Once again: Why do zebras have stripes? (And once again the answer: it’s the flies, stupid!)

February 22, 2019 • 11:15 am

Over the past three years I’ve written about research, mostly by Tim Caro’s group, dealing with the perpetual question: “Why do zebras have stripes?” (See earlier posts here, here, and here.) There used to be lots of hypotheses, including confusing predators like lions, aiding thermoregulation, camouflaging these equids, keeping groups together, and so on, but research has demolished these hypotheses one by one (for example, striped animals are no more camouflaged to predators than are solid-colored animals.)

Rather, evidence has been accumulating that stripes help deter biting flies, particularly tabanid flies (horseflies) and tsetse flies that not only draw blood but, more dangerously, carry disease that can kill zebras. It’s been shown, for instance, that the ranges of striped equids all fall within the ranges of biting flies, and that these flies can carry diseases like equine influenza, African horse sickness, equine infectious anaemia, and trypanosomiasis, all fatal. Other evidence is that tabanids appear averse to landing on striped patterns, as well as the evidence against the alternative hypotheses mentioned above.

Now, however, we have more direct evidence from Caro and his colleagues in a paper published in PLoS ONE. Click on the screenshot below, or get a free pdf here (the reference is at bottom).

What the authors did here was look at the behavior of horseflies around zebras and different-colored horses, all in a pasture in Somerset in the UK. They also dressed up the horses in zebra suits as well as in monochrome white or black suits, and watched the behavior of tabanids around the equids and the behavioral responses of zebras and horses to flies. Here’s a horse in a zebra suit from the experiment:

From USA Today

There were a number of sub-studies, and I’ll give the results briefly. The conclusion that tabanids are averse to crawling on striped patterns seems pretty solid.

Direct observations of flies. These were done by observers watching equids and flies directly, and also by video that could follow the trajectory of flies as they approached zebras and horses.

The non-video observations showed that there was no significant difference in the rates at which horseflies circled zebras versus horses, or touched them briefly, but significantly fewer flies landed on zebras than on horses. The probability of this occurring by chance, while significant, was not very impressive (p = 0.041). This could, however, reflect a differential behavior of the animals, for horses respond to flies by twitching their skin while zebras more often switched their tails (and ran away). Perhaps skin-twitching helps deter landing.

The videos showed that the trajectories of flies approaching zebras differed from that approaching horses: flies approached the horses more slowly, and decelerated very little while approaching zebras, so that they simply bumped into the zebras rather than landing on them. This difference in the proportion of approaching flies that landed was highly significant (p < 0.0001). Further, flies flew away from zebras significantly faster than from horses.

Further, once landed, a tabanid spent considerably more time on the horse than on the zebra: 10.1 versus 1.2 seconds respectively. This again is significant at the 0.01 level, but it’s not clear whether this reflected differential behavior of the equids toward flies (e.g., skin twitching versus tail switching).

Experiment with colored coats on horses. This is the most convincing part of the results. They used cloths of three types, as shown in the diagram, and you can see that the rate of horseflies landing on the cloths or touching the clothes was significantly lower (p < 0.0001) for the striped coat versus the white or black coat, with the latter two not differing from each other. In contrast, there was no difference in the rate of flies touching or landing on the horse’s heads, which were not covered by the cloth. That shows that the pattern is probably what’s important. (Sample sizes are given in the figure; they are not large.)

The upshot: Stripes don’t appear to deter tabanids at a distance, but, once close to the horse or zebra, appear to deter landing. The authors say that tabanids find equids at a distance using odor rather than visual ues, so this is understandable.

In the case of the horse-coat experiment and of the comparison of horses versus zebras, tabanids don’t land as often on stripes as on solid colors, and, once landed, they spend less time on zebras then on horses. The flies also do not decelerate so much when approaching zebras, perhaps because they don’t see a striped substrate as an appropriate place to land.

The conclusion is that if you have stripes, you’re more likely to deter flies from landing or, if they land, less likely to get bitten and thus to get any diseases carried by the flies. As the authors note,

In summary, multiple lines of evidence indicate that stripes prevent effective landing by tabanids once they are in the vicinity of the host but did not prevent them approaching from a distance. In addition, zebras appear to use behavioural means to prevent tabanids spending time on them through constant tail swishing and even running away. As a consequence of both of these morphological and behavioural defenses, very few tabanids are able to probe for a zebra blood meal as evidenced by our data.

Now we still don’t really know what it is about stripes that deter flies or drive them away. Perhaps their visual system gets confused or flummoxed. But the coincidence of the ranges of striped equids and biting flies (see here) is pretty remarkable, and supports this experiment in suggesting that stripes evolved in equids to protect them from the bites of disease-carrying flies. I’ll put up those range maps again, which are historical rather than present-day ranges (after all, selection for stripes operated in the past), and here’s what I wrote about them:

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.


Caro, T., Y. Argueta, E. S. Briolat, J. Bruggink, M. Kasprowsky, J. Lake, M. J. Mitchell, S. Richardson, and M. How. 2019. Benefits of zebra stripes: Behaviour of tabanid flies around zebras and horses. PLOS ONE 14:e0210831.


43 thoughts on “Once again: Why do zebras have stripes? (And once again the answer: it’s the flies, stupid!)

      1. The distribution maps for Tabanus seem much too restrictive to me. Various Tabanus and related tabanids occur over much of Europe and including the UK (where the experiments were carried out with the horses in zebra suits). I recognise that the maps are described as historical distributions but is there evidence that Tabanus previously had a more restricted range than now? przewalskii-like horses were of course present across western Europe 17,000 years ago as evidenced by the Lascaux and other cave paintings. I’d guess that there were blood sucking flies too. Of course the abundance of the flies and the intensity of their attacks might vary across the range and be higher in the tropics and the presence/absence of stripes on the different species could perhaps be due not simply to the presence/absence of biting flies but to the balance between the biting fly selection pressure and some other pressure selecting for unstriped coats???

  1. Biting flies must be a really huge problem for zebras for them to essentially develop targets for predators. I look forward to further research on what why the flies avoid stripes. I wonder if people can avoid being bitten by mosquitoes by wearing a particular pattern.

  2. Just a small sample:

    “…two rather stereotyped behaviors have been characterized: Drosophila’s landing response elicited during flight when confronted with an expanding stimulus in front of the animal (for example [119]), and its escape jump elicited by a sudden light-off stimulus.

    While the neural circuit underlying landing is yet to be discovered, the escape response is known to involve just a few neurons. Key players are the giant fibers — large descending neurons which receive visual input and connect to the tergo-trochanter motor neurons in the thorax via electrical synapses. Once activated, these motor neurons initiate the take-off by stimulating an extension of the mesothoracic legs. In addition, three pairs of wing muscles become activated, swinging the wings out to their flight position, and start flapping. The sufficiency of giant fiber activity for eliciting an escape jump and flight initiation has been demonstrated most elegantly by selective photoactivation of the giant fibers”

    from this intriguing paper on Insect Vision by Alexander Borst. It’s way above my head, but I take from it that an insect’s landing algorithm depends heavily on very local visual field cues of the landing area – an expanding pattern indicates to the insect that it should adjust wing flapping & undercarriage deployment in preparation for encountering the surface. It’s all pre-programmed auto-pilot stuff.

    By the same token the escape response depends on part of the visual field going dark & the ‘escape’ flight computer kicks into high gear. I can imagine that dark stripes crossing the retina while landing evokes two opposing programs at once – [1] The carrier pilot watching the deck expand below notices [2] a flicker out of the corner of the eye & initiates a ‘go around’

    1. Your speculation for a ‘go around’ sounds reasonable. Pilots are trained to watch for any anomalous appearance on or near the runway that could trigger a ‘go around’.

  3. The figure in the paper: “Behaviour of tabanid flies around zebras and horses” showing the flight paths of flies seems particularly persuasive.

  4. I think this paper comes close to clinching it. Of course, there is a lot more to be investigated, eg. how much tabanid and glossina flies carry parasites dangerous to equids in different areas. And whether these parasites are nefarious to other ungulates. And giraffes also have a kind of pattern that might make proper landing difficult.
    I really like their hypothesising about why these flies do not (or cannot) properly land on striped surfaces.
    Great article.

      1. I have some doubts there, the same was said about the zebra stripes. Giraffes always stick out (no pun), in my modest experience. If it is camouflage it is extraordinarily inefficient, meseems, at least to trichromates like us.
        I’d like to see some studies about flies landing (or not) on those net-shaped surfaces.

        1. I do not agree. Camo doesn’t need to be effective all the time to confer an advantage, but when the giraffe stands in front [or among] trees the light and dark camo blends well with dappled light.

          A bit like late-war German ‘disc’ camo designed for hiding vehicles in front of sunlit trees/bushes:

          1. Yes. Like Zebra’s standing among the reeds 🙂
            I still would like a study such as this one with giraffe patterns.

          2. My point is that giraffes are easily seen among the acacia trees. Dappled light or not, they always stick out. I’m skeptical there, in other words.

  5. Is there a reason it is stripes, and not spots or other shapes? The proposed explanations do not seem to explain the shape of the markings, only the presence of alternating colors, and perhaps the contrast from black and white versus shades of brown. Could we have had polka dot zebras?

    I am curious how expensive it is for zebras to develop stripes. It seems to me that coloration patterns are pretty easy for animals to develop, based on looking around, but I really don’t know that.

    1. “is there a reason it is stripes, and not spots or other shapes? The proposed explanations do not seem to explain the shape of the markings, only the presence of alternating colors, and perhaps the contrast from black and white versus shades of brown. Could we have had polka dot zebras?


      Plains animals with a strong herd instinct in the open plain do not benefit from camouflage, their presence is obvious before they are in sight.

      Disruptive patterning induces ‘motion dazzle’ in predators once the herd startles & takes off for the horizon.

      Cats are tuned for seeing in low light conditions & they seem hyper-tuned for seeing motion. They’re not seeing much colour – they see yellow & blue best. Cats are superb at judging distances, but can’t see much at all close up. Blurry.

      The high contrast light & dark stripes in motion on the skins of zebras probably interfere with picking out a weak zebra from the herd – cues for limping or small size may disappear in that B&W wavy, moving, stripey visual field.

      Spots may not be as effective a disrupter or it’s harder for zebras to make spots

      Why don’t all plains herd animals have stripes? Don’t ask impertinent questions! 🙂

  6. This raises the mystery of the now extinct quagga, which had a striped neck fading to a solid colored rump. What good was having half of your body protected from tse-tse flies? Better than nothing, I guess.

    1. Yes the Quagga is (was) mysterious. The brown rump had no stripes, just the head, neck and shoulders, maybe the rump was covered by a swatting tail, but their legs were just plain white. This is extra curious since some unstriped horses have some vague striping of their legs.
      Note that in Quagga areas there were no tsetse flies. Maybe their stripes were just a remnant of previously (ancestrally) needed stripes?
      I’m not even sure if their strikingly beautiful pattern: black and white head and neck, brown rump and white legs, did not at some time become part of sexual selection.

      1. And I also note that the ‘bontebok’ (Damaliscus pygargus) from the same Cape areas as the quagga, also has white legs and belly, just like the quagga, but contrary to the closely related ‘blesbok’ (Damaliscus pygargus phillipsi) or the also related ‘tsessebe’ (Damaliscus lunatus lunatus), occurring in more north eastern areas.
        Is there something about white legs in the fynbos and renosterveld areas?

  7. I wonder if a zebra suit would work against black flies In Spring/Summer hiking? Not sure any humans would be seen with me if I wore such a suit.

  8. I like the clever study design of dressing up horses in zebra suits.

    Thought experiment. I wonder if horses outfitted with zebra suits with horizontal stripes (i.e. the same vertical pattern rotated 90 degrees) would produce the same result? If it didn’t, that might suggest it does have something to do with the fly visual system assuming their flight orientation tends to be consistently “feet toward ground”.

    1. Except that horses, and many other grazers tend to graze up or down wind, while flys hunt into the wind, thus flys will tend to approach in a way that vertical stripes will present like a strobe light rather than a continuous colour band (horizontal might make landing strips!)

  9. interesting, I wonder if it works for botts?
    Now they should test pi-bald, pinto, paint and spots to see if those colour patterns also get a fly benefit.
    If stripes don’t camouflage why tigers? Or is that more like WWI battleships – disrupting range-finding.

    1. If stripes don’t camouflage why tigers? Or is that more like WWI battleships – disrupting range-finding.

      Stripes often do camouflage. You have to consider the creatures habits & environment. There is a very nice post on here somewhere from around 8, 9, 10 years ago discussing the 35? species of cat & their fur patterning. I will see if I can find it.

      1. Thanks Michael,
        I was being a little facetious with the tiger comment. I suppose if being camouflaged for tsetse allows zebra to reduce disease incidence and by being able to go into the tsetse belts reduce predation, that’s a powerful driver.

  10. The fly bump effect clinches it for me. Nice experiments, no horsing around that I could see!

    It is also easy to believe the suggested “optic flow” inherent illusion effects. And, it strikes me, the repeated structure of fly eyes may perhaps add some aliasing problems when observing repeated patterns.

Leave a Comment

Your email address will not be published. Required fields are marked *