Asymmetric interspecific competition drives shifts in signalling traits in fan-throated lizards

Abstract

Interspecific competition can occur when species are unable to distinguish between conspecific and heterospecific mates or competitors when they occur in sympatry. Selection in response to interspecific competition can lead to shifts in signalling traits—a process called agonistic character displacement. In two fan-throated lizard species—Sitana laticeps and Sarada darwini—females are morphologically indistinguishable and male agonistic signalling behaviour is similar. Consequently, in areas where these species overlap, males engage in interspecific aggressive interactions. To test whether interspecific male aggression between Si. laticeps and Sa. darwini results in agonistic character displacement, we quantified species recognition and signalling behaviour using staged encounter assays with both conspecifics and heterospecifics across sympatric and allopatric populations of both species. We found an asymmetric pattern, wherein males of Si. laticeps but not Sa. darwini showed differences in competitor recognition and agonistic signalling traits (morphology and behaviour) in sympatry compared with allopatry. This asymmetric shift in traits is probably due to differences in competitive abilities between species and can minimize competitive interactions in zones of sympatry. Overall, our results support agonistic character displacement, and highlight the role of asymmetric interspecific competition in driving shifts in social signals.

1. Introduction

Signalling traits are stunningly diverse across and within taxonomic groups in animals [1]. Biologists have long been interested in mechanisms underlying diversification of signalling traits to not only understand how signals evolve, but also because they can directly influence speciation dynamics [2]. In recent years, there has been a growing interest in the role of interspecific interactions in the divergence of signalling traits [3]. One way species interactions can influence signalling traits is through interspecific male competition [3]. Interspecific male competition can arise as a result of similarity in male signals and/or from competition over reproductive resources [4]. In closely related species, female phenotypes as well as male signals often show high degree of similarity, probably due to shared ancestry and exposure to similar selection pressures, such as predation [5,6]. When such species come into contact, via range expansions, for example, males may be unable to distinguish between conspecific and heterospecific mates and competitors, especially if they lack pre-existing species recognition mechanisms [4,7]. This can give rise to interspecific competition [4,7].

Interspecific competition can be costly [8] or adaptive [7] depending on the factors driving the competition. For instance, if males cannot distinguish between conspecific and heterospecific females, and compete over access, interspecific competition can be beneficial if it reduces the probability of missing conspecific matings, especially if the costs of heterospecific matings are low [3]. On the other hand, interspecific male competition resulting from similarity in male signals can increase risk of injury and death [9], and therefore is costly. Depending on which factor(s) drive interspecific competition and its associated costs and benefits to each species, selection can lead to rapid shifts (divergence or convergence) in signalling traits [3,7,10]. This process is termed as agonistic character displacement [3]. Since only individuals co-occurring and competing with heterospecifics show shifts in traits, this produces a distinct geographical pattern, characterized by quantitative trait differences between sympatric and allopatric populations of a species [10]. Differences in sexually selected wing coloration and male competitor recognition across sympatric and allopatric populations in damselflies Hetaerina is perhaps the best documented example of agonistic character displacement [4,11]. In the Hetaerina group, intensity and cost of interspecific male competition is driven by similarity in female phenotypes, which in turn influences direction and magnitude of shifts in wing coloration and competitor recognition function in males [7] (see [9,10] for other examples). Several theoretical models confirm the potential of such interspecific competition to result in shifts in signalling traits [7,10], but surprisingly few studies have empirically tested this, making this area of research one of the most exciting problems in evolutionary biology [10].

Despite the handful of examples of agonistic character displacement, a broader collection of studies are needed to better understand the importance and generality of interspecific competition in the diversification of signalling traits. Two fan-throated lizards, Sitana laticeps and Sarada darwini, offer an excellent system to examine the effects of interspecific competition on signalling traits. Both Sa. darwini and Si. laticeps are distributed in the southwestern part of India. In this region, Si. laticeps is widely distributed in the Deccan plateau, whereas Sa. darwini is restricted to the eastern slopes of the northern Western Ghats mountain range [12]. Across most of their range, these species are allopatric. However, in regions where the Western Ghats meet the Deccan plateau, they overlap with each other in a narrow sympatric zone [13]. Anecdotal reports and our own observations confirm the presence of interspecific male aggression in this sympatric zone. Aggression between males of these species is probably driven by a combination of similarity between female phenotypes and similarity in male signalling behaviours (figure 1). Females of both Si. laticeps and Sa. darwini are similarly sized and are morphologically indistinguishable [12] (figure 1). Courtship and agonistic signalling behaviours of males are also remarkably similar across species [14], even though the split between Sitana and Sarada genera is estimated to be about 18 Ma [13]. Agonistic and courtship signalling of both species are characterized by rapid dewlap flagging, suggesting that dewlap flagging acts as a dual-use signal [14,15]. Typically, males of both species establish seasonal breeding territories and select small rocks within territories as display sites. The presence of females (conspecific or heterospecific) and other signalling males (conspecific or heterospecific) within these territories triggers dewlap flagging from resident males. This is interesting since colour and size of male dewlaps differ significantly across species (figure 1; electronic supplementary material, figure S3). Thus, males of these species seem to have a wide species recognition function, and that differences in dewlap colour may not be sufficient to avoid interspecific aggression in these lizards. Together, these factors may make it challenging for males to quickly distinguish between conspecific and heterospecific mates or competitors [7], resulting in interspecific male competition [4].

Figure 1.

(a) Map showing locations of sampled populations. Pie charts are relative densities of Si. laticeps (grey) and Sa. darwini (blue). (b) Male, female and males with extended dewlaps of Sitana laticeps (left) and Sarada darwini (right). Images not to scale. Photos: Rajesh Sanap, Varad Giri and Amod M. Zambre. (Online version in colour.)

Our observations suggest that interspecific competition between Si. laticeps and Sa. darwini can be intense, escalating from displays to chasing and combat. Such interactions can limit access to mates and increase risk of injury and death and thus, are likely to be costly [8]. However, these costs may differ between species. Sarada darwini males are larger (approx. 20%) and heavier (approx. 30%) than Si. laticeps males. Multiple studies on lizards (and a wide range of taxa) have shown body size to be strongly correlated with competitive abilities and dominance [16,17]. Larger individuals typically have a greater probability of winning interspecific contests [18,19]. This suggests that Sa. darwini will have a distinct advantage over Si. laticeps during interspecific conflict. Therefore, costs of interspecific competition will be greater for Si. laticeps than Sa. darwini, and hence, Si. laticeps may experience stronger selection to avoid competitive interactions. If agonistic character displacement is supported, we expect shifts in agonistic signalling traits of Si. laticeps that reduce interspecific competition with Sa. darwini in the sympatric population. To test these predictions, we first determined whether males of both species distinguish between heterospecific and conspecific females. We did this by measuring the probability and intensity of inter- and intra-specific courtship signalling of Sa. darwini and Si. laticeps males in both allopatric and sympatric populations. We then quantified interspecific male competition by measuring the probability and intensity of inter- and intra-specific agonistic signalling of Sa. darwini and Si. laticeps across sympatric and allopatric populations. Additionally, we also measured dewlap size and colour of these species in allopatry and sympatry. By comparing recognition of mates and competitors, as well as agonistic and courtship signalling traits between sympatric and allopatric populations, we test the potential for agonistic character displacement.

2. Methods

(a). Sampling site

Both Sa. darwini and Si. laticeps are distributed in southwestern India. In this region, three locations were sampled: (i) allopatric population of Sa. darwini (16°35′45.95″ N, 74°11′11.98″ E), (ii) allopatric population of Si. laticeps (16°46′59.81″ N, 74°20′1.78″ E), and (iii) sympatric population of Sa. darwini and Si. laticeps (16°36′35.33″ N, 74°17′52.71″ E). Allopatric populations of Sa. darwini and Si. laticeps were 12 and 14 km away from the sympatric population respectively (figure 1) and had similar habitat types (electronic supplementary material, figure S1) and lizard densities (figure 1). By restricting our sampling to around 2 km² each at these three specific sites, we were able to minimize the possible differences in signalling environment, food availability and predation pressure, which can independently influence signals. Habitats at all of our selected sites are classified as arid grasslands characterized by short grasses interspersed with few Acacia trees (figure 1; electronic supplementary material, figure S1).

(b). Density

At each location, we measured the abundance of lizards by walking 100 × 20 m belt transects (n = 13–18 transects per location; total n = 49 transects) in March–April 2016. Fan-throated lizards (both sexes) bask on small boulders and are hence are easy to spot [20]. During each transect, we recorded species and sex-specific abundances, and used these data to calculate density.

(c). Morphological measurements

At each location, we captured lizards by hand (n = 20 individuals per sex per species per location; total n = 160 lizards) in April–May 2016. We measured snout–vent length (proxy for body size) and dewlap length (only in males) with Vernier calipers (closest to 0.01 mm), and body mass with Pesola spring scales (10 g; closest to 0.1 g). Additionally, to examine differences in dewlap coloration, we measured per cent reflectance of male dewlaps (n = 6 males per species per location; total n = 32 males) using a spectrophotometer (JAZA2474 with PX lamp; Ocean Optics). Sarada darwini males have three colours on their dewlaps; hence, we measured reflectance of each colour patch separately. Sitana laticeps dewlaps are visibly white, and hence we recorded only the reflectance of the middle of the dewlap. Collected spectra were imported into R, smoothed (α = 0.25) and corrected for negative values (function used: ‘addmin’) in ‘pavo2.0’ package in R [21]. These corrected spectra were used to calculate hue (H1), chroma (S8) and mean brightness (B2) of dewlap colours (see electronic supplementary material for details of spectrophotometry and colour measurements) [21]. All lizards were released at their respective capture locations within 30 min of capture.

(d). Male mate recognition and courtship signalling

To quantify male mate recognition functions and courtship signalling behaviour, we measured frequency and intensity of intra- and interspecific courtship for both species across their allopatric and sympatric populations. This was done using staged courtship assays, where we introduced either a conspecific or a heterospecific female into the territories of resident males. To avoid physical contact between lizards, each female was placed in a glass tank (dimensions (length × breadth × height) = 60 × 60 × 60 cm) before introduction to a resident male's territory. This tank was placed 2–4 m from resident males at or below their eye level (electronic supplementary material, figure S2). We recorded the subsequent behavioural interaction using a digital video camera (Sony HDR PJ 50E) positioned 15–20 m away. From these videos, we scored resident male responses first as 0 or 1 (0 if no response and 1 if resident male responded by flagging dewlaps) and calculated probability of intra- and interspecific courtship for both species in sympatry and allopatry. Courtship in both species involves dewlap flagging while often standing upright on the hind limbs. We also recorded the duration of courtship signalling and number of dewlap flags during signalling. The duration of courtship signalling was calculated by pooling the durations of all flagging bouts and dividing it by the total duration of the video (range of video duration 2–5 min). Flagging bouts were identified as times during which lizards flagged their dewlaps beginning with the first dewlap flag and ending 15 s after the last dewlap flag. The rate of dewlap flagging was calculated as (total no. of dewlap flags ÷ duration of signalling). In total, we performed 10 heterospecific female and 10 conspecific female trials per species per location, resulting in 40 trials in the sympatric population (20 with Si. laticeps males and 20 with Sa. darwini males) and 20 each in the allopatric populations. Resident males were tested only once, and females used during these trials were captured from different allopatric populations not sampled during this study. These females were used for a maximum of three trials and were released at their respective capture locations after the trials.

(e). Male competitor recognition and agonistic signals

We measured frequency and intensity of intra- and interspecific aggression between males of both species to quantify male competitor recognition function and agonistic signalling behaviour across both allopatric and sympatric populations. This was done using staged competition assays with introduction of either a conspecific or heterospecific male as an intruder inside a resident male's territory, in the same way described above for female introductions. Agonistic signalling, like courtship, is characterized by rapid flagging of dewlaps but can be distinguished from the former by the erection of nuchal crest and change of tail coloration to bright blue. From the recorded videos of these assays, we scored resident male responses first as 0 or 1 (0 if no response and 1 if response was dewlap flagging and erection of nuchal crest) and calculated probability of intra- and interspecific aggression for both species in sympatry and allopatry. Additionally, we also recorded the duration of agonistic signalling and number of dewlap flags during signalling (as described above). In total, we performed 10 heterospecific male and 10 conspecific male trials per species per location, resulting in 40 trials in the sympatric population (20 with Si. laticeps and 20 with Sa. darwini) and 20 each in the allopatric populations. Resident males were tested only once, with either a conspecific or heterospecific intruder male that was captured from allopatric populations not part of this study. Each intruder male was used for a maximum of three trials and released at its respective capture location at the end of the day.

(f). Statistical analyses

We compared differences in total lizard densities across the three sites using a generalized linear model with Poisson distribution (model: density ∼ site). We used a second model for each species separately to determine differences in density between the sexes across sites (model: density ∼ sex × site). For each species, differences between sympatric and allopatric populations in snout–vent length, body mass, dewlap length and dewlap coloration (hue, chroma and brightness) were assessed using standard T-tests (all data met the assumptions of normality). For probability of courtship and agonistic signalling, we initially examined differences using separate logistic regression models for each species (model: response ∼ population × stimulus species). If the interaction term was non-significant, it was dropped, and differences were examined using simplified additive models (model: response ∼ population + stimulus species). Differences in rates of dewlap flagging for courtship and agonistic signalling between sympatric and allopatric populations were also initially examined using separate linear models for courtship and agonistic signalling for each species (model: rate of flagging ∼ population × stimulus species). If non-significant, interaction terms were dropped, and simplified models were used to examine differences between populations and stimulus species (model: rate of flagging ∼ population + stimulus species). Tukey's post hoc test (‘emmeans’ function) was used to determine pairwise differences when relevant. All statistical analyses were performed in R (v. 3.6) statistical software.

3. Results

(a). Density

Total density of lizards in the allopatric populations of Sa. darwini and Si. laticeps were similar and significantly lower than the total lizard density at the sympatric population (F_2,46 = 9.20, p = 0.01; table 1). A comparison of sex-specific densities in the allopatric and sympatric populations of Sa. darwini (all z < 1 and p > 0.05; table 1) revealed no significant differences. Similarly, there were no significant differences in the densities of males and females of Si. laticeps between sympatric and allopatric populations (all z < 1 and p > 0.05; table 1).

Table 1.

Sex-specific densities (mean ± s.e./per transect) of Si. laticeps and Sa. darwini at the sampled sympatric and allopatric populations.

populations	Sitana laticeps	Sarada darwini	total lizard density
allopatric population of Sitana laticeps	male: 0.83 ± 0.21 female: 0.61 ± 0.18	0	1.44 ± 0.31
sympatric population of Sitana laticeps and Sarada darwini	male: 0.77 ± 0.28 female: 0.38 ± 0.16	male: 1 ± 0.33 female: 0.55 ± 0.16	2.72 ± 0.59
allopatric population of Sarada darwini	0	male: 0.76 ± 0.16 female: 0.69 ± 0.20	1.46 ± 0.29

(b). Morphology

On comparing the snout–vent length of males and females across populations, we found that Sa. darwini males in the sympatric population were larger than conspecific males in allopatry (t = −2.70, p = 0.01). We found no significant differences in body sizes of Sa. darwini females (t = −1.15, p = 0.25), nor in Si. laticeps males (t = −1.79, p = 0.09) and Si. laticeps females (t = 0.62, p = 0.53) between sympatric and allopatric populations (figure 2a). In terms of body mass, males and females of Si. laticeps from the sympatric population had significantly lower mass than conspecifics in allopatry (males: t = 2.24, p = 0.03; females: t = 3.03, p = 0.005). No significant difference in mass was found for either sex of Sa. darwini (males: t = −1.26, p = 0.17; females: t = 0.19, p = 0.84; figure 2b). Males of Si. laticeps in sympatry also had relatively smaller dewlaps than those from the allopatric population (t = 7.47, p < 0.01; figure 2c). There was no difference in dewlap size of Sa. darwini males across populations (t = −1.08, p = 0.28). We found no significant differences in the hue, chroma and brightness of colour patches on male dewlaps between sympatric and allopatric lizards of either species (electronic supplementary material, figure S3).

Figure 2.

Variation in lizard morphology across the allopatric population of Si. laticeps (SL), sympatric population of Si. laticeps and Sa. darwini (SL + SD) and allopatric population of Sa. darwini (SD): (a) snout–vent length (SVL), (b) body mass (c) male dewlap size standardized by SVL. Grey and white boxes are males and females of Si. laticeps, whereas dark blue and light blue boxes are males and females of Sa. darwini respectively. (Online version in colour.)

(c). Male mate recognition and courtship signalling

(i). Sitana laticeps

For males of Si. laticeps, we found no significant interaction between populations (allopatric or sympatric population) or species of female (heterospecific or conspecific) in their probability to court (z < 0.00, p = 1.00; figure 3a). Furthermore, these factors individually did not affect the probability of Si. laticeps males to court females (population: z = −1.35, p = 0.17; species of stimulus female: z = 0.00, p = 1.00). For the rate of dewlap flagging, interaction between population and species of female was not significant (t = 1.64, p = 0.11), but these factors independently affected the rate of dewlap flagging (population: t = −5.13, p < 0.01; species of stimulus female: t = −2.24, p = 0.03; figure 3b). Males of Si. laticeps from the allopatric population had a higher rate of dewlap flagging that those in the sympatric population, in response to both conspecific (t = 2.69, p = 0.05; figure 3b) as well as heterospecific females (t = 4.84, p < 0.01; figure 3b). The marginally higher rate of dewlap flagging towards heterospecific females over conspecific females is attributed to Si. laticeps' responses in the allopatric population (t = 2.83, p = 0.04; figure 3b), not sympatric population (t = 0.20, p = 0.99; figure 3b).

Figure 3.

Courtship signalling behaviour of Si. laticeps and Sa. darwini directed at conspecific (blue) and heterospecific (red) females. (a) Probability of signalling to conspecific (blue arrows) and heterospecific (red arrows) females. Numbers above the arrows are back-transformed probabilities calculated from log odds from logistic regression models. Thickness of arrows corresponds to the strength of probability. Rate of dewlap flagging, measured as the number of dewlap flags per unit display time, of (b) Si. laticeps in allopatry (SL) and sympatry (SL + SD), and (c) Sa. darwini in allopatry (SD) and sympatry (SL + SD) in response to conspecific (blue boxplots) and heterospecific (red boxplots) females. (Online version in colour.)

(ii). Sarada darwini

For Sa. darwini, interaction between population and species of female had no effect on the probability of males to court (z = 0.02, p = 0.97; figure 3a). Furthermore, neither population (z = −0.64, p = 0.51) nor species of female (z = −0.64, p = 0.51) individually affected the probability of male Sa. darwini to court females. However, an interaction between these factors (population and species of female) did significantly influence Sa. darwini's rates of dewlap flagging during courtship (t = 2.03, p = 0.056; figure 3c). In the allopatric population, there was no significant difference in the rate of dewlap flagging towards conspecific or heterospecific females (t = 1.76, p = 0.32; figure 3c). In sympatry, males of Sa. darwini had a significantly higher rate of dewlap flagging towards conspecific females compared with heterospecific females (t = 4.36, p < 0.01; figure 3b). The rate of dewlap flagging was similar across allopatric and sympatric populations when males were exposed to conspecific females (t = −2.05, p = 0.20; figure 3b) or heterospecific females (t = 0.87, p = 0.81; figure 3c).

(i). Sitana laticeps

We found that in general, males of Si. laticeps were less likely to respond aggressively to males of both species in sympatry compared to allopatry (z = −2.59, p < 0.01), and were less likely to engage in agonistic signalling with heterospecific males than conspecific males (z = 2.088, p = 0.03; figure 4a). These differences were driven by the considerably low probability of Si. laticeps to engage with Sa. darwini males in sympatry (figure 4a). For the rate of dewlap flagging, we found no significant interaction between intruder male species and population on Si. laticeps's responses (t = 0.40, p = 0.68; figure 4b). Instead, males of Si. laticeps showed significantly higher rates of dewlap flagging towards both conspecific and heterospecific males in the allopatric population compared with the sympatric population (t = −4.749, p < 0.01; figure 4b). Intruder species had no significant effect on dewlap flagging rates of Si. laticeps in either population (t = 0.091, p = 0.92; figure 4b).

Figure 4.

Agonistic signalling behaviour of Si. laticeps and Sa. darwini directed at conspecific (blue) and heterospecific (red) males. (a) Probability of signalling to conspecific (blue arrows) and heterospecific (red arrows) males. Numbers above the arrows are back-transformed probabilities calculated from log odds from logistic regression models. Thickness of arrows corresponds to the strength of probability. Rate of dewlap flagging, measured as the number of dewlap flags per unit display time, of (b) Si. laticeps in allopatry (SL) and sympatry (SL + SD), and (c) Sa. darwini in allopatry (SD) and sympatry (SL + SD) in response to conspecific (blue boxplots) and heterospecific (red boxplots) males. (Online version in colour.)

(ii). Sarada darwini

The probability of Sa. darwini to engage in agonistic signalling was similar irrespective of population (z = 0.00, p = 1.00) or species of intruder (z = 0.72, p = 0.46; figure 4a). A similar pattern was observed for rates of dewlap flagging during agonistic signalling. Males of Sa. darwini showed high rates of dewlap flagging across both allopatric and sympatric populations (t = 0.58, p = 0.56) and in response to both species of intruder males (t = −0.14, p = 0.88; figure 4c).

4. Discussion

Interspecific competition between Si. laticeps and Sa. darwini is expected to be more costly for Si. laticeps than Sa. darwini due to differences in their competitive abilities. To test whether this asymmetric interspecific competition results in agonistic character displacement, we compared signalling traits of both species across their sympatric and allopatric populations. In the sympatric population, we find shifts in agonistic signalling traits of Si. laticeps but not Sa. darwini. Compared with their allopatric population, males of Si. laticeps have smaller dewlaps (figure 2c), show strong competitor recognition (figure 4a) and reduced intensity of both agonistic signalling (figure 4b) and courtship signalling (figure 3b). These trait differences observed between allopatric and sympatric populations of Si. laticeps support predictions of agonistic character displacement [9,10] and highlight the role of interspecific competition in divergence of agonistic signalling traits.

(a). Drivers of interspecific competition between Sarada darwini and Sitana laticeps

Females of Sa. darwini and Si. laticeps are virtually indistinguishable, and we find that males of both species are equally likely to court them in their allopatric and sympatric populations. This indicates the lack of mate recognition in males of both species. Only Sa. darwini showed some mate recognition in the sympatric population as males courted conspecific females at greater intensity than heterospecific Si. laticeps females. Notably, neither Sa. darwini nor Si. laticeps rejected heterospecific females (figure 3a–c). This lack of strong heterospecific female recognition and rejection may produce conditions where females act as a ‘shared resource’, triggering interspecific male competition over access [4,7]. Heterospecific courtship behaviour can promote interspecific competition that is adaptive if it reduces the chances of missing conspecific matings and if the cost of heterospecific matings is low [4,7]. In the absence of females, we also find that both Si. laticeps and Sa. darwini in their respective allopatric populations are similarly aggressive towards heterospecific males and conspecific males (figure 4a–c). This shows that competitor recognition function is extremely broad or even lacking in males of both species [4,7], which further contributes to the persistence of interspecific competition between Si. laticeps and Sa. darwini. Taken together, it appears that interspecific competition between these two species can be a result of some combination of competition for mates and the wide competitor recognition function in males [4,7]. Irrespective of which of these factor(s) actually promote interspecific competition between Sa. darwini and Si. laticeps, continued engagement in interspecific competition will have associated costs (injury and death) and benefits (fewer loss of conspecific mating opportunities). Since Sa. darwini is larger than Si. laticeps, it is expected to dominate interspecific contests, resulting in higher costs for Si. laticeps. Selection should, therefore, promote shifts in the signalling traits of Si. laticeps that reduce interspecific aggressive interactions in sympatric populations with Sa. darwini.

(b). Case for agonistic character displacement

Agonistic character displacement can be defined as a process that leads to quantitative trait differences between sympatric and allopatric populations of a species, driven by selection against costly interspecific competition [7,9]. Hence, demonstrating that observed differences in traits do in fact reduce the frequency of interspecific competition is crucial for establishing agonistic character displacement. We find that interspecific competition is reduced by the changes in Si. laticeps and not Sa. darwini. The probability and intensity of interspecific interactions are excellent proxies for species recognition [9], and we find that Si. laticeps males have a dramatically lower probability of engaging in agonistic signalling with Sa. darwini when in sympatry compared with in allopatry (figure 4a), indicating a major shift in their species recognition function. When Si. laticeps do engage in agonistic signalling in sympatry, their signalling behaviour is less intense (figure 4b). By exhibiting strong competitor recognition and reducing intensity of agonistic signals in sympatry, Si. laticeps may effectively reduce detection by Sa. darwini. Dewlap flagging is a key signal component that influences detection in these lizards. In a closely related species Sarada superba, which has a tricoloured dewlap like Sa. darwini, experimental manipulation of dewlap colour and flagging frequency using robotic lizards found that dewlap flagging was sufficient to elicit responses from males even when dewlaps were white like those of Si. laticeps [15]. Reduction in the rate of dewlap flagging towards intruders in general may thus be an effective strategy for Si. laticeps to escape detection from Sa. darwini. In fact, Si. laticeps seems to behaviourally reduce its overall signal conspicuousness in the sympatric population by reducing the intensity of dewlap flagging during courtship as well.

In addition to the shifts in agonistic and courtship signalling behaviour, dewlap size of males in Si. laticeps also differed in the sympatric population. The mean dewlap length of Si. laticeps was about 17% smaller in sympatry compared to dewlaps of the same species in allopatry. In the context of agonistic character displacement, it is essential to rule out the role of hybridization and drift in generating such morphological differences [10]. Sitana laticeps and Sa. darwini show large differences in their hemipenal morphologies, which should ensure pre-zygotic isolation and make hybridization highly unlikely [22] (but see [23]). The effect of drift is also typically strong only when population sizes at range edges are small [24]. However, densities of sympatric and allopatric populations of Si. laticeps are comparable and therefore contribution of drift to the observed differences in Si. laticeps dewlap sizes is also likely to be minimal. More compellingly, shifts in a trait that makes signals less conspicuous to heterospecific competitors, such as dewlap size, is consistent with predictions of agonistic character displacement [10]. Smaller dewlaps, coupled with lowered intensity of both agonistic and courtship signalling, will make signals less conspicuous and should reduce the chances of Si. laticeps being detected and targeted by Sa. darwini. These trait shifts should therefore contribute to reducing interspecific aggression.

Changes to the signalling traits of Si. laticeps in sympatry could also be a strategy for males to resemble females. As Sa. darwini males show no aggression towards Si. laticeps females, shifts in the morphology and behaviour towards the female phenotype may be an effective way for Si. laticeps males to reduce competition with Sa. darwini. Similar examples of males from a subdominant species escaping aggression by a dominant species if they resemble females can be found in pied flycatcher Ficedula hypoleuca and collared flycatchers F. albicollis [25]. Given that fan-throated lizards use dewlaps and flagging for both agonistic and courtship signalling, this trait composite may be important in female choice in these lizards [15]. Thus, reduction in courtship signalling intensity shown by Si. laticeps in sympatry may have consequences for selection by female choice. Theory suggests that female choice can either accelerate or impede the rate of trait divergence depending on the strength and direction of selection [10]. Unfortunately, little is known about female choice in Si. laticeps, and thus how female choice for dewlap characteristics interacts with selection against interspecific competition remains to be determined.

In an extensive review on agonistic character displacement, eight criteria were posed to establish whether trait differences between allopatric and sympatric populations represent a case of agonistic character displacement [10]. By providing evidence of shifts in species recognition and in agonistic signalling traits in ways that can reduce interspecific competition in sympatry, our study satisfies six of the eight criteria (criteria 3–8, table 2). One of the two criteria (criterion 2) that our study does not currently satisfy is the requirement of replicate populations, which establishes character displacement as a non-random evolutionary adaptive event. Satisfaction of this criterion requires access to multiple replicate populations and similar patterns of divergence across all replicates. However, finding comparable replicate populations, especially sympatric ones, can be difficult since zones of overlap can be rare (as acknowledged in [10]). In our case, we know of only three other sympatric populations of Sa. darwini and Si. laticeps. These populations show large variations in relative densities of these species. Since competitor densities independently influence the intensity of interspecific competition, and hence the strength of selection [10], these populations could not be used as comparable replicates for our study.

Table 2.

Criteria to establish agonistic character displacement (as per [10]) and the corresponding evidence from our study.

criteria	requirement to fulfil criteria	support in this study for Sitana laticeps and Sarada darwini
(1) genetic support	common garden experiment	no
(2) not a chance event	sufficient number of replicate populations and statistical significance in trait differences across populations	partial, significant differences in signalling traits of Si. laticeps between one allopatric and one sympatric population was found; no other populations are in similar enough environments to include as replicates
(3) represents evolutionary change	displaced trait mean for sympatric populations lie outside the range of trait means for allopatric populations	yes, mean values for display probability, display intensity and dewlap size of Si. laticeps in the sympatric population lies outside the mean range in the allopatric population
(4) not an environmental effect	character displacement persists after controlling for environmental factors that could influence trait values	yes, all sampled allopatric and sympatric sites were in arid grasslands and lie within 12–14 km of each other
(5) not due to pleiotropy	displacement in competitor recognition in sympatry	yes, Si. laticeps shows reduced probability and intensity of heterospecific agonistic signalling in sympatry
(6) not due to hybridization	interacting species should belong to different genera, or there is strong evidence against hybridization	yes, Si. laticeps and Sa. darwini belong to two distinct genera and show significant differences in their hemipenal morphology, which should ensure pre-zygotic isolation
(7) phenotypic character displacement affects intensity of interspecific interactions	evidence of a shift in competitor recognition using the character that has shifted	yes, Si. laticeps shows reduced probability and intensity of agonistic responses when presented with a heterospecific in sympatry
(8) independent support for interspecific competition	natural history observations or evidence of interspecific aggression and territoriality in sympatry, or from experimental additions or removal of at least one species	yes, natural history observations show interspecific aggression in sympatry, over both females (results here) and territories; staged encounters to directly measure interspecific aggression were conducted within territories of lizards and resulted in aggression by both species to differing degrees

Some discussions of agonistic displacement have also stressed the importance of establishing genetic divergence (criterion 1) [10], for which we currently lack evidence. However, potentially plastic changes, such as difference in probability of responding to heterospecifics and changes in dewlaps sizes, as seen in Si. laticeps, could still play a role in mediating initial steps towards divergence. In this ‘plasticity-first scenario’ [26], the less competitive species (Si. laticeps) may learn to decrease responsiveness to the more competitive species (Sa. darwini), thereby reducing the likelihood of interspecific fighting between the species. For instance, after losing to a competitor, animals generally decrease subsequent aggressive behaviour (i.e. ‘loser effect’ [27,28]). Loser effects can have long-standing consequences for how males approach and court females. If such males come in contact with other allopatric populations [29], loser effects could contribute to pre-zygotic isolation between populations [30]. To tease apart the relative contribution of plasticity and genetic divergence, future experiments could use a combination of reciprocal transplants and sequencing approaches.

In summary, our results show that interspecific competition in particular and interspecific interactions in general play an important role in shaping social signals. Moreover, as interspecific competition between these two lizard species seems to be restricted to the sympatric zones at the edge of their ranges, this can have implications for peri- and parapatric models of speciation. Since shifts in signalling behaviours can directly influence speciation dynamics [2], our study highlights the need to examine speciation dynamics through the prism of interspecific interactions and signal evolution [31].

Ethics

Capture and experimental protocol has been approved by the Institutional Animal Ethics Committee at the Indian Institute of Science (CAF/Ethics/396/2014). Since this study was conducted in India, we did not need a separate approval from University of Minnesota's Institutional Animal Care and Use Committee.

Data accessibility

The dataset supporting this manuscript are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.t4b8gtj0p [32].

Authors' contributions

A.M.Z. and M.T. conceived the study; A.M.Z., A.K. and R.S. collected data. C.O. and A.M.Z scored videos. A.M.Z. performed statistical analyses and wrote the paper with inputs from E.C.S.-R. and M.T. All authors have given final approval for publication.

Competing interests

We have no competing interests.

Funding

This work was supported through grants by the Department of Biotechnology-Indian Institute of Science partnership programme to M.T., and the Ecology, Evolution, and Behaviour Department, University of Minnesota Travel and Research Awards, Frank McKinney Fellowship and Idea Wild equipment grant to A.M.Z.