Conspicuous colours reduce predation rates in fossorial uropeltid snakes

Introduction

Conspicuous colourations (colours that contrast with the background) are prevalent in the animal kingdom and have fascinated biologists for a very long time, prompting them to seek a functional explanation for such colours (Poulton, 1890; Cott, 1940; Edmunds, 1974; Cuthill et al., 2017). Such colours may be part of mate-choice or other intra-specific signals, or may be involved in predator avoidance, e.g. aposematism and mimicry (Caro & Allen, 2017). Over the last few decades, these theories have been widely tested, leading to a good understanding of bright conspicuous colours in animals (Gamberale-Stille & Tullberg, 1999; Speed, 2001; Pfennig, Harcombe & Pfennig, 2001; Tullberg, Merilaita & Wiklund, 2005; Stevens, Stubbins & Hardman, 2008).

However, the presence of conspicuous colours in many animals still remains unexplained (e.g. contrasting colours in mammals (Caro, 2009), colourations in many frogs (Bordignon et al., 2018) and caecilians (Wollenberg & Measey, 2009)) with research on new systems revealing novel insights onto the role of such colourations (Rößler et al., 2019). Subterranean organisms which live in conditions devoid of light tend to have reduced vision and lose pigmentations (Culver & Pipan, 2009). Yet, several species of fossorial snakes that spend most of their time underground exhibit bright colourations (e.g. members of the genus Cylindrophis, Anomochilus, Atractus. etc.) (Greene, 1988). Colours, especially those produced via pigments, are generally used as visual stimuli for intra-specific or inter-specific interactions and thought to be costly to produce, therefore, the presence of bright colours in subterranean animals is thought to be adaptive (Wollenberg & Measey, 2009). However, the role of conspicuous colouration in fossorial reptiles remains unexplored.

Uropeltid snakes are a family of fossorial snakes comprising ca. 55 species from South Asia (Cyriac & Kodandaramaiah, 2017). Most uropeltid species are ornamented on the ventral surface with bright conspicuous colouration, with the majority possessing varying degrees of yellow while a few species (ca. 5) possess red (Fig. 1). Although these snakes spend most of their time underground, they occasionally come to the surface during the monsoons (Rajendran, 1985). Uropeltids are mostly nocturnal but can be found actively moving close to the surface during early mornings and late evenings (Rajendran, 1985) during which times they are exposed to above ground predators (Rajendran, 1985; Gans, 1986; Kumara & Chaitra, 2001). Uropeltid snakes have characteristic morphologies adapted for burrowing into soil: a narrow head and a short distinctive tail which is tapering or rounded (Smith, 1943). The tails of many species appear obliquely cut, with hard carinate scales (Fig. S1). When attacked by predators, these snakes conceal their heads between their body coils and display their tails along with the conspicuously coloured ventral surface (Gans, 1986) (Fig. 2). The tail in uropeltid snakes, being short and rounded, resembles the head. This cephalic resemblance was found to deflect attacks of avian predators to the reinforced tail, thus allowing the snake to escape unharmed in most cases after multiple attacks oriented towards the tail (Gans, 1986). Such defenses would also potentially increase the handling time needed for predators to capture, kill and consume the prey.

Optimal foraging theory predicts that predators should avoid prey with increased handling times, especially when prey with lower handling times are abundant alternatives (Charnov, 1976; Krebs et al., 1977). Studies suggest that predators can learn to avoid unprofitable prey that are difficult to capture (Hancox & Allen, 1991; Pinheiro, 1996; Pinheiro et al., 2016), hard to process (Wang et al., 2018) or require long handling time (Cyriac & Kodandaramaiah, 2019) when such prey possess conspicuous colours, even if the prey are non-toxic and palatable (Mappes, Marples & Endler, 2005). We here postulate that bright colours in uropeltid snakes reduce predation, possibly by signalling long handling times associated with attack deflection to the tails. We first show that the yellow and red ventral colourations in uropeltid snakes are conspicuous to birds, the main predators of these snakes. In experiments involving captive chickens attacking dough models resembling uropeltid snakes, we show that attacks are directed more often towards the tail than the head. In field experiments with clay models that resemble uropeltid snakes, we show that the bright yellow and red colours of wild uropeltids decrease predation by native birds. We conclude that the bright colours of these snakes reduce avian predation, and that these colours possibly play a role in advertising long handling time.

Materials and Methods

Results

Predation rates on snake models in field

All models were recovered after 86 h. However, many models (N = 39) were destroyed due to trampling by large mammals such as Gaur (Bos gaurus), Sambar deer (Cervus unicolor) and Asian elephants (Elephas maximus), and were, therefore, excluded from the analyses. A total of 50 models (i.e. 5% of all models) across both predation trials were found to have potential avian attacks. The GLMM analysis indicated that the number of attacks varied significantly among the treatments (Fig. 3) and that the model where predation rate was affected by phenotype (i.e. model colour) was significantly better than the null model where phenotype did not affect the predation rate (ΔAIC = 16.4, X² = 24.402, P < 0.0001). The post hoc tests indicated that the number of attacks on the red (Estimate = −2.4349, z = −3.371, P < 0.01) and yellow (Estimate = −1.7275, z = −3.198, P < 0.05) models were significantly lower compared to that on the brown models. Attacks on the brown models did not differ significantly from that on either black (Estimate = 0.4834, z = 1.328, P = 0.65683) or novel coloured (Estimate = 0.6652, z = 1.745, P = 0.38739) models. Although the models with uropeltid colouration (red and yellow) experienced less predation compared to the novel models (orange), the difference was not statistically significant (red models: Estimate = −1.7697, 95% CI [−3.7821–0.2426], z = −2.369, P = 0.11423; yellow models: Estimate = −1.0623, 95% CI [−2.6060–0.4814], z = −1.854, P = 0.3248).

Captive bird experiments

A G-test on the attack position by the captive chickens on the dough models indicate that there was a significant difference in the position of first attacks (G-test: G₂ = 17.102, P = 0.00019). There were significantly higher number of attacks on the mid-body (P = 0.00034) and on the tail (P = 0.0087) compared to that on the head (Fig. 4).

Discussion

Our avian visual modelling indicates that the yellow and red ventral colourations of uropeltid snakes are highly conspicuous against the leaf litter background of all four transects and that uropeltid colours are readily visible to birds with both the VS and UVS visual systems. The field experiments indicate that clay models with the yellow and red colouration of local uropeltid snakes experience reduced avian predation compared to the brown models. However, the novel coloured models (orange) did not differ from the brown or black ones. Thus, the results indicate that avoidance is not towards all conspicuous colours but specifically towards uropeltid colours, suggesting that colouration in uropeltid snakes may have evolved as an antipredatory defense mechanism against avian predators.

Colouration in uropeltid snakes has sometimes been attributed to coral snake mimicry (although not experimentally tested) (Rajendran, 1985). South Asian coral snakes (genus Calliophis) are venomous, semi fossorial elapids, which share similar habitats with uropeltid snakes. However, all known species of coral snakes found in India and Sri Lanka have red ventral surfaces while most species of uropeltids have varying patterns of yellow. Further, coral snakes in India are uncommon across their range (Srinivasulu, Srinivasulu & Molur, 2014). Several theoretical models and experimental studies have suggested that Batesian mimicry generally becomes more perfect when the model is rare (Sherratt, 2002; Harper & Pfennig, 2007; Akcali & Pfennig, 2014). Since coral snakes are uncommon, it is expected that selection would favour high precision of coral snake mimicry, which does not seem to be the case for uropeltids. Furthermore, some uropeltids with bright colouration are found outside the range of coral snakes, making Batesian mimicry unlikely (Pfennig & Mullen, 2010). Therefore, although coral snake mimicry cannot be completely ruled out in some species of uropeltid snakes with red ventral colourations, the Batesian mimicry hypothesis does not explain the evolution of yellow colouration present in most uropeltid snakes because would-be models are red.

Overall, our experiments suggest that the conspicuous ventral colouration in uropeltid snakes could have evolved as a warning signal towards avian predators. Given that these snakes are endemic to India and Sri Lanka, where they are protected under stringent laws, laboratory experiments with live animals were not feasible. Our experiment with captive chickens and dough models resembling uropeltid snakes show that attacks were more frequent on the tail than on the head (Fig. 4). This suggests that chickens can be deceived by the cephalic resemblance of the tail and direct their attacks towards the posterior regions, rather than towards the head as many predatory birds are known to do (Smith, 1973; Curio, 1976). Such a strategy that deflects attacks to the tail would also increase the handling time required to capture, kill and consume these snakes, making them unprofitable (Humphreys & Ruxton, 2018). Although we have not specifically tested whether deflection of attacks towards the tail increases handling time in uropeltid snakes, detailed observations on the sequence of events during predation of uropeltid snakes by junglefowls and peafowls indicate long handling time associated with deflected attacks towards the tail (Gans, 1986). Gans (1986) reported that ca. 95% of the attacks by fowls were oriented towards the rounded tail of uropeltid snakes. He noted that the birds took between 22 and 40 min to consume uropeltid snakes after multiple attacks to the tail. He also reports anecdotal evidence that junglefowls and spurfowls were more hesitant to attack the brightly coloured Rhinophips drummondhayi, than the cryptic unicoloured R. philippinus. These observations provide further support for the hypothesis that diverted attacks from the vulnerable head to the tail could potentially increase handling time needed to kill and consume uropeltid snakes, which birds could learn to associate with their ventral colourations.

Startle or deimatic displays involve behaviours wherein conspicuous colours are suddenly displayed, providing prey a survival advantage (Umbers, Lehtonen & Mappes, 2015; Umbers & Mappes, 2015). Although there has been disagreement regarding what constitutes a deimatic display and how it differs from aposematism (Ruxton, Sherratt & Speed, 2004; Umbers, Lehtonen & Mappes, 2015; Umbers & Mappes, 2015, 2016; Skelhorn, Holmes & Rowe, 2016), there is general agreement that deimatism involves a momentary transient display of conspicuous signal (Ruxton, Sherratt & Speed, 2004; Olofsson et al., 2012; Umbers & Mappes, 2016) that triggers an unlearnt avoidance response in predators (Umbers et al., 2017, 2019). While uropeltid colours could also have a deimatic function, our experiments show that birds avoided uropeltid snake models that were static (pinned to the ground) and did not show any momentary display of conspicuous colourations. The snake models in our field experiment were painted such that the conspicuous colours were visible on the lateral side and thus were displayed and visible to small predators throughout the experimental duration (Fig. S2). Further, the preferential avoidance of only uropeltid colours and not all conspicuous colours suggests that this avoidance is learnt. However, as we did not design our experiments to test for a deimatic function of uropeltid colourations, we cannot preclude the possibility that uropeltid colours could also function as a startle display.

Advertising increased handling time due to deflection of attacks would, however, be advantageous only when alternative prey are abundant and when predators do not learn to recognise the deception. Repeated exposure to deflective structures and behaviours would reduce the effectiveness of deceiving predators. However, although birds regularly feed on venomous and non-venomous snakes (Guthrie, 1932), there is a risk associated with attacking snakes since many snakes are venomous. Hence, birds tend to orient their attacks towards the portion of the body with a bilateral indentation (neck) or towards the region with eye-like markings (Smith, 1973; Curio, 1976). The costs associated with attacking venomous snakes may prevent birds from learning the cephalic resemblance of the tail in uropeltid snakes. Further, it has been shown that the speed of a predator’s learning varies for different traits (Chittka & Osorio, 2007; Balogh et al., 2010), especially when other prey are available in the community (Kikuchi et al., 2019). For instance, animals show higher rates of learning towards colour signals than towards patterns or shapes (Bain et al., 2007; Aronsson & Gamberale-Stille, 2008, 2012a; Kazemi et al., 2014; Sherratt et al., 2015). Given that colours are more salient than other cues, birds may be able to associate colour with increased handling time in uropeltid snakes faster than learning the cephalic resemblance of the tail. Also, deflective traits are thought to be associated with life histories that reduce exposure to predators, thereby reducing the potential of learning such deceptive traits (Humphreys & Ruxton, 2018). Uropeltid snakes are highly seasonal and are active above ground or near the surface during the monsoons (Rajendran, 1985) when the species richness and abundance of leaf litter arthropods are high (Janzen & Schoener, 1968; Frith & Frith, 1990; Develey & Peres, 2000). High densities of arthropods during the wet seasons in tropical regions could serve as an alternative prey base for avian predators. The short activity period of uropeltid snakes would also reduce the ability of avian predators to learn to ignore deflective traits.

Although we cannot completely rule out the possibility of Batesian mimicry in some species of uropeltids, our experiments together support the hypothesis that the conspicuous colourations in these snakes act as warning signals against avian predators. Our results are also consistent with the novel hypothesis that conspicuous colourations in uropeltid snakes have an antipredatory function based on advertising long handling time associated with diverted attacks to the tail. Nonetheless, given that we have not explicitly tested whether misdirected attacks to the tail increases the handling time of these snakes, we acknowledge that a functional relationship between the colouration and tail shape and its role in reducing predation cannot be ruled out. For instance, many species of uropeltid snakes possess yellow or red blotches or stripes on the lateral sides of the tail (Fig. 2B; Figs. S1D and S1F) that could divert the attention of predators and deflect attacks towards the tail. However, further experiments would be required to determine whether colouration and tail shape interact to divert attacks and how this would influence handling time and learning in predators.

Antipredatory defenses involving both deflection of attacks and warning colourations may not be restricted to Uropeltidae. Fossorial reptiles, especially snakes, tend to have long trunks and short rounded tails (Wiens & Slingluff, 2001; Wiens, Brandley & Reeder, 2006), the latter of which many species display to divert attacks from the head to the tail (Greene, 1973, 1979, 1988; Han & Young, 2018). Several fossorial snakes (e.g. Cylindrophidae, Anomochilidae, Anillidae, Atractaspinae, Aparallactinae) exhibit conspicuous colourations potentially having a similar function as in uropeltid snakes. However, how widespread such strategies are across animals needs to be evaluated.

Conclusion

Our study highlights how combinations of defensive strategies such as deflection, which increases handling time, and warning signals, could function synergistically against predation. Different antipredatory strategies need not be mutually exclusive and can interact depending on the predators’ visual perception (Stevens, 2007). For instance, it has been shown that patterns with strong internal contrast can increase conspicuousness (Aronsson & Gamberale-Stille, 2012b) but can also generate a disruptive effect (Stevens & Cuthill, 2006; Schaefer & Stobbe, 2006). Warning colourations can also be distance dependent, providing camouflage at greater distances while being conspicuous at close proximity (Tullberg, Merilaita & Wiklund, 2005; Barnett, Cuthill & Scott-Samuel, 2018). For instance, the contrasting bands of coral snake mimics and the zig-zag pattern of many vipers act as warning signals to predators (Pfennig, Harcombe & Pfennig, 2001; Wüster et al., 2004; Niskanen & Mappes, 2005) but these bands can also function in camouflage during motion through the flicker-fusion effect (Lindell & Forsman, 1996; Titcomb, Kikuchi & Pfennig, 2014). Despite considerable advances in our understanding of animal defenses, we are yet to completely understand how different strategies interact with each other and further experiments are necessary to understand under what conditions such interactions are advantageous against predation.

Supplemental Information