The colour of success: does female mate choice rely on male colour change in the chameleon Furcifer pardalis?

Animal colouration is not as fixed as is often assumed. Indeed, many animals change colour during their development (Booth, 1990) and/or in response to environmental variation, including season (Küderling et al., 1984), food availability (Hill et al., 2002), circadian rhythms, breeding season (Keren-Rotem et al., 2016b; McGraw and Hill, 2004) or predation pressure (Hemmi et al., 2006). This type of colour change, termed morphological colour change (Leclercq et al., 2009; Umbers et al., 2014), can be achieved through the anabolism or catabolism of pigments or photonic structures, changes in the number of chromatophores in tissues, or the renewal of dead tissue (e.g. hairs, feathers, scales or cuticula) containing pigments or photonic structures through moulting (Detto et al., 2008). These morphological colour changes are typically achieved over time spans ranging from months to days.

However, some organisms can change colour in hours or even less than a second. This is referred to as rapid colour change or physiological colour change (Ligon and McCartney, 2016). Physiological colour changes can be achieved by the contraction/expansion of chromatophores (Cloney and Florey, 1968), by the mobilisation of pigments or photonic nanostructures within chromatophores, or through hydraulic infiltration into photonic nanostructures (Liu et al., 2009). These types of colour changes are typically mediated by hormones and/or neurotransmitters (Ligon and McCartney, 2016; Umbers et al., 2014). This ability has been documented in cephalopods (Hanlon and Messenger, 2018), insects (Hinton and Jarman, 1973; Key and Day, 1954), arachnids (Wunderlin and Kropf, 2013), crustaceans (Brown and Sandeen, 1948; Stevens et al., 2014), fish (Iga and Matsuno, 1986; Nilsson Sköld et al., 2013) amphibians (Kindermann and Hero, 2016; Nilsson Sköld et al., 2013), reptiles (Batabyal and Thaker, 2017; Taylor and Hadley, 1970) and even some birds (Curio, 2004). The functions of physiological colour change may differ in different taxa and include thermoregulation (Smith et al., 2016), hydroregulation (Whiters, 1995), camouflage (Allen et al., 2010; Stuart-Fox et al., 2008; Zylinski and Johnsen, 2011) and intraspecific communication (Hutton et al., 2015).

In the context of animal communication, colour change has been studied mostly during intrasexual interactions, suggesting that intraspecific communication is a prominent driver of the evolution of physiological colour change. Surprisingly, few studies have investigated the role of physiological colour change during intersexual interactions (Adamo et al., 2000; Batabyal and Thaker, 2017; Boal, 1997; Kelso and Verrell, 2002; Keren-Rotem et al., 2016a). Moreover, most of these studies focused on differences in colour change between intra- and intersexual interactions, or which colours correspond to courtship, rather than evaluating the variations in colour during the interaction itself. To date, no study has explored the role of physiological colour change in mate choice, raising questions on whether females choose mates based on male physiological colour change.

In some animals, like chameleons, sexual selection upon the ability to change colour is likely to occur since a strong sexual dimorphism both in terms of colour and the ability to change colour exists (Kelso and Verrell, 2002; Keren-Rotem et al., 2016a; Tolley and Herrel, 2013). Moreover, comparative studies have demonstrated that selection for conspicuous social signals has likely driven the evolution of colour change in a clade of dwarf chameleons (Stuart-Fox and Moussalli, 2008). Chameleons are thus an excellent model to study whether females chose mates based on male physiological colour change. Chamaeleonid lizards are famous for exhibiting striking and complex colour changes (Nečas, 1999; Teyssier et al., 2015; Tolley and Herrel, 2013) with a large repertoire (Kelso and Verrell, 2002; Ligon and McGraw, 2018). Despite previous studies describing the specific courtship colour patterns and female receptive colours (Karsten et al., 2009; Kelso and Verrell, 2002; Keren-Rotem et al., 2016a), investigations into involvement of colour change in intersexual selection in chameleons have been neglected.

Here, we explore whether mate choice outcome relies on specific aspects of male colour change in the panther chameleon Furcifer pardalis (Cuvier 1829). The panther chameleon is an ideal study species as it possesses a strong sexual dichromatism with males being brightly coloured and females being dull. Female colouration remains constant between populations while a strong variability is observed in males depending on their region of origin (Ferguson, 2004). We conducted sequential mate choice experiments and investigated whether female mate choice relies on male colour change. We predict mate choice outcome should be related to brightness changes as these changes have been shown to be involved in intrasexual interactions in chameleons (Ligon, 2014; Ligon and McGraw, 2016, 2013, 2018).

Furcifer pardalis is endemic to Madagascar and found in a wide range of habitats along the northern and eastern coasts. They are diurnal tree-dwelling lizards living in relatively intact forests and forest edges, gardens, plantations and degraded habitats. This species exhibits a strong sexual dimorphism and exceptionally large intraspecific variation in male colouration: females and juveniles are tan to brown with hints of pink or orange, while adult males are much larger and have various combinations of bright red, green, blue, and yellow. This polychromatism among males, also referred to as morphs or localities, depends on the region of origin. Local variation also appears to exist within morphs (Ferguson, 2004). It has even been suggested that most of the colour morphs in F. pardalis could be considered as separated subspecies (Grbic et al., 2015). For the present study, 28 adult (over 6 months old) captive-bred F. pardalis ‘Ambilobe’ morphs were used (N_male=19; N_female=9).

Animals were kept in a dedicated room in the Parc Zoologique de Paris in mesh terraria (46×46×91 cm, ReptiBreeze, ZooMed) outfitted with branches and plastic plants to provide hiding spots. The room temperature was maintained at 26°C and fluorescent tubes, providing 12% UVB (Reptile Lamp 12% T8, Arcadia) and a 40 W heating bulb (Repti Basking Spot, ZooMed) were suspended above each cage. The photoperiod was set at a 12 h:12 h light:dark cycle. Animals were fed thrice weekly and crickets were calcium-dusted once a week. Water was provided to the animals during three daily misting periods (09:00 h, 12:00 h and 16:00 h) using an automated misting system (Vivaria project) and 200 ml drippers. Males were individually housed, while females were kept in groups of 2–3 individuals, but all terraria were visually isolated from one another.

Experiments were carried out in compliance with French legislation and animals were given regular health checks by zoo veterinarians. Animals were alive and healthy after the experiments and showed no weight loss. In accordance to the directive 2010/63/EU of the European Parliament and French legislation, our study did not require specific authorisation because our observations did not cause any pain, suffering, distress or lasting harm.

We used a large arena (144×50×80 cm) with opaque Plexiglas sides and a front made of transparent Plexiglas 50 cm high to allow behavioural observations and photo/video recordings (Fig. 1). Chameleons are arboreal, so we provided artificial branches to mimic an arboreal environment. As chameleons have a spectral sensitivity that includes the UV (Bowmaker et al., 2005), we used a combination of different light sources, allowing us to cover a spectrum close to the solar spectrum, including the UV. The overall set up was illuminated with a combination of nine light sources placed 56 cm above the set-up: two 60 W, 4000 K, 806 lm LED bulbs (Lexman), two 60 W, 2700 K incandescent bulbs (OSRAM, Munich, Germany), two 100 W, 2800 K, 1320 lm halogen bulbs (OSRAM), one UVB fluorescent tube ReptiSun 10.0 High Output UVB Bulb (ZooMed Laboratoires, San Luis Obispo, CA, USA), one Arcadia T5 D3+ Desert 12% Reptile Fluorescent Lamp (Arcadia, Croydon, UK), and one Reptile systems New Dawn T5 LED (Aquariums systems, Sarrebourg, France).

From April to September 2018, sequential mate choice experiments were performed to assess mate choice outcomes. This experimental design mimicked the reproductive behaviour of chameleons, as perch-dwelling lizards are likely to approach mates in a sequential manner during the mating season. Males were split into four different pools of four to five individuals and females in four pools of two to three individuals. To each pool of females, we assigned one specific pool of males from a different breeder with which the sequential mate choice experiment was run. Every female from each pool was exposed to all males of the paired pool (i.e. 4–5 males). The sequential mate choice experiment was repeated four times per female. Sequential mate choice was repeated because mate guarding has been observed in chameleons (Cuadrado, 2006). Consequently, it is likely that females copulate several times with the same chosen male. However, if a female engaged in copulation, meaning that it was potentially gravid, the experiment was not repeated more than twice to prevent potential stress-related dystocia (DeNardo, 2005). Potential gravid females were isolated in specific terraria and allowed to lay their eggs. The mate choice experiments were performed at room temperature (26°C) from 10:00 h to 18:00 h, corresponding to their daily active period. The arena was sprayed with clear water and cleaned before each trial to prevent potential effects of odours remaining from a previous trial. The animals were able to interact for 1 h, but experiments were halted if the male attempted to attack the female. Each female was exposed to a maximum of three different males per day and individuals were allowed to rest for at least 90 min between each trial. The mate choice experiment was repeated, if necessary, with at least a week interval. The behaviour of the animals was recorded with an HD camera, HDCR-CX740VE (SONY, Minato-ku, Tokyo, Japan) (Fig. 1).

During the interactions, pictures of the male were taken twice every 2 min with a full spectrum converted camera (Samsung NX-1000), one picture in the visible spectrum (VIS) and one in the ultraviolet spectrum (UV). For pictures in the visual spectrum, a filter blocking ultraviolet and infrared was manually placed in front of the camera (UV/IR cut/L Filter, Baader, Mammendorf, Germany) and pictures were taken with a 1/640 s exposure. Immediately after the picture was taken, the filter was changed for a filter blocking all wavelengths except those ranging from 320 to 400 nm (Venus-U Planetary Filter, Optolong, Kunming City, China) and a picture was taken with a 1 s exposure.

For colour calibration of pictures in the visual range, pictures of a colour checker (SpyderCHECKR) placed at 13 different regions of the arena were taken once empty. Colour calibration was performed using Adobe Photoshop Lightroom 6 and SpyderCHECKR software (v.1.2.2). As individuals might be in different regions of the arena, images were cropped to isolate each individual and then according to the position of the individuals in the arena, the corresponding colour calibration was applied. Calibrated images were then used for colour measurements. As the UV filter imposes a narrow hue range resulting in a constant pink colouration, colour calibration was not required.

Colour measurements were performed by retrieving RGB values, using the RGB measure ImageJ plugin. Here, 10 squares of 16 pixels (N_VIS=5; N_UV=5; Fig. 2), describing the specific colour patterns of male F. pardalis were quantified: the bands (N=2), interbands (N=2) and lateral line (N=1). In the UV, we considered the bands, eyelid, and the head bony tubercles as colour patterns ‘absorbing’ UV (N=3), the highly ‘reflecting’ lips (N=1), and the lateral line (N=1) which either absorbs or reflects UV. RGB values were then compiled in R (https://www.r-project.org/) and converted into HSV values (H: hue, S: saturation, V: brightness) using the Colorscience package (https://cran.r-project.org/web/packages/colorscience/index.html). HSV is an alternative representation of the RGB colour model that aligns with colour-making attributes and colour perception. As pictures were taken every 2 min, each picture corresponds to a time step of 2 min with the first picture of each interaction corresponding to t=0 min.

where x represents a color value (H, S, V or DEHSV); t_min represents the beginning of the interaction; t_max represents the end of the interaction, t_xmax is the time at which x reaches its maximum value; t_xmin is the time at which x reaches its minimum value; x_tmax is the x value at the end of the interaction; and x_tmin is the x value at the beginning of the interaction. In Eqn 1, H is the hue value, S is the saturation value and V is the brightness value. For colour measurements, across all sequential mate choice experiments, we only included sequential mate choice experiments, during which the female exhibited receptive colouration in her resting state. Hence, out of the 169 interactions, we quantified images from 56 interactions only.

Here, we were unable to model the colour vision of the panther chameleon because the raw data required to calculate the cone-catch values for visual system modelling (Siddiqi, 2004; Troscianko and Stevens, 2015; Vorobyev and Osorio, 1998) are no longer available for F. pardalis (Jim Bowmaker, personal communication).

For each of these 56 interactions, female mate choice was established based on a previous study on male–female interactions in other Furcifer spp. (Karsten et al., 2009) and personal observations (Table 1). Based on this, the mate choice observed during those 56 interaction was classified into one of two categories: the male is ‘selected’ (n=27) [Table 1, preference index 2–6] or ‘non-selected’ (n=29) (Table 1, preference index 0 and 1) by the female. In the case of selected males, only images prior to copulation or copulation attempts were analysed because female mate choice should be based on male colours seen before choosing to mate with a male.

Before analyses, the distribution of each colour change variable was transformed where needed using a Box-Cox power function with ‘AID’ package to meet the requirements of normal distribution (Asar et al., 2017).

We summarised colour change information using a principal components analysis (PCA) on the centered and scaled individual values of each colour change variable with the ‘ade4’ package (Dray and Dufour, 2007). For the visible range (VIS), all variables were incorporated in the PCA; however, for the UV range only brightness and saturation were used as the filter imposed a constant hue. The number of principal components (PCs) used for the subsequent analyses was identified using the broken stick method (Legendre and Legendre, 1998) while accounting for at least 70% of the variability in the data set and for which the contribution of each variable to PCs provided a relevant interpretation of the colour change.

To determine whether the male colour change exhibited during an intersexual interaction could explain female mate choice, a generalised linear mixed-effects model (GLMM) fit with a binomial error (‘glmer’) using the ‘lme4’ package (Bates et al., 2015) was performed. We tested for the effect of colour change (i.e. PCs and maximum brightness and saturation for the UV range) either in the visible range or in the UV range, on female mate choice, considering the female and the order of males as random factors. These were tested for each body region separately because we found a significant effect of the body region on most of the colour change variables (P<0.01) in both the visible and UV range (Table S1).

We performed model selection based on Akaike's information criterion (AICc) with ΔAICc≥2, and evaluated the relative importance (RI) of each predictor variable using model-averaging approaches (Burnham et al., 2011) within model sets for each body region from each spectral range. All statistical analyses were conducted in R v.3.6.1 (https://www.r-project.org/). Additionally, model selection and model averaging were undertaken using the R package MuMIn (https://cran.r-project.org/web/packages/MuMIn/index.html).

In the visible spectrum, the first four principal components accounted for 75.72% of the total variability. The first principal component (PC1) of our PCA described hue changes and PC2 depicted saturation changes (Table 2). PC3 described brightness changes and PC4 described the overall absolute colour change (Table 2). In the UV, the first three principal components accounted for 73.68% of the total variability. PC1 represented UV brightness changes, whereas PC2 described UV saturation changes. PC3 depicted the overall absolute UV colour change (Table 2).

In the visual spectrum, model selection by AICc retained the model that included PC1 (i.e. hue changes), PC2 (i.e. saturation changes) and PC4 (i.e. overall absolute colour change) for the bands (Table S2). Multimodel averaging on the band models did uncover PC1 (RI=94%), PC2 (RI=95%) and PC4 (RI=97%) to be the best predictors of mate choice outcome (Fig. 4D). Specifically, males that exhibited fewer hue changes (PC1) but also more saturation changes (PC2) and that showed a higher overall absolute colour change were significantly more likely to be selected by females, thus having a higher mating probability (Fig. 4A, Fig. 5A). For the lateral line, model selection by AICc selected the model with PC2 and PC3 (i.e. brightness change; Table S2). Multimodel averaging for the lateral line models did uncover PC2 (RI=99%) and PC3 (RI=83%) to be the best predictors of female mate choice (Fig. 4D). More specifically, we found that males exhibiting fewer brightness changes, but more saturation changes were more likely to mate (Fig. 4C, Fig. 5C). However, the interband model selection by AICc did not allow us to discriminate between models (Table S2). Consequently, we relied on the multimodel averaging to retrieve the best predictors for the interband model. According to multimodel averaging, the best predictors were PC2 (RI=65%), PC3 (RI=67%) and PC4 (RI=74%) (Fig. 4D). This model showed that males with a higher overall absolute colour change (PC4) are selected by females (Fig. 4B, Fig. 5B). We also found that males exhibiting more brightness changes (PC3) tend to be selected less often by the females. At the level of the interbands, there seems to be no link between female mate choice and saturation changes (PC2).

In the UV, at the level of the lateral line and UV-absorbing regions, model selection by AICc selected PC1 (i.e. UV brightness changes), PC2 (i.e. UV saturation changes) and UV maximum brightness (maxV; Table S3). Multimodel averaging selected those variables as best predictors of female mate choice (Fig. 6D) for the following regions: UV-absorbing regions: PC1 RI=95%, PC2 RI=100%, maxV RI=89%; lateral line: PC1 RI=85%, PC2 RI=92%, maxV RI=79%. Our results show that selected males are those which exhibited more UV brightness (PC1) and UV saturation changes (PC2), yet lower UV brightness (Fig. 6A,C, Fig. 7A,C). At the level of the UV reflecting regions, model selection by AICc failed to discriminate among models. Hence, we used the best predictors uncovered by model averaging, PC1 and PC2. At the level of the UV-reflecting region, as in the UV-absorbing region and the lateral line, males were more likely to be selected by the females when they exhibited more brightness (PC1) and saturation (PC2) changes during the interaction (Fig. 6B, Fig. 7B).

This study presents evidence of female mate choice based on specific aspects of male colour change during courtship. Males that exhibited lower lateral line and interband brightness changes in the visible range (PC3) were more likely to be successful and engage in copulation attempts or actual copulation. This result follows our prediction that, as in the context of other social interactions (Ligon and McGraw, 2013), brightness changes should be involved in intersexual interactions. Brightness changes in chameleons are also involved in the context of camouflage in some species and are used to decrease their conspicuousness when exposed to highly visual predators (Stuart-Fox et al., 2008). Consequently, we might assume that chameleons would increase their brightness to increase their conspicuousness to communicate. This is strongly supported by our results and previous studies on chameleon social interactions, suggesting a keystone role of brightness changes in chameleon agonistic interactions (Ligon, 2014; Ligon and McGraw, 2013). However, we showed that non-selected males exhibited more brightness changes than selected males, possibly because non-selected males darken, which is similar to the submissive behaviour described for the veiled chameleon (Chamaeleo calyptratus) (Ligon, 2014). In both cases, males may increase crypsis or signal submission through an active darkening. This may prevent injuries by conspecifics and may prevent chameleons from being spotted by predators.

In addition, we found that selected males were those exhibiting greater overall absolute colour changes (PC4) at the bands and interbands. Hence, females chose males that exhibited the greatest colour difference between their initial colour and final colour (prior to copulation). Females seem to pay particular attention to the overall absolute colour change from the bands compared with other areas (Figs 4 and 5). Females also seem to choose males showing fewer hue changes at the bands and interbands and more saturation changes at the bands and the lateral line in the visible range. Even though guanine platelet translocation in iridiphores is likely causing the hue changes in F. pardalis (Teyssier et al., 2015), it is thought that xantho-erythorphores also play a role in colour change by varying the extent of light filtration by this chromatophore layer (Satake, 1980; Kotz, 1994; Oshima et al., 2001; Sato et al., 2004). Hence, both chromatophores act collectively to achieve hue changes. Xantho-erythrophores achieve colour change through the translocation of pigments granules containing carotenoid and/or pteridine pigments. However, the amount of those pigments may also impact the filtering range capabilities, with greater amounts leading to wider filtering possibilities and thus wider colour change repertoires. In general, animals with a more developed carotenoid-based colouration have a better immune system (McGraw and Ardia, 2003; Baeta et al., 2008) and antioxidant function (Henschen et al., 2016), because of immunostimulant and antioxidant properties of carotenoids and pteridines (Svensson and Wong, 2011; Simons et al., 2012; McGraw, 2005). Saturation and hue metrics of carotenoid-based colouration are good proxies of carotenoid concentration in some birds (Butler et al., 2011; Inouye et al., 2001). Consequently, it is expected that a female would preferentially opt for males exhibiting a greater overall colour change as well as greater hue and saturation changes, as it may reflect a better immune system and antioxidant function. In accordance, we did find that chosen males exhibited greater saturation changes and greater overall colour changes.

However, females also preferentially chose males showing lower hue changes. This could be possibly be explained by the fact that animals rely on numerous other compounds for antioxidant function including vitamins A, E and C, polyphenols, polyunsaturated fatty acids (i.e. omega 3 and 6), L-arginine, transferrin and ubiquinol (reviewed in Vertuani et al., 2004). Therefore, carotenoid-based colouration does not always reflect oxidative stress or immune system quality as other compounds may take over. (Schantz et al., 1999; Krinsky and Johnson, 2005; Svensson and Wong, 2011). Understanding the information conveyed by the colour change is a core question that would need further investigation. To do so, the links between fitness-related traits and different aspects of colour change need to be explored.

Another hypothesis to explain why males exhibiting lower hue changes are selected by females might be linked to the energetic cost of colour change or more precisely the cost to maintain a specific colour. However, the cost of colour change remains unknown to date. Consequently, it would be beneficial for future studies to quantify the energetic requirements of each component of colour change (i.e. melanosome translocation within melanophores, pigment granule translocation within xantho-erythrophores and guanine platelet translocation within iridophores). Even though general sexual selection hypotheses (Darwin, 1874) involve female mate choice over courting males, there is a growing body of evidence for male mate choice (Belliure et al., 2018; Edward and Chapman, 2011; Kokko et al., 2003) and even mutual mate choice (Courtiol et al., 2016; Drickamer et al., 2003; Myhre et al., 2012). Our data may reflect male mate choice where male F. pardalis expresses rejection with more hue changes. During our mate choice experiments, we did observe cases where a female exhibited active receptive behaviour, yet males did not respond to those signals (see Table 1) (N=16). In some cases, the male even fiercely rejected the female (i.e. biting or lunging at the female). Although this is suggestive of male mate choice, the female traits upon which male preference might be based remain to be investigated.

Chameleons, including F. pardalis, have a spectral sensitivity that includes the UV range (Bowmaker et al., 2005) and there is evidence that chameleons have UV patterns and UV-absorbing fluorescent patterns (Prötzel et al., 2018). However, previous studies investigating dynamic colour changes in intrasexual interactions in chameleons (Ligon, 2014; Ligon and McGraw, 2016; Ligon and McGraw, 2013, 2018) did not investigate the role of UV and the associated changes thereof. However, the literature suggests a role of UV colouration in either intrasexual (Martin et al., 2016; Whiting et al., 2006) or intersexual (Griggio et al., 2010; Lim et al., 2007; Rick and Bakker, 2008) interactions in a diversity of taxa. In the present study, we did find a relationship between UV colouration and female mate choice. Selected males showed lower UV brightness along with more brightness changes and more saturation changes. It thus appears that females choose males with higher UV absorbance and males that change more in UV colouration. Interestingly, some of the UV-absorbing colour patterns we sampled are head bony tubercles, which are known to be fluorescent under UV light (Prötzel et al., 2018). Consequently, higher absorbing UV properties might lead to higher fluorescence. Thus, females may prefer higher fluorescence rather than higher UV-absorbing properties per se. This fluorescence emits around 430 nm, which is close to the mean maximum absorbance of short wave-sensitive photoreceptors (i.e. 440 nm with a maximum at 430 nm) in F. pardalis (Bowmaker et al., 2005). Interestingly, the spectral range around 430 nm also seems to correspond to a spectral sensitivity gap in avian vision (Hart and Hunt, 2007; Hart and Vorobyev, 2005). Therefore chameleons might use fluorescence as a private communication channel (Cummings et al., 2003) to avoid predation risk from avian predators during mate choice interactions.

Females rejected the males with the lowest UV brightness and saturation changes. Non-selected males may possibly favour crypsis to avoid predation or injuries from conspecific by maintaining a constant UV colouration. In contrast, selected males should actively change UV colouration to communicate their motivation to mate. We expected UV changes at the lateral line to be more important than other UV colour patterns. UV colours at the lateral line are due to the photonic crystals, within iridophores, which chameleons can manipulate to change colour (Teyssier et al., 2015). In contrast, the UV absorption from bony tubercles is due to specific structures within the bone itself rather than the skin (Prötzel et al., 2018). Nevertheless, our results suggest that panther chameleons may be able to tune UV colouration, even at the level of the bony tubercles. How this is possible remains to be investigated, but a possible role for melanosomes present in the skin overlying the tubercles can be envisaged.

To conclude, we provide evidence of female mate choice for specific aspects of dynamic colour change in male chameleons, including UV colouration and colour change in the visible range. We showed that females rely on different aspects of colour change according to the body region to assess mates. Female mate choice consistently appears to be for colour changes between body regions. This poses the question of whether different aspects of colour change from different body regions convey different information. Moreover, whether some aspects of colour change participate more in signal design rather than to information content remains to be elucidated.

We thank the two anonymous reviewers and Dr Ylenia Chairi for helpful and constructive reviews of an earlier version of this manuscript. We thank Dr Sylvie Laidebeur, Dr Laetitia Redon, Dr Alexis Lecu, Fabrice Bernard, Morgane Denis and Mickaël Leger for assistance with chameleon husbandry and care. We thank Cedric Bordes, Denis Lebon and Loïc Laumalle-Waddy from the Ferme Tropicale for their help in providing us with materials for husbandry. We thank Hugue Clamouze and Thierry Decamps for helping us with the experimental arena. Finally, we thank the Ecole Doctorale FIRE - Programme Bettencourt for funding.