Can studies of living animal colour constrain the colours of dinosaurs? A case study with big theropods

Mighty Tarbosaurus bataar carries its Therizinosaurus cheloniformis prey through a Masstrichtian forest in the rain. Colour-wise, I've decided that a ruddy-orange dorsum, light venter, disruptive black stripes, rings and large eye masks are meant to hide this 10 m long animal during predatory acts. But without direct palaeocolour data for Tarbosaurus, could we ever evaluate how sensible this colour scheme is?

Today we’re looking at one of the most commonly asked questions about restoring extinct dinosaur appearance: colour. For centuries, queries about the colours and patterns of dinosaurs, and, indeed, most extinct vertebrates, have been effectively non-answerable, save for some arm waving about the merits of camouflage for predation and display patterns for social signalling. Nowadays, advances in analyses and understanding of fossil pigments have allowed us to reconstruct the foundation colours of several dinosaurs in detail, along with those of other popular taxa like pterosaurs and marine reptiles (see Vinther 2015 and Smithwick and Vinther 2020 for overviews). This new frontier in dinosaur science has helped to flesh out not only the life appearance of dinosaurs, but also their ecology: their habitat preferences, their daily activity patterns, their predation concern and so on (e.g. Vinther et al. 2016).

Deducing dinosaur colour to this level of precision requires exceptionally high-quality preservation of their skin, down to the microscopic level, so that their pigment cells (melanosomes) can be identified. Unfortunately, this excludes the vast majority of dinosaur specimens from such analyses. Dinosaur skin is not only rare, but often occurs as mere sediment impressions rather than films of geochemically-preserved organic matter. This preservation style applies to a great number of the most famous dinosaurs so, unless some radical new science finds a way to assess colour from skin texture alone, the colours of our favourite extinct saurian taxa will probably be lost to time forever.

But can we tackle this problem from another angle? In recent decades, biologists have made enormous strides in understanding living animal colouration, looking at how it relates to habitat preferences, camouflage, signalling behaviour, body size, posture, visuality acuity and so on. Has the science around modern animal colouration advanced to the point where we can start to make tighter predictions about the colours of extinct animals? I regard this as an important question because, as much as our depictions of dinosaur anatomy have tightened since the late 20th century, our application of colour is still pretty lawless, even among professional palaeoartists. We present the same animals with colour schemes that are totally adaptively opposed to one another — one artist’s vivid blue hadrosaur is met with another’s muted browns and reds — and yet they’re both meant to be of equal scientific credibility. But how can that be so? Colours and patterns are generally thought to be under the same adaptive pressures as other parts of animal anatomy and thus should correlate, to a greater or less extent, with aspects of behaviour and ecology. There probably is, at some level, a "right" and "wrong", or at least a "likely" and "less likely" aspect to colour restoration, just as there is with all other aspects of palaeoartistry. But how can we evaluate this without palaeocolour data? Enter, stage left, the last two decades of studies of living animal colour. Can they help constrain, even in a general way, our efforts at colouring animals from Deep Time?

What flavour Australovenator wintonensis is your favourite — red, green, blue or orange? It's strange that, as consistent as we're getting with depicting dinosaur anatomy, you could present any one of these contrasting colour variants with equal scientific validity, even though they each imply very different interpretations of Australovenator biology.

Time for a case study

To investigate this, I thought we could look at a well-known group of dinosaurs to see what, if anything, living animals might suggest about their colouration. As you’ve guessed from the article title, we're using big predatory theropods for this exercise, for several reasons: 1) they’re a popular art subject, so this article should be of wide interest; 2) as regular readers will have worked out, I’m currently involved in a few big theropod projects so have been drawing them fairly continuously for a while now; and 3), the biology and ecology of big theropods are comparatively well-researched, and that helps when plugging fossil data into models of extant animal colour. And yes, we could restrict this to a more specific theropod clade but, from what I know about giant predatory dinosaurs, I’m not sure the conclusions we’d draw for big allosauroids would be much different to those of tyrannosaurines or large megalosauroids. If we’re sticking to what we know about these animals, not what we speculate and imagine about them, they only offer so much data to compare against living species.

There are plenty of caveats with this comparison, of course. No living creature is ecologically or phylogenetically close to the largest Mesozoic theropods, and our modern environments are different to those inhabited by our case study subjects. But we might also consider the importance of uniformitarianism, the adage that “the present is the key to the past”. We can’t say whether modern animals are perfect models for the colour of Mesozoic species, but they offer the only large, statistically-viable sample size of biological colour for us to work with. We are surely better off making informed guesses about extinct animal appearance using modern species as a guide, dodging known pitfalls where we can, than simply speculating wildly.

More worrying than concerns about comparing the past with today is that the controlling factors of animal colouration are extremely complicated, and it’s not clear how we can account for this. Indeed, for all of our science and ideas around animal colour, we still have lots to learn about it. Many popular, widely communicated interpretations of animal colours and patterns are only now being experimentally evaluated (Caro 2005), which means we are still struggling to understand some foundational aspects of certain colour schemes (Caro 2013). This is especially the case for predatory species, the colours of which have been relatively unexplored compared to those of prey animals (Pembury Smith and Ruxton 2020). To that end, we must temper our expectations. As neat as it would be to pour details like extinct animal size, habitat preference and trophic level into an algorithm to receive — *ping!* — a series of likely colours and patterns, our conclusions here, if any, are going to be of a more generalistic, broader nature.

Camouflage and detectability in large living predators: what does it mean for big theropods?

For all the new work that’s been done on animal colour, we still recognise that the principal pressures on animal colouration are essentially what Darwin observed in his 1871 book The Descent of Man. This is a conflict between natural selection, which promotes colour configurations that help animals remain undetected by predators, avoid temperature stress and generally survive from day to day, and sexual selection, which promotes the adoption of bold, broadcasting colours and patterns that attract mates and deter social rivals. So the first thing we might explore for big Mesozoic dinosaur predators is how our largest living terrestrial carnivores express this conflict: are they more concerned with basic natural functionality or sexual signalling? We're specifically interested in our giant theropod ecological analogues here: big animals that hunt and kill relatively large prey items. Predators that subsist on smaller, bite-sized animals don't qualify, because their ecology isn't sufficiently similar.

A selection of the largest predatory animals of modern times and their camouflage schemes, universally showing a strong adaptive emphasis on concealment regardless of habitat type, phylogeny or locomotor method. A, Ora, or Komodo dragon Varanus komodoensis (background matching); B, lion, Panthera leo (background matching); C, tiger Panthera tigris (disruptive colouration); D, polar bear Ursus maritimus (background matching); E, saltwater crocodile Crocodylus porosus (background matching); F, golden eagle Aquila chrysaetos (background matching); G, great white shark Carcharodon carcharias (countershading). All images from Wikimedia: A, Yuliseperi2020, CC BY-SA 4.0; B, Bernard DUPONT from FRANCE, CC BY-SA 2.0; C, Charles J. Sharp, CC BY-SA 4.0; D, Andreas Weith, CC BY-SA 4.0; E, fvanrenterghem, CC BY-SA 2.0; G, Juan Lacruz, CC BY-SA 3.0.

Across vertebrate groups, and across habitat types, our biggest modern predators are pretty consistently (maybe entirely consistently) primarily coloured for concealment: that is, they have camouflaging colours and patterns which hide their presence from their prey. This applies as much to mammals, which are a relatively drab group overall on account of several ecological and physiological factors (Caro 2013), as it does to clades that have the adaptive capacity to produce the most brilliant and striking colour schemes in nature, such as lizards, snakes, birds and fish. So maybe that’s our first note: big predators in the modern day are all about cryptic colouration, with little in the way of conspicuous display patterning.

Research on the impact of body size on predatory ecologies sheds light on why big predators seem to be consistently camouflage-coloured, and it’s a simple explanation: bigger animals are generally more conspicuous than smaller ones, even when they're trying their best not to be seen. The relationship between predator size and concealment capacity is still being investigated but a trend between size and conspicuousness seems to apply widely across Animalia, even in species with famously adept camouflage adaptations, like chameleons (Cuadrado et al. 2001; Pembury Smith and Ruxton 2020). Size doesn’t just affect detectability, either: it also correlates with prey response. Bigger predators instigate more vigorous reactions than smaller ones, such that prey species react sooner, flee further, or initiate more aggressive counter-responses (Stankowich and Blumstein 2005). There are strong pressures, therefore, on big predators to do what they can to remain hidden. Their size already puts them at a disadvantage for stealthily approaching prey, and they are going to have to run further or fight harder once they give up their hiding spot. Given that the largest theropods are the biggest terrestrial predators that have ever lived, we have to wonder what this link between body size and cryptic capacity implies for their colouration. Is one obvious inference that big theropods needed all the help they could get to remain inconspicuous? Would predators already handicapped by their greater detectability and exaggerated prey responses really have some of the signalling-dominant, hyper-obvious colour schemes we've given them from time to time?

Giganotosaurus adapted for the open county with high-bodied, sharply marked countershading, from my recent post about the possible facial anatomy of this animal. But note the ornament on this animal's head: I feel I gave it a pretty meaty set of soft-tissues around its snout and eye, but Giganotosaurus is still pretty undecorated compared to some theropods. Is this something we can read into — does the extent of cranial ornament tell us something facial colouration?

While fossils do not tell us anything about this correlation directly, I wonder if some anatomical evidence points to larger predatory dinosaurs aiming to be less conspicuous. Mid-and large-sized theropod fossils tend to have bony cranial ornaments more often than smaller ones (Gates et al. 2016), but in my estimation (by which I mean, this hasn't been verified by any study), the ornamentation in very large species is generally reduced and less spectacular than that of their smaller cousins. In tyrannosauroids, for instance, we see a general shift away from tall midline cranial crests in smaller, earlier species towards low-relief rugose surfaces, small horns or blunt bosses in larger taxa (Gates et al. 2016). Indeed, the very largest theropods are some of the dullest-looking, at least in terms of cranial ornament. Consider the flattened orbital bosses and rostral rugosities of Tyrannosaurus and Tarbosaurus, or the low, corrugated textures over the snouts of giant carcharodontosaurids. We can only speculate on what impact these ornaments might have had on theropod camouflaging efforts, but it’s well-established that distinctive body outlines can increase detectability, to the extent that modern predators attempt to hide them from their prey where possible (see below).

Whatever their adaptive significance, these reduced facial ornaments give us grounds to think about cranial colouring. Faces are often sites for signalling patterns and colours in modern species (e.g. Caro et al. 2017) and a reduction in bony facial ornament could indicate a lessened emphasis on this behaviour, possibly including muted facial colouration. A caveat here is that elaborate osteological features are only ever suggestive of striking colours and patterns, not directly correlated. But part of the palaeoart game is looking for clues about the nature of these animals wherever we can, and an absence or reduction of showy features is something we can factor into the reasoned speculation we must utilise when creating colour schemes.

I don't think we make enough of how display-adapted Spinosaurus aegyptiacus was, and how weird that is for not only a giant dinosaur predator, but any giant predator. Here, a gaggle of Spinosaurus show off their sails and tails, display structures (well, probable display structures, in the case of the tail) that use almost every inch of their axial length for showing off. So how does this fit into your "big theropods were camouflage-colour dominant" narrative, smart guy?

The elephant in the room here, of course, is Spinosaurus, which is highly unusual for being a giant apex predator with the same tailor as a peacock. This was a carnivore with an unprecedented disregard for remaining inconspicuous or having an anonymous body profile. For all the controversy over this animal, one aspect we all agree on is that its enormous sail was a sociosexual display device (see Hone and Holtz 2021 for references and discussion). Doesn’t this doesn’t torpedo the wider point being made here about predator size and possible camouflage needs? On the contrary, it might support it. As something straddling the terrestrial-aquatic realm, normal rules about camouflage and crypsis may not have applied to Spinosaurus. We see this evidenced in modern times in that the "rules" of camouflage in terrestrial settings are not the same as those of aquatic habitats (Caro 2013), and we should probably allow for, or even expect, some weirdness from animals operating at that interface. The atypical ecology of spinosaurids may have liberated them from the adaptive pressures experienced by purely terrestrial dinosaur predators, allowing them to become more ornamental and spectacular. Perhaps the fishy prey of Spinosaurus barely saw the full outline of their largest predator, an especially viable idea in the (I think, superior) “giant heron” ecological model favoured by several authors (e.g. Hone and Holtz 2021; Sereno et al. 2022).

Pigment availability

Moving on, can we get a sense of the skin pigmentation available to giant theropods, thus letting us know which paints/colouring pencils/digital palettes to crack open? Here, we have to think about the availability of environmental pigments, like carotenoids. Many readers will know that animals cannot create all the pigments used in their integument and that some are obtained through eating plants or microbes. Carotenoids and other environmental pigments create some of the most vivid colours seen on animals today, including hot reds, bright oranges and canary yellows. But environmental pigments are hard to source in terrestrial settings, to the extent that even tiny songbirds compete with one another to source them (Blount 2004; Biard et al. 2005). Outside of specialist ecologies, the most famous being that of flamingoes, larger terrestrial animals tend to make do with pigments they can manufacture themselves, such as melanin. This is one reason why so many terrestrial animals are earthy tones, such as greys, blacks, browns, orange-reds, and white (where pigment is withheld). But structural colour, features of skin, scales and feathers that manipulate light to create colour without pigmentation, has also been developed across all vertebrates and is exploited to produce greens and blues. In all probability, it’s from these basic pigment and structural palettes that giant theropods were deriving their hues. Unless conditions of the past were very different to those of today, it’s hard to imagine multi-tonne terrestrial animals finding enough carotenoids to develop large patches of particularly intense pigmentation.

Specifics of patterning

Our discussion raises a notch in complexity as we move to consider giant theropod skin patterns, even if we stick within the camouflage-dominant framework outlined above. Concealment strategies are adapted to specific habitats, predation styles and prey types because no one system is universally effective. Indeed, one of the few constant rules of camouflage — that, no matter how perfectly a crypsis strategy works on a stationary animal, movement always gives the game away (Pembury Smith and Ruxton 2020) — is of little use to us here because we don’t know where and how big theropods hunted. The concealment strategy of an endurance predator, one that simply hounds its prey tirelessly, waiting for it to become vulnerable from exhaustion, might be different to that of an ambush predator that relies on surprise, springing at its prey at the last moment for a short chase.

These are only the first factors to consider. Predator colours are also modified by the time of day the predator tends to operate, as well as their position in the food chain: some have to be worried about being prey items themselves. And that, in turn, is altered by the colour schemes that can be created by different integument types (e.g. fibres vs. naked skin vs. scales), as well as the functional impacts of pigmentation. Darker pigments, for instance, can protect skin from harmful UV rays and may have antibacterial properties but, conversely, also absorb more solar heat and increase an animal’s thermal load (Walsberg 1983; Caro 2005; Caro and Mallarino 2020). There’s a lot to think about here, and the fact we still can’t account for these and other variables reliably in living animals is why biologists still consider our knowledge of animal colour to be fairly limited. It goes without saying that, if we’re still working out what’s happening among living species, robust predictions of camouflage patterning in extinct animals are way off.

The colour schemes we give our dinosaurs have functional implications beyond interacting with other animals. This dark, adult Tyrannosaurus would be well-protected from solar radiation by its dark skin, but it would absorb a lot of heat in direct sunlight.

Nevertheless, we may be able to narrow down some possibilities for giant theropods by looking at what works for large modern predators. Most employ background matching, where their skin tone approximates that of their surroundings, or else they use countershading, where dark upper regions and lighter undersides disrupt the formation of shadows, diminishing contrast with the background (note that this is disputed by some, there is actually a fair amount of controversy around countershading function: see Ruxton et al. 2004; Rowland 2009). Other predators use disruptive colouration, where high-contrast colours break up body outlines and disguise distinctive features such as eyes. Unlikely strategies for big theropods are masquerading tactics: attempts to match unexciting objects like rocks or twigs. To pull off this illusion, masqueraders have to resemble something of equivalent size and shape, and that becomes harder at larger sizes, perhaps explaining the absence of this method among large terrestrial predators today. This strategy is distinct from mimicry, where an organism adopts the appearance of another species to be misleading about its true nature (Skelhorn et al. 2010).

With several patterning options on the table, progressing further with this discussion is only possible if we start making assumptions about giant theropod ecology, pushing us further into the realm of inference and speculation. But we can ground ourselves by considering the results of studies into camouflage function and performance. For instance, if countershading does indeed work to disrupt shadowing, then studies show that a sharp, high-body colour transition would work better in an open setting than a more gradual colour change lower on the flank, which obscures animals more effectively in forested settings (Vinter et al. 2016). We generally see more uniform, low-contrast colours on big animals in open habitats because large patches of colour generally don’t conceal animals as well in woodlands (Pembury Smith and Ruxton 2020, although flat-grey elephants are reportedly remarkably difficult to find once they enter forests — see Caro 2013). Conversely, high-contrast patterns seem to work better at hiding animals in vegetated or otherwise busier environments.

Baby tyrannosaurs, barely a metre long, with cryptic colours that help them blend into the forest floor. In all likelihood, baby tyrants were at high predation risk and it seems reasonable to assume they used camouflage tactics to avoid being eaten. But the colour schemes of infants may not have worked so well for their gigantic parents, nor even older juveniles or subadults. Might tyrannosaurs and other giant theropods have tracked through multiple colour morphs en route to somatic maturity?

We can consider things like the age of our restoration subjects, too. In scaly animals (the only skin type we currently have direct evidence for in giant predatory theropods, even if we can’t rule out the possibility of some protofeathering), colour vividness tends to reduce with age (Olsson et al. 2013). This change may not just be physiological, but also adaptive. The juveniles of all animals, including apex predators, are targetted by carnivores and their colouration has to be multi-functional, hiding them from predators as well as — in precocial species — their prey. This is often achieved with disruptive patterning. Stripes, spots, bars and other features may serve an additional role, achieving a “motion dazzle” effect that confuses predators about animal speed and direction, or draws focus to less critical anatomies, like tail tips (Murali & Kodandaramaiah 2016). Dazzling capacity diminishes at lower speeds and agility, and is thus less useful in larger animals (Pembury Smith and Roxton 2020), perhaps partly contributing to the dulling of living reptiles as they approach adulthood. We should not imagine that juvenile theropods transitioned to their adult colours straight away, however. It took decades to grow gigantic theropodan predators and, in all probability, the route to adulthood was via several different ecological niches (e.g. Holtz 2021), each of which may have had different adaptive pressures on colouration. So maybe giant theropods had several colour schemes throughout their lives, and we should render them as being colour-adapted to their various age-specific lifestyles? We could go on listing the adaptive aspects of different animal skin patterns all day, but you get the idea. There's a lot of camouflage science we could factor into our reconstructions, even if we can't ever know the real colours and patterns of our subject species.

So... does animal colour science help us in palaeoart?

Background-matching, age-dulled Tyrannosaurus rex takes on a countershaded, partly disruptively-coloured Edmontosaurus annectens. These guys are mainly here to stomp about and wake people up with some Hardcore Dino Action™ in case anyone has drifted off when reading this long, long post.

Let’s conclude by returning to our main question: can studies of living animals constrain our speculations about the colours of dinosaurs, or will colour restorations forever remain a crapshoot when we don’t have palaeocolour data? Here, we've extrapolated the findings of predator-specific colour studies to giant, terrestrially-hunting theropods and, based on these, we've suggested that large dinosaur predators...

• were likely under very strong pressures for crypsis
• probably didn't load their skin with many environmental pigments
• likely expressed background matching, countershading or disruptive patterning, depending on their specific ecologies
• may have had several colour schemes throughout their lives as their ecology changed with age

Down the line, we can discuss the merits of these predictions — do remember that you're reading a secondary take on all the science discussed here and that you may come to different interpretations based on your own literature crawl. But it's not these specific findings that are most important here. Rather, it's that this case study shows what animal colour science can offer to the process of restoring one type of extinct animal, as well as its broader potential for focusing our loosely-constrained applications of colour within palaeoart. The points made above, or others like them, do not give a colour scheme for giant theropods, but they do suggest that some concepts are more likely than others, and even rough guidance isn't to be sniffed at when we're otherwise running virtually blind. It's strange that palaeoartists are often able to point to core palaeontological studies for interpreting fossils, and core anatomical studies for depicting anatomy, but we don't generally talk about or know the same literature on animal colour. I wonder if it'll eventually be worth keeping up with developments in this field as much as we do new fossil and anatomical data — if we’re not at this point already. The result can only be more scientifically credible and realistic artwork, and that's a win for everyone.