Defensive Posture & Eyespots Protect Caterpillars

Pastry "caterpillar" used to examine the protective effect of eyespots and defensive posture of caterpillars.

You may recall that Tom Sherratt and I published a paper last year in Animal Behaviour on the protective value of eyespots to model caterpillar (see more about this work here). In this paper we found that eyespots could protect prey, but the protective effect was dependent on other aspects of caterpillar body colour. Specifically, the protection conferred by eyespots was less when caterpillars were a uniform green than when those caterpillars were two-toned (i.e., countershaded). Overall the protective effect of eyespots was much weaker than we had anticipated, perhaps because our pastry caterpillars don't exhibit any of the behaviours that real caterpillars do when they are attacked by birds. This led us to consider another feature of these caterpillars that may augment the protective effect of eyespots - their defensive posture.

The vast majority of the caterpillars with eyespots that I have had a chance to examine have some form of defensive posturing upon perceiving a threat. Some species tuck their head under their body and puff up their thoracic body segments (e.g., Papilio canadensis, Papilio troilus), while others pull their head telescopically into their body which also puffs up their thoracic segments (e.g., Xylophanes sp.). The suggestion has been that this defensive posture augments the protective effect of eyespots.

Papilio canadensis caterpillar on a branch in its defensive posture.
This is my photo of a caterpillar that I reared for another project.
Xylophantes juanita (Sphingidae) caterpillar in defensive posture.
Photo from Janzen & Hallwachs database.

There are a few reasons why the defensive posture might increase the protective effect of eyespots. First, it could help emphasize the markings to an attacker by drawing attention or emboldening the spots. Second, it may help the caterpillar appearing larger enabling it to buff predators away from attack. Finally, this posture might make the anterior body segments resemble the head, perhaps increasing overall resemblance to a snake by placing the false eyes (i.e., eyespots) in the context of a "head". Interestingly, several caterpillars engage in a similar defensive posture, including some with poorly developed eyespots or where eyespots are lacking altogether. Thus "head-shape" might be a mimetic signal in and of itself. Our goal was to test whether eyespots and the defensive posture protect caterpillars from birds, and if so, does having both traits increase protection even more?

Materials & Methods

It wasn't possible to deploy real caterpillars in the field, so we used artificial caterpillars made of flour and lard. We had observed previously that eyespots were more effective when the caterpillars were two-toned (i.e., countershaded), so we made all our prey two-toned by combining dark and light green dough (dyed with food colouring). All caterpillars were formed into 4 cm long cylinders, then to make the "defensive posture" prey we pressed these half of these caterpillars into a mould made from plaster of Paris. This ensured that all model prey (rest or defensive posture) the same volume, just like a real caterpillar that has shifted its posture.

Preparing to make the mould for this experiment
The finished mould. It cracked in half when I was removing it.

Over the course of a summer we made and deployed 576 model caterpillars, divided evenly among four treatments:

  1. No eyespots - Resting posture
  2. Eyespots - Resting posture
  3. No eyespots - Defensive posture
  4. Eyespots - Defensive posture

No eyespot - Resting posture treatment
No eyespot - Defensive posture treatment
Eyespot - Defensive posture (above) and Eyespots - Resting posture (below) treatments.

Model prey were deployed in the field by pinning them to tree branches. Each tree had four caterpillars - one of each treatment. We deployed caterpillars 6 times (i.e., 96 prey deployed each time), always in a new location, then tracked their survival three times a day over the next 90 h. Sometimes prey were gone, other times they were riddled with peck marks. Here are some examples of pecked "caterpillars" from the field:

Results & Interpretation

Our results showed that both the defensive posture and eyespots reduced the number of attacks, but having both traits didn't further increase the protection. This was surprisingly consistent with another recent study that examined rates of predation on model snakes with a viper-like head shape and/or zigzag patterning (Valkonen et al. 2011). They too found that possessing either trait conferred protection, but having both traits didn't increase that protection further. This seems to make sense - if either trait indicates risk from a threat, having both traits may help better identify that threat but doesn't change the risk associated with it. It is still unclear though how predators decide which cues reliably indicate a threat and which do not. A better understanding of this learning process might also help us understand why so many mimics only vaguely resemble their model.

Figure 2 from Valkonen et al 2011 showing number of raptor attacks on snake replicas

See the full Valkonen et al 2011 paper for free here:
Valkonen JK, Nokelainen O, Mappes J (2011) Antipredatory Function of Head Shape for Vipers and Their Mimics. PLoS ONE 6(7): e22272. doi:10.1371/journal.pone.0022272

We also looked at the number of pecks per caterpillar as well as where they were pecked. Caterpillars with both eyespots and the enlarged "head" shape received more "head"-directed pecks that caterpillars with either trait alone. Assuming birds direct their attacks toward a prey or predator's head, this might mean that birds still do perceive the anterior segments as more head-like when they have both traits - just not enough to deter their attack/curiosity.   

The next step for this research is to see if we can determine why the wide "head" defensive posture works. Of course the leading hypothesis is that this posture makes the caterpillar look more snake-like, or perhaps more like a viper. To make this argument convincingly you would need to show that these caterpillars look more like snakes (or perhaps vipers specifically) when they are in their defensive posture compared to when they are at rest. We are currently working on a project that examines whether or not this is the case using photos of preserved snakes and live caterpillars.


Hossie, T. J., Sherratt, T. N. (2012) Eyespots interact with body colour to protect caterpillar-like prey from avian predators. Animal Behaviour 84(1): 167-173, doi:10.1016/j.anbehav.2012.04.027

Hossie, T. J., Sherratt, T. N. (2013) Defensive posture and eyespots deter avian predators from attacking caterpillar models. Animal Behaviour [Early View]


  1. Hi Thomas,

    Interesting stuff! I'm sorry I didn't across your blog sooner.

    I'm playing with colourful orb-weavers at the moment, and have been slowly wading through the mimicry literature, so I'm interested in the caterpillar-as-snake-mimic idea. With the disclaimer that I don't know anything about these guys - it just seems hard to believe that birds would regularly misclassify a ~ 5 cm caterpillar as a > 30 cm viper, even with the markings. What leads you to think this may be the case? Are there tiny green vipers running around near these caterpillars? (I don't know much about snakes either...)

    Keep up the good work!

    Tom (@TomEdWhite)

  2. Hello Thomas,

    Thanks for the kind words and interest in my blog/research. We can be fairly confident that at least some of these caterpillars are snake-mimics. Have a look at caterpillars from the Hemeroplanes genus:

    There are a number of arguments that various people have used to answer this question of size asymmetry between models and mimics in this system. Most of these are outlined in Janzen et al (2010) PNAS

    Here is a partial list of the arguments:

    -Only a subset of insect-eating birds need to be fooled some of the time in order for selection to favour mimicry

    -Newborn vipers fall within the size range of large caterpillars and pose considerable risk. Check out these snake photos for example:

    -Total length may not be an important cue because the length of the caterpillar is obscured by foliage

    -Many arboreal snakes that pose important risk to small insect-eating birds are quite thin (e.g. Oxybelis sp.), and large caterpillars may be equally thick.

    -Selection strongly favours those birds that flee quickly upon perceiving predator stimuli so much so that many birds are genetically programmed to quickly flee upon seeing any eye or face like stimulus.

    -Snakes may not be the only model - predator eye mimicry more generally is favoured because startlingly close proximity to other eyes or faces can also be dangerous to small insect-eating birds.

    -Size may not be as important a cue to birds as it is to humans. See hoverfly work by Dittrich et al 1993 Proc B 251: 195-200

    I’d be interested to hear which of these you find most/least convincing.

    Kind regards,


  3. This comment has been removed by the author.

  4. Wow, those Hemeroplanes are stunning!

    You make a good case, much appreciated. My question was driven by naivety of the system. I can certainly see how those (quite tiny) snakes would be a problem for insectivorous birds. Presumably the caterpillars have to deal with a community of invertebrate predators too (mantids, spiders), which I imagine would be similarly wary of the eyespots and posturing.

    Final quick question, if you don't mind - do the swallowtails that employ these eyespots and behavioural tactics also use osmeteria and chemical defenses? I'm just interested to know if they've traded off one defensive strategy for another, or if they've 'doubled-down' and use both, if that makes sense.

    And thanks for the papers, I'll check them out.

    All the best,

    (I deleted the previous response because it was a bit unclear)

  5. Hi Thomas,

    I’m not sure the how much protection these defences (head shape & eyespots) offer against non-avian attackers. I don’t see mantids as an important predator in this system, and I know from experience that spiders aren't afraid of Papilio canadensis caterpillars – See this post:

    The osmeterium is a very interesting feature of Papilio caterpillars. Many naturalists speculate that it aids in protection by resembling a snake tongue. While, it does resemble a snake tongue at least to human observers, its role seems primarily for defence from insects. The osmeterium secretes a volatile compound which helps to ward off parasitoid flies and wasps, as well as ants (See Damman 1986 Ecol Entomol 11:261-265). Assassin bugs may be also to circumvent the osmeterium defence (Berenbaum et al 1992 J Insect Behav 5: 547-553). The osmeterium seems to be less effective against birds though, and probably doesn’t play a major role in determining palatability (see Leslie and Berenbaum 1990 J Lep Soc 44: 245-251).

    Hope this helps!


  6. Thanks for all the info Thomas, I appreciate the effort! There's a guy in our lab playing with assassin bugs, maybe I should rustle up some swallowtails...

    All the best,