Why are there imperfect mimics?

A few colleagues and I recently had a paper published in Nature on “A comparative analysis of the evolutionary of imperfect mimicry”. Those of you fortunate to have a Nature subscription can read the paper here.  Alternatively, you can email me and I’ll send you a copy.  Unfortunately, I can’t make the paper available due to issues with copyright from Nature (see elsewhere for details of scientists’ love-hate relationship with publishers…) but I can summarise the paper here.

Mimicry in nature

There are lots of ways in which animals and plants evolve to look like something else in order to gain an advantage.  I’ve produced a little gallery of some of my favourites below (click to embiggen – they are pretty big images):

A & B – The stone flounder, Kareius bicoloratus, which can change its colour in order to blend in with the sea floor.  Both of these pictures are the same species, demonstrating the effectiveness of this colour flexibility.  [Public domain]

C – What looks at first glance like an ant is actually a spider, Myrmarachne japonica.  The genus name literally means “ant-spider”: “myrm-“=”ant” and “-arachne”=”spider”. [Photo by Akio Tanikawa]

D – A second classic example of mimicry is the genus of moths called Hemeroplanes.  The caterpillars of these moths have evolved a startling similarity to snakes which they use (we presume) to deter bird predators.  The “head” of the snake is actually the tail of the caterpillar, which is inflated when the animal is alarmed.  [Photo from Tom Hossie’s Caterpillar Eyespots Blog, and go there for more details].

E – The tawny frogmouth, an Australian bird which is more closely related to the nightjars than the owls that it resembles, rests on dead trees during the day.  The patterning of the bird’s plumage helps camouflage the animal against the tree bark.  [Photo by C.Coverdale]

All of these examples exhibit fairly extraordinary similarities to whatever they are trying to match, demonstrating the capacity of natural selection to produce near-perfect mimics.  However, there are many instances where this “mimetic fidelity” falls far short of perfection and a good example of this is the range of mimetic fidelities that occur within the hover flies (Diptera: Syrphidae).  Hover flies are a group of “true flies” (order Diptera) so they are related to midges and mosquitoes.  The adults mostly feed on nectar and pollen, while the larvae feed on a range of foods (including aphids, and they are recognised as a useful biocontrol agent for these crop pests).

Imperfect mimicry in nature

Of primary interest to evolutionary biologists, however, is their strong similarity (in some cases, at least) to stinging wasps and bees.  I’ve produced a few photographs below to show the variation in mimetic fidelity:

The wasp on the left is in a completely different order of insects (Hymenoptera) but the hover fly Spilomyia longicornis (second left) bears a striking resemblance to it.  The next hover fly, Metasyrphus corolla, is still noticeably similar to the wasp, but lacks the clearly-defined, complete, bright yellow stripes of the wasp.  Finally, Syritta pipiens (far right) has a different body shape and patches of yellow rather than stripes.  So the key question here is:

If natural selection can produce excellent mimics why doesn’t it always do so?

This is the puzzle of imperfect mimicry, and there have been a number of hypotheses to explain it:

  1. Eye of the beholder hypothesis – Some people have proposed that the hover flies are all good mimics but they only appear poor within the context of human vision.  If a bird was looking at these hover flies, it would see all good mimics.
  2. Multi-model hypothesis – Perhaps the species are not trying to resemble one wasp, but a number of different wasp species?  By not resembling any one species, but partially resembling many, a hover fly would be imperfectly mimetic but still gain benefits through many mimetic associations.
  3. Kin selection hypothesis – This hypothesis is a little more complex.  If you have a group of wasps and a group of hover flies that are perfectly mimicking those wasps, a predator will attack everything because it cannot differentiate.  It has to eat, and a certain proportion of the time it will come across a tasty hover fly (in between being stung by nasty wasps).  This means that there is actually no benefit from mimicry at all!  However, if some of the hover flies are not perfect mimics, the predator can distinguish hover flies from wasps and so will eat the worst mimics while the best mimics survive.  This suggests that there will be an equilibrium resemblance that is close to perfection without ever reaching it.
  4. Constraints hypothesis – Maybe there is some other pressure that is pushing back against selection for mimetic perfection?  A possible candidate might be thermoregulation, as the patterns of black colouration on the hover flies will affect the amount of heat that is absorbed from the sun.  Perhaps this heat absorption is more important than the avoidance of predation, and so colour patterns move away from those that most resemble wasps.
  5. Relaxed selection hypothesis – Finally, some scientists have predicted that the intensity of selection might be reduced closer to mimetic perfection and that this might happen earlier in some species rather than others.

What we did to test them

So how do we test these hypotheses?  We used the “comparative method”, which involves looking at many species and comparing their traits in the context of their evolutionary relationships.  We asked 21 human volunteers to rate 38 species according to how closely they resembled a wasp, a honey bee or a bumble bee.  We then measured specimens of those species (hover flies, wasps and bees) to calculate differences in their shape and size.  This gave us two measures of similarity: human rankings and measurements.  We also knew how abundant the different species were from previous field studies.

I’ll take the hypotheses one by one to show what we found:

  1. Eye of the beholder hypothesis – We compared human rankings and the similarity based on measurements and found that there was a strong correlation.  We already knew that human rankings correlate with bird rankings of fidelity (from a previous study) so this suggests that it isn’t human vision that is giving the appearance of poor mimics – those mimics really are poor.
  2. Multi-model hypothesis – Comparing measurements of species, we saw that all of the hover flies closely resembled one another, all of the wasps closely resembled one another and all of the bees closely resembled one another.  There was no evidence that any of the hover flies are intermediate between two models.
  3. Kin selection hypothesis – First of all, the kin selection hypothesis requires that closely related individuals remain close together (so offspring don’t move far from their parents, and siblings from their siblings).  However, hover flies fly considerable distances making this unlikely.  Second, we would expect that the mimetic fidelity of a species would decrease as abundance increased.  This prediction arises because a larger number of hover flies would make it more likely that avian predators would try to eat them.  There is, therefore, a greater pressure on maintaining mimetic imperfection so that the benefits of some similarity are retained.  We found that there is no evidence of a negative relationship between abundance (measured in field studies) and mimetic fidelity.
  4. Constraints hypothesis – Due to the diversity of potential explanations for constraints, this is really a series of hypotheses and we couldn’t demonstrate that constraints were not playing a role somehow.  However…
  5. Relaxed selection hypothesis – When we looked at the relationship between body size and mimetic fidelity there was a very strong relationship.  Species that had larger bodies were much better mimics than species that were smaller.  This is easily explained by looking at which species of hover fly would be most profitable for a bird to attack.  If you are a big, juicy hover fly, birds are going to attack you more because it is worth chasing you.  If you are a small hover fly then birds will not bother.  As a result, selection is relaxed on smaller species and so those species do not evolve a high degree of mimetic fidelity.


We demonstrated that variation in mimetic fidelity in hover flies is very likely due to lower predation on smaller, less profitable species leading to relaxed selection for mimetic perfection.  There are a few alternative hypotheses tied up in the “constraints” hypothesis that we cannot discount, but the strength of this relationship suggests that they are relatively minor.  Of course, the next step is to test this theory in other systems, including the evolution of eye spots in caterpillars.  We already have some anecdotal evidence that larger caterpillars tend to be better mimics, but we are busy testing that right now so watch this space!

PS. It’s likely that I haven’t explained everything as clearly as I could have done, so feel free to ask questions or point out mistakes in the comments!


17 thoughts on “Why are there imperfect mimics?

  1. Pardon me if I’ve missed something, but it seems to me that there is an operating assumption that natural selection should produce perfect mimics, whereas it is possible that even imperfect mimics can survive. Also, it isn’t a given that the organisms are “trying” to mimic anything; it just so happens that they do, in an accident of evolution, with no specific purpose. There just happen to be numerous forms that just happen to fit a niche.

    • Hi Jim,

      There is a (slightly simplistic) assumption that natural selection should be driving species to ever greater degrees of mimetic fidelity. The benefits to mimicry are obvious in terms of decreased predation by birds and we know that some species evolve to be extremely good mimics. So why aren’t all species good mimics? Of course poor mimics can survive, but they should survive at a lower rate than better mimics so that the genes for better mimicry spread through the population. We looked at reasons why this is not the case.

      I agree that species are not “trying to mimic” anything. If I anthropomorphise, it is only a colloquialism based on teleonomy, not teleology. Animals exhibiting the natural variation that exists within the population experience differential mortality such that better mimics survive to produce more offspring.

  2. I thought your paper was interesting and well done except for the fixation on bird predation. I agree that birds may be an important selective factor in the evolution of the larger fly mimics of wasps and bees – not only do these hoverflies look very wasp or bee-like (and the same with asilids like Laphria), but they tend to sound like their models. Also, the big impressive mimics, e.g. hoverflies like Spilomyia sayi or stratiomyids like Stratiomys, act as self-confident as their vespid models on do flowers. So much so that they are easy to collect by hand.

    You can’t say the same about the smaller ‘imperfect’ hoverfly mimics: they are very wary, very fast, and highly manoeuvrable. My anecdotal observations suggest that social vespids and specialized crabronids (e.g. Ectemnius spp.) are more likely to be strong selective forces. Both hunt flies including the smaller hoverflies aggressively at flowers (I don’t think there is any evidence that birds do unless caged and having no other choice). So, rather than selection being ‘relaxed’ among the mimics that look less perfect to us and birds, it is coming from another direction. Perhaps yellow flashes discourage vespids from pursuit, but in any case, I’m doubtful that birds are part of it.

    • Hi Dave,

      I think a lot of what you have said is absolutely correct. Our finding is based on the fact that birds are less important as predators in smaller species of hover fly. Birds ignore smaller species and so those smaller species have less pressure to evolve to look like something that birds won’t attack. We were only interested in wasp-/bee-mimicry, which is a defense against birds.

  3. Nice blog! I think the Kin selection hypothesis needs a bit more explanation. I guess the idea is that the imperfect mimics reached an optimum on the resemblance scale – which is imperfect resemblance -, and owe their selective advantage to their kin being sacrificed. Or: by being eaten by birds the sacrificed flies enhance the fitness of their kin. The kin relationship is essential to avoid evolution to creep towards more successful mimics.
    I think this hypothesis is ingenious even if implausible. Are their any convincing examples supporting this theory at all? One would expect cheaters to spoil the game immediately.

    • Hi Louis,

      The kin selection hypothesis was one of the most difficult to explain to our reviewers in the paper, and I struggled even more to get it in plain English here… Johnstone (2002) states the hypothesis like this:

      “Signal detection theory predicts that predators will modify their level of discrimination adaptively in response to the relative frequencies and similarity of models and mimics. If models are rare and/or weakly aversive, greater local similarity of mimics can thus lead to greater attack rates.”

      In other words, as a population of mimics moves closer to mimetic perfection, they lose the protection that mimicry provides if the models are rare or not too toxic/dangerous because the predators will attack that mimic population more.

      As far as I know, Johnstone only provided a few hints that some of the predictions of his model were met: (i) lower benefits to mimicry at higher densities, and (ii) poor mimics are more abundance. Of course, we demonstrate that poor mimics are not rarer than good mimics, which is one of the reasons that we discount the kin selection hypothesis in this system.

      • In your paper I found one phrase about this hypothesis: “third, kin selection, such that imperfect mimicry is maintained through its benefit to conspecifics carrying the same trait”.

        Well, it sounds plausible that mimics won’t work if the model is rare relative to the mimics and/or relatively harmless. But that does not imply that less perfect mimics would perform better than perfect ones, nor is the ‘kin’ part of selection obvious.
        Maybe the missing part is: the imperfect mimics have reached a density and level of mimicry where selective pressure towards more perfect mimicry has become low or negligible. Maybe an optimum has been reached weighing mimicry benefits against costs. Or the imperfect mimics are on the way to become perfect ones, or they have given up the effort .. (the Disequilibrium hypothesis in the Supplement, this should be allowed theoretically?!).
        And then .. how does mimicry – perfect or imperfect – benefit kin (rather than conspecifics)?

  4. Questions from a layperson:

    Why are there so few insects that mimic bees/wasps through natural selection when they would obviously benefit? I’m thinking of house flies and flesh flies in particular. Their size is almost identical to bees and wasps, and they inhabit the same environments, yet they don’t seem bothered enough by birds to select for mimicry.

    What is different about the species that use mimicry that causes mutations in the genes that alter colour/size? Could some species stop improving their mimicry because they develop genetic stability?


  5. Hi Chris,

    Sorry I missed this comment. Here are some attempts at answers to your questions:

    1 – “Why are there so few insects that mimic bees/wasps through natural selection when they would obviously benefit? I’m thinking of house flies and flesh flies in particular. Their size is almost identical to bees and wasps, and they inhabit the same environments, yet they don’t seem bothered enough by birds to select for mimicry.”

    There are a few potential explanations for why there isn’t more mimicry, and it is likely that these explanations act together rather than one being correct and the others incorrect. The first is that other species do not possess the evolutionary tools to accomplish mimicry. Species need to possess a raw genetic blueprint that can be acted upon by natural selection to produce the similarity to another group. Maybe other species of flies simply do not have the capacity to evolve “yellow stripes”. A second explanation is that when too many species mimic something nasty like a wasp, there are too few nasty individuals to educate the predators. This means that suddenly everybody gets eaten. We call this “density-dependence”, because an increase in the density of mimics reduces the effectiveness of that mimicry. A third potential explanation is that the hoverflies we discuss are particularly vulnerable to avian predation as a result of their habit of flying around flowers. Perhaps other species are less conspicuous or inhabit areas that are less visited by birds. A fourth (off the top of my head, though there are certainly more) is that there are traits that are more beneficial than mimicry but that cannot be maintained along with mimicry. It has been proposed that little black flies are black because it enhances their ability to absorb heat from the sun. If you`re suddenly bright yellow, it reduces how much heat you can absorb.

    2 – “What is different about the species that use mimicry that causes mutations in the genes that alter colour/size?”

    All of the responses to your first question are relevant here: evolutionary constraints (no genetic raw material), competing traits (e.g. thermoregulation), the biology of the species (susceptibility to avian predation)… Body size is a very different trait, and we see a lot of rapid changes in body size (for example, island dwarfism in elephants over just a few thousand years), so this is likely the more flexible of the two traits.

    3 – “Could some species stop improving their mimicry because they develop genetic stability?”

    We think that the species in our system are all in a state of equilibrium with respect to their mimicry.

  6. […] First, though, what does behavioural mimicry look like?  In the system that we studied, hover flies (which can’t sting) have evolved to look like wasps and bees (which can sting). The theory is that birds learn to associate the characteristic yellow and black patterns of wasps with a nasty sting and so leave yellow and black insects alone.  Hover flies have exploited this and so get eaten less even though they don’t have any form of defence. Here are a few photos (taken by Heather Penney and which I mentioned in an earlier post on imperfect mimics): […]

  7. Fantastic blog! Just one small comment… the Hemeroplanes snake mimic you mention actually inflates its anterior section, not its tail (as Hossie explains in his blog, and which can also be seen quite clearly in videos of it). The head of the “snake” is the upside down head of the caterpillar.

Leave a Reply to Jim Dominic Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s