Why do zebras have stripes? II. It’s the flies, stupid!

February 3, 2016 • 10:30 am

Referring to the above, it may be flies; we still don’t know for sure. What I wrote above was clickbait inspired by James Carville.

Four years ago (has it really been that long?), I reported about a paper in the Journal of Experimental Biology showing that, on Hungarian horse farms, striped patterns are repugnant to horseflies (tabanids), for the flies prefer to land on solid colors than on stripes. That in turn suggested that maybe zebras are striped to reduce the number of fly bites they get, bites that can not only produce a serious loss of blood, but also spread disease (see below). There were, however, problems with that paper, and so the results were very tentative.

A few days ago I reported on a paper in PLoS ONE testing an alternative theory for zebra striping: the popular idea the pattern breaks up the outline of the animals, making them less visible to predators like lions and hyenas. That theory wasn’t supported by the data, but other theories for stripe evolution remain, including confusing predators in a herd by presenting them with a mass of strange patterns, “warning coloration” (a pattern that tells animals like hyenas to “stay away because I can bite and kick you”), thermoregulation, recognition of individuals of your herd or members of your species, and, of course, reduction of fly bites.

A 2014 paper by Tim Caro and colleagues in Nature Communications seems to eliminate most explanations in favor of the “fly hypothesis”.  The study I mentioned earlier, and two that have also appeared, give experimental evidnce that both horseflies and tsetse flies are averse to landing on striped surfaces. The Caro et al. paper doesn’t give direct experimental evidence, but supportive correlational evidence.

What the authors did was examine the historical Eurasian ranges of zebra species and subspecies, as well as those of other equids, and then match those up with the ranges of horseflies, tsetse flies, temperature (for the thermoregulation hypothesis) and the historical ranges of predators (lions, hyenas, tigers, and wolves). They were looking for a selective factor whose historical ranges would correspond to the ranges of the zebra before humans changed range sizes. (Stripes evolved before humans changed the face of the planet.) The factor Caro et al. looked at was not the ranges of the species themselves, but of the ranges of striping on various parts of the equid body, for it’s the stripes themselves that are thought to result from selection.

The major results were these:

a. The striping patterns of zebra subspecies and species correspond more closely by far to the ranges and climate preferences of tabanid and tsetse flies than to any other factor, although lion ranges are also associated with a few measures of striping, like leg stripes.

Here’s the association between the historical (not present!) ranges of equids and of tabanids and tsetse flies; equids at top (zebra ranges striped!) and flies at bottom. Note that tsetse flies (Glossina) aren’t found outside Africa. E. kiang is an unstriped wild ass, E. africanus is the African wild ass, having thin stripes on its legs, E. hemionus is the onager, an unstriped wild ass, and E. ferus przewalskii is Prezewalski’s horse, a rare wild horse thought to be the closest living relative of the domestic horse.


The correspondence is pretty good, although not perfect, since flies live in some areas where zebras don’t. The crucial observation, though, is that biting flies always occurred in areas where zebras lived.

Note, too, that unstriped equids don’t generally coexist with either kind of fly, though the African wild ass, which does have thin striping on its legs, does live in areas with horseflies.

Here’s another figure showing the degree of association of leg striping among the equid species with activity of tabanids in the species range. The outer circles show the intensity of tabanid fly activity (see key), the next row in gives the intensity of leg striping, and the lines show the evolutionary relationship of the species. The quagga is extinct, but is part of the same clade as zebras, showing that full body striping evolved only once (this means that the correlation between striping of the species and presence of flies may not reflect independent evolution in each group, which is a problem for the authors’ conclusion, though one they admit). But the key observation here is that stripe intensity is highest in species that experience more tabanid activity.



b. No other factor supposedly associated with striping, including group size, thermal highs, or presence of predators, was as consistently associated with striping as was the presence of flies. The authors thus rule out the species and individual recognition hypotheses, the thermoregulation hypothesis, and the predator confusion or camouflage hypotheses as forces promoting the evolution of stripes. The predator hypothesis was also largely ruled out by the paper I posted about last week (Caro was also an author of that one). This points to the fly hypothesis as the most viable one. But that raises an immediate question:

c. Can flies really be a significant selective factor in the evolution of striping? Apparently yes. The authors note that tabanids can take significant amounts of blood from horses and cows, and that flies can also carry diseases that kill equids:

At an ultimate level, blood loss from biting flies can be considerable. Calculations show that blood loss from tabanids alone can reach 200–500 cc per cow per day in the United States. For example, in Pennsylvania, mean weight gain per cow was 37.2 lbs lower over an 8-week period in the absence of insecticide that prevented horn fly (Siphona), stable fly (Stomoxys) and horse fly (Tabanus) attack, and in New Jersey, milk production increased by 35.5 lbs over a 5-week period with insecticide. Milk loss to stable flies was calculated at 139 kg per cow per annum in the United States. Similarly, blood-sucking insects have been shown to negatively affect performance in draft horses.

That’s a lot of blood and a lot of milk loss (which could, of course, reduce offspring survival). But the authors favor the disease hypothesis, although there’s no reason that both blood loss and disease could act as joint selective factors:

Alternatively or additionally, striped equids might be particularly susceptible to certain diseases that are carried by fly vectors in sub-Saharan Africa. We collated literature on diseases carried by biting flies that attack equids in Africa (Supplementary Table 1) and note that equine influenza, African horse sickness, equine infectious anaemia, and trypanosomiasis and are restricted to equids, all are fatal and all are carried by tabanids. Currently, we are unable to distinguish whether zebras are particularly vulnerable or susceptible to biting flies because they carry dangerous diseases or because of excessive blood loss, but we are inclined towards the former because Eurasian equids are not striped, yet demonstrably subject to biting fly annoyance.

Here are those nasty flies, with a blood-engorged tsetse fly at the top and a tabanid below (a video of a biting tabanid is here):



d. Stripe width in zebras, especially the thin stripes on the legs, is small enough to deter flies. The graph below shoes the limits of stripe width in zebras (colored vertical lines) versus the preference of three kinds of flies for difference stripe width. Flies like to bite less when the striped pattern is thinner, and zebras are all in the stripe range that flies don’t like. Note that the thinnest stripes are on the face and legs, and that flies like to bite on the animal’s legs, as it’s shaded underneath the belly. (Remember that the African wild ass has thin stripes on the legs.):


This is getting long, so one more point:

e. Why, among the many African grazing mammals, are only the zebras so stripey? After all, other species of horses and asses, as well as many antelopes and artiodactyls, also have to deal with tabanids too, but only zebras have stripes. The answer may have to do with hair length and density. The authors show that among grazing animals, zebras have the shortest hairs and smallest hair depth, and that enables the flies to bite more easily through the coat to the flesh. That could mean either that the zebras evolved stripes because their short coats exposed them to more biting, or they evolved their short coats because it’s advantageous to have such short coats for other reasons (perhaps thermoregulation?), and the earlier evolution of zebra stripes sufficiently deterred flies to enable the evolution of shorter coats.

That’s the upshot. I think the fly-bite hypothesis is a good one, and now has both experimental and correlational evidence to support it, but there may be other advantages to being striped, like helping you find other zebras when you’re lost (granted, the authors found no correlation between stripiness and zebra group size). One crucial experiment, which can’t be done, is to release equids dyed with stripes into zebra territory, along with unstriped controls, and see if the former suffer fewer bites. Or dye zebras gray and see if they get bitten more often. For the time being, though, we’ll just have to say that we’re approaching the explanation for zebra striping slowly and asymptotically.

h/t: Diane Morgan

Caro, T. A. Izzo, R. C. Reiner, Jr., H. Walker, and T. Stankowich. The function of zebra stripes. Nature Communications, doi:10.1038/ncomms4535.

65 thoughts on “Why do zebras have stripes? II. It’s the flies, stupid!

    1. We can get striped cattle AND avoid the PR problems of GMO by simply putting some striped sticks in sight of where the cattle mate. At least, according to the Bible…

  1. Greetings,

    Apart from blood-loss and disease, I would have thought a related factor would be that the scent of blood on the wind would attract predators to “weakened” prey – thus a more pertinent form of natural selection would be at work.

    Kindest regards,


    1. Yes, this all begs the question of whether there is a selection pressure that discourages flies from landing on anything stripey.

    2. Presumably, the flies can more easily simply dine on unstriped animals. If there were a paucity of alternatives, there would be pressure to evolve. I suspect the flies reluctance with respect to stripes has to do with a fairly rudimentary fear of being trapped between objects.

    3. Good question. So if the idea is right, there is a degree of selective pressure for the flies to counter it. But they have other hosts so the selection pressure may not be that strong. Also if the mutations don’t happen then the evolution won’t happen. Traits do not always evolve just because there is selection to do so.

      1. “…Also if the mutations don’t happen then the evolution won’t happen.”

        That was the first thing that occurred to me.

      2. Good point. …but the flies probably have a short generation compared to the zebras, and greater numbers of individuals (I’m just supposing that is likely) so good chances for mutations to arise?

        1. Good points!

          I wonder if anyone’s studied whether Oxpeckers have color/pattern preferences? (Maybe it’s easier to follow a bug on a solid-color surface, e.g.)

  2. One experiment we could do is to get supporters of (say) Newcastle, West Brom, Sunderland and Birmingham to go to the appropriate bits of Africa and see how much each gets bitten.

  3. Do we know the mechanism by which stripes deter flies? I thought horse flies find their targets by body warmth, carbon dioxide, and movement. Perhaps they use visual cues. It would be ironic if stripes are camouflage against flies but not against mammal predators.

    1. Horses have a well-developed cervical panniculus, or subcutaneous striated muscle layer over the neck and withers (human equivalent is the platysma muscle), which they can twitch extensively to dislodge flies and other biting insects. I wonder whether thin stripes visually mimic the ripple effect created in the overlying skin, when a horse twitches its panniculus? Fly thinks, “Surface is already moving, I’m not going to land on it.”

      Stripes would be of most value as fly deterrents on the lower legs, since that area is difficult to reach with a swishing tail (yours or a friend’s) or mane, and isn’t covered by a panniculus. Horses have to repeatedly stamp their feet to dislodge flies from lower legs.

      1. Good answer, like your idea.

        One generally shoo a fly with movement, so the stripes adds further to the appearance of motion- note that the stripes are lined vertically on the body in relation to forward motion, and horizontally on the legs to enhance the stomping motion.

    2. Probably because too many flies were wasting time debating where to dine with conversations along these lines:

      Darth: Come over to the dark side.
      Luke: Nah. White meat is healthier.

    3. Hypotheses: (1) the edge of a pigment stripe looks (to a fly) like the edge of a solid body, and (2) a fly aiming to land on a sold body that smells like food will not head for its edge because there’s a good chance of missing altogether, especially if it moves.

      Presumably flies follow rules of thumb like (example from one of Dawkins’ books) a moth navigating by distant stars but fooled by a candle, and simple rules may work well enough for them (in a world of diverse and abundant megafauna) to be beaten by a simple kluge like striping. The experiments to determine the 3-D capabilities of fly-eyes and the algorithms for their flight trajectories are left as an exercise for the reader.

  4. With the new (well, 2014) data it looks to me that theory may fly.

    other species of horses and asses, as well as many antelopes and other artiodactyls, also have to deal with tabanids too, but only zebras have stripes.

    Doesn’t okapi count as partially striped? [ https://en.wikipedia.org/wiki/Giraffidae ]

  5. In the spirit of devil’s advocacy, the correlation could also be explained as: (1) these species first evolve stripes for some unknown reason, then (2) the stripes create some positive environmental factor for the flies, so that (3) over a few generations the flies’ range changes to match the range of the striped equines’.

    There’s also the possibility that unknown factor X worked simultaneously on fly and equine evolution, producing both successful fly populations and equine stripes in the same geographical areas but with no real relationship from one to the other.

    I don’t really believe either of those possibilities. The difference in hypotheses could be tested by seeing what sort of animals the flies drink from most. My examples just highlight problem of correlation not being causation.

    1. I think the stripes are interesting, but does there have to be a reason. They are different looking. Physically interesting, maybe for mating or distinguishability.

      I can say for sure that facial hair, for example, might have been important to some fellow long ago, but I sure could do without it. There’s an example of evolution getting something wrong; from my perspective, of course.

      1. I talked about nonadaptive explanations in my last article. And yes, stripes could, as I noted, be a “spandrel” or nonadaptive byproduct, but my hunch is that they are so unusual that they invoke an adaptive explanation.

  6. I have a couple of supporting bits of anecdote from the cattle industry. in light-skinned cattle such as Charolais, skin pigmentation around the eyes is correlated with a reduction of pinkeye (infectious conjunctivitis)which can lead, even if promptly treated, to loss of visual acuity or blindness. Not a good thing for a prey species. I also note that a number of species, both predator and prey, mammals and birds, have dark lines through the eye.
    Tabanid fly bites are quite painful, as anyone who has been bitten by a horsefly can attest. Cattle alter their behaviour to avoid the flies, lounging in shady bush rather than grazing. Bulls in particular are targeted by the flies, probably by scent cues, and in particular their scrotums ( thin skin and lack of long hair, for thermoregulation purposes). This effect is large enough that tabanid traps are designed to mimic the appearance. Bulls are so affected by this that high tabanid populations lead to lower conception rates, as bulls seek refuge from flies rather than receptive cows.
    Hair length and structure have substantial effects on insect bites, I have seen young calves whose eyes are almost puffed shut from the swelling from mosquito bites, and so much inflammatory response that whorls and parts in their hair coat begin shedding hair. I wonder if that is why horses have manes, to protect the backline where a natural part occurs in the hair coat?

      1. I can only imagine. Deer fly bites are bad enough. Which makes one wonder how such painful bites could be adaptive.

        1. Pain is adaptive to the extent that it motivates critters to avoid something harmful. It wouldn’t be adaptive to the flies, however, to cause pain. They’re just in it for a chance of flesh to chow down on. The victim’s discomfort is not a concern.

          Kind of like Republican politicians, now that I think of it.

          1. I hope you weren’t thinking that I was referring to the bitee. 😀

            “Kind of like Republican politicians, now that I think of it.”

            LOL. A meme that needs spreading!)

            (As you probably know, many biters do have essentially unfelt bites…blackflies, fleas, even vampire bats, IIRC.)

  7. Concerning point g) first sentence (in bold), there is a typo in “arTiodactyls” (the T is missing). However, you may consider changing the term for “large herbivores” as zebras are perissodactyls.

    Just a minor suggestion, I really appreciated your text.

  8. and E. ferus przewalskii is Prezewalski’s horse, a rare wild horse thought to be the closest living relative of the domestic horse.

    I would have thought that was a well settled question by now?
    Seems that there is some work on this. Hmmm, but that (fig 1 B) implies that E.caballus is less derived (genetically) then E.przewalskii, which is not the normal tale. A bottleneck effect?

  9. The fly hypothesis is looking pretty good. It is interesting that the Quagga was striped in the anterior area, which is the area that it could not defend against biting flies.

  10. Present ranges of tsetse and stripey equids.

    It’s interesting that Glossina once occurred in the Americas — beautifully preserved fossils tsetses are known from the end-of-Eocene Florissant beds in Colorado. Equids were still small browsers, but perhaps museum dioramas should be repainted with pajama-striped brontotheres and entelodonts??

    1. I have a fossil slab from the Green River formation that has numerous large fly larvae. I bought it from a dealer and it was labelled to be Tsetse fly larvae. I thought it was pretty cool to think that they were over here once. More recently I was wondering if they were really horsefly larvae. But now, not sure…

      1. Very likely to be tabanid larvae. Glossina [living species, certainly] is ‘pupiparous’, giving birth to mature larvae that immediately pupate inside the larval skin [puparium, characteristic of the higher flies].

        Glossina are pretty choosy in dropping their bundles of joy, so almost no chance an isolated larva would get into the lacustrine sediments of the Florissant.

        Tabanid larvae ARE characteristically aquatic or semiaquatic, active predators that are abundant in shallow marshy situations with lots of leaffall, so much like the probable Florissant sites..

        1. Of course! And I knew that! Tabanids are more likely, then. I had been thinking of sending pix of my fossil and bone collection to Jerry, and this would be included.

  11. That theory [camouflage] wasn’t supported by the data

    Unless I misunderstood the earlier paper, it wasn’t about actual experimental data; it was about inferences drawn from a mathematical model of predator vision.

  12. Couldn’t ambitious experimenters simply spend time among the biting flies themselves, with and without striped clothing, then count the number of bites they get?

    Obviously with whatever vaccines are available first, and very regular medical attention to deal with the nastiness that such fly bites can result in without treatment.

  13. I can’t disagree with the paper, but I could not help thinking while reading it, maybe it’s just because they like stripes.

  14. An alternative theory (which is mine):
    Could it possibly be a “spandrel”?

    For example, perhaps zebras that were all white or all black were easier to spot by predators than some mix of the two colors, or all-black ones grew too hot and thirsty in the summer, or something like that… But the particular shape of the colored areas wasn’t really critical as long as they were about evenly split.

    In zebras’ case, then, this theory which is mine goes, evolution just happened to stumble on the “stripes” divot on the evolutionary potential surface first (perhaps only a really simple genetic deviation was needed to repurpose existing pigment genes with just a little bit of jiggering to make them produce stripes) and since the striped offspring did better than the all-one-color phenotype, this change was more likely to survive to be passed on…

    (Hmm, to establish this I might also have to demonstrate that this happened recently enough that the lucky stripes pattern genes haven’t had much time to drift into spots or blobs that would work just as well…)

    1. Reading more carefully, I see that the first part of my theory was specifically investigated and rejected. These guys are smart.

  15. The distribution maps are described as historical distributions but when exactly in the past do they refer to? The range of E ferus certainly extended far across to the west in the past, as indicated by cave paintings in France and Spain. Flies of the genus Tabanus and related tabanids occur (and presumably occurred)much further north in Europe than indicated e.g. T sudeticus has been recorded in northern Scotland. This seems to me to contradict the assertion that non-striped equids don’t co-exit with either Tabanus or Glossina.

  16. seems my post never made it … some ignorant questions:

    wondering if the flies are always “in season”, or if they feast only during a spawning period?

    also wondering how the lifespan of an organism matters when discussing an evolutionary process – for instance, I was thinking about how the flies probably live for a year, but the zebra lives for maybe 10 – but then wondered if that’s irrelevant when considering evolutionary time scales.

Leave a Reply