Author: Terry Ord

I am an evolutionary ecologist interested in how natural selection and sexual selection contribute to the evolution of diversity, and how present-day ecological communities are formed in nature

Dewlap Displays Supersede Headbobs, Yet Again

By Terry Ord

On March 11, 2023

In All Posts, New Research

The dewlap is probably the most noticeable thing about anoles. For me, the best way to spot an anole is by the flash of color from the dewlap as a lizard displays. Without that, many anoles would remain cryptic amongst the vegetation. This seems to be the case for the lizards themselves as well. The burst of color and movement as the dewlap is rapidly extended is a wonderful device for attracting the attention of rivals and mates. It’s possible that the dewlap originally evolved as an attention-grabbing flag to augment an existing sequence of elaborate headbob movements in forested environments. These days, the dewlap is a complex signal component in its own right, often with a dizzying array of colours and displayed using a variety of movements.

Anoles aren’t the only ones with a moveable dewlap. The Southeast Asian Draco lizards have a dewlap, and again to back up the headbob movements that make up their main channel of social communication. There are many other parallels between Draco and Anolis lizards, but the similarities in how they communicate is something that particularly fascinates me.

Early on in my fieldwork with Draco, I started discovering species that didn’t seem to use headbobs as part of their social display. It seemed these species had lost the headbob entirely and instead concentrated all of their communication through the dewlap display. These species are a minority, but not by much. It was a puzzle. These Draco had lost a central and complex element of their communication in favour of something that was seemingly more basic. Communication biologists are often fixated on trying to explain how animal communication becomes more elaborate over evolutionary time, but less attentive to why complexity subsequently becomes lost. These Draco lizards were an excellent case study.

Draco melanopogon (photo above) only communicates using the dewlap, whereas Draco sumatranus (opening banner photo) relies on both headbobs and the dewlap, just like anoles.

After nearly a decade of fieldwork on numerous species of Draco throughout Malaysia, Borneo and the Philippines, my trips stalled in 2020, as did the rest of the world. Celebrities had nothing better to do than write biographies, but my lockdown project was to focus on using the data I already had at hand to finally solve the curious case of the missing headbob.

It felt like an endless series of stay-at-home orders in Australia, and well into 2021 too. While the celebrities had gone on to finish their books and were now doing the zoom promotion circuit, my progress was hurdled by home-schooling two young children. We survived home-schooling in the end, and my attempt at figuring out why some Draco have lost the headbob has finally been published.

The evolutionary history of visual displays in agamid lizards

The first discovery is the headbob display is very ancient, evolving something like 130 million years ago or more. That’s before the evolution of Draco, and before the evolution of the anoles, in an evolutionary ancestor to both the iguanid (new world) and agamid (old world) lizard families. This was back in the age of the dinosaurs. Today, virtually all iguanid and agamid lizards use a headbob display or some variant of it in social communication. Which means the absence of the headbob in a handful of Draco species is very unusual.

The loss of the headbob from the social display of Draco is effectively a loss of complexity. A loss of complexity means a loss of “information potential.” Try writing a biography with half the alphabet. You might manage the following or something a little longer: “I was born. I paid taxes.” Thirteen unique letters in total. Obviously not the rich backstory you might hope for. Not because you hadn’t lived a fulfilling existence; rather you don’t have the language complexity to convey it in detail.

There are various reasons animals might lose complexity in their social signals. Perhaps the original need for a complex signal is no longer present. Perhaps the invasion of a new environment puts a brake on the level of complexity that can be accurately perceived. Or perhaps natural selection on other things, like body size, has made performing a complex signal too costly.

The beauty of having spent so much time in the field is the accumulation of a large library of data. By leveraging this information, I was able to test each of the above scenarios. The short of it is, Draco that have lost the headbob are unusually large species. Physically moving the head and body in a headbob display is more energetically expensive than pumping the dewlap in and out. It seems, then, that the physiological cost of performing the headbob became too great for these large species and they shifted to relying only on the dewlap for communication. This implies the communication system of these species is compromised, unless they have made up the loss of information potential somewhere else.

Draco without the headbob have more complex dewlap colour patterns. Each dot is a different species.

In fact, the dewlap itself tends to be more complex in Draco that have lost the headbob. Stealing a method for measuring complexity of anole displays, the dewlap of these Draco are more elaborately coloured than the average Draco. Unfortunately, this is unlikely to have been enough to fully cover the loss of the headbob. This means Draco that no longer use the headbob are relying on a constrained communication system.

The idea that the headbob is likely to be more energetically expensive than the dewlap was originally proposed for the anoles. It was used to explain the physiological basis for why Jamaican anoles might have evolved an innovation that allowed them to move away from a headbob-centred display in favour of one focussed on the dewlap. To be clear, the Jamaican anoles do still rely on headbobs in their social displays. But a rapid series of dewlap pumps features more prominently in their displays compared to the typical anole, like those on Puerto Rico for example.

It seems the dewlap has begun to supersede the headbob in anoles as well.

If you’d rather not slog through the paper itself, you can view a 12 minute video summary instead. If you would like to slog through the paper and can’t access it behind the paywall, drop me an email and I’ll forward you a free copy (t.ord@unsw.edu.au).

Dewlap Size Is Not What We Thought

By Terry Ord

On October 14, 2022

In All Posts, New Research

The large, colourful dewlap is an obvious defining characteristic of the anole. Understandably, then, there has been a lot of investigation (and speculation) on what the dewlap is used for. Without doubt it’s for social communication, but to communicate what. Historically, the dewlap was thought to be used for species recognition, which remains a reasonable explanation today. But a typical assumption made by many anole researchers and evolutionary ecologists alike is the dewlap, and specifically its size, is effectively an ornament used to attract mates or advertise potential fighting ability among territorial rivals. In other words, the evolution of the dewlap is the product of sexual selection.

If that’s the case, then dewlap size should be linked to some aspect of an individual’s ‘quality’ or physical condition, especially in males who seem to be the ones courting females (not vice versa) or defending territories. This is because a male’s quality or condition can be hard to assess by general appearance alone, unless there is a key feature that provides an honest indicator of that quality. In anoles, this is assumed to be a large dewlap that’s physiologically costly to produce.

One easy way that has been proposed to test for sexual selection in the origin of a morphological structure like the dewlap, is to look how it scales with body size. Structures that are honest indicators of condition will be costly to develop and maintain. Large males are often in better condition than small males because of the underlying factors that result in bigger bodies (e.g., a history of successful foraging, superior growth rate, having ‘good’ genes). This means larger males can invest more in exaggerating the size of the dewlap than smaller males. There would be a clear evolutionary incentive to do so as well, because having a larger dewlap would attract more mates and appear more threatening to male rivals. The outcome of this should be disproportionately larger dewlaps in larger males. This is called positive allometry or hyper-allometry. If dewlap size has a hyper-allometric scaling relationship with body size, then it probably resulted from sexual selection. Or at least that’s the idea. And you can find this out by just measuring a bunch a males.

The dewlap of anoles featured heavily in the original formulation of this idea, with the conclusion being that dewlap size was hyper-allometric and assumed to be the product of sexual selection. Anoles have therefore become a classic example of how sexual selection drives hyper-allometric scaling in ornament size.

Tom Summers

Tom Summers was a graduate student who thought about hyper-allometric scaling a lot. He looked at the scaling relationship of ornaments that he had confirmed experimentally to be the target of sexual selection in fish, and found they were hyper-allometric…sometimes. Tom found natural selection on ornament size can often work in the opposite direction to sexual selection. This is because large ornaments can interfere with locomotion and often be conspicuous targets for predators. When these pressures are high, species tend not to show hyper-allometry in ornaments. Those ornaments were still the product of sexual selection, but their allometric scaling was dampened by opposing natural selection.

Tom turned this attention to the anoles, and found overwhelmingly that dewlap size was not hyper-allometric but hypo-allometric. That is, larger males have disproportionately smaller dewlaps than smaller males. He even looked at another group of lizards that have independently evolved a dewlap, the southeast Asian Draco, and found the same pattern. His results have just been published in the Journal of Evolutionary Biology.

The scaling relationship of the dewlap in both groups varied from one species to another, but never was it hyper-allometric. In the case of the anole dewlap, this variation in dewlap size was predicted by factors important in signal detection (receiver distance and habitat light). This was consistent with the general hypo-allometry of the dewlap as well.

The effectiveness of a visual flag (like the dewlap) in attracting the attention of a receiver (another lizard) is dependent on the gross size of that flag, not how big it is relative to the signaller’s body (i.e., allometric scaling is irrelevant). Beyond a particular threshold size, which is dependent on the visual acuity of the animal in question, there are diminishing returns for detection with increasing size. Even a large increase in dewlap size beyond a certain point wouldn’t really improve signal detection, a phenomenon known as ‘Weber’s Law’. The resulting pattern when comparing dewlap size among males is hypo-allometric scaling. Larger males have generally reached the size threshold for reliable detection, so there’s little point in further elaboration.

It also fits with the extensive amount of work showing that the dewlap is likely to be most important in signal detection, rather than a cue of quality.

So why such a dramatically different finding to earlier investigations of the anole dewlap? All studies prior to Tom’s measured dewlap size by catching the lizard and manually pulling out the dewlap using forceps. Simon Lailvaux has discovered that the skin of the dewlap varies in its elasticity. Larger dewlaps are going to be more stretchy than smaller dewlaps. This means you can probably pull the dewlap out to a larger size in larger males. This would subsequently generate the artifact of hyper-allometric scaling when comparing dewlap size across males of different size.

Tom had measured dewlap size from high-definition videos of free-ranging males fully extending their dewlaps during display. There are various analyses in his paper that confirm this approach provides an accurate measure of dewlap size. His logic at the time was this view of the dewlap would be how lizards actually see and evaluate the size of the dewlap relative to body size. It also meant animals didn’t have to be caught, so the approach was less intrusive for the animal (always a plus). It just happened he avoided the potential problem of over stretching the dewlap if he had caught the animals and manually extended the dewlap by hand.

What does this mean for all that data that has been based on researchers pulling out the dewlap using forceps to measure its size? Honestly, I don’t know. Maybe nothing depending on what the data are being used for. Maybe everything if the data are being used in allometry studies.

Do Anoles Display with Greater Complexity than English?

By Terry Ord

On July 10, 2022

In Explainers, New Research

Above: the territorial display of a male Anolis stratulus on Puerto Rico

Take the time to watch the displays of an anole and you might appreciate how elaborate those signals seem to be. And by comparison to other lizard species, the anole display is arguably one of the most complex. Not only do anoles communicate with the up/down movement of head-bobs, but with the repeated extension of a large dewlap that in itself is often spectacularly colored. These displays are used to convey a variety of messages, from advertising the ownership of territories to the attraction of mates. We know the display is packed full of detail on species identity too.

But how do we go from gut impressions of what is complex to properly measuring the complexity of lizard displays, or any form of animal communication for that matter, including human language? The main way scientists have done this is by essentially counting the number of different components and using that to estimate an animal’s communication repertoire. There are various problems with this, such as deciding which components are different enough to count. It’s the most common method probably because it’s the easiest, but it is also the crudest. It offers only a basic view of signal complexity, missing the complexity inherent within components making up the repertoire.

An alternative approach is to apply some math from physics to measure the information potential of a signal. It is better than simply counting things because it measures the complexity of the entire signal, including the number of different components and the elaborations within those individual components as well. Best of all, it doesn’t require any decisions on what parts of a signal might be worth counting. It also provides a common, repeatable index of complexity that can be used to compare signals from very different animals, such as anole displays and human language.

So, how do anole displays stack up?

Want to know whether Anolis pulchellus on Puerto Rico has the most complex display? Read the paper and find out!

First, let’s consider some other lizards. The head-bob displays of sagebrush lizards (Sceloporus graciosus) are fairly representative of other species of fence lizards, and they clock in at 4.26 bits of information per display. “Bits” is a general unit for the amount of ‘information’ (think data) that can be “potentially” encoded in a signal, or its “information potential” for short. The number is largely meaningless by itself. The songs of birds would probably be the most obvious rivals of complexity in nature. Chickadees have 4.64 to 5.79 bits per song. But is that effectively the same or way more complex than sagebrush lizards? We need more benchmarking.

How about the famous waggle dance of the honeybee? The waggle dance was first uncovered by Karl von Frisch who found that it was a highly accurate signal conveying the direction and distance to an outside nectar source to worker bees inside the hive (google it, it really is super interesting). This discovery later contributed to von Frisch winning the Nobel Price alongside Konrad Lorenz and Nikko Tinbergen. von Frisch was also one of the first scientists to apply information theory to animal communication. The honeybee waggle dance comes in at 7.43 bits per dance. Bees are more complex than birds!

If we apply the same method of measuring complexity to written English, we find it has about 8.12 bits per word. Now let’s recap: sagebrush lizards are 4.26 bits per headbob display, chickadees are 4.64-5.79 bits per song, the honeybee waggle dance (my personal favourite) is 7.43 bits per dance, and written English is about 8.12 bits per word. Those comparisons in themselves are very interesting, but what about anole displays?

We’ve comprehensively measured the male territorial displays of eight different species of Puerto Rican anole and published our findings in Behavioral Ecology. When I say comprehensive, I mean just that: we measured dewlap colour pattern, the way in which the dewlap is repeatedly extended and retracted during the display, the pattern of movement of both push-ups and head-bobs, and a variety of other behaviors often seen accompanying territorial displays (e.g., tail curls and flicks).

The least complex part of the display is the dewlap colour pattern. At best, it encodes 1.02 bits per dewlap pattern. That’s for a dewlap with at least four different colours. The movement of the dewlap during the display—the timing of the in/out movement, how much the dewlap is extended—has far more information potential with as much as 3.87 bits per display. The sequence of head-bobs is the most complex aspect of the anole display and can be as high as 5.11 bits per display. Considering the entire display, the complexity of the territorial display ranges from 6.54 bits per display in Anolis poncensis to a whopping 15.40 bits per display in Anolis pulchellus.

15.40 bits per display! Does this mean anole displays are more complex than written English? Yes! And no. The estimate for written English—8.12 bits per word—was for single words, not a sentence, a paragraph or an encyclopedia. But the fact that anole displays are as complex as they are and might outclass songbirds is truly amazing.

It is contentious as well. During peer-review of our paper, some scientific referees found the reported values hard to digest. All of them thought our numbers for anoles were correct, but couldn’t accept that signals of mere lizards might be more complex than those of songbirds. The comparison to written English drew so much heat that we had to remove comparison to it from the paper entirely. The referees had various reasonable points. One referee highlighted that the value for written English was for single words, not whole texts (fair enough). Another referee suggested our application of information theory was more comprehensive than how it has been previously applied. The implication being other studies have tended to focus on measuring the easiest things and not the full breadth of a song (hmm…).

If you want to find out which Puerto Rican anole species varied most in display complexity and the adaptive explanations of why, or what might have driven the evolution of such complex signals in anoles to begin with, you’ll have to read our paper. Email me and I’ll send you a free copy.

Anoles outclass songbirds? Why not, I say. Perhaps in communicative complexity, but certainly on many other scales.

Anoles Are Powerful Educators, Use ’em!

By Terry Ord

On March 31, 2022

In All Posts, Anole Videos, Education and Anoles

Did you ever read those choose-your-own-adventure books as a kid? I had a whole collection. What if lectures were like that too? Check this one out on anoles (above).

This lecture came about from the need to update a lecture on ecological competition for a second year undergraduate course. In the past, someone might have handed me a textbook and I would have quickly shelved it, never having opened its cover. As a student I hated textbooks and things really haven’t changed for me now as an educator. The real challenge isn’t the content, it’s presenting that content effectively. We’re now on the other side of “the great digital shift of 2020,” but this challenge of engagement remains the same, if not more so. Does this choose-your-own-adventure lecture offer the solution?

Let’s step back for a moment so I can first make the case for anoles…

Anoles first came into view for me way back in my first year of graduate school. Not in real life, of course — there were no anoles at any of my field sites in Sydney. Instead, I happened across a remarkable paper appearing in one of the weekly tabloids. It recounted how researchers had returned to some tiny islands in the Bahamas where a bunch of lizards had been introduced a decade or so before. I couldn’t make head nor tail of the PCA plots or Tables. But the scatterplot later in the paper was clear to even a dunce like me. These lizards had adapted their limbs over a matter of years (years!) to cope with living on spindly bushes. Evolution happening in real-time? Holy cow, this was revolutionary for me. Why am I only seeing this now?

I’d never really thought about adaptation outside of centuries or millions of years. But then my undergraduate experience was the usual, tired textbook fodder of ecology and evolution that never came to life, regardless of how glossy the graphics might have been. My undergraduate experience was mostly about memorising facts and figures, and there was a great mental chasm between those and the real world around me. What I actually saw in nature were animals doing weird and crazy things, so I ultimately gravitated towards animal behaviour for my PhD. But when I discovered this paper, I had just finished reading Richard Dawkin’s “selfish gene” and Dan Dennett’s “Darwin’s dangerous idea,” and I was now fascinated by evolution.

And here was some character named Jonathan Losos, along with his mates Ken Warheit and Tom Schoener, reporting in a glossy magazine called ‘Nature’ years before (in 1997 no less) that evolution happens now, not in the past… Now! If only I had been exposed to this and other stuff like it as an undergraduate. [NB: Jonathan gives a great backstory in his book about how this study almost never left the bottom drawer].

These days I am towards the other end of the student-teacher continuum and I make a point of not teaching from a textbook. First, they are WAY too expensive for students. Second, they are out of date by the time they are published. Third, if classic works are covered (like those on anoles), the format of a textbook makes even the most exciting example remote and dull. My approach has always been to go directly to the source. And anoles offer such a rich collection of content for educators.

But what of this new “choose-your-own-adventure” style format? What is really being achieved here? My sales pitch to you is that it prompts student engagement at strategic points. By doing so, it maintains an active connection between the student and the content. In other words, it should stop students cognitively dropping out while writing copious amounts of notes that they will only ever read just before the exam and promptly forget soon afterwards. By forcing students to direct their own learning experience, they are being subtly pushed to reflect on the content explicitly and intuitively, and they might not even realise it. The hope is they not only grasp the concepts being presented more effectively, but retain (and apply) that comprehension outside the bounds of the course and into the future. And it’s fun too.

Convinced?

The danger is the format could just be a gimmick that’s great as a one-off, but quickly becomes annoying or distracting. The analogy I think of here is the transition from slides to powerpoint in my early conference days at the start of the 2000s. For the ancients among you who remember that time, you might recall having to sit through a plague of animated slide transitions with cheesy swirly sounds as presenters explored the seemingly infinite number of options on offer. Oh, the liberation of going digital! Then most of us eventually realised how annoying and distracting it all was and went back to simpler presentations. Perhaps “choose-your-own-adventure” lectures are the same? Would you have an entire course with choose-your-own-adventure lectures?

Huge thanks to Mike Kasumovic and Arludo for both the hideous yellow shirt and putting the lecture together for me. I use a lot of Arludo’s interactive digital games in my teaching as well – they’re free, the students love them, and they have clear educational outcomes. Evolution, ecology or behaviour, whatever you need, they’ll have something you can engage your students with. Do check them out.

How Do Anole Species Tell Each Other Apart?

By Terry Ord

On March 26, 2022

In New Research, Research Methods

When it comes to finding a mate or defending a territory, animals need to recognise members of their own species. The reasons are intuitive: you only want to mate with your own species to ensure viable offspring, and you should only invest the effort in being territorial when confronted by rivals from your own species. There are exceptions and these are interesting—hybridization or territorial competition between species—but generally animals need a system for species recognition.

The large, often spectacularly coloured throat fan or dewlap of anoles seems like an obvious way to evaluate species identity. Taxonomists have historically thought so, too. Each species appears to display a dewlap that’s unique in colour and pattern. But there are various Anole Annals posts highlighting this is not always the case. Instead, the colour of the dewlap is often an adaptation to the light environment for enhancing the detection of territorial displays.

So what about those territorial displays? Might anoles use the complex movements of the head-bob and push-up display to figure out species identity?

Classic work by Charles Carpenter and Tom Jenssen revealed how often the head-bob movements of lizards, and anoles in particular, seemed specific to each species. Pioneering experiments using video playback by Joe Macedonia in the ’90s has also provided evidence that anoles are able to distinguish displaying rivals of their own species from those of other species. But what is it about the pattern of movements used in the head-bob and push-up display, or even how the dewlap is extended and retracted, that conveys species identity? Is there one feature that varies the most among species that anoles commonly rely on to identify species?

Display-Action-Pattern graphs (above) showing the complexity of movements used by Puerto Rican anoles for territorial advertisement displays

These are hard questions to answer. Anole displays are complex, using many different types of movements, so there’s a huge number of possibilities. One approach would be to isolate and manipulate each type of movement and use video or robot playbacks to ask the anoles themselves. But doing that would take an entire career. There are a seemingly infinite number of combinations to consider. In fact, it would be impossible without a way to narrow things down.

Claire Nelson is a creative (and courageous!) graduate student who had an eye for solving the challenge. She figured it was possible to leverage the large archive of footage I’d accumulated over many, many years. These videos were of free-living male lizards performing territorial advertisement displays. Her idea was to develop an objective method for identifying which movements used in the head-bob, push-up or dewlap display had the potential to convey species identity. She’s just published her solution in Animal Behaviour.

Claire (above) doing a balancing act with some non-anoles

Claire used this archive of display videos to create Display-Action-Pattern graphs, a method developed by Carpenter back in the 60s. These track the up-and-down movement of head-bobs and push-ups as well as the extensions and retractions of the dewlap during the territorial display. To keep it manageable, she limited her efforts to anole species on Puerto Rico, and graphs of about 10 territorial advertisement displays per male. But there was an important biological reason for selecting this number of displays as well. It effectively mimicked the number of displays an anole might typically see on first encountering another lizard. That is, anoles likely make judgements on species identity from only a handful of displays.

From these Display-Action-Pattern graphs, Claire took a host of measures, ranging from the duration and number of movements used, to variation in amplitude and pauses between movements. She also noticed that anoles tend to perform certain combinations of movements together in what she came to call ‘motifs.’ After many many hours of effort, Claire accumulated a huge amount of data for nearly 20 different types of display movement for eight Puerto Rican Anolis species, and in many cases, for different populations of the same species.

Claire asked me for advice on how to analyse it all. I have to admit I was completely useless on this front. I muddled something about using coefficients of variation and some other nonsense, but really I had no idea. I was still in shock over how much data she had accumulated, and the novelty (and implications) of discovering motifs in the displays. She knew what she was doing, though. Her analytical solution was vastly superior to anything I could have suggested.

Claire investigated a variety of approaches, but in the end she settled on the method of random forest tree classification. It’s a sophisticated machine learning algorithm that, in a nutshell, takes data and groups like with like. It doesn’t require any prior direction or preconceived notion on how data should be grouped. It just uses the variation in the data itself. You could view the algorithm as an anole brain using basic rules of variation to make judgements on which displays are likely to be different and which displays are likely to be the same.

The outcome was impressive. The algorithm correctly assigned the vast majority of lizards to their correct species based on just a handful of displays. Where errors occurred, it was partly because lizards were assigned to the right species, but the wrong population. This means anoles from different populations tend to share some display features because they’re still from the same species. Yet the algorithm was able to correctly assign most lizards to the right population. In other words, there was still enough variation in the displays between populations of the same species to identify them as belonging to separate populations. This is very interesting!

Random forest tree classification (above) can assign over two thirds of displaying lizards to their correct species.

The evolution of new species begins with individuals of the same species starting to segregate from each other in some way. Often it’s physical separation (on opposite sides of a mountain range), but changes in social signals can also prompt behavioral separation as well. This could be the case for some anoles on Puerto Rico. Once individuals stop recognising each other as the same species, they no longer reproduce with one another, and the door to speciation is propped open.

The other discovery Claire made was the apparent lack of any common display feature that could be used to identify species (and population identity). Instead, different features were important for different species. The duration and number of headbob movements were features that could be used to identify the territorial displays of Anolis poncensis—a species that is striking in its use of lots of extremely rapid, up and down body movements—whereas the way the dewlap was extended was influential in identifying different populations of Anolis gundlachi—a species that has an unusually long dewlap display. Other species like Anolis pulchellus and Anolis krugi were best identified by effectively considering features of the entire territorial display.

Whether or not anoles actually use the features identified by the algorithm in species recognition remains an open question. But Claire has managed to identify the potential candidate cues that could be used. It is now possible to develop a focussed research program to test whether, and how, anoles used these features to identify species. Again, the obvious way to do this would be to ask the lizards themselves using robot playbacks.

Random forest tree classification sounds awfully complicated, and it is very sophisticated, but it’s actually easy to implement. Any dummy can do it. I taught myself how and wrote a step-by-step tutorial so you can as well. We’ve published this tutorial alongside Claire’s paper in Animal Behaviour. Give it a whirl!

More Astonishing Parallels between Draco and Anolis

By Terry Ord

On July 9, 2021

In New Research

Despite being on opposite sides of the world and separated by millions of years of evolution, Draco and Anolis lizards have converged on many common adaptive solutions.

It has long been suspected that the Draco lizards of Southeast Asia were ecological analogues of the Anolis lizards in the Caribbean. But it has only been recently that we’ve started to truly figure out just how similar these two groups actually are. Especially exciting has been the discovery of how lizards in both groups have converged on remarkably similar adaptations: natural selection appears to repeat itself, time and time again.

The classic textbook scenario in anoles is adaptive convergence in ecomorphology, which reflects where lizards tend to hang out in the environment. This is particularly obvious in the adaptation of Anolis limb morphology to different types of perches. I’m not sure we can say whether Draco exhibit the same ecomorphs as Caribbean anoles just yet, but Draco do exhibit the same adaptive changes in limb morphology as the anoles of the Caribbean.

Then there’s the obvious case of the dewlap, which has evolved separately in both Anolis and Draco as a key component of their territorial and courtship displays.

Most Anolis and Draco lizards also rely heavily on elaborate sequences of head-bobs in these displays as well.

This in itself isn’t unusual. Many lizards head-bob and push-up over territory and mates. Up to now, however, Caribbean Anolis lizards were the only ones known to tailor their display movements to ensure detection. For any visually communicating animal, visual “noise” from windblown vegetation and poor ambient light make it difficult to see visual signals. Anolis lizards were exceptional because they actively monitored environmental conditions and exaggerated their body movements when it was visually noisy (e.g., on windy days) or extended the duration of their displays when light levels diminished (e.g., on cloudy days or deep inside a forest).

There are a countless ways lizards might produce a conspicuous signal. Many Australian and Chinese lizards add arm-waves or elaborate tail-flicks or simply rely on colourful ornamentation that stands out well in the environment. Many North American lizards rely on performing lots of head-bobs or positioning themselves to accentuate a colourful badge on their throat or sides. But anoles seemed unique in both the complexity of their visual displays and their capacity to modify their behaviour to the prevailing conditions in the environment.

But after years of studying Anolis on Jamaica and Puerto Rico, and then even more years studying Draco in the Philippines, Borneo and Malaysia, we have now discovered more astonishing parallels between the two lizard groups that extends beyond just morphology.

Southeast Asian Draco lizards exhibit virtually identical strategies for coping with visually difficult environments as do the Caribbean Anolis lizards. Draco use the dewlap in the same way as the anoles, and change the speed (or exaggeration of movements) and the duration of displays in the same way as anoles, and this capacity to tailor displays to the conditions of the environment has also tended to precede what seems to have been adaptive divergences in display behaviour among species.

To discover all this, we had to study many different species of Draco and Anolis (11 and 12 species respectively), including hundreds and hundreds of lizards (727 to be precise), and then conduct thousands and thousands of hours of video analysis (13,310 hours – !!).

To be perfectly honest, what I was hoping to document from all this work was how differences in evolutionary history between the Anolis and Draco had shaped the trajectory of display evolution. Sure, Draco had evolved a dewlap like Anolis, but how that dewlap has been morphologically constructed was quite different between the two groups. I had become quite interested in these so-called “many-to-one, form-to-function” outcomes in evolution, and I was aiming to show something similar for display behaviour.

To be clear, there were differences in how Anolis and Draco lizards responded to environmental conditions, and how plastic changes in behaviour have contributed to display differentiation among species. In fact, the head-bob component of the territorial display has been entirely lost in some Draco species.

But the similarities were stunning and outweighed the differences by a large margin. Even the loss of headbobs in some Draco have intriguing similarities to how some Anolis species have shifted their display effort to the dewlap, which seems to be a more energetically efficient means of producing a conspicuous, complex visual display, than the more tiresome head-bob and push-up movements.

We have also confirmed experimentally in anoles that the manner in which Anolis lizards tailor their displays does actually improve display detection in visually difficult environments. This took a lot of work in itself and required the development of a robotic playback system, but this is now ancient history.

But to complete the loop, a similar type of playback experiment needed confirm the same adaptive benefit in Draco.

Some years ago we had conducted a lengthy field experiment using robot playbacks that were designed to test the response of Draco lizards to different coloured dewlaps. That experiment showed little effect of dewlap colour on detection, but a tangible effect that once lizards saw the dewlap, they used it to evaluate the species identity of the signaller.

I was in the lab one day looking at these old Draco robots to get some inspiration for designing a new system for some other crazy idea I had. As I was fiddling with the mechanism, I noticed that the robots weren’t exactly the same, with the lever controlling the dewlap of one being slightly longer than another. This meant the display probably differed in speed between the robots. These things happen and I didn’t think much of it at the time. The treatments used in the field experiment were systematically inter-changed across the robots to make sure this type of thing didn’t cause any problems.

Later, however, it occurred to me that perhaps this might offer a serendipitous opportunity to confirm the adaptive benefit of at least one of the key convergences exhibited by Draco lizards. I downloaded the data from the original study from its dryad repository, extracted the response times of lizards to the two robots that differed in dewlap speed, and sure enough, detection times were much quicker to the robot with the faster dewlap display.

The top panel (a) shows the differences in dewlap speed between the two robots, while the bottom panel (b) shows the detection time of free living Draco melanopogon.

If you’re interested in a short video introduction to this work, or want to know more about how these findings relate to our general understanding of adaptation and animal communication, you’ll find some answers in this 5 minute video below.

Robot Lizard Army versus Deadly Predators

By Terry Ord

On June 9, 2021

In New Research, Notes from the Field, Research Methods

Over the years, there has been a lot of discussion on Anole Annals about the large, conspicuous dewlap. And rightly so because it is arguably the most evocative feature of the anoles. Much of this discussion has focussed on its function, such as its role in species recognition, mate choice, and territorial communication. But is there a cost to having such an audacious visual signal?

We needn’t isolate this question to just Anolis lizards. All socially communicating animals need to produce a signal that will be obvious to conspecifics. There’s little point producing a mating or aggressive signal if females or rivals never detect it. But there is a cost to being conspicuous and it can be a matter of life and death: the unintended attraction of predators.

Generally, the assumption has been that animals just incur the potential risk of predation for the sake of successful communication. But just how risky is it? The dewlap is often large and brightly coloured, but when it’s not being used in display, you’d never know anoles even had one.

There are also at least two other independent origins of the dewlap, including in the gliding lizards of Southeast Asia, the Draco. In these lizards, the dewlap is again large and often conspicuously coloured.

For both Anolis and Draco, one of the best ways to find lizards in the wild is by the quick flash of colour as males rapidly extend and retract the dewlap during their territorial displays. In fact, it is often the only way to find Draco, which are camouflaged and extremely difficult to spot, even when you happen to be staring right at them.

I had this crazy idea a few of years ago… Would it be possible to build an army of robotic Draco lizards with plasticine bodies that could retain impressions of predator attacks and measure the risk of predation from performing a conspicuous dewlap display?

It really was a ridiculous thought, but my long-time collaborator Indraneil Das was game.

And it worked, with the results just published.

Robotic lizards compared to the real thing in (a) morphology and (b) behaviour (robots were modelled on Draco sumatranus from Borneo).

It was an awful experiment to do. Building the robot army turned out to be the easy bit. To be clear, it took months of development and manufacture, all of which I did in my garage (long story). It then took years to run the experiment, with multiple replications across two continents because the data was puzzling. There were bushfires, floods, battles with swarming wasps and kamikaze leafcutter ants, chipped teeth, falls from ladders, bogged car rentals, hammered thumbs, and in the end I only just managed to get it finished before the world turned side-ways in 2020.

Left: fresh-faced and optimistic in June 2018; Right: brave-faced but really a little shellshocked with the retrieval of robot 2,120 in February 2020 (NB: batteries have a habit of failing and parts started to corrode so only 1,566 robots were fully functional in the experiment).

It turns out that prey that can produce a signal intermittently — effectively turning their conspicuous display on and off at strategic moments, like the dewlap — can drastically reduce their risk of predation. In fact, attack rates by predators on dewlapping robotic lizards were no different to robots that remained unmoving and cryptic in the environment. Which means there doesn’t really seem to be a large cost from increased predation for animals that perform bouts of conspicuous behaviour.

But this wasn’t the biggest surprise.

The experiment included robotic lizards that kept the large, conspicuously coloured dewlap permanently extended so it was always visible. Think of peacocks with their massive tail trains or other animals that are spectacularly ornamented. These features are always visible and are not signals that can be turned on and off. My assumption was that these robotic lizards would be the hardest hit by predators.

This wasn’t the case at all. Predators actually avoided these robotic prey and to such an extent that the probability of attack was lower than the robotic lizards that remained cryptic and didn’t perform any conspicuous behaviour.

Photo montage of predator attacks left in the plasticine body of the robotic lizards

At first, I found this to be confusing and replicated the experiment over and over again. I even called in my partner Katrina Blazek who is a biostatistician to blind the data and independently perform the analyses (Katrina is also a skilled tailor and made all the robot dewlaps). I also dragged in my colleague Tom White who is an expert on animal colour discrimination to confirm that the dewlap really was as conspicuous to predators as I thought it was.

The data was robust.

This type of predator phobia actually helps explain the evolution of a completely different type of animal signal in nature: aposematic signals or warning signals that some prey evolve to explicitly advertise their location to predators to warn them against attack, usually because they’re toxic. Conspicuous poison dart frogs are an obvious example, so are ladybirds (or ladybugs).

The paradox is how these warning signals could evolve in the first place given the first individuals that tried to advertise their warning would be quickly eaten by predators that had no idea the signal was meant to advertise unprofitably until after the attack.

One of the key hypotheses that has been proposed to resolve this evolutionary paradox is that predators are highly conservative in the types of prey they go for. That is, they tend to avoid prey that look unusual in some way, even if those prey are more easily detected.

This is exactly what happened in this experiment. The robotic lizard with the permanently extended dewlap was ‘weird’ and so predators instead targeted the robotic lizards that either displayed intermittently or remained cryptic, both of which were more typical of their familiar prey.

The take home message is:

Follow your ridiculous idea and call on your friends to help.

(But don’t hold metal tools between your teeth. Your dentist will be very annoyed with you.)