The problem of warning coloration

Many species have bright and conspicuous colors. When these colored individuals are dangerous or distasteful, the pattern is called warning coloration or, as biologists call it, aposematic coloration.   These patterns act (and presumably evolved) to cause predators to avoid attacking the individuals, which either taste dreadful or could hurt the predator.  Predators (often birds) do indeed learn to avoid the colors after an initial experience with them.  You, for example, have learned to avoid insects having black and yellow stripes, so you won’t go near anything that looks like this:

Here are some examples of aposematic colors and patterns in diverse species:

The striped skunk (Mephitis mephitis), which squirts a nasty fluid from anal glands

The poisonous coral snake (Micrurus fulvius)

The ladybug (Coccinella septempunctata); not a bug but a beetle.  When grabbed by a predator, they secrete a foul-tasting fluid from their leg joints

The noxious caterpillar of Eupackardia calleta (the calleta moth).

The blue-ringed octopus (Hapalochlaena lunulata). It’s one of the deadliest sea animals, with a venomous bite that’s killed many people.

The yellow banded poison dart frog (Dendrobates leucomelas), which secretes toxic alkaloids through the skin.

Now scientists have shown that lots of these species are indeed toxic or dangerous to predators, and that the predators do indeed learn to avoid the patterns and colors (sometimes, as we’ve discussed before, the predators can even evolve a hard-wired aversion to the pattern).  So this system is good for both the aposematic species and its predator.  But there’s always been a big problem with this system: how did the coloration evolve in the first place?

You can show from phylogenetic analysis that many of these brightly-colored species evolved from camouflaged (“cryptic”) ones.  This lead to the evolutionary problem:  the first mutant individual that was toxic but had a new, bright color would call attention to itself.  Since no predator has yet learned, since the pattern is new, that individual would be more likely to be eaten than its camouflaged conspecifics, and so there would be a penalty attached to the new mutation.  Only when those aposematic individuals were sufficiently common—and thus could be encountered often enough by predators who would remember the pattern and avoid the next aposematic individual—could they enjoy the protection afforded by their color.  The problem of aposematic coloration, then, is that at very low frequency it would seem to be disadvantageous and could not invade a population. Only when it gets above a certain frequency would natural selection then act to fix the trait in the species. (This kind of natural selection, in which the advantage or disadvantage of a mutation depends on its frequency in the population, is called frequency-dependent selection.)

Various solutions have been offered to this puzzle, but all of them suffer from problems. I won’t recount these, but want to briefly highlight a paper that offers a possible solution.  In the June issue of Evolution, Michael Speed and his colleagues suggest a way to get aposematism off the ground.

The solution is this: although the first mutant aposematic individuals are more likely to be detected by predators, they have another feature that more than offsets this advantage: they forage more openly, enabling them to get resources that their cryptically colored (and presumably hidden) conspecifics don’t share.  This could increase the fitness of aposematic individuals above that of their cryptic kin.  Speed et al.’s paper conducts numerical simulations, involving “optimization” of the color and foraging traits, that show (as you might guess), that if these foraging advantages (which of course make them more conspicuous to predators) are sufficiently large, they can outweigh the deleterious effects of being seen and attacked by a predator.

Under this model, then, the evolution of warning coloration is associated with a niche shift.  Speed et al. show that their model makes several predictions, which preliminary data seem to support:

  • Species that are aposematic should expose themselves more often to predators (because they’re out seeking resources) do their non-aposematic relatives.  Speed et al. show that there is indeed evidence for this in poison-dart frogs and Lepidoptera: aposematic species forage during the day more often than do their cryptic relatives.  And, of course, skunks are perhaps the noisiest and most conspicuous foraging mammals in North America!
  • Aposematic species should, because they are able to get more resources, be able to evolve larger body sizes than their relatives.  There are preliminary data supporting this in insects and amphibians
  • For species with very different life stages, like Lepidoptera, the mobility of a life stage should correlate with its degree of warning coloration.  Speed et al. cite some supporting evidence: “Pupal forms of lepidopteran species that are aposematic in larval or adult stages are often cryptic. . . consistent with the fact that they are completely stationary and hidden away from predators.
  • Aposematic species should “make use of a greater range of food plants than closely related, but cryptically colored species.”  This prediction has not been tested.

This model sounds good, and may well obviate the problems with getting warning coloration off the ground, but there’s one problem.  It is an optimization model in which several traits are simultaneously “optimized” by natural selection.  But evolution has to work on genes, and a gene-based model of Speed et al.’s theory seems a bit problematic.  Which mutation came first, the one for bright color or the one for foraging more openly? (Presumably these two types of traits—behavioral and morphological—are based on different genes.) If the aposematic mutation came first, it would still encounter the problem of being more visible to predators without the compensating benefit of getting more food.  If the foraging-openly mutation came first, it would be disadvantageous because it would expose itself to predators more (and if there were an advantage to foraging more openly if you were cryptically colored, the species would already be doing it).

In other words, the theory seems to depend on the concatenation of new mutations for two traits, each of which is individually disadvantageous at low frequencies but which become advantageous when they occur together.  But, given the rarity of mutation, this seems to be improbable. Indeed, I know of only a few cases of evolution that involve two mutations that are useful together but individually disadvantageous. (For you evolution fanatics, one example is the evolution of compensatory mutations in the loop structure of transfer RNA.)

This shows a persistent problem with optimization models: what looks optimal from an evolutionary standpoint may not be attainable when you deal with the messy genetic details.  And, after all, anything that evolves has to obey the rules of population genetics.

I may be missing something, and perhaps the authors will correct me, for I really want this idea to be true. It fits the facts about nature so well!

_______

Speed, M. P., M. A. Brockhurst, and G D. Ruxton.  2010.  The dual benefits of aposematism: predator avoidance and enhanced resource collection.  Evolution 64:1622-1633.

47 Comments

  1. Alex M
    Posted July 5, 2010 at 12:40 pm | Permalink

    Maybe the two mutations only had to occur together in one species. Once that was established, any subsequent species acquiring bright colouration would be avoided by predators who were now avoiding the first species, it maybe being likely that “don’t eat brightly coloured animals” is a better description of the predator’s behaviour than “don’t eat this one particular type of brightly coloured animal”.

    • Andrew
      Posted July 5, 2010 at 4:46 pm | Permalink

      Mimics often closely resemble the unpleasant or dangerous animal they look like; the Monarch and Viceroy butterflies for example. If predators were just avoiding bright colours, this would be unnecessary.

      • TheBlackCat
        Posted July 6, 2010 at 9:16 am | Permalink

        I read a summary of a paper a while back that argued that the so-called monarch mimics are distasteful themselves, and so aren’t really mimics.

        • Andrew
          Posted July 6, 2010 at 3:24 pm | Permalink

          So if the Viceroy is also distasteful, the question becomes why have the Viceroy and Monarch butterflies converged onto a common colouration? Whatever the answer, I don’t think it would contradict the idea that predators are avoiding specific patterns and colours and not just bright colours in general. In fact, I think a distasteful Viceroy would support it more.

  2. Iain Brown
    Posted July 5, 2010 at 12:47 pm | Permalink

    Don’t we need to have 3 mutations working in parallel? Surely the creation of poison is evolutionary disadvantageous, unless it means you get eaten less, and that will only happen once you develop the distinctive colouration? Or do we think the toxins came first?

    • whyevolutionistrue
      Posted July 5, 2010 at 12:53 pm | Permalink

      Good comment, but it seems likely that the toxins came first. That’s because there’s no apparent disadvantage to being toxic: if the predator lets you go without killing you (as it sometimes does with poisonous amphibians or insects, and obviously with the skunk), there would be a flat-out individual advantage to being toxic. In the case of the blue-ringed octopus the venom probably evolved to help it kill prey, but can be used against predators. And in bees and wasps, the advantage of having a sting is obvious.

      • Andrew
        Posted July 5, 2010 at 4:54 pm | Permalink

        I’ll add that it could also be a consequence of feeding strategies. If you can figure out how to eat a poisonous plant without dying, you have a definite advantage over the competition. If you then learn how to store the toxin instead of dumping it right away, you have another advantage. An example is Monarch butterflies and milkweed.

      • articulett
        Posted July 5, 2010 at 4:56 pm | Permalink

        Some bees die after they sting, so the advantage cannot be to themselves.

        • jeff
          Posted July 6, 2010 at 3:56 pm | Permalink

          Because of the particulars of honeybee genetics, workers are more closely related to their sisters to any descendants they could have, and ‘suicidal’ stinging could evolve. When the stinger rips out of the bee, the venom gland is still attached and continues to pump venom. Harder to brush off than a whole insect.

        • JesseS
          Posted July 6, 2010 at 4:16 pm | Permalink

          You’re right, the advantage is to the colony. Remember, the majority of bees in a colony don’t actually breed, so any advantage to the group-level supersedes disadvantages at the individual level.

          Social insects are better thought of as either super-organisms or in terms of group selection.

  3. Ray
    Posted July 5, 2010 at 12:55 pm | Permalink

    I’d say the order of mutations would be as follows

    1)poison — seems beneficial all around
    2)agressive foraging behavior — it’s possible that even with camo it would be easy for the creature to be seen (especially if the agressive foraging takes the animal to an environment with differing coloration from where the camoflage evolved)
    3)the predator has now learned to avoid the animal, but it’s better at avioding versions of the animal with more extreme colors. The colors then get progressively brighter.

  4. Posted July 5, 2010 at 1:21 pm | Permalink

    Won’t any alternative model make those very same predictions? However it evolved, once a species is aposematic, then it can afford to have all those features mentioned in the predictions.

    • whyevolutionistrue
      Posted July 5, 2010 at 1:27 pm | Permalink

      Yes indeed; if you’re already aposematic you can afford to forage more broadly, noisily, etc. But then that leaves the problem of how to get aposematic coloration in the first place. I suppose one observation on the authors’ side is the correlation between aposematism and mobility among life stages of single species. You wouldn’t expect that if aposematism evolved first and that simply permitted open foraging. If that were true, why wouldn’t immobile stages, like pupae, also be aposematic?

      • Posted July 5, 2010 at 1:57 pm | Permalink

        Just to be clearer, my point was that for the model to be favored because of a prediction it makes, the prediction has to be one that any other model would not make.

        “If that were true, why wouldn’t immobile stages, like pupae, also be aposematic?”

        I don’t know anything about developmental biology, but are the pupae of species usually of the same color as their adult stage? If no, then the coloration only kicks in at the adult stage, and so the gene for aposematism has an effect only then. Presumably, other models would not depend solely on aposematism during the pupae stage, and therefore this prediction also does not favor the model.

  5. Divalent
    Posted July 5, 2010 at 1:41 pm | Permalink

    I’m not sure I see the problem. Most species, poisonous/distasteful or not, have a distinct coloration, although it may be suble. But as long as the predators can discern the difference (even weakly), then NS will tend to accentuate the coloration pattern. (and once the pattern is strikingly obvious, they can then forage openly).

    (Prior to that point, selection for poison/distastefulness would seem to require that at least some of the time it results in an aborted predation.)

  6. Sili
    Posted July 5, 2010 at 1:49 pm | Permalink

    WATERLOO!!

  7. Thanny
    Posted July 5, 2010 at 1:49 pm | Permalink

    It seems very likely to me that all the predictions are worth little, since they are being tested against species which already have the toxicity and advertisement – there’s nothing to say they didn’t acquire the wider foraging traits after the anti-predatory adaptations.

    I would think the toxicity evolved first, for accidental reasons (such as a side-effect of diet). Being species-wide, would-be predators would have plenty of samples to develop an aversion to. Given that state of affairs, any marking which allowed said would-be predators to more quickly recognize the noxious meal would be favored.

    It’s a mistake to think of an individual which might be plucked out of existence before the marking could serve its purpose. It would be a small population with just slightly more prominent markings, that on average would leave more descendants, as the predators in the area recognized them as a bad morsel more quickly (or, more accurately, under more diverse conditions) than versions which remained more cryptic.

    There’s a fine line between calling attention and increasing recognition of something already being avoided, yet I think it’s there.

    • Posted July 5, 2010 at 2:06 pm | Permalink

      I think you are right. The predator would already have evolved/learned some way of recognizing and avoiding the toxic species. Then the toxic species evolves more pronounced coloration that makes it easier for the predator to recognize it.

  8. Eddie Janssen
    Posted July 5, 2010 at 1:57 pm | Permalink

    Is there any possibility of sexual selection playing a part in this proces (at least in some of these species)?

  9. Posted July 5, 2010 at 3:07 pm | Permalink

    I’m no biologist and perhaps this is a silly suggestion but what if the selection for brighter colouration isn’t about attracting or repelling predators at all but is instead all about attractiveness for mates? The adaptation for poison or other defensive benefits favoured by evolution might be explained by simply allowing for more mating by those with these more successful defensive mechanisms who also happened to have brighter colouration to occur.

    I’m sure colouration for mating preference has already been tested in these species, but just in case…

    • whyevolutionistrue
      Posted July 5, 2010 at 3:27 pm | Permalink

      It’s not a silly suggestion but probably isn’t true for two reasons. First, aposematically colored species are usually colored the same in both sexes, while in most sexually selected species it is the males who are much brighter or more colorful. Second, aposematically colored species are harmful or poisonous, or else they are mimics of harmful or poisonous ones. You wouldn’t see that correlation unless it had evolved as a warning.

      • Posted July 5, 2010 at 5:02 pm | Permalink

        You see? I didn’t know that. Thanks for the explanation.

        • Posted July 5, 2010 at 5:03 pm | Permalink

          Oh, and how the hell did I get a high school diploma without ever having to take biology? It’s a gap I’ve been trying to fill ever since.

          • Hempenstein
            Posted July 5, 2010 at 6:00 pm | Permalink

            If only many more in your category would do the same!

  10. whrrr
    Posted July 5, 2010 at 5:01 pm | Permalink

    Are bright colors needed to initiate predatory avoidance of poisonous species?

    I automatically assume all wild mushrooms are poisonous, for example.

  11. Andrew
    Posted July 5, 2010 at 5:04 pm | Permalink

    As an example of linked changes in behaviour and colouration, wasn’t there a program in Russia to try and breed a docile silver fox for the fur trade? They got docile foxes all right, but they also lost the silver coat colour.

    • Hempenstein
      Posted July 5, 2010 at 5:58 pm | Permalink

      I believe that was with mink. The docile mink did lose their pigmentation.

      • Andrew
        Posted July 5, 2010 at 8:19 pm | Permalink

        Not only did they attempt to breed docile foxes, they succeeded.

        http://www.sibfox.com

        Hey, buddy. Wanna buy a fox?

        Looks like I was wrong in that the silver coat didn’t disappear completely but it does look like a lot of variations crept into the gene pool.

  12. Posted July 5, 2010 at 6:05 pm | Permalink

    A good friend of mine did their PhD work on pop gen models addressing the feasibility of such sequential mutations, so I’ve thought a little about this question. In that work, even moderately deleterious mutations can sometimes persist in populations due to stochastic effects. So patchy populations might have a good go of it?

    So small population effects and weakly deleterious intermediates might get cryptic populations “over the hump.”

    Now I’ll talk off the top of my head, so no promises any of this holds water… 😉

    Non-constant selection pressure might also overcome frequency dependent fitness effects. This work seems to ignore predation changes and coevolution or learned behavioral adaptations? Ex: imagine a region with similar venomous and non-venomous snakes. Predators might learn/evolve to avoid all snakes if they can’t distinguish. This releases selection pressure to be cryptic. Variation may come in and provide enough to later allow them to distinguish some snakes as edible, and others dangerous, etc.

    Main point here: a temporary lack of strong selective pressure (esp. in smaller, variable populations) might be enough allow enough variation to get things over the hump.

    Last thought: what if these mutations are recessive or codominant (I’m thinking of the genetics of snake coloration and pattern here – most variants known in captivity are not dominant)? Does that sufficiently reduce fitness costs for heterozygotes?

  13. Posted July 5, 2010 at 6:06 pm | Permalink

    Perhaps this will be a case that shows that a mutation doesn’t have to be beneficial AT FIRST in order to become fixed in a population?

    My “rookie” understanding is that not all mutations have to be beneficial right away. But this is a lovely problem!

  14. Josef Uyeda
    Posted July 5, 2010 at 6:28 pm | Permalink

    Jerry, I think you’re thinking about this too much in a few-locus model rather than a quantitative genetic model. These are complex traits (certainly foraging behavior is, as is conspicuousness). There is ALWAYS standing genetic variation for complex traits influenced by many loci (large mutational target size, weak selection on any one locus). Consequently, the population will always have individuals with alleles that on average are more conspicuous and forage more brazenly. These will be selected for in this model, and can rise to frequency, without any simultaneous mutation, and without any physical linkage or pleiotropy between alleles.

    One thing I’ve noticed is that the warning coloration of aposematic amphibians is never for the species that they evolved the poison to defend against. For example, the Taricha newts and poison dart frogs presumably evolved toxicity in an arms race with snakes. Yet snakes don’t care much about coloration, and it seems they evolve the aposematic coloration only after they have reached a level of toxicity many times over stronger than what they would need to resist predation by the actual target their aposematic coloration (usually birds).

  15. Donovan
    Posted July 5, 2010 at 7:52 pm | Permalink

    I have to agree with many of the previous commenters. I think there is a big problem in trying to think why a frog turned from dusty green to bright yellow. Evolution can work in seemingly large leaps, but it most often happens in much smaller baby steps.

    Richard Dawkins was explaining the evolution of a stick bug while denouncing the idea of an arms race with specific predators. In Dawkins’s account, the stick bug does not have to evolve from something like a grasshopper to something that looks like a stick to have some success. It only needs to evolve to look more like a stick than the other bugs. I call this the “I don’t have to outrun the bear, I just have to outrun you” rule of evolution.

    If we take this same concept an apply it to any of the aposematic colorations, I think we’ll quickly see the same possible derivation. I think the foraging idea is silly for both what you, Jerry, pointed out, and because it would have needed to occur in species of highly differing feeding styles in the same way, each also confronted with the same problems you stated. But slight changes that offer only a tiny advantage explains all species coloration with no special pleading.

    1) A frog evolves to be toxic. 2) This frog has a distinctive camouflage pattern. 3) The more distinct the pattern, the more likely a predator is to recognize that frog as belonging to a toxic species. 4) Frogs with the most easily recognizable patterns become the dominant members of that toxic species.

    I would also like to point out that this would explain why the aposematic coloring, unlike the sexual coloring, is so often reminiscent of camouflage, sort of like the brightly colored camouflage things sold in trendy clothing stores, the clothes that are anything but “subtle.”

  16. Posted July 5, 2010 at 10:23 pm | Permalink

    I don’t quite follow your reasoning here:

    “The solution is this: although the first mutant aposematic individuals are more likely to be detected by predators, they have another feature that more than offsets this advantage: they forage more openly, enabling them to get resources that their cryptically colored (and presumably hidden) conspecifics don’t share.”

    I can understand foraging more openly being worth the downside of being noticed more readily. I can understand how being poisonous is generally useful, and how it can make being easy to spot worthwhile to begin with.

    What I don’t understand is, given that foraging openly is worth the trade-off in this case, why is being aposematic in the presence of ignorant predators beneficial? It seems like being more aggressive in foraging and keeping your existing camouflage would still be better than being easier to spot by predators who don’t yet know any better.

    Is there something about being aposematic that makes conspicuous foraging more effective than if you’re camouflaged, regardless of how predators approach you? It feels like I’m missing something really obvious here.

  17. M8R-qr4wb21
    Posted July 6, 2010 at 12:38 am | Permalink

    sexual reproduction allows two mutations for recessive traits to occur at two different times, persist in the genome for a while without expression, and then occasionally show up later combined in the same organism. it allows evolution to descend into just the kind of multi-dimensional minima you’re talking about. it sounds like you’re imagining a bacterial model of evolution maybe, that can only descend into minima at…sortof…right angles? one dimension at a time?

    evolution spends a good bit of its power evolving the tools of evolution itself: ex., development (the growing of new organisms, the D in MCDB) is structured in a way that makes the genetic code more evolveable, NOT in the way that produces a replica of the parent most cheaply. and the positions of genes on chromosomes—whether they’re on the same chromosome or different ones, and how far apart they are if on the same one—controls how likely they are to be inherited together, so maybe if some marking genes and etiological genes like risk-taking ones or circadian-activity-level ones are close together on a chromosome then this argument would get stronger?

    there is also genetic drift: the predators could have disappeared for a while, then come back.

    i’m not an expert, but it seems there are many ways that the genetic computation does not have to happen in a single instant.

    there are also a lot of stupid “just so” stories floating around related to evolution with no basis in reality other than that they sound plausible, which it sounds like is exactly what you’re looking for, so…watch out, if you are around any molecular biologists who do not go for that kind of cheap story.

  18. ChrisZ
    Posted July 6, 2010 at 1:28 am | Permalink

    I was going to come and suggest what others already have, namely that the predators’ aversion to the species came before the markings, but thinking about it more I wonder if that doesn’t create a new problem, the “poisonous” or “icky-tasting” problem. Evolving that trait would not be useful to an individual unless it’s predators already knew they were not worth preying on. Until then they’re just wasting resources creating the poison or whatever. I suppose these could have happened accidentally, but is that the hypothesis for all such defense mechanisms?

    • Posted July 6, 2010 at 6:14 am | Permalink

      I can imagine scenarios in which being poisonous is beneficial without being aposematic:
      – Kin selection, in cases of lots of offspring so that one bad taste will save many copies of the gene.
      – If other predators are already capable of recognizing the species even when they’re camouflaged, the poison would be useful against them.

      A lot of the “which came first” problems seem to go away with this one if you think of a large pool of predators with a range of abilities to detect you.

    • Donovan
      Posted July 6, 2010 at 7:14 pm | Permalink

      The poison could easily evolve for smaller predators. A small bird trying to peck at a frog notices a sick feeling shortly into the attack. This poison would strengthen the more beneficial it showed itself to be. The same goes for foul tasting creatures or creatures with odor attacks. It starts with only a small advantage, but enough that over generations (a few years for most of these animals) it proves to be just that little bit more successful than its cousins.

  19. origin
    Posted July 6, 2010 at 3:05 am | Permalink

    “one example is the evolution of compensatory mutations in the loop structure of transfer RNA.”

    Can anyone recommend any literature covering this topic?

  20. Dan L.
    Posted July 6, 2010 at 8:11 am | Permalink

    I don’t know enough about population genetics to create a model myself, but I was wondering whether aposematism could develop under similar conditions that have caused so many male birds, for example, to evolve bright, attention-getting plumage. I’m imagining a situation in which a camouflaged species has developed a mutation in a recessive, maybe sex-linked gene that causes some low percentage to be born with defective camouflage (assuming the species is already toxic here).

    The individuals with defective camouflage will tend to get eaten, keeping their incidence in the population low, but if the trait is recessive, it could still remain in the population long enough for predators to develop a distaste for the toxic creatures with bad camouflage. Once the predators have learned to avoid the individuals with bad camouflage, it becomes advantageous rather than disadvantageous to have the recessive gene that breaks the camouflage, and so the trait becomes expressed with greater frequency throughout the population.

  21. W Benson
    Posted July 6, 2010 at 9:18 am | Permalink

    Interesting post on the Speed, Brockhurst, Ruxton study.
    However, I don´t think we need be so pessimistic. There seem to be several good ways to get warning coloration ‘off the ground.’ My favorites follow.
    Nearly all animals have small peculiar markings that can be used by interested parties to distinguish them from similar appearing species. This is what field guides are all about. A distasteful animal with distinctive markings will, if the predator attacking it is smart enough, educate that predator against attacking, in the future, other similar-looking prey. However, if the next distasteful prey individual encountered by the predator has less developed identifying features, it will be less likely to be recognized and deter an attack. On the other hand, if the next prey has a slightly enhanced warning pattern, it will tend not only be easier to recognize, but also easier to recognize at a distance, before the predator launches its attack. After all, a predator learns to really pay attention the bad-tasting prey when it is really close, like in his mouth, and a warning pattern learned from that perspective might be remembered as really BIG. Experienced predators may therefore recognize and thus avoid to a greater degree distasteful prey with patterns that are enhanced relative to those actually sampled. (Actually, it is not necessary for more conspicuous variants to receive greater protection, only that they have lower attack rates than do forms that are less conspicuous that the population average). No one to my knowledge has collected data on this. Ronald A. Fisher long ago proposed that warning colors might be favored because they reduce confusion with palatable cryptic species. I would bet that a reasonably decent field experiment would show that slightly enhanced aposematic markings are better deterrents than average ones.
    There are other factors that seem to aid the evolution of aposematism. One is that the cryptic distasteful precursor is abundant enough (or easily detected because of movement) that essentially all predators encounter them occasionally and have learned to avoid them. Predator attack decisions will depend largely on a predator’s ability to recall correctly what distasteful prey eaten in the past look like. If a pattern enhances predator recall by being a slightly more colorful (and probably more distinctive: the two are difficult to separate) version of the original, warning coloration will tend to evolve.
    I agree that it is a fallacy to argue that a trait should be favored because the end-state is advantageous. Finding a niche advantage is not by itself evidence that the advantage can become a cause. A moment’s thought will show that aposematic species, no matter how warning coloration first arose, will tend to evolve to be large, plump, and fecund, and to search with bold efficiency for food and host plants, for being awkward and unwary imposes little cost to a species with almost no risk of predation. (Moreover, if predators tend to spit out distasteful prey, it pays to be robust enough to survive the ordeal; aposematic prey are often ‘rubbery’). Thus the facts surrounding behavior and resource use predicted by Speed’s model seem to be well explained by simpler considerations. A similar case can be made regarding cryptic pupae. The fourth prediction, that aposematism should promote more generalized diets, does not mesh with my understanding.
    Finally, there may be a simple way to test between models based on the assumption of major color pattern shifts and those premissing that warning patterns build gradually on continuous variation. The continuous variation model predicts that warning patterns recently evolved by cryptic ancestors (and do not involve Mullerian mimicry, for mimicry can select color-pattern mutations of large effects if these occur) will be under polygenic control. The new model assums control by ‘major’ genes. Kin selection (KS) is another mechanism that may favor fixation of major color changes when prey offspring stay within the home-ranges of individual predators. With KS prey aggregation helps, but it is not necessary. And it is trivially true that some animals, like big butterflies, that are always conspicuous as they move around, may have practically no costs to becoming colorful, although there may be costs for not being distinctive. In my view the Monarch butterfly, using objective criteria, may not be aposematic (i.e., has coloration that makes it more conspicuous in the field than it would be if less colorful, or colored like a palatable species, such as the tiger swallowtail). It is unsafe to assume that warning coloration evolves through one universal mechanism, thereby forcing all cases into the same ecological mold.
    It is certainly an interesting bone to gnaw on. Too bad there are not more good tests between alternatives.

  22. TheBlackCat
    Posted July 6, 2010 at 9:21 am | Permalink

    What about the possibility that it evolved first as startle strategy, perhaps even in a color-changing organism like a cephalopods that also happened to be poisonous?

    Also, why aren’t there more mimics? Wouldn’t it be highly advantageous to piggyback off the instinctive avoidance of bright colors and patterns?

  23. Torbjörn Larsson, OM
    Posted July 6, 2010 at 11:23 am | Permalink

    How about selective sweeps? AFAIU they are able to link otherwise uncorrelated (or correlated) genes over what seems to me large clusters of them. Wouldn’t once in a while several alleles get over “a non-advantageous hump” by riding on something else?

    Now I see that Josef Uyeda notes that some of these traits (foraging behavior, coloration) may be complex gene wise. This could be a problem for sweeps, at a guess. But the massive predictiveness of Speed et al model means there is something there. If it isn’t a population genetic mechanism, what would it be?

    Not offering an alternative model for actual predictions (if they bear up) is a weak criticism at best, most often non-existent in practice.

    [Btw, I’m a bit surprised that specific gene-for-gene pathways needs to be demonstrated, as I thought there were a massive freedom in these mechanisms. But I can maybe see that “a non-advantageous hump”, if “large”, can be a slight problem for the general selection mechanism.]

    • Torbjörn Larsson, OM
      Posted July 6, 2010 at 11:30 am | Permalink

      “Not offering an alternative model for actual predictions (if they bear up) is a weak criticism at best, most often non-existent in practice.”

      Oops, forgot that is of course only if they aren’t ad hoc vs the theory of the area; otherwise they are pattern matching which will always have results in the same way that one can always fit an algorithm to any finite sequence of stochastic numbers. Since optimization models aren’t germane here, the observation isn’t.

      Still, if they are _relatively_ suitable, it is an intriguing theory (observation) that could do with an explanation of its own.

  24. Posted July 6, 2010 at 1:22 pm | Permalink

    Someone might have already said this- might a lull in predation (maybe there’s only one major predator on a population and its population is temporarily diminished for some reason) allow both disadvantageous mutations to flourish, boosting the chance that they’ll occur together by inheritance? Would this kind of scenario make a difference?

  25. IanW
    Posted July 7, 2010 at 4:19 am | Permalink

    “The poisonous coral snake”

    I don’t think the coral snake is poisonous; it is venomous. I suspect you could eat it without being poisoned – but of course you’d have to catch it first….

  26. Delusional
    Posted July 15, 2010 at 7:02 pm | Permalink

    Mayhaps sexual selection was involved with at least one of these poisonous, brightly-colored creatures? The male peacock is colored the way it is to communicate that even despite its flashy coloring, it can still evade predation long enough to mate. Maybe one of these creatures started in a similar way.

  27. bill molloy
    Posted June 14, 2011 at 5:23 am | Permalink

    The original study by Speed et al. appears to have assumed that an increase of open foraging in animals occurred simultaneously with the development of warning colouration. Perhaps it’s improbable, but maybe not impossible. The feeding advantage would then perhaps eventually outweigh the depradation caused by the conspicuous markings, and the depradation would in time become less as the predators learned to avoid this type of prey (or the number of predators dropped as a result of eating toxic prey). The point at issue is the simultaneous development of warning coloration and open foraging. To me as a non-scientist this does seem a difficult assumption. My own theory, for what it is worth, like some others who have posted comments is that potential prey survived in sufficient numbers to become established, because predators after one bite got sufficient toxin to put them off leaving the prey to recover and survive.


One Trackback/Pingback

  1. […] Jerry Coyne has an excellent article about species which are toxic (when touched or eaten) and are very brightly colore…. There are some cool photos and a discussion of the problem. Basically it is this: Assumptions: The […]

%d bloggers like this: