Author: Jose Ramos

Figure 1 – Long-nosed (Gowidon longirostris) dragon performing a territorial.

In the Greater Antilles, lizard radiations have produced the same suite of ecomorphs on different islands as a consequence of adaptations to similar environments. In the same way, species that use motion-based signals, and occur in sympatry, would be expected to develop similar adaptations to enhance signal efficacy as they are frequently exposed to the same environment (e.g. background noise). Additionally, sympatric species often develop mechanisms to ensure they can distinguish between conspecifics and heterospecifics, particularly if they are closely related. This means that potentially opposing selective pressures might be at work for such systems.

Agamid lizards are widespread across the Australian mainland, and species distributions regularly overlap, especially in arid and rocky habitats. We analysed the motion-based signals of two pairs of sympatric species of Australian agamids to consider how they maintain reliable communication, while at the same time they avoid misidentification during signalling interactions. We calculated the speed distributions of the motion produced by lizard signals, and also by the environment (i.e. background noise). We then compared these two sources of motion to obtain a measure of signal-noise contrast, which indicates how much the signals stand out from the background and is therefore a proxy for signal efficacy (see Ramos & Peters 2017a).

The ring-tailed dragon (Ctenophorus caudicinctus) and the long-nosed dragon (Gowidon longirostris; Figure 1) are often found in sympatry in south Northern Territory and southeast Western Australia, around gorges and rocky outcrops. We recorded territorial displays at West MacDonnell National Park, in Northern Territory. The two species differed in display complexity (example of displays by all four species) and motor pattern use, as well as overall morphology (Figure 2). Interestingly, the speeds produced during their displays (Figure 3) and their signal-noise contrast scores were strikingly similar. Not only that, but their scores indicate that the signals from both species are highly effective in the context of the plant environment. These results demonstrate similar adaptations to their shared environment, while maintaining species recognition cues through morphology and overall display appearance.

Figure 2 – Habitat, average snout-vent length and known repertoire of core motor patterns for both species pairs. The core motor patterns refer to HB = head bob, LW = limb wave, PU = push up, TC = tail coil, and TF = tail flick. Ctenophorus caudicinctus has been observed performing limb waves, but this motor pattern is not present during its territorial displays. (Figure adapted from Ramos & Peters 2017 Journal of Comparative Physiology A)

The military mallee dragon (Ctenophorus fordi) and the painted dragon (Ctenophorus pictus) are very common in arid and semiarid sandy areas of northwest Victoria, South Australia, and southwest Queensland. We recorded displays at Ngarkat Conservation Park in South Australia, where they are often found in sympatry. These two species are much closer in appearance, but their display complexity and motor pattern use were just as contrasting as in the previous pair of lizards (Figure 2). In addition, the speeds produced during their displays and their signal-noise contrast scores were considerably higher in the painted dragon (Figure 3). We suggest this difference is related to the lack of territoriality in mallee dragons. This species is not known to protect territories or perform aggressive displays, so the motivation to produce conspicuous signals is likely to be reduced compare to its territorial relatives.

Figure 3 – Comparisons of the motion speed distributions for all species. Kernel density functions for a) Ctenophorus caudicinctus (red) and Gowidon longirostris (black), and b) C. fordi (red) and C. pictus (black), averaged within species. (Figure adapted from Ramos & Peters 2017 Journal of Comparative Physiology A)

In this study we were able to show that the ring-tailed and long-nosed dragon perform displays with almost identical motion speed distributions and signal-noise contrast scores, despite utilising very different territorial displays (see Ramos & Peters 2017b for more details). In the case of the other sympatric pair, motion speed distributions and signal-noise contrast scores appeared to be much higher in the painted dragon than in the non-territorial mallee dragon. This difference in social behaviour could be key to explaining why the signals of the sympatric C. caudicinctus and G. longirostris seem equally well adapted to their local environmental noise, as evidenced by their equally high signal-noise contrast scores, but the signals produced by C. fordi and C. pictus do not. Thus, the selective pressure to generate signals with high efficacy appears to be mediated by signal function, at least in this context.

Habitat characteristics influence the efficacy of animal signals, which means that populations of the same species occurring in distinct habitats are likely to show differences in signal structure as a form of local adaptation. This kind of variation in signal structure has been well-studied for sound and colour signals, including in several species of anoles, but had not been reported for motion-based signals until recently.

Jacky dragons (Amphibolurus muricatus) are Australian agamid lizards well-known for the complex motion-based displays performed by males. These displays comprise five distinct motor patterns utilised in sequence: tail flicks, backward limb wave, forward limb wave, push up and body rock (A. muricatus display video). A study conducted by Barquero et al. (2015) found evidence of temporal and structural variation in the core display of three populations of A. muricatus. These differences were not related to genotypic differences between populations, so they suggested they might be a consequence of local habitat structure.

The Jacky dragon

Concurrently, Richard Peters and I were developing a methodology to accurately quantify the effect of background noise on the motion based signals of different Australian agamids (see Ramos & Peters 2017a; b). Our approach calculates the speed distributions of the motion produced by lizard signals and the environmental noise independently. It then compares these distributions to obtain a measure of signal-noise contrast. This is accomplished by recording lizard behaviour and reconstructing its motion in three dimensions before comparing it against the motion produced by the surrounding windblown plants, which are the main source of noise for motion based lizard signals. This methodology stands out from other approaches for quantifying motion signals because it does not assume that the camera is ideally placed when recording the displays, but instead provides an accurate representation of the motion from any angle or viewing position.

Building upon the work by Barquero et al. (2015), we applied our novel approach to a couple of populations of Jacky dragons with distinct habitat characteristics. Croajingolong National Park in Victoria (Australia) is densely vegetated coastal heath with tall grasses and shrubs on a sandy substrate. Conversely, Avisford Nature Reserve in New South Wales (Australia) is mostly open woodland with an understory of scattered grasses and small shrubs, and rocky outcrops spread throughout the park.

The habitats of (a) the Jacky dragon. (b) Croajingolong National Park, in coastal Victoria, Australia. (c) Avisford Nature Reserve, in New South Wales, Australia.

Our results revealed that lizards from the densely vegetated habitat (Croajingolong NP) performed displays of longer duration and introductory tail flick components, and also produced a significantly greater amount of high speeds. However, when we calculated the signal-noise contrast for both populations at their respective habitat, we found no difference. This means that the signals from both populations are equally effective when used within their intended habitat, regardless of their structural differences.

Differences in signal structure between populations. (a) Mean bout and tail flick durations for both lizard populations. (b) Mean tail flick to bout ratio for both lizard populations. (c) Average kernel density functions for both lizard populations.

As mentioned before, our approach records animal signals and environmental noise independently, which allowed us to consider signals not only in the environment where they were filmed, but also in the habitat of the other lizard population. Consequently, to highlight the effects of the environment on lizard signals, we calculated signal-noise contrast for the signals belonging to one population in both habitats (densely vegetated vs. open woodland). As expected, both lizard populations performed worse in densely vegetated habitat, probably because the complex understory is producing greater motion noise and negatively affecting signal efficacy. Another way of looking at these data, but this time focusing on the displays rather than the habitat, was to compare the signal-noise contrast of both lizard populations in a single habitat. Lizards originating from the densely vegetated habitat produced higher contrast scores in both habitats, indicating that their displays are more effective overall.

Taken together, our results are consistent with the local adaptation hypothesis. Lizards from Croajingolong NP produce displays with longer durations and characterised by faster speeds in order to communicate effectively in a dense and noisy habitat. Conversely, lizards from Avisford NR have adapted to a less noisy environment and do not require such lengthy or energetically expensive displays. Such population level differences in signal structure due to habitat variation represent novel findings for motion-based lizard signals.