SEARCH

SEARCH BY CITATION

Keywords:

  • anti-predator defence ;
  • birds ;
  • butterfly ;
  • deflection hypothesis ;
  • eyespots ;
  • lepidoptera ;
  • predator;
  • prey interaction ;
  • speed;
  • accuracy trade-off

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Small eyespots on butterflies have long been thought to deflect attacks, and birds are the presumptive drivers selecting for these patterns; however, evidence of this function is still ambiguous. Marginal eyespots typically consist of a UV-reflective white pupil, surrounded by one black and one yellowish ring. We have recently shown that Cyanistes caeruleus (blue tits) attack such eyespots, but only under low light intensities with accentuated UV levels: the increased salience of the eyespots relative to the rest of the butterfly probably explains this result. Possibly the background against which the butterfly is concealed may deceive birds to make similar errors. We therefore presented speckled wood butterflies decorated with eyespots (or controls without eyespots) to C. caeruleus against two backgrounds: oak and birch bark. Our results show that: (1) eyespots, independent of background, were effective in deflecting attacks; (2) the time elapsed between a bird landing and the attack was interactively dependent on the background and whether the butterfly bore an eyespot; and (3) the speed at which a butterfly was attacked predicted the outcome, with faster birds being more prone to errors than slower birds. This underscores a speed–accuracy trade-off in the predators, and that background plays a role in the defensive qualities of marginal eyespots. © 2013 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 109, 290–297.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Eyespots are common patterns on the wings of many butterfly species, and are customarily thought to function in their anti-predation repertoire. Depending on the size and position of the eyespots, two hypotheses on their adaptive function have traditionally been proposed. Larger eyespots, often dorsally placed, are suggested to thwart predators from attacking through intimidation (the ‘intimidation hypothesis’; Blest, 1957; Stevens, 2005), whereas small eyespots, often ventrally placed, are suggested to attract attention, thus provoking predators to misdirect their attacks, which gives the butterfly a chance of escaping (the ‘deflection hypothesis’; Ruxton, Sherratt & Speed, 2004; Stevens, 2005). A repelling function of large eyespots is substantiated by solid evidence (e.g. Blest, 1957; Vallin et al., 2005; Stevens, Hardman & Stubbins, 2008; Brilot et al., 2009; Kodandaramaiah, Vallin & Wiklund, 2009; Merilaita et al., 2011; Blut et al., 2012), but whether small, marginal eyespots on butterfly wings really function to divert attacks remains largely ambiguous (Stevens, 2005; Kodandaramaiah, 2011).

Birds are generally assumed to be the predators targeted by the diverting function of marginal eyespots in butterflies (Blest, 1957; Stevens, 2005; Kodandaramaiah, 2011). Theory postulates that predators should launch incapacitating attacks to prevent prey from escaping. In particular, predators confronted with prey as elusive as a butterfly will probably see the insect escape if the first strike fails. For this reason, we expect prey to evolve morphological traits that entice predators to aim their first strike at less vulnerable body parts (Poulton, 1890; Edmunds, 1974; Powell, 1982; Ruxton et al., 2004). Furthermore, an inherent extension of the ‘deflection hypothesis’ is that eyespots may confuse a predator by presenting ambiguous information, thereby increasing the time from detection until the predator has made a decision on just where to target its attack (cf. the ‘false-head hypothesis’ Robbins, 1981; Tonner et al., 1993). Until recently, evidence that patterns located at a prey animal's body periphery can change the strike distribution of predators comes from laboratory experiments using mealworms with one painted-on ‘eyespot’ at the posterior (Blest, 1957), or butterflies, Pieris rapae, with a small painted-on black circle at the ventral surface of the whitish hindwing (Wourms & Wasserman, 1985); however, these studies are hampered by the low sample sizes of predators (five and four birds, respectively; Olofsson et al., 2010; Vallin et al., 2011). There is also indirect evidence that deflection is involved in the defence repertoire in certain butterfly species: Hill & Vaca (2004) demonstrated that marginal wing patches with a pattern assumed to have a deflective function tore off more easily, compared with the corresponding wing area of closely related species that lack such patterns.

Eyespots are common pattern features in the largest butterfly family, the Nymphalidae (Nijhout, 1991). Butterflies in the nymphalid subfamily Satyrinae often possess marginal eyespots that are, as a rule, more accentuated on the ventral wing surface that is exposed during rest. These eyespots are typically conspicuously composed of a white central UV-reflecting dot (conventionally referred to as a ‘pupil’) surrounded by one black and one yellowish ring (Monteiro, 2008), and have the purported function of diverting attacks (Poulton, 1890; Blest, 1957; Edmunds, 1974). At tropical latitudes, satyrines often exhibit polyphenism with regard to their ventral wing patterning: dry-season forms typically possess no or only vestiges of eyespots, whereas wet-season forms often possess fully developed eyespots (Brakefield & Larsen, 1984; Brakefield et al., 1996). Great efforts have been undertaken in investigating the genetic basis for this polyphenism, and have substantially increased our understanding of the machinery controlling the formation and inhibition of eyespot development (e.g. Brakefield et al., 1996; Monteiro, Brakefield & French, 1997; Beldade & Brakefield, 2002; Beldade, Koops & Brakefield, 2002; Monteiro, 2008). However, the selection pressures responsible for the origin and maintenance of eyespot polyphenism are not fully understood. It has been clearly demonstrated that eyespots are costly in a withered, brownish environment because of the increased discovery rate by birds, which helps explain why crypsis is selected for in uniform dry-season conditions (Lyytinen et al., 2004). The remaining issue is why prominent marginal eyespots are selected for during the wet season, as well as in non-tropical areas where many satyrines also possess marginal eyespots. Experiments using live squinting bush brown butterflies, Bicyclus anynana, have not supported the idea that bird or lizard predators direct their attacks differently when confronted with eyespot-ornamented wet-season individuals, compared with when confronted with camouflaged dry-season individuals (Lyytinen, Brakefield & Mappes, 2003; Lyytinen et al., 2004; Vlieger & Brakefield, 2007). Interestingly, Lyytinen et al. (2004) found that naive juvenile pied flycatchers, Ficedula hypoleuca, were less successful in capturing and killing eyespot-ornamented B. anynana than were adult birds, whereas the cryptic form was attacked with the same success (Lyytinen et al., 2003). However, whether this difference should be attributed to a deflective function of marginal eyespots remains to be investigated, as there was no evidence to indicate that misdirected attacks towards the marginal eyespots explained the lower success rate of naive juveniles.

We recently demonstrated that the eyespots of the woodland brown butterfly, Lopinga achine, were effective in deflecting attacks from Cyanistes caeruleus (blue tits), but only when the butterfly was presented in a low light intensity with accentuated UV light levels. However, when presented under low light intensity without UV or under high light intensity with prominent UV an overwhelming majority of the birds attacked the head region of the butterfly. It seems likely that the strong deflective effect could be attributed to a contrast shift, which accentuates the eyespots (that have a UV-reflecting white pupil at the centre), and fades the rest of the butterfly when the light intensity was low but the UV level was high. With this result we suggested that this phenomenon possibly occurs under light conditions when shorter wavelengths are accentuated, such as during the dawn (Théry, Pincebourde & Feer, 2008). Contrast shifts may occur for other reasons than changes in light composition, however, and the most apparent situation would be if a butterfly with a conspicuous eyespot is perceived against a matching background (Vallin et al., 2011). Although background has been acknowledged as a possible factor in influencing the deflective properties of marginal eyespots (Lyytinen et al., 2003; Vallin et al., 2011), it has hitherto never been investigated with real butterflies. To the best of our knowledge, the only study that has attempted to test the influence of background used artificial eyespots printed on triangular paper targets with C. caeruleus as experimental animals, and the birds were shown to be more prone to lift the triangles from the side of the midline, where a small eyespot had been printed; however, this finding was independent of whether the target matched or mismatched the background (Vallin et al., 2011). Vallin et al. (2011) concluded that small eyespots on prey (such as butterflies) can also deflect bird attacks under broader conditions (i.e. daylight conditions) than those proposed by Olofsson et al. (2010). However, as all previous experiments using real butterflies have reported negative results concerning whether bird attacks could be misdirected towards the eyespots when using lighting regimes corresponding to daylight (Lyytinen et al., 2003, 2004; Vlieger & Brakefield, 2007; Olofsson et al., 2010), it remains to be tested whether there are circumstances in daylight conditions when eyespots on real butterfly wings can deflect attacks. Therefore, in this study we test the influence of background on the deflective function of a single marginal eyespot by presenting two types of manipulated butterflies against two natural backgrounds: oak and birch bark. We do this by investigating where C. caeruleus direct their first strike when confronted with dead specimens of the speckled wood butterfly, Pararge aegeria, the lower hindwing margin of which we either provide with an eyespot or a uniformly coloured control pattern, both coming from the ventral forewing of the closely related wall brown butterfly, Lasiommata megera. Furthermore, we also measure the time from landing until the bird launches its strike towards the butterfly: we do this to investigate whether the presence of an eyespot distracts the bird's attention, and so delays the time from detection until the butterfly is attacked, and also whether the time from landing until the attack predicts whether an attack is deflected or not.

Material and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Study species and animal husbandry

Predator

Cyanistes caeruleus were chosen as the predators in our experiment because they are omnivorous generalists and include insects in their diet. Although C. caeruleus are not specialized on hunting butterflies, they are extremely skillful at assessing the presence of a butterfly and almost invariably attack its most vulnerable body part – the head – more or less in a stereotypical manner (Olofsson et al., 2010). For this reason, C. caeruleus are highly suitable for testing whether the presence of an eyespot could change the direction of a bird's strike. Cyanistes caeruleus were caught in mist-nets and were housed indoors individually in cages (80 × 60 × 40 cm). The cages were furnished with two perches and commercial litter on the floor. Water and food (suet and sunflower seeds) were supplied ad libitum. Additionally, birds were fed a few mealworms (Tenebrio molitor) each day. The lighting regime in the housing room was kept on a light/dark cycle adjusted weekly to correspond to that of the prevailing season, including half an hour for dusk and for dawn. In total, 88 birds were used in the experiment and each bird only participated in one trial.

Prey

We used two closely related species of satyrine butterflies in this study: Pararge aegeria and Lasiommata megera. The two species differ notably in eyespot patterning: L. megera has a range of small, but fully developed, eyespots on its ventral hindwings and a large apical eyespot on its forewings. In contrast, P. aegeria has only vestiges of eyespots (small yellowish dots) on its ventral hindwings, and a considerably smaller apical eyespot on its forewings. The butterflies, P. aegeria and L. megera, were reared communally, but in species-specific colonies as larvae on the grasses Poa annua and Festuca ovina, respectively. Emerged butterflies were placed singly in glassine envelopes and were killed by freezing (–25 °C). The butterflies were thereafter allowed to dry for a couple of weeks.

The dead butterflies were used to fabricate two different forms of specimens, allowing us to test the deflective function of a single marginal eyespot. First, we made P. aegeria specimens eyespotless by detaching their forewings. Second, using a razor, we carefully cut out the ventral apical eyespot from the forewing of L. megera specimens, and, from the same specimen, an equally large circular piece below the eyespot that functioned as a control marking. The cut-out circles, comprising a conspicuous eyespot or a dull brownish control, were glued onto the forewingless P. aegeria specimens a few millimeters from the lower end of the hindwing margin (Fig. 1).

figure

Figure 1. Schematic picture of the two butterfly forms: the butterfly (Pararge aegeria) to the left has been provided with an eyespot from the ventral surface of the forewing of the closely related wall brown butterfly, Lasiommata megera, and the butterfly to the right has been provided with a control marking that consisted of the ventral surface area just below the eyespot in L. megera. All attacks distributed to the left of the dashed line were considered as anterior attacks (i.e. ‘not deflected’); attacks distributed elsewhere were invariably directed exactly towards the provided eyespot or control marking, and were considered as having been ‘deflected’.

Download figure to PowerPoint

Experimental procedures

Experiments were carried out at Tovetorp Research Station located in the south-east of Sweden, between 16 March and 1 April 2011, and between 12 January and 11 February 2012. All trials were performed in a cage made of fibreboard (55 × 90 × 70 cm, W × H × D), which was illuminated from the ceiling by two daylight-mimicking, high-frequency fluorescent lamps (15 W, BIOlight, Narva). The experimental set-up consisted of a small plank (11.5 × 36 cm: W × L) that had two perches at slightly differing heights: one higher, in front of the butterfly (∼5 cm), and one lower that was mounted further away from the butterfly on a piece of oak bark that was affixed to the plank. The butterfly specimen was presented in a resting position at a right angle to the closest perch, glued onto a horizontally mounted piece of bark. Immediately behind the butterfly (∼1 cm), a 4 × 8 cm (W × H) piece of birch or oak bark was erected, against which the butterfly was exposed or concealed, respectively. The experimental box was placed in a dark room with a constant temperature (16 °C). In order to encourage the birds to attack the butterfly from the closest perch, the background bark piece was protected with two small sheets of blasted acrylic glass on each side and an opaque plastic tube on the top (Fig. 2). All experimental trials began with a training procedure. A decapitated mealworm was offered in front of the butterfly, which at that time was hidden from the bird's view by a small piece of brown paper. As soon as a bird had consumed the mealworm and had flown off from the perch, we turned the light off for a second or two and simultaneously removed the piece of paper via an attached nylon line from outside the cage. Thereafter, we turned the light on and the bird was now allowed to attack the butterfly. All trials were recorded with a digital video camera (Sony DCRVX1000E) that was mounted close to the ceiling on the opposite wall of the set-up. Furthermore, a small perch was available for the birds and was mounted below the ceiling on one of the long side walls, close to where the camera was mounted. The camera was zoomed so as to focus on the butterfly and the last perch, which allowed us to assess where the birds aimed their initial strike (i.e. whether the attack was deflected or not), as well as the time (measured in seconds) it took the birds from landing until their first strike was launched [see Fig. 2 for details on the experimental set-up; also see Dimitrova et al. (2009), who used the same experimental cage for the camouflage experiments]. A small bowl of fresh water was always provided for the birds, placed on the floor in front of the small plank.

figure

Figure 2. Outline of the experimental cage.

Download figure to PowerPoint

Experiments were conducted on two different occasions (see above). Most of the trials in which the butterflies were presented against the oak bark background were carried out in 2011 (n = 36), and were continued in 2012 (n = 8), and all trials with birch bark as the background were carried out in 2012 (n = 44) following the same protocol as that used in 2011. Ideally, the two different backgrounds should have been used in both years to rule out possible differences between years in the birds' motivation to perform in our experiment. For this reason, we compared 2011 and 2012 with respect to the time it took the birds to attack the mealworm that was provided during the training phase (i.e. time from start of a trial until the mealworm was attacked). The results show that the birds' motivation to attack a provided prey (i.e. mealworm) did not differ between years: birds in 2011 attacked the mealworm after the same period of time (593 ± 166 s; mean ± 95% confidence interval) as birds in 2012 (686 ± 129 s; Welch's two-sample t-test: t = −0.8717, d.f. = 72.085, P = 0.386).

Statistical analysis

The point of attack (i.e. whether an attack was deflected) was analyzed using a generalized linear model with logit link as function and binomial response (attack deflected/attack not deflected), and with treatment (eyespot/no eyespot) and background (birch/oak) as factors. Additionally, we included ‘time from landing until attack’ (measured in seconds) as a continuous predictor.

‘Time from landing until attack’ could be considered as a variable either predicting the outcome of an attack (i.e. the location of the attack (as above)) or as a response variable (i.e. that the time required performing an attack depends on the treatment and/or the background). To investigate whether ‘time from landing until attack’ was influenced by treatment (eyespot/no eyespot) and background (birch/oak) we performed an ANOVA (type II) on all the birds whose attacks were not deflected. The requirement of homogenous variances was met by log transforming ‘time from landing until attack’.

All analyses were performed in R 2.10.1 (R Development Core Team 2009).

Ethical note

The experiments described herein comply with the current laws of Sweden, and the experimental procedures have been reviewed and approved by the regional ethical committee (Linköpings djurförsöksetiska nämnd, Dnr 11-11).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

More birds attacked the posterior of the butterfly when confronted with a butterfly decorated with an eyespot (located on the posterior part of the wing, Fig. 1; Video Clip S1), compared with when confronted with a butterfly provided with a control marking (Tables 1 and 2; Video Clip S2). In total, ten birds misdirected their initial strike towards the posterior of the butterfly, and invariably struck exactly towards the provided eyespot marking (nine birds) or control marking (one bird) (Video Clip S1), whereas the remaining birds (n = 78) attacked the anterior region of the butterfly (i.e. left of the dashed line in Fig. 1; Video Clip S2). There was no significant interaction between background and treatment on whether the birds misdirected their attacks (Tables 1 and 2). Interestingly, birds that launched their attack quickly after landing on the perch in front of the butterfly were more likely to misdirect their attack compared with birds that took a longer time before launching their attack (Table 2).

Table 1. Distribution of bird attacks on the two butterfly forms (with eyespot or with control marking) when seen against two different backgrounds: oak or birch bark
BackgroundTreatmentAttack not deflectedAttack deflected
  1. All attacks that were ‘not deflected’ were distributed on the anterior part of the butterfly (Fig. 1). Attacks that were ‘deflected’ were invariably directed exactly towards the provided eyespot or control marking.

BirchEyespot203
BirchControl201
OakEyespot186
OakControl200
Table 2. Results from a generalized linear model with ‘point of attack’ as a binomially distributed response (two levels: eyespot/control marking or anterior; Fig. 1), with treatment (two levels: eyespot or control marking) and background (two levels: oak or birch) as categorical factors, and the interaction between treatment and background. Additionally, ‘time from landing until attack’ was incorporated as a continuous predictor
Factorsχ2d.f.P
Treatment (T)5.13510.0234
Background (B)2.060210.151
Time from landing until attack13.957710.000187
T × B2.348210.125

For the birds that did not misdirect their attack (n = 78) the time from landing on the perch in front of the butterfly until the attack was launched was interactively dependent on the background and whether the butterfly bore an eyespot (Table 3, Fig. 3). This suggests that birds required more time to attack butterflies that bore an eyespot, but only if the rest of the butterfly was concealed against the background.

figure

Figure 3. Time from landing until attack measured in seconds (mean ± standard error), depending on treatment (eyespot/control) and background (oak/birch). Note that the birds have been divided into two groups: those whose attacks were deflected (to the left of the dashed line) and those whose attacks were not deflected (to the right of the dashed line). Numbers above each mean value denote the number of birds.

Download figure to PowerPoint

Table 3. ANOVA results for birds (n = 78) whose attacks were not deflected with ‘time from landing until attack’ (log-transformed) as the response, with treatment (two levels: eyespot or control marking) and background (two levels: oak or birch) as factors, and their interaction
FactorsSSd.f.FP
Treatment (T)0.000810.01010.92
Background (B)0.437615.8480.0181
T × B0.41815.5860.0207
Residuals5.537174  

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We have demonstrated here that a butterfly's eyespot can deflect attacks from a generalist bird predator, C. caeruleus, under daylight-mimicking conditions. In our experiment, 19% (nine of 47) of the birds that confronted a butterfly with an eyespot located at the posterior margin of the wing attacked the eyespot instead of the anterior region, whereas only 2% (one of 41) misdirected their attack when confronted with a butterfly provided with a control marking. It is noteworthy that all attacks that were misdirected were launched exactly towards the eyespot (nine birds) or control marking (one bird). It is also noteworthy that the nine birds that were deceived by the eyespot, and directly targeted their attack towards the eyespot, were quicker to attack [0.94 ± 0.17 s (SE)] compared with the 38 birds that were not deceived, and who directed their attack towards the anterior of the butterfly (2.33 ± 0.29 s). Theory postulates that animals pay accuracy costs with decreasing decision time (Chittka, Skorupski & Raine, 2009), and indeed birds that forage on elusive prey such as butterflies inevitably face a trade-off when detecting their prey: a too hasty strike may increase the risk of misdirecting the attack, whereas too much time spent contemplating on how the prey should be incapacitated may result in a loss of the prey altogether. This relates directly to the ‘speed–accuracy trade-off hypothesis’, and may explain why some of the birds attacked the eyespot and overlooked the real position of the head of the butterfly.

Previous attempts to test whether eyespots on butterfly wings can make predators misdirect their attacks have not given support for the idea (Lyytinen et al., 2003, 2004; Vlieger & Brakefield, 2007), and it is relevant to note that 81% (38 out of 47) of the birds in this study were not deceived by the eyespot when attacking the butterfly. The only previous study of a real prey animal that has given strong support for the idea [but see Vallin et al.'s, (2011) study of the deflective function on spots on artificial prey] is Olofsson et al.'s (2010) study, in which 12 of 14 C. caeruleus misdirected their attacks towards the marginal eyespots of Lopinga achine specimens, but only under low light conditions with a prominent UV component. The difficulty with which prey with eyespots such as satyrine butterflies seem to have in deceiving bird predators under natural light conditions underscores the cognitive skill of many birds, and probably explains why it has been so difficult to demonstrate that these eyespots can have an adaptive function as a defence against predation.

Although it seems reasonable to assume that a conspicuous eyespot would be more likely to deflect an attack from a bird if the rest of the butterfly blends into the background, our study did not detect such an effect (Vallin et al., 2011); however, as the ‘deflection hypothesis’ posits that eyespots should attract the attention of predators (Poulton, 1890; Blest, 1957; Edmunds, 1974; Ruxton et al., 2004; Stevens, 2005; Kodandaramaiah, 2011), an inherent extension of the hypothesis would be that an eyespot may confuse a predator by presenting ambiguous information, and thereby increasing the time from detection until the predator decides where to launch its attack (cf. the ‘false-head hypothesis’; Robbins, 1981; Tonner et al., 1993). The current study supports such an effect with reference to the background: birds that attacked the anterior of the butterfly took about twice the time to attack butterflies with eyespots when these were presented against a concealing oak bark background (3.16 ± 0.49 s), compared with when attacking butterflies with eyespots against an exposing birch bark background (1.59 ± 0.26 s), whereas control butterflies (without eyespots) were attacked equally fast, independent of the background.

It is apparent that birds such as C. caeruleus are not only skilled at finding camouflaged prey but also at making an impressively accurate assessment of how they should aim their strikes (Olofsson et al., 2010; this study). The non-random attack behaviour of C. caeruleus demonstrates that one or more butterfly features govern the birds' decision on how the prey should best be incapacitated, although further experiments are needed to establish what these cues are. Nevertheless, this substantiates that prey animals may gain fundamentally by leading astray a predator's perception of where the essential body parts are located. However, given that a predator that misdirects its initial strike then launches a second accurate strike, we would expect deflective eyespots (or equivalent traits) to evolve primarily in species that have the ability to escape immediately after the first attack (Cooper & Vitt, 1991). In butterflies, this is accomplished by escape flights or, when temperatures are unfeasible for flight, by dropping from the roosting place into the undergrowth (Dennis, 1986). As such, we can only speculate about the underpinning mechanisms of why some of the C. caeruleus birds aimed their strike at the eyespot; however, an appealing idea is that a circular, contrasting pattern may attract attention by resembling a real eye (or head), which otherwise is likely to function as a fail-safe cue in guiding the predator towards the most vulnerable body parts of the prey (cf. the ‘false-head hypothesis’; Robbins, 1981; Tonner et al., 1993).

In their capacity as generalist predators birds are unlikely to possess innate specific hunting tactics, such as perfectly attacking the head of a butterfly; rather, they should be expected to have some crude innate ‘predation’ behaviours that are subject to learning through repeated encounters with prey (Schlee, 1983). Quite surprisingly, however, it has been demonstrated that prey-attack behaviour may change dramatically during a predatory bird's ontogeny. For example, Smith (1973) investigated the prey-attack behaviour of the loggerhead shrike, Lanius ludovicianus, and found that naive juveniles switch from attacking the tails to attacking the neck of living mice as they grow older, whereas older individuals that were kept naive with respect to mice invariably struck at the neck at the first encounter. This suggests that very specific hunting tactics can indeed be innate, hard-wired behaviours in predatory birds (Smith, 1973). Regardless of whether attacking a butterfly's head is learned through experience or programmed ontogenetically, it is likely to play a major role in the evolution of deflective patterns in prey (cf. Lyytinen et al., 2003; Lyytinen et al., 2004), and it has been suggested that the high incidence of naive predators during the tropical wet season may condition marginal eyespots as a better defence than crypsis (Lyytinen et al., 2004). Nevertheless, presumed differences in prey-attack behaviour in juvenile and adult birds when confronting butterfly prey needs more substantiation through experimental investigation: specifically, such experiments should focus on whether juvenile birds are more likely than adult birds to misdirect their attacks towards the eyespots.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Karl Gotthard for statistical advice and valuable comments on the article. We also thank Bertil Borg and two anonymous reviewers for their helpful comments.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
bij12063-sup-0001-si.wmv8068K

Video S1. The video shows a blue tit that misdirects its attack towards the butterfly's eyespot.

bij12063-sup-0002-si.wmv6732K

Video S2. The video shows a blue tit that attacks the butterfly's head.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.