina, (iv) proximal retina, and (v) concave mirror are indicated.

Here’s a colored cross section that makes identification of the parts easier (see the caption). The two retinas are olive-green and orange, and the mirror is bright green:

Low-resolution cryo-SEM micrograph of an eye cross section after high-pressure freezing and freeze-fracturing. The lens (blue), distal retina (yellow), proximal retina (orange), and concave mirror (green) are shown in pseudo-colors. The cilia and microvilli of the photoreceptors were used to identify the locations of the distal and proximal retinas.

What’s truly remarkable about the eye is the “mirror”, composed of a tiled “floor” of guanine crystals in the shape of squares (not their natural crystalline configuration—how does the scallop do this?). Each “floor” is a sheet, and there are 20-30 of them set one above the other, interspersed with cytoplasm. Here’s a cryo-SEM image (see caption), and isn’t it remarkable?

The ultrastructure of the multilayer mirror. (A to C) Cryo-SEM micrographs of high-pressure–frozen, freeze-fractured cross sections through the eye of P. maximus. (A) The mirror viewed perpendicular to the eye axis. White arrow, direction of on-axis incident light. (B) The tiled mirror viewed from above. (C) Crystals in adjacent layers, stacked directly on top of one another, viewed in a fracture through the mirror. (D) TEM micrograph of a single, regular square crystal extracted from the eye. The crystals are 1.23 × 1.23 ± 0.08 μm (N = 20) with internal corner angles of 90.16 ± 2.78° (N = 28) (means ± SD).

Why the multiple layers? As this site from Duke University explains:

. . . .by carefully choosing the thickness of each layer, one can arrange for light that reflects at each interface to interfere constructively such that all incoming light (within a certain range of wavelengths) is reflected back toward its source, that is the layers act as a high quality mirror. It is fairly easy for biological creatures to secrete such alternating clear layers with slightly different properties, rather harder for humans to do so using chemical and mechanical engineering.

That is a remarkable feat of natural selection.

Equally remarkable is the calculation (from the authors’ simulation) that the light best reflected is in the blue-green spectrum, with a wave length of about 500 nanometers: just the wavelength of light that reaches the sea-floor environment of the scallop.

But wait! There’s more! As suggested by another of the authors’ models, light from different parts of the eye reaches the two retinas differentially. Light entering the center of the eye is preferentially directed toward the distal, or outer retina, while light coming in from the sides of the eye goes to the proximal or inner retina. Thus the two retinas give information about different parts of the habitat. Why would this be useful? As the authors suggest, the peripheral vision could help the scallop guide its movement while swimming and help it to find a new settling site, while the central vision could give information about a predator approaching them.

Finally, the data coming from the different eyes is integrated and sent to a scallop “brain”, or, as the authors describe it, “the lateral lobes of the parieto-visceral ganglion (PVG), the site of visual processing in scallops.” Thus there’s no independent data from each eye, which isn’t really needed here given that the retinas single out different parts of the scallop’s environment.

The mirror reflecting light onto an image-detector is precisely the way reflecting telescopes work, though human-constructed mirrors are very different from those of the scallop. In fact, I don’t think humans are capable of making mirrors like this bivalve does. As Leslie Orgel once said, evolution is cleverer than you are.

h/t: Gregory


Palmer, B. A., G. J. Taylor, V. Brumfeld, D. Gur, M. Shemesh, N. Elad, A. Osherov, D. Oron, S. Weiner, and L. Addadi. 2017. The image-forming mirror in the eye of the scallop. Science 358:1172-1175. (pdf here)

67 thoughts on “The fantastic eye of the scallop revealed in a new paper

    1. Since the topic has come up, I still don’t have a clear idea for the meaning of “sub” or “subscribe” either. I have always assumed it was in approval or appreciation of an essay etc. However I am still uncertain how it is used. Would appreciate being enlightened. I come to this site to learn – and it is not always science. Thank you.

      1. To use the “Subscribe” feature, which notifies you when additional comments are posted to an article, you have to make a comment in the article. When a person doesn’t have anything to say at the moment but wishes to “Subscribe” to the article it has become customary for them to leave a comment with a simple “sub” or “subscribe.”

      2. Think of it as an easier (for you to write and the rest of us to read) to say:

        “I need to check the checkbox to get comments emailed to me but I don’t have anything useful to say. So ignore this uninteresting but necessary comment.”

        IOW, nothing to do with agreement or disagreement with anything.

    1. Peter Singer made that point eloquently in the early 1970, if not before.
      If you appreciate the ability of animals to perceive and react to their environments, then veganism follows as night follows day.

      1. Well, there’s plenty of evidence that Singer is wrong about that, unless you rely on a no-true-scotsman definition of “appreciate”.

      2. “If you appreciate the ability of animals to perceive and react to their environments, then veganism follows as night follows day.”

        It does not.

        1. Not only that, but we also know that plants have the ability to perceive and react to their environment.

          Plants can even hear sounds and react appropriately to them. Arabidopsis for example can detect the sounds of caterpillars munching on a neighbor and can produce defensive chemicals in response; neutral sounds like wind do not elicit the same response.

  1. Integrating all of those data into a form that the scallop can react to in real time must be a challenge. Multiple eyes, probably flopping about each with two retinas (retinae?). I would assume limited visual acuity but enough for a significant selective advantage given the considerable resources that have been invested.

    Very cool…..I’ll definitely continue to eat them, but I’ll try to make sure I’m not being watched!

    1. My guess is that these are not producing hi-res images, but that each eye delivers no more than a handful of pixels to the visual ganglion, perhaps as few as one pixel per retina.

    2. From the penultimate paragraph above:

      …data coming from the different eyes is integrated and sent to a scallop “brain” […] there’s no independent data from each eye, which isn’t really needed here given that the retinas single out different parts of the scallop’s environment

      My interpretation of this is – the scallop isn’t ‘seeing’ [no image is formed]. The scallop is a filter feeder & it needs to know when there’s particles of food suspended in the water nearby [& perhaps other local conditions such as light levels?].

      The scallop feeds when an “eye” is triggered & goes into a startle response [closes up] when light is blocked across a few consecutive “eyes”. The scallop doesn’t need info on the direction of the food or the threat. The above is only my guesswork.

  2. The next question – is there an eyeless/PAX6 homolog in scallops? That is the regulatory gene that triggers eye development in (at least) vertebrates, flies and cephalopods; don’t know if it’s ever been looked for or found in a mollusk.

    1. Good question. I thought of that but forgot to look it up. Eyes in gastropods evolved independently of those in insects and vertebrates, but they may use the same Pax-6 signal. Perhaps a reader knows–or can look it up.

      1. BMC Evol Biol. 2017; 17: 81.
        Published online 2017 Mar 16. doi: 10.1186/s12862-017-0919-x
        PMCID: PMC5356317
        Ancestral and novel roles of Pax family genes in mollusks

        “The Pax6 expression pattern in Aculifera largely resembles the common bilaterian expression during CNS development.”

      2. Seems like Pax6 (and homologs) go back a long way – found in cephalopod eyes. Expression in Drosophila of Pax6 homologs from the mollusk Loligo (and even from C. elegens – which has no eyes) can apparently induce ectopic eyes.

        That’s the limit of my searching for this morning!

        1. That kind of thing never ceases to be amazing to me. For example, taking a gene that regulates limb bud development from a skate and putting it in chicken embryos which proceed to develop nearly normal limbs. The same model genes kicking around for, in some cases, hundreds of millions of years is very, very cool.

          1. It’s largely to do with how these regulatory genes work. They (generally) don’t produce proteins involved in the structure or function of the broader cell, but instead produce proteins involved specifically in gene regulation, and thus work as master control switches for a whole slew of other genes. Pax 6 in vertebrates, for example, will ‘turn on’ or ‘off’ hundreds of other genes, each one tweaking the behavior of the cell in a certain way. Thus if a pax-6 homologue is substituted in, provided it has similar enough affinity to the binding sites of the native pax-6, it will work largely the same

  3. They (or their bivalve cousins) certainly react quickly when a diver approaches with a camera (or knife or pry bar), and rapidly “clam” up.

    1. Indeed. I’m a scuba diver who occasionally takes scallops while diving. On California reefs scallops tend to attach themselves in crevices in rock structure with a holdfast and are normally hinged in the open position to filter feed. As a diver reaches toward the scallop the shell snaps shut. I’ve always assumed it has an incredibly sensitive sense of changes in water pressure as the hand slowly approaches it. How could it see? I guess I know better now – it does see me. I must say it gives me pause when I consider ever going scallop hunting again.

  4. Is this another example of “irreductible complexity” demonstrating that the Designer’s favorite creatures are not Humans but Invertebrates 😉

    More seriously, the cells on the retinas seems very different to those of vertebrates. The DAPI-labeled section doesn’t permit to appreciate their morphology.

  5. Now each layer doesn’t reflect the light back to the retinas; rather, each layer bends the light impinging on it a bit, and then the next layer bends it a little more, and so on until the sequence of layers has turned the light right around and reflected it back to focus on the retinas (see this explanation from Duke University).

    That’s not what the linked explanation says. What it says is that each layer reflects a portion of the incident light and lets the rest through. Layers are spaced so that the reflected portions add up and the transmitted portions cancel out, so in aggregate nearly all of the light is reflected.

    1. Sorry; wrong reference. What I said is what Carl Zimmer said in the NYT, who presumably talked to the authors:

      The layers of tiles are separated by thin layers of fluid, and as a ray of light passes through them, it gets bent further and further from its original direction. Eventually the light gets turned completely around, heading back toward the front of the eye.

      1. You’re right that that’s what Zimmer says (correct link here), but I think he just got that part of it garbled. Everything else about his description — the alternative layers of tiles and fluid (with different refractive index), and the fact that it’s optimized for a particular wavelength — is characteristic of the kind of dielectric mirror described in the Physics 162 explanation and in this Wikipedia article.

  6. guanine crystals in the shape of squares (not their natural crystalline configuration—how does the scallop do this

    In general, this is no news. I recall “news” reports from the late 18xx that table salt (NaCl, halite) would form cubic forms when grown from solution in pure water, but if grown from water with (IIRC) 1% urea in solution, it would form octahedral form crystals. This requires no change in the internal, atomic even, structure of the crystals, but the energy of the crystal surface relative to the solution in which it grows.
    The faces you see on a crystal are “special”, but in a way that would distress “special snowflakes”. What you see are the SLOWEST-growing surfaces.
    Consider a single-crystal irregular blob of NaCl in a solution supersaturated for NaCl. On the {100} form faces (perpendicular to each of the 4-fold symmetry axes), you add 2 layers of NaCl in a unit of time; on the same crystal in the same solution, you only add 1 layer on the {111} faces (the “octahedral” faces). The difference is clearly related to the energy released by so ion attaching in either face. More energy is released by the ion attaching to the {100} face than to the {111} face.
    It appears counterintuitive but is simple thermodynamics.
    A cell can influence the crystal FORM adopted by a growing crystal by manipulating the concentration of some component that changes the energy release. IIRC (and it is decades since I studied the numbers), the differences are in the order of tens of Joules per sq. [m/mm/nm] of face area, on the order of “percent” of the total latent heat of crystallization per unit volume (previous units, cubed) of the bulk crystal.
    The faces you see on a free-grown crystal are the laggards.
    To return to the guanine crystals – the cell has a (several?) compounds in solution that make the growth of one pair of faces slow (implied – guanine in these conditions crystallises with a centre of symmetry) which forces it into a tabular form. D’Arcy Thompson (“On Growth & Form”, centenary this year!) deals with the arrangement of tabular crystals in a membrane.
    Why guanine? Murray’s Answer. “Because it’s there.” It’s a common, small biomolecule. Minor stress could cause it to crystallise like calcium oxalate in a gourmand’s hallux (PCC(E), beware!).High R.I.? Cue selection.
    Subsidiary points…. ?
    I could go on for ages about crystal shapes. Oscillatory conditions of [P,T,Conc] just happen frequently in (e.g.) stick-slip fault movement in the earth. Or in animal hormones. Or body temperature. Plenty of range there to give a ‘ratchet up Dawkin’s Mount Improbable.

  7. Potentially, this poses an additional layer of complication for bivalvegans, those vegans who eat bivalues on the assumption that bivalues, including scallops, don’t feel pain (or feel less pain than other animals). A rudimentary brain and the propulsion has been taken into consideration, but when this much detail is known about eyes, I wonder if the opinion on eating them will change for those who include them among animals for which eating is not causing that much suffering.

    Granted eyes may have little to do with pain, but when you think about the level of sophistication, are we sure that scallops don’t suffer when killed?

    1. Of course, not all bivalves are built with he same features. To my knowledge, mussels don’t have eyes. But then again, I didn’t know that scallops did until I ran across that Science article!

      1. Doggone it, no wonder my pet mussel never responded to my hand signels! His name was Popeye and I brought him home form Martha Hart Pond in my home town.

        Hey, he outlived all my fish.

    2. My reading is: While the mechanism of the scallop eye is complex it’s only transmitting a simple signal ‘back’ so the scallop is triggered to feed or to ‘clam up’

      She doesn’t differentiate the direction of the signal – for improved survival she only needs to sense if a shadow is large enough for it to be beneficial to go into shut down mode. She has no sense of direction of the opportunity or the threat 🙂

      The pain thing: too philosophical for me, we’ll never know what it’s like to be a scallop.

    3. Hmmm, never heard of bivalvegans, but I’ve wondered if vegans would eat bivalves since they don’t have nervous systems. So I guess you answered my inner-query, thanks.

      Do vegans eat insects? What about honey?

      1. Do vegans eat peanuts or popcorn or potato chips? If they do they’re eating insects.

        Of course, almost all plant products have some insect matter in them.

  8. Potentially, this poses an additional layer of complication for bivalvegans, those vegans who eat bivalues on the assumption that bivalues, including scallops, don’t feel pain (or feel less pain than other animals). A rudimentary brain and the propulsion has been taken into consideration, but when this much detail is known about eyes, I wonder if the opinion on eating them will change for those who include them among animals for which eating is not causing that much suffering.

    Granted eyes may have little to do with pain, but when you think about the level of sophistication, are we sure that scallops don’t suffer when killed?

  9. I take an opportunity to share :

    Yummy Food :

    mushroom stump dusted with Old Bay, sautéed.

    … it’s a scallop substitute. It’s not bad.

  10. Remarkable! Guanine was the first base discovered, and it was discovered in guano- hence the name. I actually learned that on Jeopardy.

    So does the outer retina react to light and the inner retina react to darkness? I was a little confused with the distinction.

    1. What it looks like to me from the diagrams is that the two retinas lie in different focal planes, with one imaging distant objects and the other imaging nearby objects. Sort of like bifocal eyeglasses, but instead of two lenses projecting onto one focal plane, the scallop eye has one lens and two focal planes.

    2. @Mark R.

      From the BACK of the eye going forward:

      [1] The curved [bowl shaped] mirror
      [2] In front of this, coating the entire mirror surface is the Distal Retina [DR] – it approximates to our own retina that coats the back of our eyeballs
      [3] In front of this, coating ONLY the central region of the DR surface is the Proximal Retina [PR]

      The DR & the PR have their own photoreceptors that convert photon energy into chemical/
      electrical energy

      In one species of scallop that lives in blue ocean water…
      PR absorbed light best at a wavelength of 488 nm
      DR absorbed light best at a wavelength of 513 nm

      In one species of scallop that lives in green inshore water…
      PR absorbed light best at a wavelength of 506 nm
      DR absorbed light best at a wavelength of 535 nm

      Notice that the DR is optimised for slightly longer wavelengths than the PR in both species. There are two reasons for this [it is supposed]

      The different optimisations of frequency means the PR doesn’t screen the photons that are most sensitive to the DR.
      The PR gets photons first & steals the wavelength range it is tuned to.
      The longer wavelength photons carry on through the PR & are captured by the DR which is tuned to the longer wavelength range of those photons
      The remaining uncaptured photons hit the mirror & the mirror transmits new photons that have to run the gauntlet of DR & PR
      The result is a stronger signal is produced for a given number of photons & thus eyes that operate at low lighting conditions

      To correct for the “longitudinal chromatic aberration of the lens” [I’m quoting that] – I don’t fully understand this! It reminds me that expensive telescope lenses are made of two types of glass which bend light of different amounts. This is done to remove the rainbow colours around observed objects in cheap lenses or to reduce blur. [there’s two types of aberration, so I’m being very hand wavy here]

      I speculate that the PR & DR not only contain different pigments, but also they are slightly different densities

      I also read somewhere that the PR & DR treat polarised light differently. Worth thinking about.

      1. Thanks for this analysis. I’m still confused, because you created more puzzle pieces, but my original question is answered (sort of).

  11. I don’t think humans are capable of making mirrors like this bivalve does

    The earliest reference to a common dielectric mirror that I can google up in a snap is from 1982:

    “Mirrors of this type are very common in optics experiments, due to improved techniques that allow inexpensive manufacture of high-quality mirrors. Examples of their applications include laser cavity end mirrors, hot and cold mirrors, thin-film beamsplitters, and the coatings on modern mirrorshades.”

    [ ]

    And the multi-part mirror reminds of TI:s active mirror arrays [ ].

  12. What a superb post, and article!

    And not even *once* did the authors suggest that the scallops developed eyes because they were striving to see! (see here, if you’re reading Jerry’s posts in order).

  13. And I thought I understood the scallop eye! The distortion caused by the mirror is corrected by a Cartesian lens (another feat of evolution) and that was that, but now I see my understanding had hardly scratched the surface.
    Great post, thanks!

  14. It would be interesting to know how the retinas work and how their signals are used together. I’m thinking of the human eye, with its center-surround receptive field, where the eye only sends on to the brain the difference of information from the center of the field and the surrounding area. Could the scallop do something similar with the signals from the two retina, or would there be no point in that? (My knowledge of this field is quite limited.)

    In any case, this is an amazing paper. Thanks for posting it!

Leave a Reply