Why blue tongue? A potential UV-based deimatic display in a lizard
Deimatic displays are a type of anti-predator behaviour that startles the predator. They have received much recent theoretical attention, enabling the empirical study of this phenomenon within a predictive framework. It has long been known that bluetongue skinks (Tiliqua spp.), when approached by predators, open their mouth widely and expose a conspicuously coloured tongue. Here, we test whether such ‘full-tongue’ displays are triggered by an imminent predatory attack in the Northern Bluetongue skink Tiliqua scincoides intermedia and examine whether this display behaviour is consistent with the predictions from deimatic display theory. First, we demonstrate that luminance at the rear of the tongue, which is only exposed during full-tongue displays, is almost twice as high for lizard and bird receivers compared to the tip of the tongue, and that tongue colouration is generally more conspicuous to a bird than a lizard visual system. Second, staged predatory encounters using model predators reveal that lizards primarily exhibit full-tongue displays in the final stages of a predatory attack. Lizards performed full-tongue displays congruent with the predictions associated with deimatic displays, i.e. rapid exposure of conspicuous elements from a previously inconspicuous state concurrently with aggressive defensive behaviour, most frequently during the final stages of a predatory encounter. Surprisingly, we also found that lizards vary the area of the tongue exposed during chemoexploratory tongue-flicks depending on whether a predator is present or absent.
Bluetongue skinks have long been known to expose their large blue tongue in response to predatory threats. However, this behaviour has never been investigated empirically. Here, we use Northern Bluetongue skinks (Tiliqua scincoides intermedia) to test whether this behaviour is consistent with predictions associated with deimatic displays. We show that the rear of their tongue is UV-blue and more conspicuous to predators compared to the tip and that this ‘full-tongue display’ is only triggered in the final stages of a predatory attack.
KeywordsDeimatic displays Anti-predator behaviour Reptile Coloration
From camouflage to aposematism, anti-predator strategies are ubiquitous in nature (Humphries and Driver 1970; Edmunds 1974; Stuart-Fox et al. 2006; Speed and Ruxton 2007; Bradbury and Vehrencamp 2011; Maan and Cummings 2012; Caro et al. 2016). Many of these strategies aim to deter predatory attacks. However, when a predatory threat is imminent, some prey exhibit a specific type of anti-predator behaviour such as deimatic displays. Deimatic displays are momentary, transient, and highly conspicuous, inducing a startle response and/or overloading the senses of an attacking predator such that the predator pauses, slows, or stops the attack (Umbers et al. 2015, 2017; Umbers and Mappes 2016). For example, mountain katydids (also called bush crickets Acripeza reticulata), which are chemically defended and normally have a cryptic colouration, flash their spectacular abdominal colours by lifting their wings as soon as a predator attempts subjugation (Umbers and Mappes 2015). This behaviour, combined with the katydids’ chemical defences, is believed to startle the predator, allowing the katydids to survive the attack.
Recent contention regarding deimatic display theory (Umbers et al. 2015, 2017; Skelhorn et al. 2016) has helped establish a useful theoretical framework for understanding the function and efficacy of these displays. Deimatic display theory is based on a fundamental tenet, that the effectiveness of deimatic displays depends on a sudden transition from an inconspicuous state (e.g. camouflage) to a highly conspicuous display when a predatory attack is imminent, causing a reflexive recoil in the predator (Umbers et al. 2015, 2017; Skelhorn et al. 2016). Selection is expected to create a conspicuousness-camouflage trade-off that is by necessity behaviourally regulated (Umbers et al. 2015, 2017). Unlike other predator-deterrent strategies (e.g. pursuit deterrent signals or aposematism, but see Humphries and Driver (1970)), deimatic displays do not require any learning from the predator to be effective (Umbers et al. 2017). Overall, deimatic displays are unique anti-predatory behaviours that have, unfortunately, received little empirical attention; their mechanisms and evolution are far from fully understood (Umbers and Mappes 2015; Umbers et al. 2015, 2017; Skelhorn et al. 2016).
Our aim was to identify what triggers bluetongue skink anti-predatory displays and investigate whether responses conform to the predictions corresponding to the production of deimatic displays. We used the Northern Bluetongue skink (Tiliqua scincoides intermedia) as a model to test the following predictions: (i) that tongue colour is conspicuous to a predator visual system and reflects a conspicuous-camouflage trade-off indicative of deimatic displays; (ii) that ‘full-tongue’ displays represent a rapid transition from relative crypsis to full display, most often deployed at a late stage during encounters with model predators; and (iii) that lizards use additional anti-predator behaviour to amplify the effect of the full-tongue display.
Material and methods
Northern Bluetongue skinks (T. s. intermedia) are omnivorous, ground-dwelling lizards, and one of the largest scincoid species. This species is restricted to the wet-dry tropical region of northern Australia, inhabiting semi-open habitats that include human-modified environments (Koenig et al. 2001). Bluetongue skinks have a cryptic body colouration that includes broad brown bands across the back. When threatened, they perform a full-tongue display, during which they widely open their mouth and protrude their large blue-coloured (most likely UV—Abramjan et al. 2015) tongue (Fig. 1). Birds, snakes, monitor lizards, and potentially carnivorous marsupials (e.g. Northern Quoll) constitute the primary natural predators of bluetongue skinks (Belcher 1995; Olsen et al. 2010; Fitzsimons 2011). All these predators are believed to have UV-vision, although it has not been shown in Australian quolls (Cuthill et al. 2000; Loew et al. 2002; Deeb 2010). Foxes, which lack UV vision (but see Douglas and Jeffery (2014)), are a more recent introduction in Australia (1870) and occasionally prey upon bluetongue skinks (Nielsen and Bull 2016).
Data have will be made publically available in the figshare data repository. https://figshare.com/s/ca22f4243e5a6961aece
Tongue measurements of conspicuousness
We used a USB-2000 portable diode-array reflectance spectrophotometer with a R200-7-VIS-NIR reading illumination probe and a PX2 strobe light source (Ocean Optics Inc.) to obtain spectral reflectance measurements of the back, the middle, and the tip of the 13 lizards’ tongues in order to characterise the ‘colour’ of the tongue. We used a notebook computer with SpectraSuite (Ocean Optics) to obtain the spectral data. We repeated each measurement twice and used the average. Spectra were recorded at 0.37-nm intervals and expressed as the percentage of light reflected relative to a certified Spectralon 99% white diffuse reflectance standard (Labsphere, Inc.). The illumination and reflectance probes were held at a constant distance of 5 mm to the tongue surface, at an angle of 90°. Spectral data were analysed in R.v.3.1.1 using the package PAVO (Maia et al. 2013). We extracted three colourimetric variables: luminance (sum of individual wavelength amplitudes between 300 and 700 nm, R300–700), UV-chroma (R300–400/R300–700), and λmax (wavelength at the maximum reflectance peak). We used visual modelling (Vorobyev-Osorio model) to test for differences in tongue conspicuousness between conspecifics and avian predators (Vorobyev et al. 1998). Because visual systems of terrestrial lizard species are conservative (Loew et al. 2002), we used the visual system of Platysaurus broadleyi (Fleishman et al. 2011) as a lizard representative. Platysaurus broadleyi is the closest relative of bluetongue skinks (Pyron et al. 2013) for which visual modelling is possible; Pavo cristatus was used as an avian predator with UV-photoreceptors (Hart 2002). As relative cone abundance for visual modelling, we used 1:2:2:4 (UWS/SWS/MWS/LWS) for the lizard and 1:1.9:2.2:2.4 (VWS/SWS/MWS/LWS) for the bird and a value of 0.1 for the Weber fraction for both visual systems. For each visual model, we calculated the chromatic and achromatic contrasts, in units of just noticeable differences (JND, Vorobyev et al. 1998), for each region of the tongue against a soil substrate background using standard daylight illumination (D65). We used the average of three reflectance measurements of soil, which we assessed would constitute the main background against which full tongue displays would be seen.
Behavioural experiments and tongue area
We conducted behavioural assays using 13 wild-caught adults (6 males and 7 females) of the Northern Bluetongue Skink (T. s. intermedia) from the Kununurra region, in Western Australia (15° 51′ 17″, 128° 43′ 42′ E). Behavioural assays took place in an outdoor circular arena (∅ = 2.5 m, h = 1.15 m). A piece of cloth was strung above the arena to provide some shade for the lizards. However, model predators were consistently presented in full sun. We only used diurnal predators, since evidence for nocturnal predators is weak in this species. Predator treatments consisted of a snake (frozen Western Brown, Pseudonaja mengdeni—snout-vent length [SVL] = 108 cm), a bird (stuffed blue-winged kookaburra, Dacelo leachii), a monitor lizard (mounted yellow-spotted monitor, Varanus panoptes, SVL = 55 cm), a stuffed fox, and a piece of wood (75 × 9 cm) as a control. Resources to create a predatory treatment corresponding to a carnivorous marsupial or a dingo were not available. Each lizard participated in two experimental trials a day during morning hours, which is the peak activity period for this species (Price-Rees et al. 2013). We randomly determined the order in which we tested lizards. We used a pole with metal prongs to attach and move model predators towards the lizards during four stages of predation in which predation risk was escalated. All trials were filmed. In stage 1, we placed the lizard in the arena, beneath a bucket, and subsequently introduced the static model predator 1.5-m away from the lizard. After 2 min, we removed the bucket and recorded the lizards’ responses for 30 s. In stage 2, we slowly moved the predator 75 cm towards the lizard at a constant speed (25 cm every 10 s). In stage 3, we abruptly moved the predator closer (50 cm in 1 s) to simulate a single attack and left the predator immobile for 2 min approximately 20 cm from the lizard. In stage 4, we simulated intensive attacks by repeatedly and rapidly presenting the predator to the lizard (no physical contact) for 15 s. From the recorded footage, we scored the number of tongue flicks, full-tongue displays (full-tongue displays are easily identifiable and cannot be mistaken with tongue-flicks), and all other anti-predatory behaviour (hissing, body inflation, lateral presentation), using the software JWatcher (Blumstein et al. 2006). During video analyses, observed tongue-flicks performed during stages 3 and 4 (i.e. immediately before or after an attack) differed in comparison to tongue-flicks displayed when the predator is farther away. Tongue flicks during stages 3 and 4 were directed towards the predator and exposed a greater proportion of the tongue. We therefore compared how much tongue area was exposed during normal chemosensory tongue-flicks (i.e. when the predator is not attacking) compared to during predatory encounters (i.e. when a predator attacks). For all recorded tongue-flicks, we took a good still frame of the tongue at its maximum expansion during a tongue-flick and used ImageJ (Schneider et al. 2012) to estimate the area of the tongue extruded during tongue-flicks. We used the most adequate of the three known head measures of each lizard (i.e. length, width, and depth) as a reference scale depending on the angle with which the tongue was presented in the picture.
We analysed the data using the lme4 package (Bates et al. 2014) and multcomp package (Hothorn et al. 2008) in R v.3.1.1 (R Development Core Team 2014). To explore differences in tongue colour, we used linear mixed models including luminance, λmax, and chromatic and achromatic contrasts using both the lizard and the avian visual systems as response variables, region of the tongue as a fixed factor, and individual identity as a random factor. The distributions of the variables for luminance and achromatic contrast (bird and lizard visual system) were right-skewed and were thus log-transformed to achieve normality. The variable λmax was left-skewed, so we used a fourth-root transformation for the same reason (Quinn and Keough 2002). We then performed post hoc Tukey tests to compare each tongue region. Finally, we used independent t tests to compare chromatic and achromatic contrasts between a bird and a lizard visual system.
To explore the differences in full-tongue display across stages and predator treatments, we used linear mixed models. No full-tongue display occurred during stages 1 and 2, so we only considered stages 3 and 4. As a response variable, we used the number of full-tongue displays corrected by time (seconds). This variable was left-skewed, so we used a fourth-root transformation to conform to normality. As fixed factors, we included stage, predatory treatment, and the interaction between both, and we added lizard identity and trial order as random factors. When the interaction was not significant, we considered each term separately. We used post hoc Tukey tests to detect differences between treatments and/or stages.
To explore differences in tongue-flick rate and tongue area exposed during tongue-flicks in stages 3 and 4, we ran two linear mixed models with tongue area and tongue-flick rate as the response variables. We included stage, predatory treatment, and the interaction between both as fixed factors and trial order and lizard identity as random factors. We also performed post hoc Tukey tests to explore the differences between each predatory treatment. Our tongue area variable was previously alpha-winsorized (∝ = 0.02), that is, transformed to limit extreme values to avoid undue influence of outliers. When the interactions were significant in our models, we used the same models within each stage (3 and 4), discarding the fixed factor stage. During behavioural assays, lizards exhibited several other typical anti-predatory behaviours, including hissing, body inflation, and lateral presentation. We created a variable grouping the frequencies of these defensive behaviours, and because this variable was left-skewed, we performed a fourth-root transformation to conform to normality. We used it as a response variable in linear mixed models, to which we added predatory treatment, stage, and the interaction between both as fixed factors, and trial order and lizard identity as random factors. Because the interaction was not significant, we discarded it from the model and only considered stage and predatory treatment as fixed factors. We then performed post hoc Tukey tests to explore the differences between predatory treatments.
Results from linear mixed models and post hoc Tukey tests exploring how luminance, λmax, chromatic contrast for a lizard’s (∆S-lizard) and bird’s (∆S-bird) visual system, and achromatic contrast for a lizard’s (∆L-lizard) and a bird’s (∆L-bird) visual system vary between different regions of the tongue (back, middle, and tip). Significant p values are represented in italics
Estimate ± se
χ2, p value
Post hoc Tukey tests
Luminance ~ tongue region + (1|lizard ID)
Back (int): 3.128 ± 0.134
Mid: − 0.354 ± 0.077
Tip: − 0.915 ± 0.077
χ2(2) = 48.556
p < 0.0001
Mid–Back: p < 0.0001
Tip–Back: p < 0.0001
Tip–Mid: p < 0.0001
λmax ~ tongue region + (1|lizard ID)
Back (int): 4.256 ± 0.002
Mid: − 0.001 ± 0.002
Tip: 0.002 ± 0.002
χ2(2) = 2.162
p = 0.339
Mid–Back: p = 0.959
Tip–Back: p = 0.523
Tip–Mid: p = 0.361
∆S-lizard ~ tongue region + (1|lizard ID)
Back (int): 14.421 ± 0.496
Mid: − 0.319 ± 0.565
Tip: 0.415 ± 0.565
χ2(2) = 1.782
p = 0.410
Mid–Back: p = 0.839
Tip–Back: p = 0.743
Tip–Mid: p = 0.396
∆L-lizard ~ tongue region + (1|lizard ID)
Back (int): 1.357 ± 0.174
Mid: 0.579 ± 0.153
Tip: 1.273 ± 0.153
χ2(2) = 34.259
p < 0.0001
Mid–Back: p = 0.000401
Tip–Back: p < 0.0001
Tip–Mid: p < 0.0001
∆S-bird ~ tongue region + (1|lizard ID)
Back (int): 17.618 ± 0.678
Mid: − 0.469 ± 0.405
Tip: 0.760 ± 0.905
χ2(2) = 1.968
p = 0.374
Mid–Back: p = 0.862
Tip–Back: p = 0.678
Tip–Mid: p = 0.363
∆L-bird ~ tongue region + (1|lizard ID)
Back (int): 1.357 ± 0.173
Mid: 0.579 ± 0.153
Tip: 1.273 ± 0.153
χ2(2) = 34.259
p < 0.0001
Mid–Back: p = 0.000424
Tip–Back: p < 0.0001
Tip–Mid: p = 0.0001
Behavioural responses to predators
Results from a linear mixed model aiming to detect how predatory treatment and stage affect the occurrence of full-tongue displays. The last column corresponds to the output of post hoc Tukey test applied to the variable predatory treatment. Levels of significance are indicated after p values (. < 0.1, * < 0.05)
Fixed factor (χ2(df), p value)
Estimate ± se
Post hoc Tukey tests (estimate ± se, p value)
Full-tongue display ~ treatment × stage + (lizard ID) + (trial)
Treatment × stage (χ2(4) = 7.407, p = 0.116)
Intercept: − 2.170 ± 0.845
Fox: 0.364 ± 1.181
Monitor: 1.743 ± 1.181
Control: 1.829 ± 1.206
Snake: 1.592 ± 1.181
Stage: 0.831 ± 0.236
Fox × stage: − 0.001 ± 0.334
Monitor × stage: − 0.625 ± 0.334
Control × stage: −0.646 ± 0.341
Snake × stage: − 0.596 ± 0.334
Fox – Bird: 0.015 ± 0.169, p = 1
Monitor–Bird: − 0.442 ± 0.169, p = 0.068 .
Control–Bird: − 0.430 ± 0.174, p = 0.096 .
Snake–Bird: − 0.492 ± 0.169, p = 0.030 *
Monitor–Fox: − 0.457 ± 0.170, p = 0.0543.
Control–Fox: − 0.445 ± 0.174, p = 0.078 .
Snake–Fox: − 0.507 ± 0.169, p = 0.023 *
Control–Monitor: 0.012 ± 0.174, p = 1
Snake–Monitor: − 0.050 ± 0.169, p = 0.998
Snake–Control: − 0.062 ± 0.174, p = 0.997
Treatment (χ2(8) = 24.688, p = 0.002 *)
Stage (χ2(5) = 23.706, p < 0.001 *)
Results from the linear models and post hoc Tukey tests exploring differences in tongue area exposed during tongue-flicks, tongue-flick rate, and frequencies of all other anti-predatory behaviours across predatory treatment and stages. Levels of significance are indicated (. < 0.1, * < 0.05, ** < 0.01, *** < 0.001)
Fixed factor(χ2(df), p value)
Estimate ± se
Post hoc Tukey tests (estimate ± se, p value)
Tongue-flick area ~ treatment × stage + (lizard ID) + (trial)
Treatment × stage (χ2(4) = 20.601, p < 0.001 ***)
Intercept: − 0.334 ± 0.206
Stage-4: 0.543 ± 0.286
Fox: 0.709 ± 0.182
Monitor: 0.017 ± 0.181
Control: − 0.050 ± 0.221
Snake: 0.088 ± 0.148
Stage-4 × fox: − 0.995 ± 0.391
Stage-4 × monitor: − 0.721 ± 0.382
Stage-4 × control: − 1.055 ± 0.385
Stage-4 × snake: 0.114 ± 0.349
Fox–Bird: 0.388 ± 0.158, p = 0.099 .
Monitor–Bird: − 0.156 ± 0.154, p = 0.847
Control–Bird: − 0.552 ± 0.162, p = 0.006 **
Snake–Bird: 0.192 ± 0.132, p = 0.588
Monitor–Fox: −0.545 ± 0.150, p = 0.003 **
Control–Fox: −0.940 ± 0.165, p < 0.001 ***
Snake–Fox: − 0.197 ± 0.155, p = 0.707
Control–Monitor: − 0.397 ± 0.161, p = 0.099.
Snake–Monitor: 0.348 ± 0.143, p = 0.104
Snake–Control: 0.744 ± 0.153, p < 0.001 ***
Treatment (χ2(8) = 60.434, p < 0.001 ***)
Stage (χ2(5) = 20.747, p < 0.001 ***)
Tongue-flick rate ~ treatment × stage + (lizard ID) + (trial)
Treatment × stage (χ2(4) = 3.798, p = 0.434)
Intercept: 0.201 ± 0.051
Fox: − 0.074 ± 0.069
Monitor: − 0.042 ± 0.069
Control: − 0.050 ± 0.071
Snake: 0.074 ± 0.069
Stage-4: − 0.176 ± 0.068
Fox × stage-4: 0.068 ± 0.097
Monitor × stage-4: 0.067 ± 0.097
Control × stage-4: 0.102 ± 0.099
Snake × stage-4: − 0.062 ± 0.097
Treatment (χ2(8) = 6.879, p = 0.550)
Stage (χ2(5) = 23.933, p < 0.001 ***)
Anti-predatory behaviours ~ treatment + stage + (lizard ID) + (trial)
Treatment (χ2(4) = 9.576, p = 0.048 *)
Intercept: 0.650 ± 0.203
Fox: − 0.193 ± 0.165
Monitor: − 0.169 ± 0.165
Control: − 0.313 ± 0.171
Snake: 0.162 ± 0.164
Fox–Bird: − 0.193 ± 0.165, p = 0.768
Monitor–Bird: − 0.169 ± 0.165, p = 0.843
Control–Bird: − 0.313 ± 0.171, p = 0.357
Snake–Bird: 0.162 ± 0.164, p = 0.862
Monitor–Fox: 0.024 ± 0.167, p = 0.999
Control–Fox: − 0.120 ± 0,171, p = 0.956
Snake–Fox: 0.355 ± 0.163, p = 0.191
Control–Monitor: − 0.144 ± 0.172, p = 0.919
Snake–Monitor: 0.331 ± 0.165, p = 0.266
Snake–Control: 0.475 ± 0.170, p = 0.042 *
Stage (χ2(1) = 50.205, p < 0.001 ***)
Anti-predatory behaviours were not affected by the interaction between stage and predatory treatments (χ2(4) = 4.139, p = 0.388). Therefore, we considered the effects of stage and treatment separately. We found significant differences between treatments and stage with anti-predatory behaviours being more frequent in the presence of a snake than control predatory treatment, and that they were more frequent during stage 4 than stage 3 (Table 3).
Deimatic displays have remained elusive to address, most likely because of the difficulty with studying the behaviour of a receiver in response to standardised deimatic displays, in a controlled environment. Although we did not examine predator (i.e. receiver) responses to lizard display behaviour, our results are consistent with the predictions of deimatic display theory. First, UV-blue tongue colouration is particularly conspicuous to an avian visual model and exhibits a marked gradation in luminance such that luminance at the rear of the tongue, only visible during full-tongue displays, is almost twice as high as the tip of the tongue.
Second, this tongue colouration pattern, which is primarily concealed, allows for a sudden transition between an inconspicuous state (i.e. a lizard with its mouth closed or extruding only the tongue tip during chemosensory tongue-flicking) and a highly conspicuous display (i.e. a lizard exhibiting a full-tongue display; Umbers et al. 2017). Additionally, lizards restricted the use of full-tongue displays to the final stages of a predation sequence, when they were at most risk (i.e. subjugation would follow), and did so in concert with aggressive defensive behaviours that amplify the display (i.e. hissing, lateral posture, body inflation; see Vallin et al. (2005) for a similar multi-component display in a butterfly). This late deployment of the display is predicted to enhance the startle effect on a potential predator. The close proximity would also likely amplify the effect of overwhelming the sensory system of the receiver.
The timing of this display is crucial. If performed too early, a display may break the lizard’s camouflage and attract unwanted attention by predators and increase predation risk. If performed too late, it may not deter predators (Bateman et al. 2014). Lizard responses, therefore, appeared timed in accordance with this prediction. The timing of this display is in contrast to that expected for pursuit deterrent signals. Pursuit deterrent signals, common in other lizards, are deployed earlier to signal detection by receivers and deter the predator from further pursuit (e.g. Font et al. 2012). Also, in contrast to aposematic signals, deimatic displays are deployed at a specific moment during a predator encounter while aposematic signals are displayed continuously (although there are some exceptions). The crucial difference between aposematic signals advertising prey unprofitability and deimatic displays is that the latter does not require learning from the predator while the former does (i.e. negative reinforcement). The effectiveness of deimatic displays depends on startling and/or overloading the sensory system of the predator; this effect would be lost if predators learned to ignore it (Umbers and Mappes 2016; Umbers et al. 2017). In very rare instances, bluetongue skinks lunged and bit the attacking predator, which opens the possibility that full-tongue displays are in fact aposematic and signal prey unprofitability, in that the predator could get hurt during the predation event. However, given that in the great majority of cases the lizards did not attack the predator at all, it is unlikely that predators would learn to avoid preying on them. Attempting to bite a predator is likely the next sequence of an anti-predator response in the event that a deimatic display fails to deter a predator.
Similar to previous studies on other lizard species (e.g. Abramjan et al. 2015; AB personal observations among lacertid lizards), T. s. intermedia in our study had a maximum reflectance peak of 328 nm on their tongues. These other lizard species are not known to use their tongue during displays; yet, they have the same reflectance peak (i.e. ~ 330 nm), and some of them also have a secondary peak in the red part of the spectrum. However, this applies to only a handful of species and the tongue colour of more lizard species needs to be characterised for a better appreciation of how common this UV pattern is in lizards. In addition, UV-sensitive photoreceptors have a maximum absorbance peak of 360–370 nm in lizards (Loew et al. 2002; Fleishman et al. 2011) and 367 ± 6 (SE) nm in birds (Hart 2001, 2002; Endler and Mielke 2005). Although lizards and birds are very likely able to see this UV pattern, this mismatch between tongue reflectance peak and both lizard and bird visual systems is unusual in species that use UV colours as displays (e.g. Leal and Fleishman 2004; Pérez i de Lanuza et al. 2014). For these reasons, it seems unlikely that the UV colouration of the tongue evolved primarily as a visual display to conspecifics or predators. A possible explanation is that UV reflecting tongues emerged as a by-product of tongue histology driven by other functions, such as the mechanical role played by collagen fibres (Prum and Torres 2003). This UV reflecting surface therefore may have been exapted as a component of a visual deimatic display. Comparative studies examining the co-variation of lizard tongue colouration and anti-predatory displays would be invaluable for testing this hypothesis.
As a result of our post hoc analysis of tongue-flicks, we found that tongue-flicks performed immediately after a predatory attack by a model predator (i.e. stage 3 and 4) exposed a greater area of the tongue, albeit at a lower tongue-flick rate than they did when the predator was not attacking (e.g. stages 1 and 2). Furthermore, during a predatory attack, tongue-flicks were exhibited in association with other anti-predatory behaviours (i.e. hissing, body inflation, and lateral posture) directed towards the predator. While tongue-flicks during exploration and foraging should intuitively be under selection to be less conspicuous (e.g. by exposing only the less-conspicuous tip of the tongue), the role of increasing tongue exposure following an attack from a predator is less clear. Here, we propose three working hypotheses that should be considered by future studies aiming to explain the function of such behaviour. First, increasing the tongue area during tongue-flicks may have a chemosensory function. Lizards may expose a larger area of their tongue to acquire more chemical information about the attacking predator (Schwenk 1995), although the lower tongue-flick rate seems to contradict this hypothesis. There is evidence to suggest that both snakes (Shine et al. 2003) and lizards (Carazo et al. 2007) can extract complex information with just a few flicks of their tongue. Subsequent tongue-flicking may be related to localisation of chemical stimuli sources rather than actual discrimination (Cooper 1998). Given that the localisation of a predator is obvious immediately after an attack, this also does not support the ‘chemosensory’ hypothesis. Second, these tongue-flicks may act as visual signals to threaten the predator of an imminent defensive attack from the lizard, as it is the case in some snake species (Gove 1979). Indeed, in rare instances during our behavioural assays, lizards lunged with an open mouth towards the predator (i.e. bite) in response to the predatory threat. Similarly, they could act as visual pursuit-deterrent signals, which would explain why the exposed tongue area only increased after an attack the lizard successfully ‘escaped’ from. Third, a lower tongue-flick rate in the presence of a predator could be a way to avoid chromatic adaptation and thus preserve the startling effect of full-tongue displays until necessary. However, the fact that a larger part of the tongue is exposed contradicts this hypothesis, especially given the colour pattern of the tongue. Lastly, the observed variation in tongue flicks (i.e. rate and area of tongue exposure) may not have a specific adaptive function, and the observed differences may be unrelated to any chemosensory, communicative, or anti-predator function. For example, responses could be a consequence of an altered psychological state caused by the simulated attack or a mechanical side-effect of full-tongue displays on tongue flicks (e.g. tongue muscles may become overstretched in such a way that tongue flicks expose more of it). While it is worth considering these alternative hypotheses, none of them have any significant support at this stage.
In conclusion, we have established that full-tongue displays in this species are consistent with a deimatic display at least in their production mechanisms and suggest that the next logical step for future studies would be to focus on receiver responses. We suspect that such a highly conspicuous display deployed at close range to a predator will induce a reflexive startle response that will deter predators. We also hypothesise that this type of display will be particularly effective against aerial predators, for which an interrupted attack would not be easily resumed due to loss of inertia. Deimatic displays may be more common in lizards than previously believed, and future studies which focus on receivers and their sensory system will be invaluable in contributing to our understanding of how deimatic signals have evolved and how they are maintained. As Umbers et al. (2017) emphasise, deimatic signals are not well understood, and we need more empirical studies to test predictions and refine associated theory underpinning this unique behaviour.
We are grateful to Grant Napier for his invaluable field assistance, Sarah Pryke for loaning us a reflectance spectrophotometer, and Bill Stewart and Corrin Everitt for making their property available for our study. We also thank two anonymous reviewers for improving this manuscript.
MJW and PC conceived the study. SJPR, PC, and MJW conducted the experiments and collected the morphological and colour measurements. AB scored the behaviours. MFB took the tongue measurements. AB and PC carried out the statistical analyses. AB, PC, and MJW drafted the manuscript, and all authors provided feedback.
This work was supported by funding to MJW from Macquarie University. AB was funded by an iMQRES doctoral scholarship awarded by Macquarie University (2014166), and PC was funded by an Endeavour fellowship.
Compliance with ethical standards
For the handling of animals, we followed the ABS (Animal Behavior Society)/ASAB (Association for the Study of Animal Behaviour) ‘Guidelines for the treatment of animals in behavioural research and teaching’. Our research protocols were approved by the Macquarie University Animal Ethics Committee and University of Sydney Animal Care and Ethics Committee, Parks and Wildlife Commission of the Northern Territory, and the Western Australian Department of Environment and Conservation.
Conflict of interest
The authors declare that they have no conflict of interest.
- Bates D, Maechler M, Bolker B, Walker S (2014) lme4: linear mixed-effects models using Eigen and S4. R package version, 1(7). https://CRAN.R-project.org/package=lme4
- Blumstein DT, Evans CS, Daniel JC (2006) JWatcher 1.0. http://www.jwatcher.ucla.edu
- Bradbury JW, Vehrencamp SL (2011) Principles of animal communication, 2nd edn. Sinauer Associates, SunderlandGoogle Scholar
- Dutson G, Dutson L (2016) Microhabitat niche differentiation in sympatric eastern blue-tongued lizard Tiliqua scincoides and blotched blue-tongued lizard Tiliqua nigrolutea in Melbourne, Victoria. Vic Nat 133:55–58Google Scholar
- Edmunds M (1974) Defence in animals: a survey of anti-predator defences. Longman, HarlowGoogle Scholar
- Fitzsimons JA (2011) Predation on a blotched bluetongue lizard (Tiliqua nigrolutea) by a highlands copperhead (Austrelaps ramsayi) in the Blue Mountains, Australia. Herpetol Notes 4:259–260Google Scholar
- Gove D (1979) A comparative study of snake and lizard tongue-flicking, with an evolutionary hypothesis. Ethology 51:58–76Google Scholar
- Maia R, Eliason CM, Bitton PP, Doucet SM, Shawkey MD (2013) Pavo: an R package for the analysis, visualization and organization of spectral data. Methods Ecol Evol 4:906–913Google Scholar
- Nielsen TP, Bull CM (2016) Impact of foxes digging for the pygmy bluetongue lizard (Tiliqua adelaidensis). Trans R Soc S Aust 140:228–233Google Scholar
- Shine R, Phillips B, Waye H, LeMaster M, Mason RT (2003) Chemosensory cues allow courting male garter snakes to assess body length and body condition of potential mates. Behav Ecol Sociobiol 54:162–166Google Scholar