Larval fish mimic unpleasant, unpalatable, or nutritionally worthless invertebrate zooplankton

Mimicry is rife not just among animals, but among plants. But one group that’s been neglected in such studies is the juvenile stages of marine organisms. (This isn’t the case for juveniles of terrestrial animals, as seen by the numerous studies of caterpillar mimicry.)

This has begun to be remedied by a brand new study by Adam Greer and his colleagues published in Marine Ecology Progress Series (free pdf here). The authors give pretty convincing evidence that the larval stages of many fish have evolved to resemble either invertebrate zooplankton that are unpalatable or dangerous (larvae of tunicates and ctenophores), or don’t offer much of a food reward (the larvae of crustaceans). The mimicry, which involves shapes, ornamentation, and behavior of the larval fish, would be a form of Batesian mimicry, in which an edible species (the larval fish) comes to resemble either a distasteful, dangerous, or nutritionally worthless prey item (crustaceans and some invertebrates), because this kind of resemblance reduces your chance of being eaten. That, of course, is mediated by fish predators, who learn to avoid the worthless or dangerous prey and thereby select for larval fish to resemble those prey. It is a phenomenon that requires a learning predator.

Here are two examples of the “models” (“salps” are free-swimming larval tunicates) and putative mimics (larval fish), taken from the paper:


Screen Shot 2016-06-14 at 9.00.06 AM

Here’s a figure with photographs, and I’ve put the caption below the figure. All the models on the left are invertebrates, and all the mimics on the right are larval fish.  The authors note the various features of fish that presumably have been molded by natural selection to resemble the models: elongated body shape, stalked eyes, long fin rays, guts that trail behind the animal, and so on. The paper notes the various traits of fish that have evolved to look like invertebrates:

The resemblances of different larval fishes to members of the gelatinous zooplankton community are striking when viewing general body shape, transparency, and positioning of fin rays. Gelatinous zooplankton often have long, delicate tentacles that can be readily observed using in situ imaging. These tentacles typically contain pigmented or translucent ‘notches’ that are budding cormidia or nematocysts in siphonophores and hydromedusae or, in ctenophores, branches that increase cross-sectional area and improve prey capture ability. Many larval fishes have evolved delicate and elaborate fin ray extensions that closely resemble the tentacles of siphonophores and ctenophores, including small swellings that resemble the ‘notches’ (Fig. 1A−D). In the Liopropoma genus of groupers, these long fin rays have been taken to an extreme, with 2 rays extending sev- eral times the body length (Fig. 1D). The general shape and positioning of the dorsal fin ray on many flatfishes resemble salps and the projection of their tunics (Fig. 1E,F). The closest resemblance between fish larvae and gelatinous zooplankton is seen in the leptocephalus larvae that are almost identical to a cestid ctenophore (Fig. 1G,H).

Screen Shot 2016-06-14 at 9.01.19 AM

Fig. 1. Example images of morphological and behavioral mimicry showing less palatable models (from the perspective of a vi- sual predator) and corresponding mimics in their natural orientation: (A) ctenophore Euplokamis dunlapae; (B) calycophoran siphonophore; (C) flatfish larvae (Paralichthyidae), note pigmented swellings similar to A and B; (D) grouper larva (Liopropoma spp.), note pigmented swellings similar to A and B; (E) salp (Thalia spp.); (F) flatfish larva (Bothidae); (G) cestid ctenophore; (H) leptocephalus eel larva (Muraenidae); (I) leptocephalus larva in curled posture with similar appearance to a salp; (J) narcomedusa Solmundella bitentaculata; (K) flatfish larvae in curled posture; (L) chaetognath (Sagitta spp.); (M) anchovy larva (Engraulidae); and (N) clupeid larva (Clupeidae) vertically orienting. Note changes in scale bars among images

The extensions and ornaments that make the fish resemble the larvae are thought to be costly, because they reduce swimming ability. But if the fitness loss due to slower swimming is more than offset by the survival advantage of resembling an animal that predators avoid, then the mimicry will evolve.

The authors also note that some of the mimicry, as it often does, involves behavior: larval fish that are mimics like these tend to float around like their models, with their bodies curled, something that nonmimetic larval fish don’t do.

Now the evolution of mimicry is inferred, but there’s pretty good evidence for it. First of all, the larval fish are located in the same areas as the models, generally (but not always) a requirement for Batesian mimicry to evolve (predators must apply their learned avoidance to fish evolving mimicry). In this case, the models and mimics were trawled from the ocean in the Gulf of Mexico as well as off New England in the U.S.  Second, predation by fish has led to other cases of mimicry (camouflage) in adult rather than juvenile fish, so the selective pressure—fish predation—is present. Further, there is high abundance of the invertebrate zooplankton in the place where mimetic fish larvae live, and a high density of models is usually thought to be necessary in most cases of Batesian mimicry (predators must encounter the model fairly frequently, so that their aversion is reinforced compared to when they encounter the tasty mimic, which reduces aversion).

As I said, the mimicry hypothesis is intriguing but tentative. One experiment that could be done, but hasn’t been, is to expose predatory fish in the laboratory to some of the nutritious models as well as to palatable prey, let them learn to avoid the former, and then test the schooled fish (pardon the pun) by putting them with mimetic larval fish. If this is truly mimicry, predators will then avoid the larval fish mimics a lot more than they would avoid non-mimetic larval fish. In control experiments in which fish haven’t learned, they should nom up the mimetic fish larvae.

Finally, the authors do a computer simulation to show that even a tiny selective advantage of fish larvae, making them more closely resemble the models, will lead to a rapid evolution of mimicry. (We already knew this from early work on population genetics, but it’s useful to reprise.) Since larval fish are produced in huge numbers, with the vast majority dying before reproduction, the authors assumed a mortality of 0.999999 among 1 million larvae released. They they added two traits, a behavioral one and a morphological one, that reduced the mortality slightly—to 0.9999937 when both traits were present. Running this simulation over only 5000 generations showed a rapid increase in the frequency of both traits. (We used computer simulations like this one in my evolution course: you just plug in the survival values and generation numbers, and get a printout of the results.) 5000 generations, of course, is just an eyeblink in evolutionary time.

While the evidence for Batesian mimicry in larval fishes isn’t absolutely conclusive, it’s pretty strong based on the strong resemblance—involving both morphology and behavior—and the fact that the models are, according to the authors, avoided by predators. A few laboratory experiments like those I suggest above would make the argument nearly watertight—excuse the second pun.


Greer, A. T., C. B. Woodson, C. M. Guigand, and R. K. Cowen. 2016. Larval fishes utilize Batesian mimicy as a survival strategy in the plankton. Marine Ecology Progress Series 551:1-12.


  1. GBJames
    Posted June 14, 2016 at 11:11 am | Permalink


  2. Ken Elliott
    Posted June 14, 2016 at 11:13 am | Permalink

    From the edges of the universe down to the discovery of a particular boson, the world is immensely fascinating. I’m beginning to find I very much love the ability to mimic as one of the most fascinating exhibits in nature. My learnin’ on this topic has come from WEIT posts.

  3. Posted June 14, 2016 at 11:20 am | Permalink

    I wonder if one could look for selective sweeps surrounding the genes that would have most likely been involved producing the ornaments- DLX etc. I suppose this would only work if the mimicry evolved recently

  4. Posted June 14, 2016 at 11:29 am | Permalink


    So sometimes one borrows someone else’s “form most beautiful”.

  5. Loic
    Posted June 14, 2016 at 12:26 pm | Permalink

    A very interesting point.

    Out of those simulations, do we have a math formula that correlates:

    1-Probability of survival with trait.
    2-Probability of survival without trait.
    3-number of descendent per generation.
    4-duration of generation

    And other parameter that I’m not thinking about.

    Result being impact on prevalence of the trait.

  6. Gregory Kusnick
    Posted June 14, 2016 at 12:32 pm | Permalink

    In control experiments in which fish haven’t learned, they should nom up the mimetic fish larvae.

    I’m not clear on why this step is necessary to demonstrate evolved mimicry. Hypothetically, if a predator species had an innate (rather than learned) tendency to avoid larvae of a particular shape, wouldn’t that create just as strong a selection pressure on larvae to mimic that shape? Or are there technical, definitional reasons why we wouldn’t call that Batesian mimicry?

    • Posted June 14, 2016 at 1:33 pm | Permalink

      Indeed, you make a good point: is the aversion innate or learned? Every case except one that I’ve heard of (birds avoiding coral snake pattern) is learned. But when you eat something distasteful, you don’t suffer virtually any fitness loss (at least compared to getting bit by a snake), so there’s little if any selective pressure to evolve innate recognition of distasteful prey that don’t kill you or harm you much.

      These experiments were done with blue jays and monarch butterflies, and the aversion (which used to be thought of as affording protection to Batesian mimetic Viceroys, though that story is dubious now) was clearly not innate but learned. Eating a monarch made the jays sick, and they learned quickly.

      • Gregory Kusnick
        Posted June 14, 2016 at 1:57 pm | Permalink

        But when you eat something distasteful, you don’t suffer virtually any fitness loss

        This suggests that perhaps there’s a second level of mimicry at work here: the bad-tasting model is itself mimicking the flavor of some genuinely harmful substance, and the aversion to that flavor presumably is innate for sound selective reasons.

  7. teacupoftheapocalypse
    Posted June 14, 2016 at 12:36 pm | Permalink

    Such a shame we can’t persuade our plastics to degrade down to similar forms.

  8. Posted June 14, 2016 at 12:47 pm | Permalink

    Something I’ve never fully understood about mimicry. Why would evolution select organisms to look like something that tastes bad rather than just selecting the organisms to taste bad?

    • Gregory Kusnick
      Posted June 14, 2016 at 12:51 pm | Permalink

      Just guessing, but maybe tasting bad has significant metabolic costs. Mimicry of something that’s already absorbed those costs may be the cheaper option for some organisms.

    • Posted June 14, 2016 at 1:34 pm | Permalink

      There are several possible answers; Gregory’s is one. Another is that there’s more genetic variation in populations for your body shape or color than there is for being toxic, so selection will snap up the first variation that protects you. (That’s a speculation, not a fact.)

    • Torbjörn Larsson
      Posted June 14, 2016 at 1:58 pm | Permalink

      Maybe it isn’t “just”, but I believe there is oily fish that has that trait.

      In any case, it seems to me main fitness difference lies in having a model. Being the model or being the mimicker wouldn’t make a crucial difference in fitness at a guess. The model would survive some predator mistakes that the mimicker would not. (But I don’t know if this is tested, I can’t mimic mimicry expertise.)

      If so the model/mimicry evolution could start anywhere, a change that tends towards becoming distasteful, poisonous, looking or behaving as a present model, et cetera.

      • Torbjörn Larsson
        Posted June 14, 2016 at 2:10 pm | Permalink

        So Jerry commented while I composed my comment. So we could look at relative frequencies of variation change?

    • Mark Sturtevant
      Posted June 14, 2016 at 3:03 pm | Permalink

      What Jerry said. I can only add that selection can only use whatever variation is offered. So if a form (mimicry) comes up that imparts fitness, then form it will be. If variation happens to offer a real defense against predation (like toxicity), then that is the direction natural selection may take. Of course whatever direction things go in will also depend on other factors like cost.

    • Posted June 14, 2016 at 4:22 pm | Permalink

      Yes, those answers make sense to me. Thank you.

  9. DrBrydon
    Posted June 14, 2016 at 1:37 pm | Permalink

    I’ve wondered for some time if the language used to talk about evolution doesn’t occasionally create problems for acceptance/understanding. I think that, to a novice, saying the larvae mimic something implies intentionality or at least application on the part of the larva, when, of course, the larva is the passive receiver of the effect. Wouldn’t it be better (for comprehension if not for grammar) to say something like “natural selection has lead to the evolution of larval forms that are mimics of”? A person not familiar with the idiom might ask, How could a larvae choose to imitate anything?

  10. Posted June 14, 2016 at 2:48 pm | Permalink

    Very cool, and the discussions here as always! Thanks for the free education PCCE! 🙂

  11. Mark Sturtevant
    Posted June 14, 2016 at 2:57 pm | Permalink

    How interesting! And no small matter to have possibly discovered a huge realm of mimicry. This could be a major result.
    I am open to finding that the predators do not necessarily ‘learn’ to avoid low nutrition models. They might just be hard wired to do so, as an instinctive behavior. No matter. The predators are still predicted to avoid the models and their mimics.

  12. Redlivingblue
    Posted June 14, 2016 at 3:08 pm | Permalink

    A welcome diversion among the dreadfully disturbing national news. Thanks for the science post!

  13. Paul S
    Posted June 14, 2016 at 3:46 pm | Permalink

    Very cool. Always nice to read science articles that I can grasp.

  14. Michael Hart
    Posted June 14, 2016 at 5:26 pm | Permalink

    I teach this area of invertebrate zoology, and I think there is a pretty big hole in the reasoning of this paper. One could just as easily switch the two columns of photos (like the ones Jerry posted, from the Abstract) and ask why the adult salp (above) and the adult ctenophore (below) have evolved as Batesian mimics of the fish larvae. After all, they have the same sorts of complex life cycles of the fishes, with a similar range of larval and adult morphologies. In these cases, the larval salps don’t look like the adult salps (they look like little tadpoles, complete with notochord and dorsal hollow nerve cord, both lost at metamorphosis into the adult salp). And the larval ctenophore doesn’t look at all like the adult ctenophore in that photograph (it looks like other more typical ctenophores). Using the same kinds of (weak) evidence in the paper, one could just as easily argue that it’s the adult salps and ctenophores that have evolved to mimic fish larvae. Everything else in the study seems to be conjecture (with a little fish-biased cherry-picking thrown in).

    Also Jerry just to nit-pick, salps are definitely not larval tunicates, they are more like planktonic tunicate colonies; they are morphologically and functionally much more like the adult tunicate than like the tunicate tadpole larva.

    • Posted June 14, 2016 at 6:11 pm | Permalink

      The way to solve your question, which is a good one, is to look at outgroups, or animals in an area without fish predators, to see which group (or neither, or both), has evolved a novel larval morphology.

      • Michael Hart
        Posted June 14, 2016 at 6:47 pm | Permalink

        Yes, that’s a good idea. It would help in telling the difference between the mimic and the model.

        It’s difficult to come up with a good test along those lines because novel larval morphologies are common in marine animals. In the case of the salps and tunicates and their relatives, it’s likely that the tadpole larval form is ancestral, and the adult form (never mind the details) is highly derived. The sister group to the tunicates (Ascidiacea) plus the salps and their relatives (Thaliacea) is a clade of planktonic gelatinous zooplankton called the Larvacea that are solitary (rather than colonial like the salps) with a form like the tunicate or salp tadpole larva but with gonads. That would be more consistent with the idea that the adult salp (the derived form) is a mimic of fish larvae. Similarly for that specific ctenophore group (the Cestidae), the adult form is highly derived compared to the other ctenophore taxa (most of which more closely resemble jellyfish in a superficial way), and maybe consistent with the idea that the adult cestid is a derived form that is a mimic of fish larvae.

        There is another weakness of the paper, related to the first. It simply assumes that gelatinous zooplankton are unpalatable or nutritionally worthless in comparison to yummy, economically important vertebrates (fish larvae). Possibly this is just a bias toward the importance of fish larvae on the part of fish ecologists (that seems to be the background of all of the authors). I don’t know of data to compare those taxa in those ways, although they might exist and I’m just ignorant of them. But in general it’s not true that gelatinous zooplankton like salps and ctenophores are not worthless or unpalatable. There are certainly lots of predators that specialize in catching and eating gelatinous zooplankton (turtles, some fishes, other gelatinous zooplankton).

        And (sorry to go on and on) many of the gelatinous zooplankton that are proposed to be models for Batesian mimicry by fish larvae (especially the chaetognaths and the ctenophores) are themselves famous as voracious predators on fish larvae. I am not sure about the logic of prey that mimic their predators – is there any precedent for that? Again I might just be ignorant of many examples like this but I can’t think of one.

        • Michael Hart
          Posted June 14, 2016 at 6:51 pm | Permalink

          Er, “it’s not true that gelatinous zooplankton…are worthless or unpalatable.”

    • Gregory Kusnick
      Posted June 14, 2016 at 6:26 pm | Permalink

      Another test might be to do experiments with naïve predators (on the assumption that the aversion behavior is learned).

      Put untrained predators in a tank with the putative model species, plus some other palatable source of food, and see which food they prefer. Then repeat with a fresh batch of predators, the putative mimic species, and the palatable food source. If the predators develop an aversion to the model but not the mimic, then the hypothesis is confirmed.

    • Posted June 15, 2016 at 11:32 am | Permalink

      I had assumed – I guess naively – that there would be some way to tell which had arisen first.

      (Of course, it could be an “arms race” or first change in one, then in the other, repeatedly.)

%d bloggers like this: