Gallery and print store

Thursday, 31 October 2019

Megafuzz under the microscope: how credible are restorations of giant fluffy extinct animals?

Images of giant prehistoric animals covered in thick, fluffy coats are par for the course in modern palaeoart, including lots of my own (image above shows Therizinosaurus, from 2015). But... hey: just how warm are these multi-tonne animals under all that fuzz?
Rendering giant prehistoric animals with extensive hairy coats or thick feathery coverings is a convention now so well established within palaeoart that few of us give it a second thought. While this practise is well-grounded in fact for some cold-adapted Pleistocene megamammals, such as woolly mammoths and woolly rhinoceros, our treatment of other giant species - giant sloths and giant coelurosaurs - has a greater basis in tradition and expectation than fossil data. We have, after all, mostly lacked detailed insights into the skin of these giant extinct animals, and have thus relied on scraps of soft-tissues and phylogenetic bracketing to inform our art.

My artistic history firmly places me on the megafuzz bandwagon. Earlier this year I painted a shaggy Megatherium and since 2013 I've painted woolly Pachyrhinosaurus, several extensively feathered tyrannosaurs and a Therizinosaurus with more feather coverage than most modern birds (above). But I was recently given pause to question these reconstructions when Dennis Hansen, one of my excellent patrons, asked about the possibility that giant sloths, such as Megatherium, were largely or wholly devoid of hair because of issues with thermal energetics. At that size, wouldn't giant sloths be far too warm? This idea has been promoted by some sloth researchers (Fariña 2002, Fariña et al. 2013) but it's rare to see it expressed in palaeoart. Megasloths are, near-universally, restored with the same shaggy fuzz first given to them by Benjamin Waterhouse Hawkins in 1854 and it now seems shocking and wrong to see one without that characteristic pelt. Should you want to draw one, you have to fight your hand - Evil Dead II style - to force those strange, hairless contours onto the canvas.

When pondering this query I came to realise how little I really know about thermoregulation in large animals in general. By this, I don't mean the generalities of surface area:volume relationships, or different mechanisms of homeothermy: I'm talking about the preferred temperature ranges and ideal climatic conditions of large living endotherms. At what temperatures do species of a given size and shape start to feel hot or cold? How does that vary across clades, body shapes, and sizes? How sensitive are they to changes in ambient temperature? What difference does a coat of fur or feathers make to the thermal tolerance of a giant animal? This seems like a major hole in my knowledge as a palaeoartist, and I don't think I'm alone in not having a firm grounding in this topic. I gather from online conversations that most of us are shooting from the hip when putting fur, fluff and fat onto our reconstructions, applying what seems 'right' given the phylogenetic position and palaeoenvironment of our subject species, but without specific reference to models of thermal energetics, the temperature tolerances of analogous animals, or any other form of quantified data.

Tyrannosaurus rex: megafuzz edition, from 2016. This was pre-Bell et al. (2017), obviously. They were different times.
So, for the last few weeks, I've been dipping into technical papers on this subject whenever I've had a spare few moments. I've found this a very useful exercise and encourage other palaeoartists to do the same. There's heaps of literature on the thermal energetics of endotherms and many enlightening, sometimes surprising results to ponder. While this exercise does not address the many unknowns of extinct animal physiology that are essential to understanding their strategies for thermoregulation or heat dissipation (e.g. metabolic rate, activity level, conductivity of skin etc.) it makes for an excellent palaeoart 'calibrating activity' or reality check. After all, if we don't know, in a measured and quantified sense, how size influences the thermal tolerances and integument of living animals, how can we be expected to make credible reconstructions of their fossil relatives?

Into the Thermal Neutral Zone

There are several different concepts we can use to investigate thermal energetics. One of the most enlightening and useful mechanisms is thermal neutrality. Endothermic organisms are thermally neutral when their environment is warm enough that their Basal Metabolic Rate (BMR) is sufficient to maintain their core temperature without additional energetic investment or water loss. This can be given as a single value, which represents the thermal neutral temperature for a specific configuration (e.g. a certain pose and hair or feather arrangement etc.) or it might be given as a range - a Thermal Neutral Zone (TNZ). We define the TNZ as the temperatures at which very minor adjustments to an animal's posture or integument control core temperature rather than changes to BME. While the TNZ does not exactly equate to an animal's thermal 'comfort zone' (Kingma et al. 2014) this is also not the worst layman's summary of the term: if an animal has to invest more than minimal energy to maintain a steady core temperature (e.g. exposing a heat-radiating body part, or altering insulation depth by raising/lowering hair or feathers), it's outside the TNZ.

Principles of the Thermal Neutral Zone. This graph is based on an excellent diagram included in this lecture, but I've been unable to find the original source.
The TNZ is bounded by two thresholds, Lower and Upper Critical Temperatures (LCT and UCT, respectively - see diagram, above). These are the ambient temperatures at which an animal has to take action (e.g. invest energy above BMR) to keep itself at a desired core temperature. Below the LCT, animals use energy to keep warm (e.g. by shivering or exercising), while exceeding UCT instigates cooling responses, such as seeking water, sweating or panting. Some species are well adapted for survival outside of their TNZ, or are capable of tolerating huge temperature fluctuations without changes to BME. Others are specialised to live in a narrow ambient temperature band and react inefficiently when subjected to cooler or warmer conditions.

What's neat about the principle of thermal neutrality is that it allows us to explore the effects of body size, metabolism, insulation and temperature in a quantified manner. Thermal neutrality is applied widely to all manner of biological studies: just a few applications include animal husbandry, understanding animal responses to climate change, and the evolution of organismal physiology. For our purposes, it's helpful that well-established scaling trends have been recognised from studies of endotherm thermal neutrality. They're based on pretty fundamental physical factors such as animal mass, ambient temperatures, animal core temperature, and skin conductivity, so we can be pretty confident that they should apply to fossil endotherms too.

Generally speaking, the smaller the animal, the closer their thermal neutral temperature is to core temperature. Small animals have narrow TNZs, higher LCTs, and - owing to their lessened thermal inertia - sharper increases in metabolic rate when ambient temperature takes them away from thermoneutrality. These facts describe the well-known phenomena of small animals generally being more concerned with staying warm than keeping cool. The inverse is true for large animals, which have broader TNZs, lower LCTs, and lower metabolic costs to warm themselves below LCT: in other words, they're less sensitive to cool temperatures.

Whatever size an animal is, excessive heat is more dangerous than excessive cold. Endotherms can tolerate ambient temperatures much lower than their LCT before reaching dangerous levels, but their tolerance to temperatures above UCT is much lower: just a fraction of their potential LTC response range. While a cold animal can generate a lot of additional heat from exercise and increased metabolic rates, hot animals have to rely on raw physical processes - conduction, radiation, evaporation and convection - to cool down. We can only enhance these processes so much and, as most endotherms run within 3-6°C of critically high core temperatures, we have a low margin for error when exposed to very high temperatures. An organism's thermal neutrality is not fixed, and can be altered by anything which affects heat production and loss (e.g. wetting the skin, humidity, air movement), so we have to consider a range of environmental factors, not just temperature, when discussing this concept.

My very conventional take on Megatherium, a four-tonne sloth restored almost exclusively as extensively hairy since the mid-1800s. I feel safe and cozy with this image, and the idea of hairless megasloths is downright weird to me. Good job I've not tried to draw one or anything.
Values of thermal neutrality have been reported for numerous animals, including humans. A lot these stem from research into livestock welfare, wherein farmers and breeders need to know what temperatures their animals are comfortable in (for an extensive summary, see the 1981 findings of the National Research Council). Thus, the TNZs of horses, cows, sheep, chickens and so on are well documented and easy to find outside of technical literature. A complication to these figures is that they often lack details such as animal weight, breed, and environmental specifics, so they are - at best - a rough introduction to livestock TNZs. Nevertheless, these are useful species to discuss because they're so familiar to us, and I've attempted to summarise representative values from several sources here.

Unsurprisingly, smaller animals like chickens (c. 2 kg) feel the cold relatively easily and have a relatively high and narrow TNZ of c. 18-23°C. A freshly hatched chick has an LCT of 34°C. Larger birds, like emus (on average, 30 -40 kg), have a lower LCT of 10°C (Maloney 2008). Dairy cattle (450-800 kg) are less sensitive to temperature changes, with a TNZ of 5-25°C, though some dairy cows are reported as having LCTs of -15°C. This range seems to apply to certain beef cattle breeds as well, though not all: some (presumably smaller and leaner?) have LCTs of c. 10°C. Horses have a TNZ of 5–25°C (Morgan 1998), although they can reportedly tolerate freezing temperatures comfortably with unshorn hair. Cattle with full, dry winter coats can also tolerate freezing temperatures, down to -7°C. Animal condition and food intake are important variables: well-fed animals with access to food have lower LCTs than those that are fasting. For cattle, the difference between fasting and full-feed equates to a 19°C difference in LCT, from -1°C in full-feed to 18°C in fasting (National Research Council, 1981).

Naked humans are thermally neutral around 27°C, making us - perhaps counterintuitively - most comparable to the smaller species mentioned above. This relatively high temperature reflects both our long-term hominid reliance on clothing as well as our ancestral climate. Habitat and climates influence the temperature tolerances of endothermic animals in terms of both short-term acclimatisation and longer-term adaptation (Scholander et al. 1950; Scholander 1955). Arctic animals have amazingly broad TNZs of many tens of degrees. Resting arctic foxes, for example, show little change in BME whether they are in 30°C or below -30°C. They achieve this by mixing high-performing insulation around their bodies with thinner insulation on their extremities so that, by simply changing posture, they create an 11-fold difference in heat retention or loss. Tropical animals - which includes human ancestors - have relatively narrow zones of thermal neutrality and begin to feel cold when exposed to temperatures of even 25°C. They also respond more energetically to changes in temperature, raising their metabolic rates far quicker, relative to temperature change, than their polar equivalents. The bodies of tropical species can be seen as specialised for continuous high temperatures, while those of colder climates are adapted to deal with extreme fluctuations in daily conditions.

The impact of integument and body shape on TNZ

Data are also available regarding the impact of insulating tissues - fur, fat, feathers etc. - on animal heat loss. One very familiar source on this topic are sheep in their fleeced and shorn state. The National Research Council (1981) reports that a sheep with a 10 cm thick fleece has a LCT of -120°C(!), but this lowers to -15°C when the fleece is trimmed to 7 mm. That's a remarkable change in temperature tolerance, and shows the enormous impact that integument thickness has on animal energetics. In a wet, windy setting, that LCT of our 7 mm fleece sheep raises even more, to 13°C.

We can also explore the scaling effects of adding insulation using digital models. Calculating TNZ at various animal sizes and body shapes, and both with and without a standardised insulation, shows that insulating layers have increasing impacts on TNZ at large size (Porter and Kearney 2009). The addition of insulation only lowers the LCT of very small animals (e.g. rodent, microbat or songbird sized) a few degrees, but LCT drops exponentially quicker in larger animals. Applying the same grade of insulation to a one tonne animal lowers LCT by about 65°C, from ~25°C in a naked-skinned animal to below -40°C in a fuzzy one. Again, I have to remark on how big that shift is: this is the sort of difference that could adapt a species to a whole new biome.

The impact of insulation, body shape and wind speed on LCT values in endotherms. Note how LCT generally falls with increased body size, but how the introduction of insulation compounds this effect dramatically. From Porter and Kearney (2009).
Porter and Kearney (2009) also show that changes in body shape - which could reflect either different anatomical bauplans or changes in posture - have an impact on LCT values too. Unsurprisingly, longer and thinner animals have higher LCTs than more compact animals, but the impact of increasing surface area becomes less pronounced at large size, and any impact they have is vastly overshadowed by the addition of insulation. This is an important point for those of us thinking that the body shapes of extinct animals might allow for fibres and fluff at larger sizes than we'd predict from living animals. Yes, body shape has an influence, but perhaps not as much as we intuitively expect, and with less and less impact as animals scale to gigantic proportions.

Thermal neutrality in giant animals

One frustration of current literature on thermal neutrality is a lack of specific data on our largest living species, such as rhinos and elephants. Though some literature discusses the TNZs of these species, I was unable to find their LCT and UCT values. Nevertheless, a wealth of studies have been performed into the thermoregulation of elephants that give hints about where their TNZ might lie. This research has been catalysed by both simple scientific curiosity as well as concern for zoo elephants in climates far removed from their naturally hot ranges in Asia and Africa. Elephants provide some of our best insights into the thermoregulatory challenges facing large extinct land animals, but these data come with important caveats. As discussed in my post on indricotheres, elephants have thermoregulatory issues beyond simply being huge: they are unusually compact, live in climates which are routinely warmer than their core temperature, and they cannot sweat or pant (Myhrvold et al. 2012). They still provide useful insights into the issues of maintaining a steady internal temperature at multi-tonne masses, but they are probably not biologically 'typical' giant animals.

Elephants are noted for tolerating a wide range of temperatures in their natural habitats, from sweltering daytime heat of over 40°C to overnight cools of freezing or sub-freezing temperatures. Their size and thermal inertia permits them to endure freezing nights without issue and, in discussions about the controversial subject of keeping zoo elephants in cooler climates, their handlers often suggest they are happy in snowy and icy conditions, at least for short periods, provided they can keep active. It seems one of the biggest problems elephants face in freezing temperatures is frostbite on their ears and trunks, not the cold itself. This probably indicates a very low LCT (close to or below freezing) for elephants, which is what we'd expect from the scaling trends outlined above. Estimates of thermal neutrality in multi-tonne fossil species (see below) point to similar values.

Desert elephants, such as these Namibian bush elephants, are specialised populations adapted to life in extreme heat and aridity. They have several anatomical adaptations to desert life - some specifically influencing to their thermal energetics (smaller bodies, longer legs) - and avoid extensive exercise during the day, especially in warmer seasons, to avoid risk of hyperthermia. Their nomadic lifestyle is mostly achieved by long walks at night, not during the day. Image by Ron Knight, from Wikimedia, CC-BY-2.0.
Elephants may also spend a lot of time at or above their UCT, reflecting their struggles with heat dissipation. Monitoring elephant body temperatures during moderate exercise (walking) in a range of weather conditions (averaging 8 to 35°C) shows that their tissues accumulate heat 2.2 - 5.3 faster than it can be dissipated, depending on conditions (Rowe et al. 2013). This is in part because very large animals have a thick thermal boundary layer - a region of air adjacent to the skin which is warmed by heat radiating from their bodies. Larger animals effectively carry their own warm microclimate wherever they go, and face the challenge of trying to shed heat through it. This, combined with the heat produced by large-scale muscle activity, means exercise can be thermally stressful to elephants, especially in hot, windless conditions. Rowe et al. (2013) predict that four hours of continuous walking in very warm conditions would be fatal to an elephant, perhaps explaining why elephants living in their natural, warm habitats limit their daily exercise, routinely seek shade and water, and are often more active at night. Elephants spend much of their lives with internal temperatures close to the critical mammalian limit, even tolerating extending periods of near-lethal hyperthermia, to the extent that climate change may push wild elephants over the edge of their adaptive capacity to endure elevated temperatures. They are not entirely alone in this: other large mammals of very warm tropical settings - such as rhinoceros - also employ elephant-like behavioural adaptations when faced with high ambient temperatures. Rhinoceros have a more conventional mammalian capacity to deal with heat - they can pant and sweat - and yet they still seek shade and water during hot periods (Rowe et al. 2013). We have to view the thermal stresses faced by multi-tonne animals as defining physiological and behavioural factors in their lives, and as major selection pressures on their anatomy.

Thermal energetics in extinct giants

Having just learned a little about thermal neutrality in living species, can we make some broad predictions about the energetics of extinct giants? Many researchers have applied these principles to fossil animals and their findings are in line with the general points made above: there are strong indications that extinct giants - seemingly regardless of metabolic rate - had major issues with heat loss. It's quite reasonable to assume that this could have influenced aspects of their anatomy and appearance.

One such study is the article which catalysed this blog post: Richard Fariña's (2002) estimates of giant sloth thermoneutrality, with a strong focus on Megatherium*. Fariña (2002, later summarised by Fariña et al. 2013) calculated that a hairless 4-tonne sloth with a typical placental metabolism would be thermally neutral at -17°C. As a mid-latitude creature living in a semi-arid temperate climate (Bargo et al. 2001), this result paints Megatherium as having elephant-like issues with staying cool. The environments inhabited by Megatherium are similar to those of modern northern Patagonia, and thus rarely, if ever, dropped to -17°C, and we have to wonder if the shaggy pelt traditionally applied to Megatherium would be cooking an already very warm animal. Given the arid settings inhabited by this sloth, water loss through panting would soon become a major problem for a heat-stressed Megatherium. We must also consider that a similarly sized-sloth, Eremotherium, lived in tropical temperatures in what is now Florida: if it had a similar thermal neutrality to Megatherium (which it almost certainly did), Eremotherium must have been pretty hot and bothered most of the time, even if it largely or wholly lacked fur.

*It's worth stressing that, contrary to popular belief, we do not have any skin preserved from a megasloths: all the skin specimens we have stem from smaller ground sloth species.

Here's your reminder that I'm posting this on Halloween 2019: behold the horror of a near hairless ground sloth. A century and a half of seeing giant sloths with long, shaggy hair makes images like this hard to swallow, but there's a legitimate scientific case to be made for megasloths looking like this. We need to be careful that we don't let tradition and expectation blind us to what might be a more tenable hypothesis of life appearance.
Modern sloths, of course, have an unusually slow metabolism, and it's appropriate to ask how much that might affect thermal neutrality of their giant cousins. Fariña (2002) calculated that halving the metabolism of a naked Megatherium leads to thermoneutrality at 10°C, a figure comparable to animals that inhabit temperate settings today without the need for long, shaggy fur. This being so, perhaps mass alone might have been enough to keep Megathium warm, even if it had an unprecedentedly low metabolic rate for a mammal.

Fariña (2002) also computed the thermoneutrality of a two tonne Mylodon darwinii in both naked and shaggy configurations. His estimates give thermal neutrality at -4°C without fur, and -28°C once a 4 cm thick hairy covering was applied. This matches expectations that fur makes a large difference to the thermal neutrality of large animals and also implies that, even without hair, Mylodon was pretty cold tolerant. Of course, fossil evidence shows that Mylodon was hairy, suggesting that we have a species adapted for dealing with extreme cold. This seems sensible given what we know of ground sloth distribution. Mylodon lived much further south than Megatherium, at the southern tip of South America, and also at high altitude. Unlike Megatherium, it would have routinely experienced sub-freezing temperatures and probably needed extra insulation to survive harsh winters. There's more work to do with Fariña's sloth calculations (both his 2002 and 2013 contributions to this topic are short and don't play with as many variables as I'd like) but these results are certainly thought-provoking as goes our considerations of the life appearance of sloths, and perhaps other giant extinct animals too.

Turning our attention now to extinct giant reptiles: I'm not aware of any studies that calculate thermal neutrality for large dinosaurs, but the sort of figures suggested for multi-tonne sloths are probably reasonable assumptions if we assume a mammal-like metabolic rate. Some vindication of this stems from studies suggesting that large dinosaurs had elephant-like issues with overheating. Rowe et al. (2013) questioned how long it would take a 3655 kg Edmontosaurus to overheat with continuous exercise and, even though their model assumed a sub-mammalian metabolic rate, just 3.5 (endothermic) or 4 (ectothermic) hours of walking in daytime temperatures typical to mid-latitude Late Cretaceous settings would elevate Edmontosaurus core temperatures to lethal levels. They suggested that, like large mammals, giant dinosaurs might have relied on panting, finding shade and water, resting during the warmest parts of the day, and nocturnal behaviour to avoid heat stress.

How quickly would it take for Edmontosaurus to overheat when subjected to low-grade exercise during the daytime? Not that long, despite it not being the largest dinosaur, nor the most insulated. What might this graph look like for a hypothetical larger, fluffier dinosaur living in the same habitat? Modified from Rowe et al. (2013).
This is some major food for thought given that the Edmontosaurus model of Rowe et al. (2013) lacks an insulating skin cover and is considerably smaller than some dinosaurs we routinely coat with thick layers of fluff. If 3-4 tonne scaly dinosaurs were already experiencing elephant-grade issues with heat build up during exercise, surely 6 tonne coelurosaurs living in the same climates would experience similar issues, even if only covered in scales? Everything we understand about the scaling of thermal energetics suggests that heat retention problems would get worse, not better, in larger animals, and it might be unrealistic to assume coelurosaurs twice the mass (or more) of Edmontosaurus were comfortable wandering around with a thick, insulating blanket of feathers.

Would body shape - such as having a dinosaurian-grade long necks and tails - have helped avoid the issues of heat retention? Seemingly not. Don Henderson’s (2013) models of sauropod thermoregulation found that skin area does not scale rapidly enough with increased body size, even with proportionally long necks and tails, to effectively cool their bodies. Sauropods are probably our best bet for dinosaurs using body shape to dump unwanted heat and, if even their skin area can't keep pace with internal heat production, other dinosaurs were unlikely to have managed either. This seems to match expectations from Porter and Kearney (2009) that elongate body shapes affect thermal neutrality, but that the effects of elevated body mass are difficult to circumvent.

Although some of the most extreme neck and tail proportions exist in the largest sauropods, such as Dreadnoughtus, these anatomies do not augment their surface area quick enough to counter their increase in bulk and heat production. If sauropods - animals famously observed as being thin at one end, much much thicker in the middle and then thin again at the far end by A. Elk (1972) - couldn't rely on their necks and tails for this task, other dinosaurs likely couldn't either.
We needn't just rely on equations and computer models for evidence of high heat loads in large dinosaurs: we also have direct fossil evidence suggesting them. Brand new research by Ruger Porter and Lawrence Witmer (2019) has noted that large dinosaurs had enhanced vascularity in their skulls related to shedding heat. Like other reptiles, dinosaurs likely used panting and cooling sinuses in their heads to shed heat, and they seem to have increasingly relied on these mechanisms at large size. Porter and Witmer's study shows a strong correlation between body size and development of these cranial cooling mechanisms in all three major dinosaur groups, suggesting that superior cooling anatomy was acquired independently in large-bodied dinosaurs regardless of the body shapes or integuments common to their clades.

We can also - perhaps more controversially - look at our current understanding of dinosaur skin as matching expectations of thermal energetics. And yes yes yes, our data here is less than perfect, taphonomic issues abound and we still have large gaps in our understanding of dinosaur skin. But it's nevertheless interesting that - as I write this in October 2019 - we're still finding indications of scales in virtually all dinosaurs above the 1.5 tonne mark ("the Yutyrannus threshold") regardless of whether that group is phylogenetically likely to sport fibres or not. We typically consider coelurosaurs in this context (e.g. Bell et al. 2017) but perhaps we should also consider ornithischians as evidence of this relationship too, given that at least some smaller ornithischians were covered in fuzz (e.g. Godefroit et al. 2014) but scales dominate in all sampled multi-tonne species. So yes, while our dataset of dinosaur skin configurations might just reflect a number of preservational and taphonomic factors, we have to be open to the possibility that we're actually seeing how dinosaurs adapted to large size. It's also worth stressing that, given estimates of thermal energetics in large extinct animals, an extensively fuzzy giant dinosaur would actually be pretty surprising. 

We don't discuss it much, but the well-documented scaly hides of large ornithischians, such as Triceratops, might represent the same thermal responses postulated to explain the presence of scales in large theropods. We need a lot more data on the skin of smaller ornithischians to test this, but it's a hypothesis consistent with our understanding of heat retention in animal scaling.
A response to this last paragraph might be that certain Mesozoic settings were actually a lot colder than we generally assume, and that maybe even large dinosaurs needed extra insulation to stay warm. While this argument has some merit, we need to be careful not to overstate how cold these settings were, as well as consider some of the temperature values associated with thermoneutrality in very large species. It's true that the Mesozoic was not the global tropical hothouse we once assumed it was, but temperatures were still generally warmer than today: many so-called 'cold' Mesozoic climates would be quite tolerable to modern temperate species. Maastrichtian Mongolia, for instance, which many artists are now populating with woolly Deinocheirus, shaggy Therizinosaurus and fuzzy Tarbosaurus, had a mean annual temperature of between 5-10°C and a climate similar to Shijiazhuang, northeastern China (Owocki et al. 2019). This predicts average daily temperatures above 10°C for most of the year but only a month or so of average daily temperatures around freezing. It seems doubtful to me, given what's outlined above about the thermal energetics of 3-4 tonne animals, that six-tonne (or more) coelurosaurs would need thick, full body insulation to live in this climate. To the contrary, modern cattle or horses could have lived in these settings without discomfort. Similar statements can be made about Maastrichtian Alaska, which was cold in the winter, but lacked sustained freezing temperatures (Spicer and Herman 2010). If two-tonne animals are thermally neutral at c. -4°C (Fariña 2002), the 2-4 tonne hadrosaurs and ceratopsids of these habitats might have survived untroubled by the winter months without needing extra insulation.

Yet more art of extensively fuzzed large dinosaurs, which I once assumed would be sensible given the cold temperatures of Maastrichtian Alaska. Given everything outlined here, I'm now looking at these Pachyrhinosaurus from 2015 as being over-insulated for their chilly, but not deeply-cold habitat.
We thus have to be careful not to get carried away when we hear that palaeotemperatures of a given ancient setting have been revised down. A "cool" Mesozoic climate has yet to equate to modern-grade tundra or polar desert, and we needn't start restoring animals as looking suited to such habitats. It also pays to remember that most Mesozoic climates were warmer, sometimes significantly warmer, than these cooler examples. O'Connor and Dodson (1999) suggest that a modern elephant dropped into the climates of Late Cretaceous North America would experience the same issues, or worse, with overheating as they do today. This being the case, perhaps a lot of big dinosaurs spent much of their lives feeling pretty darn warm.

So... about those giant shaggy coelurosaurs and sloths...

Let's bring this long article into land by returning to our original question: how likely is it that giant fossil animals, such as giant sloths and giant coelurosaurs, were covered in extensive fuzz for the purpose of insulation? To me, our discussion of the thermal energetics, heat production and dissipation, and data from the fossil record suggest a few key takehomes:
  1. Animals do not need to be gigantic nor super shaggy to be tolerant of cool temperatures. Species weighing several hundred kilogrammes, and with only moderate insulation, are thermally neutral at temperatures approaching freezing, and those that exceed a tonne or so have TNZs extending below 0°C. Simply being large is a very effective way to stay warm, regardless of body shape or phylogenetic affinity.
  2. Near-naked multi-tonne animals struggle to shed body heat even in cool conditions because, when engaged in any activity, they generate more heat than they can easily lose. Hypothetical structures that would inhibit heat loss further - such as thick fur or feathers - seem maladaptive and unlikely for such species.
  3. We seem to lack data on the thermal energetics of the very largest fossil land animals, but there's no reason to think they would have escaped the the challenges of heat dissipation outlined above. If anything, these issues would be more far more pronounced than that of the taxa discussed in this post, on account of their increased body mass.
These points considered, perhaps reconstructions of large animals that have jettisoned most or all of their fibrous integuments are viable interpretations of fossil giants, regardless of their ancestral integument. I stress "most or all" fibres because, of course, we also have data suggesting that sparse, short fibres can help negate thermal boundary issues (see Myhrvold et al. 2012), and there's no reason to assume this wouldn't apply to giant sloths, indricotheres, coelurosaurs etc.

I'm certainly now looking at some of my own portfolio with new eyes: I find it hard to believe that my super-fluffy Therizinosaurus, Pachyrhinosaurus and even my traditionally hairy Megatherium aren't sweltering to death under all their fluff. Fariña's naked sloths might be weird and scary to us after centuries of depicting them with shaggy fur, but - so far as I can tell - his models fit our understanding of animal energetics and Megatherium habitat far better than the established model. It's worth remembering that a counter case for Megatherium requiring extensive hair has never been made, and that our standard reconstruction is, from the perspective of basic physics, actually far more outlandish than the seemingly radical Fariña model. It may seem shocking, but the case for a hairy Megatherium is less developed than the case for a hairless one.

Putting my artistic money where my mouth is, here's the giant, six-tonne ornithomimosaur Deinocheirus restored with just a sparse, localised set of filaments around the head, shoulders and tail tip. After researching this article, images like this read as more plausible to me than the general 'walking haystack' guise Deinocheirus has attained in palaeoart.
And yes, we might want to apply the same caution to giant coelurosaurs, even species we're used to clothing in extensive fuzz such as therizinosaurs and deinocheirids. Perhaps even the 'feather capes' we sometimes drape on giant tyrannosaurids are thermal overkill. Tyrannosaurus and similarly-sized tyrants are as large or bigger than living elephants, and it seems unlikely to me that animals of this size, mostly living in temperate or subtropical climates, needed even a small feather blanket to keep them warm. Our reconstructions of these animals which lean towards mammoth-like pelts seem especially untenable, given the impacts that even light feather coats would have had to species of their size. This does not rule out sparse fibres for heat loss or display of course, but a thick, shaggy pelt would surely be stifling - maybe even dangerously warm - given their body mass and the temperatures they experienced.

I'm aware that the argument I've presented here is a very broad brush, 'woods for trees' approach to this topic: I don't doubt that there are nuances and details to get into, and that there are many questions left to answer. For instance, what about the role of air sacs in dinosaurian cooling? What about climates which have high precipitation rates or strong winds? These are good points worthy of exploring, but - without wishing to add lots more detail to this already long article - I wonder if they're going to overturn the general arguments outlined here. With all indications being that giant animals are thermally neutral at very low temperatures, and that body mass seems to be the dominant effect on thermal neutrality, we're asking a lot of these additional factors to overturn the points made above. Note, for example, how Porter and Kearney's (2009, also above) assessment of wind speed on LCT follows the general trend of larger animals being less affected than smaller ones, and how it has very little impact on LCT for the largest animal in their study. We should assume even lesser impact in gigantic species.

I'm expecting a certain amount of harrumphing about this article from some quarters, especially from those who - like me - quite like seeing big, shaggy animals in palaeoart. They look cool, give off that 'new palaeo' vibe and provide us with lots of fun and exciting looks to explore. But, of course, palaeoart isn't really about what we like, it's about creating tenable, data-compliant takes on fossil species. So I'm going to end this article with a request: for those of us who want to continue restoring giant fossil animals with thick layers of hair and feathers, we need to demonstrate how the data presented here is wrong, and to the same calibre as the cited studies. What large modern animals deviate from well-established energetic scaling trends? What models of extinct animal physiology show that multi-tonne animals were immune to expected issues of thermal storage and heat dissipation? What are the flaws in papers arguing for low thermal neutrality in giant endotherms? Such a discussion would at least give us some actual data, and not just arm waving and intuition, to make predictions about how much fuzz extinct giant animals actually had. It's our job, after all, to ensure that the fluff in our palaeoart is kept on the bodies of our carefully researched restoration subjects, and isn't also a description of our approach to research.

Enjoy monthly insights into palaeoart, fossil animal biology and occasional reviews of palaeo media? Support this blog for $1 a month and get free stuff!

This blog is sponsored through Patreon, the site where you can help online content creators make a living. If you enjoy my content, please consider donating $1 a month to help fund my work. $1 might seem a meaningless amount, but if every reader pitched that amount I could work on these articles and their artwork full time. In return, you'll get access to my exclusive Patreon content: regular updates on upcoming books, papers, painting and exhibitions. Plus, you get free stuff - prints, high-quality images for printing, books, competitions - as my way of thanking you for your support. As always, huge thanks to everyone who already sponsors my work!


  • Bell, P. R., Campione, N. E., Persons IV, W. S., Currie, P. J., Larson, P. L., Tanke, D. H., & Bakker, R. T. (2017). Tyrannosauroid integument reveals conflicting patterns of gigantism and feather evolution. Biology letters, 13(6), 20170092.
  • Fariña, R. A. (2002). Megatherium, the hairless: appearance of the great Quaternary sloths (Mammalia; Xenarthra). Ameghiniana, 39(2), 241-244.
  • Fariña, R. A., Vizcaíno, S. F., & De Iuliis, G. (2013). Megafauna: giant beasts of pleistocene South America. Indiana University Press.
  • Godefroit, P., Sinitsa, S. M., Dhouailly, D., Bolotsky, Y. L., Sizov, A. V., McNamara, M. E., ... & Spagna, P. (2014). A Jurassic ornithischian dinosaur from Siberia with both feathers and scales. Science, 345(6195), 451-455.
  • Henderson, D. M. (2013). Sauropod necks: are they really for heat loss?. PloS one, 8(10), e77108.
  • Kingma, B. R., Frijns, A. J., Schellen, L., & van Marken Lichtenbelt, W. D. (2014). Beyond the classic thermoneutral zone: including thermal comfort. Temperature, 1(2), 142-149.
  • Maloney, S. K. (2008). Thermoregulation in ratites: a review. Australian Journal of Experimental Agriculture, 48(10), 1293-1301.
  • Morgan, K. (1998). Thermoneutral zone and critical temperatures of horses. Journal of Thermal Biology, 23(1), 59-61.
  • Myhrvold, C. L., Stone, H. A., & Bou-Zeid, E. (2012). What is the use of elephant hair? PLoS One, 7(10), e47018.
  • National Research Council. (1981). Effect of environment on nutrient requirements of domestic animals. National Academies Press.
  • O'Connor, M. P., & Dodson, P. (1999). Biophysical constraints on the thermal ecology of dinosaurs. Paleobiology, 25(3), 341-368.
  • Owocki, K., Kremer, B., Cotte, M., & Bocherens, H. (2019). Diet preferences and climate inferred from oxygen and carbon isotopes of tooth enamel of Tarbosaurus bataar (Nemegt Formation, Upper Cretaceous, Mongolia). Palaeogeography, Palaeoclimatology, Palaeoecology, 109190.
  • Porter, W. P., & Kearney, M. (2009). Size, shape, and the thermal niche of endotherms. Proceedings of the National Academy of Sciences, 106, 19666-19672.
  • Porter, W. R., & Witmer, L. M. (2019). Vascular Patterns in the Heads of Dinosaurs: Evidence for Blood Vessels, Sites of Thermal Exchange, and Their Role in Physiological Thermoregulatory Strategies. The Anatomical Record. In press.
  • Rowe, M. F., Bakken, G. S., Ratliff, J. J., & Langman, V. A. (2013). Heat storage in Asian elephants during submaximal exercise: behavioral regulation of thermoregulatory constraints on activity in endothermic gigantotherms. Journal of Experimental Biology, 216(10), 1774-1785.
  • Scholander, P. F. (1955). Evolution of climatic adaptation in homeotherms. Evolution, 9(1), 15-26.
  • Scholander, P. F., Hock, R., Walters, V., Johnson, F., & Irving, L. (1950). Heat regulation in some arctic and tropical mammals and birds. The Biological Bulletin, 99(2), 237-258.
  • Spicer, R. A. and Herman, A. B. 2010. The Late Cretaceous environment of the Arctic: A quantitative reassessment based on plant fossils. Palaeogeography, Palaeoclimatology, Palaeoecology, 295, 423–442.

Friday, 30 August 2019

We need to talk about teratorns

Very awesome take on the teratornithid Teratornis merriami by Charles Knight. Like virtually all illustrations of teratorns, the implication of this image is that Teratornis is a scavenger, arriving to steal parts of this American camel (Camelops hesternus) from noble Smilodon. But how accurate is this widely portrayed view? Image © AMNH, borrowed from Gizmodo.
Teratorns (formally known as Teratornithidae) are a group of large to gigantic raptorial birds that roamed the Americas for much of the Neogene, only becoming extinct about 11,000 years ago. On account of their large size, carnivorous habits and association with charismatic mega mammals, they are some of the most famous of all fossil avians. The most widely known teratornithid is surely Argentavis magnificens, a Miocene Argentinian species often regarded as the largest flying bird of all time, but our most complete picture of their anatomy stems from Teratornis merriami,, a 3.5 m wingspan taxon from the Pleistocene La Brea Tar Pits. Exactly where teratornithids fit into avian evolution is not entirely resolved but they likely have close affinities with New World vultures, Cathartidae (Mayr 2009). Traditionally, this would have made teratorns relatives of storks and herons, but recent shifts in avian phylogenetics have seen cathartids reclassified as Accipitriformes, a large raptor group only excluding falcons and owls. This being the case, Teratornithidae should be regarded as Accipitriformes as well.

Teratornithids occupy an unusual status in popular palaeontological culture. They are legitimately popular animals, but fundamental aspects of their palaeobiology are poorly known to non-specialists. Indeed, it’s accurate to say that the two things most people ‘know’ about teratornithids - 1) that they were enormous, 7-8 m wingspan giants and 2) that they lived as condor-like scavengers - conflict with modern interpretations of their palaeobiology. There's clearly a need to bring folks up to speed on what science actually thinks about these awesome but frequently mischaracterized fossil birds, and that's what I aim to do here.

1) No, teratornithids were not the largest flying birds (at least, in terms of wingspan)

Teratornithids were large animals which routinely attained sizes beyond those of living fliers. Even the moderately-sized Teratornis merriami likely massed around 14 kg (Chatterjee et al. 2007), a figure comparable to masses of the largest modern flying birds, and upper size estimates for Argentavis are staggering: up to 6-8 m wingspans and body masses of 65-120 kg (Palmqvist and Vizcaino 2003; Chatterjee et al. 2007). If these estimates are correct, Argentavis was the largest flying bird we know of by a comfortable margin. The only birds rivalling it, the pelagornithids, have comparable 6-7 m wingspans but only 16-40 kg body masses (Mayr and Rubilar-Rogers 2010; Ksepka 2014). These predictions of Argentavis size have shaped our understanding of its flight and ecology. Generally assuming a 70-80 kg mass and 6-7 wingspans, several authors have suggested that Argentavis could only launch under favourable conditions and relied on strong winds for soaring flight (e.g. using downward slopes and headwinds - Campbell and Tonni 1981; Vizcaíno and Fariña 1999; Chatterjee et al. 2007). Vizcaíno and Fariña (1999) presented calculations of Argentavis energetics, range and ecology based to these size estimates and concluded that only a scavenging lifestyle could sustain such enormous birds.

It’s paramount to ask, therefore, how reliable these mass and wingspan predictions are. Argentavis, after all, is only known from a few limb bones, some shoulder material and a lower jaw, and this means we're extrapolating data from other birds to get our size estimates. We should probably tackle this question in two parts.

Like many large extinct fliers, we don't have a great skeletal inventory for Argentavis magnificens. This means that any size predictions of this species are just that - predictions - and only as reliable as the assumptions they're based on. For those very large Argentavis wingspan estimates, that's a critical point. Argentavis skeletal from Chatterjee et al. (2007), white bones are known elements. I'm pretty sure that this figure is an (uncredited) mash-up of teratornithid skeletals by Greg Paul (2002).
Firstly, those 7-8 m wingspan estimates are definitely looking shaky. It’s difficult to know what the skeletal wingspan of Argentavis was as we do not have any complete wing bones but, using a projected humeral length of 57 cm, Mayr and Rubilar-Rogers (2010) regressed a surprisingly small wing spread of just 366 cm. This is way under the 4-5 m skeletal spans measured for some Pelagornis species and an immediate red flag for those 7-8 m Argentavis wingspan estimates. To attain such sizes Argentavis would need primary feathers reaching unprecedented lengths of 1.5-2 m (Chatterjee et al. 2007), entirely unlikely proportions given that primary feathers actually scale negatively to wingspan (in other words, big birds actually have proportionally small flight feathers - Kspeka 2014). 7-8 m wingspans are thus extremely unlikely for Argentavis skeleton and lower values - around 6 m - are more realistic total wingspans.

The wing bones of various large birds, as illustrated by Campbell and Tonni (1983). That's the incomplete Argentavis humerus at top, a critical bone for estimating its wingspan. Alas, it's not 100% complete and we have to estimate how much was missing, especially from the proximal region.
And this brings us to our second point: although 6 m has been the generally accepted wingspan estimate in teratorn studies, it might actually be the upper bound for Argentavis size, not an average or middling value. Kspeka (2014) modelled Argentavis wingspans using several different means of predicting feather length and found 6.07 m as his top value, with all other equations suggesting wing spreads of 5.09-5.7 m (Kspeka 2014). If accurate, these predictions suggest Argentavis might have been more likely to hit wingspans of 5-5.5 m than 6 m. This definitely takes Argentavis out of the running for having the largest wingspan of any bird. Wingspan estimates of big Pelagornis are 6-7 m and, in being based on much more complete material, we can be confident that we aren't overestimating their proportions (Mayr and Rubilar-Rogers 2010; Ksepka 2014).

The actual winner of the Grand Cenozoic Wingspan-off, Pelagornis (species shown here is P. chilensis, other species may have been a little larger). Of course, giant pterosaurs look at this competition to reach 6-7 m wingspans with rolled eyes and a bemused smile.
These revisions set our 70-80 kg Argentavis mass estimates in a new and interesting light. The mass of Argentavis was calculated using measurements of the hindlimb (specifically, the circumference of the tibiotarsus) which correlate well with total mass in living birds (Campbell and Tonni 1983). 70-80 kg sounds tenable for a 7-8 m span bird, but is very heavy for a 5-6 m one, especially if it’s meant to be a soaring species. According to my own mass/wingspan regressions*, 25-40 kg is a more likely mass for a bird this size. Might this suggest that Argentavis was either a heavyset bird with proportionally small wings (and maybe more swan-like than raptor-like in flight?), or might its legs be sending a skewed signal on account of being proportionally robust? There may be something to the latter idea (see below) but, in any case, this conflicting data is something we could investigate using other means of predicting mass, ideally those which estimate body volume instead of relying on scaling equations. Regressing mass from linear measurements can be useful but is also easily thrown when fossil animals are outside the size range or body shapes of living animals. There are already Argentavis skeletal reconstructions out there (e.g. Paul 2002, Chatterjee et al. 2007) waiting for this approach, and the results would shed light on which of those mass values (if either) is more likely.

*Based on 90 bird species with values taken from various literature.

2) Teratorns probably weren't giant Neogene vultures

This conflicting mass data has bearing on the other widely known ‘fact’ of teratornithid palaeobiology: that they were vulture-like scavengers. This idea is hugely influential in teratorn palaeoart where they are unwaveringly restored with condor- or vulture-like integuments and colours. It may be surprising to learn that, while vulture-like lifestyles are not without support (e.g. Palmqvist and Vizcaíno 2003; Fox-Dobbs et al. 2006), since the 1980s most studies of teratornithid functional morphology have suggested they were actually poorly suited to scavenging, and were instead active predators. Much of this research focuses on the best-known teratorn, Teratornis, but there's little reason to think that what's said for this taxon does not apply to the group as a whole.

It’s true that, at first glance, teratornithids seem like ideal scavengers. After all, flight studies and their phylogenetic affinities suggest that teratornithids were exceptional soarers, using updrafts to travel vast distances across American mountain ranges and open plains (e.g. Campbell and Tonni 1983; Vizcaíno and Fariña 1999; Chatterjee et al. 2007). Their upper jaws terminate in a hook that seems suited to ripping into carcasses and, as likely relatives of cathartids (and once considered relatives of storks), habitual scavenging would seem to be in their blood. The recovery of many Teratornis bones from the La Brea Tar Pits, to which they were presumably attracted by the promise of dead or dying animal flesh, is the cherry atop this particular palaeoecological cake.

Teratornithid skull material, so much as it is known, contrasts markedly with that of scavenging birds. The overall construction is more consistent with raptors that take live prey and, some details (the distended palate) are albatross-like, perhaps an adaption for prey restraint. From Campbell and Tonni (1981).
There are several compelling reasons to reject the scavenging hypothesis, however. To start with, Campbell and Tonni (1981, 1983) and Hertel (1995) noted numerous differences in skull structure between Teratornis and scavenging birds and linked these to foraging mechanics. Raptor skull shape is strongly influenced by dietary preferences (Hertel 1995) so the distinction between vulturine and teratornithid skulls is not to be shrugged off. Vulture skulls have low, narrow, largely inflexible and strongly hooked rostra which work essentially like meat hooks: they latch into chunks of flesh and pull them from carcasses using strong neck movements. They’re also mechanically weak against all but vertical forces, this probably reflecting the immobile nature of vulture foodstuffs and, for some species, limited options for lateral head motion when throngs of vultures feed at one carcass (Hertel 1995). Campbell and Tonni (1981) report that Teratornis, in contrast, has a highly flexible and broad skull with a deep, dorsoventrally parallel rostrum. Though possessing a well developed rostral hook, its size and association with a robust and straight jaw better matches raptorial grabbing aids than a vulturine 'meat hook', and thus seems ill-suited to tearing flesh (Hertel 1995). While scavenging is not precluded by this configuration, the bulky but loosely built teratorn skull does not match predictions of skull morphology for a habitual scavenger.

A better morphological match for teratorn skulls are birds which dine on living prey, such as large eagles and - more surprisingly - albatross (Campbell and Tonni 1981; Hertel 1995; Paul 2002). Like albatross, Teratornis has a low-slung palate which nestles neatly between the rami of the lower jaws when the mouth closes. This configuration grips prey by pinching it between the lateral surface of the palate and the inner margin of the mandible. Intriguingly, Hertel (1995) also found a maritime connection with Teratornis skulls, noting some skull proportions uniquely matching those of piscivorous raptors. Combined with the albatross-like jaw structure, we might wonder if aquatic prey was a routine part of teratorn diets (an idea also suggested by Paul 2002). I’m not fully convinced of this because the biometric signal of piscivorous raptor skulls is not strongly separated from those with less specialised diets (data in Hertel 1995), teratornithid skeletons lack features we’d expect of habitual waders or fishers and - perhaps most tellingly - Teratornis bone chemistry indicates a diet of terrestrial animals (Fox-Dobbs et al. 2006). These skull features are nevertheless evidence of teratorns being live-prey carnivores, not scavengers. Their strongly kinetic skulls - which included a loosely jointed mandible, and a jaw joint that expanded their gape 10% when the mouth is opened - implies a great ability to swallow prey whole at the expense of capabilities to tear it apart (Campbell and Tonni 1981). We might thus summarise their skull morphology as being suited to grabbing, holding and swallowing small animals.

The pelves of teratornithids weren't like those of other raptors, but more akin to those of birds which spend a lot of time walking around. From Campbell and Tonni (1983).
Working out what sort of prey teratorns preferred is aided by examining their skeletons. We can immediately rule out the use of their feet in capturing and restraining prey, these lacking the long, robust talons and indicators of a powerful grip characteristic to many birds of prey (Campbell and Tonni 1983). Teratornithids also have pelves which differ markedly from raptors that use their legs in prey capture. Eagles, falcons and similarly adapted birds have strongly bent posterior pelvic regions which optimise the orientation of their hindlimb musculature for powerful leg action. Teratornithids, in contrast, have relatively straight pelves that recall those of storks and other birds adapted for walking more than those of their raptorial cousins. Combined with their non-raptorial feet, we can probably rule out teratornithids gripping and carrying prey with their legs (Campbell and Tonni 1983), but can assume that they'd be much more comfortable striding around the ground than most other raptors.

Put together, these hindlimb features, the functional signature of their skulls and terrestrially-derived bone chemistry has seen many authors agree that teratornithids must have been caracara-like ground predators of smallish prey (e.g. Campbell and Tonni 1981, 1983; Vizcaíno and Fariña 1999; Paul 2002; Chatterjee et al. 2007). Their stork-like pelves would have facilitated more efficient walking than those of other raptors and, without large claws to imbue their locomotion, their strong feet and long legs would be ideal for sustained bouts of terrestrial activity. This ecology might play into their mismatched leg and wingspan proportions because strong legs have clear advantages for terrestrially hunting birds. Perhaps teratornithids used their legs for occasional powerful ground activity, such as stamping or standing on prey (suggested by Campbell and Tonni 1981), providing bursts of speed or digging for hiding animals? Strong legs could have also facilitated rapid landing if prey was spotted from the air, or allowed for explosive launches to avoid danger. Teratornithids may have been large, but some species lived alongside even larger predatory mammals. Their leg proportions were not suited to fast running (Campbell and Tonni 1983) and rapid escape to the air was surely necessary on occasion. Around 80-90% of avian launch power stems from their hindlimbs so, if teratornithids wanted to get airborne rapidly, having substantial, strain-resistant leg bones would be a good start. I'm curious to know what the launch prospects of Argentavis are if we factor extremely robust hindlimbs at the lower body masses proposed above: could these birds perhaps launch from a standing start?

My take on Teratornis merriami as we probably need to start picturing it: a ground-stalking hunter of small prey, such as brush rabbits (seen half-eaten on the right of the image). Teratornis is deliberately reconstructed to look more 'predatory' than vulturine here, for reasons made clear in this post. A mew gull and turkey vulture are included as a nod to the rich avian fauna that once lived alongside teratorns: it's odd to think that species we have around today once lived alongside these very large and unusual birds.
With jaws suited to eating essentially any animal they could fit in their mouths, we might imagine teratornithids as stalking across Neogene plains, water margins and mountains looking for all manner of small vertebrate prey - lizards, snakes, frogs, smaller mammals, ground birds and so on. Campbell and Tonni (1981) suggest that the 24 cm long Teratornis skull would put a 9 cm diameter limit on prey size, while the much larger Argentavis - with an estimated skull length of up to 55 cm - could have swallowed 15 cm wide prey. That means hare-sized animals, including small sloths, armadillos, and the notoungulate Paedotherium borrelloi, could have been regularly going down the throat tubes of Argentavis. Sounds like some fun concepts for palaeoart to me.

So there we have it: teratornithids, household names for many of us interested in palaeontology, may have been both smaller and ecologically very different to how we've mostly imagined them. All this said, in researching this piece I was struck by how much of our work on teratornithid size and ecology is now decades old. This doesn’t invalidate the points outlined here, but there's probably scope for bringing modern techniques to teratorn studies, both to pin down their lifestyles further as well as to explore that interesting mass/wingspan issue. Teratornithids seem like pretty awesome birds, so hopefully modern insights into their anatomy and lifestyles won’t be long coming.

Enjoy monthly insights into palaeoart, fossil animal biology and occasional reviews of palaeo media? Support this blog for $1 a month and get free stuff!

This blog is sponsored through Patreon, the site where you can help online content creators make a living. If you enjoy my content, please consider donating $1 a month to help fund my work. $1 might seem a meaningless amount, but if every reader pitched that amount I could work on these articles and their artwork full time. In return, you'll get access to my exclusive Patreon content: regular updates on upcoming books, papers, painting and exhibitions. Plus, you get free stuff - prints, high-quality images for printing, books, competitions - as my way of thanking you for your support. As always, huge thanks to everyone who already sponsors my work!


  • Campbell Jr, K. E., & Tonni, E. P. (1981). Preliminary observations on the paleobiology and evolution of teratorns (Aves: Teratornithidae). Journal of Vertebrate Paleontology, 1(3-4), 265-272.
  • Campbell Jr, K. E., & Tonni, E. P. (1983). Size and locomotion in teratorns (Aves: Teratornithidae). The Auk, 100(2), 390-403.
  • Chatterjee, S., Templin, R. J., Campbell, K. E. (2007). The aerodynamics of Argentavis, the world's largest flying bird from the Miocene of Argentina. Proceedings of the National Academy of Sciences, 104(30), 12398-12403.
  • Fox-Dobbs, K., Stidham, T. A., Bowen, G. J., Emslie, S. D., & Koch, P. L. (2006). Dietary controls on extinction versus survival among avian megafauna in the late Pleistocene. Geology, 34(8), 685-688.
  • Hertel, F. (1995). Ecomorphological indicators of feeding behavior in recent and fossil raptors. The Auk, 112(4), 890-903.
  • Ksepka, D. T. (2014). Flight performance of the largest volant bird. Proceedings of the National Academy of Sciences, 111(29), 10624-10629.
  • Mayr, G. (2009). Paleogene fossil birds. Springer Science & Business Media.
  • Mayr, G., & Rubilar-Rogers, D. (2010). Osteology of a new giant bony-toothed bird from the Miocene of Chile, with a revision of the taxonomy of Neogene Pelagornithidae. Journal of Vertebrate Paleontology, 30(5), 1313-1330.
  • Palmqvist, P., & Vizcaíno, S. F. (2003). Ecological and reproductive constraints of body size in the gigantic Argentavis magnificens (Aves, Teratornithidae) from the Miocene of Argentina. Ameghiniana, 40(3), 379-385.
  • Paul, G. S. (2002). Dinosaurs of the air: the evolution and loss of flight in dinosaurs and birds. JHU Press.
  • Vizcaíno, S. F., & Fariña, R. A. (1999). On the flight capabilities and distribution of the giant Miocene bird Argentavis magnificens (Teratornithidae). Lethaia, 32(4), 271-278.

Thursday, 25 July 2019

The science of the Crystal Palace Dinosaurs, part 4: The mammals of the Tertiary Island

An 1853 illustration of one of the Crystal Palace mammals, Palaeotherium magnum, imagined next to the plesiosaurians it would eventually share the Geological Court with. The Palaeotherium sculpture that this image is based on is now lost, just one of many misfortunes to befall the Crystal Palace mammals. From Die Gartenlaube, archived at Wikipedia.
It's time for our final visit to the prehistoric animal sculptures of Crystal Palace Park. Having toured through the Palaeozoic and Mesozoic exhibits in the last three posts (part 1, part 2, part 3), today we turn our attention to the Cenozoic section of the Geological Court, one containing exclusively mammalian palaeoart subjects. As with previous entries in this series, these words stem from writing palaeoart notes for the Friends of Crystal Palace Dinosaurs charity - please check out part 1 for additional context about their work and the ongoing need for care and maintenance of the Crystal Palace sculptures. You can help conservation efforts by donating money or volunteering your time to keep the Geological Court maintained.

At risk of being melodramatic, I find it difficult to escape a sense of tragedy when concerning myself with the Crystal Palace mammals. They are the least documented, least discussed and most suffering of the sculptures, with several bearing obvious hallmarks of neglect and low-quality repair work. Even in the 1850s they were being sidelined to make way for the more spectacular fossil reptiles, a fact all the more tragic because extinct mammals were the inspiration for having prehistoric animals at Crystal Palace in the first place (McCarthy and Gilbert 1994). Initially, Hawkins planned to restore a woolly mammoth and other large mammals but, when developing his display, his attention was drawn to the dinosaurs and fossil reptiles which would ultimately consume most exhibition space and public interest. Once located on their own 'Tertiary Island' (Doyle and Robinson 1993), the mammals are today situated on the 'mainland' component of Crystal Palace and it's hard not to view them as being a little tucked away. Although close to the reptile displays (within sight of the Mosasaurus), most of the mammals are located on a separate path to the reptilian sculptures and they are often obscured by foliage. It's quite easy to miss them on a casual walk around the park.

A sense that the mammals were being overshadowed by reptiles may explain why Hawkins wanted to expand this component of the park. He wrote to Owen in 1855 with a plan to augment his Cenozoic fauna considerably, including models of a mammoth, a bathing Deinotherium, glyptodonts, Sivatherium and extinct bison, as well as moa, dodo, turtles and snakes (Doyle 2008; Dawson 2016). This letter was dated a full month after Hawkins was no longer working for the Crystal Palace company however, who thought he had constructed enough models and would not even let him finish a half-completed mammoth sculpture. The official reason for terminating Hawkins' work was allegedly an artistic one of “less being more” (McCarthy and Gilbert 1994), but the abrupt termination of Hawkins’ project surely reflected the financial struggles of the Crystal Palace Company shortly after the park opened (Dawson 2016). It’s certainly difficult to believe that the same company who filled their park with reconstructed ancient buildings, expanded the Crystal Palace to incredible dimensions and built fountains rivalling the biggest in Europe would suddenly be concerned about artistic excess. Some idea of what Hawkins’ mammoth, Sivatherium and turtles may have looked like may be taken from his later drawings and paintings, including his 1860s poster series that borrowed heavily from his Crystal Palace designs (Rudwick 1992).

Hawkins' grand plans for the Tertiary Island, drawn on the back of a letter to Owen in 1855. He wanted it packed with Cenozoic mammals, reptiles and birds, but the Crystal Palace Company thought this was excessive (or, more likely, couldn't afford to pay for the work). From Doyle (2008).
To my knowledge, very little information survives regarding Hawkins’ construction of the four Crystal Palace mammal species. They were built and installed at the same time as the other models but were not mentioned in Owen’s (1854) Geological Court guidebook despite his interest in fossil mammals (e.g. Owen 1846). Perhaps this is further evidence of Owen’s general disinterest in the Crystal Palace project? Victorian visitors had to make do with a very brief and incomplete overview of the mammal fauna provided in the general Crystal Palace Park guide (Phillips 1854). Later versions of this book would tweak their text on the mammals to provide short, but often historically important, insights into their composition and display. So lacking is the documentation of the mammals that we're sometimes reliant on the throwaway text in these guides to tell us how many sculptures were originally installed!

From a palaeoartistic perspective, a clear distinction between Hawkins’ task with the mammal reconstructions and his more famous reptilian efforts was the availability of anatomical information. Mammalian palaeontology was considerably more advanced than studies of fossil reptiles in the early 1800s. Complete skeletons had been known for several of the Crystal Palace species for several decades, allowing scholars to describe, illustrate and restore the osteology of these animals in detail. Hawkins surely benefitted from Owen being an authority on the anatomy of mammals (e.g. Owen 1846), including the Crystal Palace species, and probably also made use of several pioneering skeletal reconstructions, muscle studies and life restorations published by Georges Cuvier. Neglected and somewhat forgotten as they are, the Crystal Palace mammals are actually pretty good takes on the form of their subject species, and clearly demonstrate Hawkins as the equal of later palaeoart masters.


The surviving Palaeotherium sculptures in their original site at Crystal Palace, photographed in 2013. These models were temporarily moved for a period in the mid-20th century, which may explain the damage to the sitting model and loss of a third, larger sculpture.
The Eocene equoid Palaeotherium was one of the first discovered fossil mammals and was studied in detail by Georges Cuvier during the early 1800s. Its entire osteology was understood from more or less the moment it was found thanks to near-complete skeletons being recovered from French gypsum deposits at the turn of the 19th century. These brought several Palaeotherium species to the attention of early palaeontologists and led to it being the subject species for some of the oldest palaeoartworks. Both its skeleton and body outline were restored by Cuvier and artists in his employ in the early 1800s (Rudwick 1992, 1997) and these images - after some initial hesitation from Cuvier - were eventually widely published in European literature. With so much data available, Hawkins probably had little difficulty restoring Palaeotherium in three dimensions for the Geological Court.

The Crystal Palace Palaeotherium have an unfortunate history. Two models survive today but photographs from 1958 (see McCarthy and Gilbert 1994), 19th century illustrations, and later editions of Crystal Palace Park Guide (Anon. 1871) indicate that a third model once existed. It was clearly larger and anatomically distinct from the surviving models, but at present no-one seems to know what happened to it - a most regrettable circumstance. It may have been relocated or destroyed when the Tertiary Island site was taken over with a small zoo in the 1950s (we know that parts of the zoo directly encroached into the space for the models - the base of the Megatherium was part of a goat enclosure, for example (see McCarthy and Gilbert 1994)) or else when the smaller mammal models were temporarily moved in the post-war period (Doyle and Robinson 1993). I hope it hasn't been destroyed and may still turn up in some neglected part of the park or emerge from someone's garage.

Two of the three Palaeotherium models photographed in 1958, printed by McCarthy and Gilbert (1994). The standing model shown here is remarkably different in form and size from the surviving Palaeotherium sculptures and almost certainly represents a different species (P. magnum?). Its whereabouts is unknown today.
The surviving Palaeotherium have not escaped misfortune either. The sitting sculpture lost its head at some point in the late 20th century and has been fitted with a replacement, but photographs show that the original head was quite different to the one it has now (compare image above with that below). The replacement is probably a replica or cast from the other surviving Palaeotherium. Both heads are very similar in the snout, ear and eye region, and the differences - the abbreviated cheek and braincase in the sitting statue - are likely results of marrying the head of the standing animal to a sitting one. The neck has also evidently been lengthened since the 1950s and the head/neck join lacks the well-executed muscle contours characteristic of Hawkins' work. It is not the only example of strange, somewhat crude, restoration work on the Crystal Palace mammals (see below).

The sitting Palaeotherium as it appears today - note the different head to the photo from 1958 above, and the slightly awkward manner in which the head replicated from the standing animal has been grafted to the neck.
Questions about lost models and restoration work are not the only uncertainties about these models. I’m not aware of any literature that identifies the Palaeotherium sculptures beyond generic level, but my assumption is that at least one of the surviving models represents P. minus. This sheep-sized species was well described and illustrated by Cuvier and others in the early 1800s, providing Hawkins with ample reference material. I'm uncertain whether both existing sculptures represent P. minus given their historic differences in head shape and other anatomies, but each was clearly distinct from the missing third model, which was significantly larger and of contrasting form. From photographs and illustrations I estimate that the missing model was about the size of a small horse, and this almost certainly labels it as P. magnum, another species that was well illustrated in literature of the early 1800s.

I see you, cryptic P. magnum, hiding in plain sight within P. H. Delamotte's 1853 illustration of Hawkins' workshop. I've long wondered what this sculpture was given that it didn't quite fit anything on display at Crystal Palace, but it's a perfect match for the missing P. magnum model - note the concave back, upright head, straight forelimbs and Gonzo-esque nose. Image modified from Wikipedia.
It’s difficult to evaluate the Palaeotherium models against the science of their day because of modifications made since their installation. Scholars of the early 19th century regarded Palaeotherium as a tapir-like animal with a short proboscis. Cuvier went as far as to suggest that some Palaeotherium species would, should we see them alive, be virtually indistinguishable from modern tapirs (Rudwick 1997). Hawkins evidently followed this suggestion with his smaller standing model, giving it a long, tapir-like face, an arched back, a podgy, creased torso recalling the Malayan tapir, and short, round ears. He opted to give the feet a more horse-like appearance however, which is appropriate to Palaeotherium limb anatomy. The sitting Palaeotherium model also has a tapir-like body, but it lacks the obvious creases of the other surviving model. Perhaps they are, indeed, meant to be different species. The previous head of this sculpture was certainly very different in being much shorter and smaller, and doesn't bear a strong resemblance to any living animal.

Stranger still is the lost model, which was far removed from a tapir-like form except for its short trunk. This large sculpture rather recalls African bush elephants, including a concave back, wrinkled skin, stocky limbs, a deep, short face, prominent brow and conspicuous orbital margins. The skull of P. magnum was not entirely known in the early 1800s and Hawkins may have taken this as an opportunity to be creative with the facial form of the larger model. His apparent referencing of elephants may seem unusual but, even though Palaeotherium was regarded as being related to horses even in the early 1800s, it was also considered it a member of Pachydermata. Today, the term ‘pachyderm’ is best known as being an obsolete taxon for elephants, rhinoceros and hippos, but in the early 1800s it included many hoofed mammals too. Under this classification, it might not have seemed much of a stretch to include some elephantine anatomy in a P. magnum restoration.

Palaeotherium magnum as we might reconstruct it today: essentially a robust, compact horse.
Hawkins’ surviving take on Palaeotherium - muddied as they’ve been by time - are not too far off how we regard this creature today: a browsing hoofed herbivore that must have looked something like a tapir or small horse. The now-lost short faces of the sitting and large sculpture are admittedly peculiar as all Palaeotherium have long, somewhat horse-like skulls (Rémy 1992), a fact well established by the 1850s. The introduction of elephant features into the P. magnum reconstruction is, of course, questionable. Elephants are now considered very distant relatives of hoofed mammals and there is no reason to think that they are a good soft-tissue analogy for Palaeotherium. The depiction of trunks is also probably erroneous. Short trunks have evolved repeatedly in perissodactyls and can be predicted for fossil species through a range of bony correlates (Wall 1980). Palaeotherium bears features indicating a particularly fleshy set of lips, but it lacks the full suite of features we associate with having a proboscis.


A parade of Anoplotherium commue hanging out at the water's edge in 2013. These remain, a few details aside, pretty darned good takes on Anoplotherium anatomy. Note the impression of musculature in the tails: even though they're hanging low, they look powerful and mobile.
Like Palaeotherium, Anoplotherium was an early subject of palaeoartistic reconstruction at the hands of Georges Cuvier. Two incomplete skeletons of this peculiar hoofed mammal were recovered from Eocene gypsum deposits adjacent to Paris at the turn of the 19th century and, with these, Cuvier was able to reconstruct most of its osteology in a series of papers published from 1804 to 1825 (see Rudwick 1997 and Hooker 2007 for discussion and references). Cuvier's skeletal reconstructions and basic life restorations of Anoplotherium were widely reproduced and would have been well-known among Victorian scholars. Cuvier privately developed muscle studies based on the same illustrations but did not publish them out of concern that they were too speculative for scientists of the early 19th century (Rudwick 1992, 1997). Cuvier was clearly ahead of his time in this regard, foreshadowing a practice that would become important to studies of functional morphology as well as an essential part of the palaeoartistic reconstruction processes.

Cuvier is thus very much the architect of the Crystal Palace Anoplotherium, and Hawkins followed his vision fairly faithfully. He deviated by giving the Anoplotherium camel-like facial details, including large lips, small, rounded ears and a sloping skull roof. Referencing camel anatomy was not Hawkins’ whimsy but informed by early ideas of where Anoplotherium sat in mammalian systematics. This depiction was a departure from scientific credibility however, and Cuvier’s take, with its lower snout and modest lip tissues, was more in keeping with the underlying skull and inferred soft-tissue anatomy of Anoplotherium. This seems to be another example of Hawkins transferring anatomy from living species rather than, as he often did, reconstructing it objectively from fossil bones. Another error is the reconstruction of four toes on each foot. Anoplotherium feet actually had three toes each: two hoofed main digits, and single, somewhat opposable ‘thumbs’ on the inside of each limb (Hooker 2007). Cuvier was aware of there being three digits on the forelimbs at least, and it’s possible that Hawkins added more toes because he thought the fossils were incomplete or otherwise somehow anomalous. After all, Anoplotherium is meant to be an even-toed hoofed mammal, and even today it's an oddball for its unusual toe counts. But other than these relatively minor errors, the Anoplotherium sculptures are compelling reconstructions that are still used to illustrate the form of this taxon today (e.g. Prothero 2017). I particularly like the strong, flexible-looking tails and the form of their muscular torsos.

Anoplotherium commune, the middle portion of a Venn diagram containing Bambi, Lassie and Rory Calhoun, in its characteristic feeding posture. Our anatomical interpretation of this animal is very similar to how Cuvier and Hawkins reconstructed it, but we have different ideas about its lifestyle.
Hawkins created three Anoplotherium sculptures: one standing, one resting, and one in a curious half-crouched pose with an outstretched neck and head*. I’m not entirely sure what behaviour the latter is meant to depict. In the early 1800s Anoplotherium was regarded as a swimming animal that used its powerful tail to propel itself through water, perhaps like an otter or coypu (e.g. Owen 1846). Maybe the third animal is meant to be shaking itself, dog-style, to dry off as it emerges from the water surrounding the Tertiary Island? The soft-tissues of the neck are inconsistent with the other models, and this is not the result of damage or poor conservation. Might it represent deformation of the skin as the neck is shaken about? Today, Anoplotherium is interpreted as a fully terrestrial animal adapted for high browsing (Hooker 2007). Peculiarities of its pelvis are shared with mammals that regularly stand upright on two legs, and it’s probable that Anoplotherium adopted this pose to browse above the feeding envelope of contemporary mammals (Hooker 2007). The strong tail, in this hypothesis, becomes a stabilising organ rather than a swimming aid.

*The 1854 Routledge's Guide to the Crystal Palace and Park at Sydenham suggests there are meant to be four Anoplotherium, but I'm not aware of any other documents indicating this. Is there another missing model, or was this a typo?

"Anoplotherium" gracilis - or, more appropriately, Xiphodon gracilis. Some authors suggest that some of the Crystal Palace Anoplotherium sculptures represent this species, but I strongly doubt this. gracilis has a completely different shape to commune and this was understood early in the 1800s. This image is by Georges Cuvier, and it pre-dates the Crystal Palace project by several decades. From Rudwick (1997).
According to McCarthy and Gilbert (1994) and Doyle and Robinson (1993), two species of Anoplotherium are represented at Crystal Palace: the two standing individuals are A. commune, and the reclining sculpture is “A”. gracilis. It’s not clear to me that this is accurate, however. Firstly, by the time Crystal Palace Park opened A. gracilis was well-differentiated taxonomically from A. commune. Cuvier placed gracilis in a subgenus, Xiphodon, in 1822, and this was erected to a full genus by M. Paul Gervais in 1845. Owen agreed with this change and stopped referring to “Anoplotherium gracilis” at some point between the mid-1840s and mid-1850s (see Owen 1846, 1856, 1857). By the time the Crystal Palace sculptures were being commissioned the disassociation between Anoplotherium and X. gracilis was thus well established, and we have to assume that Hawkins and Owen were aware of it. A complication to this is that Hawkins still referred to A. gracilis in the early 1860s (judging by the labelling on one of his 1862 posters), but this brings us to our second point: that Hawkins evidently knew how different gracilis and commune were anatomically. His 1862 posters show commune as reconstructed at Crystal Palace while his gracilis are the long-legged, long-necked, and short-tailed creatures of Cuvier and other artists in the early- and mid-1800s (see Rudwick 1997 for Cuvier’s own detailed accounts on the anatomy of this species). This isn’t surprising: in the early 19th century commune and gracilis were regular fixtures in palaeontological texts, and skilled, intelligent artists like Hawkins would not readily confuse them. This is not to say that claims of gracilis being featured at Crystal Palace are baseless, but they neither marry up with the Anoplotherium statues we have today nor the history of Anoplotherium research. As with the uncertainty about the taxonomic representation of the Palaeotherium sculptures, the poor records and deficit of historic interest in the Geological Court mammals do little to help resolve this confusion.


Towering above surrounding vegetation is Hawkins' Megatherium americanum, shown here as it was in 2013. There's always some vegetation obscuring this sculpture and, last time I visited the park, it was near impossible to see Megatherium at all. This obscures, among other things, the expertly sculpted feet, legs and tail. Controversy still reigned about the foot posture of ground sloths in the 1850s, but Hawkins was spot on in his depiction.
The research history of Megatherium americanum began nearly six decades before Hawkins commenced work on his model. Despite this long research lead, Hawkins’ rearing, tree-grasping Megatherium was ultra-progressive for the time. It was one of the first depictions of Megatherium in a pose that chimes with our modern understanding of giant sloth habits and actually pre-dated publication of ideas that it was capable of such feats. Megatherium was, for much of the early 1800s, regarded as Georges Cuvier imagined it in the latest 1700s: a flat-footed, trunked quadruped with particularly dextrous forelimbs (Rudwick 2005, Argot 2008). Even into the 1850s scholars were still confused over aspects of how this animal lived, with authors like François Jules Pictet-De la Rive writing long discussions about its capacity for burrowing, climbing, and harvesting vegetation. Owen's studies on another species, Mylodon darwinii, elucidated many aspects of ground sloth lifestyle and anatomy that would inform Hawkins model. Owen showed that Mylodon walked on the sides of its feet and the shorter, clawless fingers of its hands. He viewed giant sloths as browsers and tree fellers based on numerous lines of anatomical evidence (Owen 1842) and, under his direction, a Mylodon skeleton was restored as a tree-rearing biped at the Hunterian Museum in the late 1830s/early 1840s. In contrast, contemporary European mounts of Megatherium and illustrations of its skeleton retained fully quadrupedal, flat-footed stances.

The Crystal Palace Megatherium being manufactured in an image published by Die Gartenlaube in 1853. Note the size of the hands, which are much larger than the surviving hand on the model today. From Wikipedia.
Owen eventually wrote at length about tree-rearing giant sloths when discussing the anatomy and habits of Megatherium, but only well after the Crystal Palace models were completed. An article in the German magazine Die Gartenlaube shows that Hawkins’ Megatherium model was completed in 1853, a year that could - if Hawkins had been strictly following available Megatherium reconstructions - have seen the Crystal Palace ground sloth restored as a Cuvierian quadruped. The fact that Hawkins avoided this implies that he either combined Owen’s ideas on Mylodon with the anatomy of Megatherium, or else that Owen tipped him off about the direction of his future research. Either way, the portrayal of an elephant-sized mammal rearing into a tree has to be regarded as extremely progressive for the 1850s. Other early 19th century scholars assumed such animals were confined to quadrupedal poses (Argot 2008) and, while it was not a stretch to imagine the bear-sized Mylodon routinely rearing onto its back legs, it was a bold prediction to portray an enormous Megatherium doing the same. As demonstrated in many Crystal Palace models, Victorian scholars often imagined fossil animals as variants on recognisable modern forms - paleotheres as tapirs, dicynodonts as turtles, pterosaurs as birds - but Hawkins’ Megatherium was boldly different from anything alive today, and foreshadowed the way we would start visualising prehistoric species as our science and data improved.

The Crystal Palace Megatherium is of further note for being one of the oldest life restorations of a ground sloth. Though several skeletal reconstructions were published prior to the 1850s, few, if any, restorations with a complete suite of soft-tissues were attempted. (It’s curious that none have been found among Cuvier’s archives, given his links to Megatherium and his habit of restoring the anatomy of fossil mammals). Despite its vintage, Hawkins’ Megatherium has held up well as a portrayal of ground sloth form. A commendable portrayal of proportion and musculature is buried under layers of long, shaggy hair, with the muscular, relatively slender shoulders contrasting appropriately against the wide and robust pelvic region. The feet are appropriately inturned and the arms are depicted as if grasping a tree to access vegetation or push it over, entirely in accordance with Owen’s interpretations of sloth behaviour. Hawkins’ depiction of shaggy hair anticipated the discovery of giant sloth hair by almost half a century (Woodward and Moreno 1899) and, although it remains unclear whether Megatherium itself was covered in such fur, this take is certainly consistent with the fossil skin of several giant sloth species.

Megatherium americanum as we know it today - really not so different from how Hawkins envisaged it 165 years ago.
Two major difference between Hawkins’ Megatherium and our modern reconstructions are obvious. The first is the presence of a short proboscis, a Cuvierian interpretation also endorsed by Owen (1842). Hawkins’ restoration would have pleased Victorian scientists, but trunked sloths have not withstood modern scrutiny. Today, it is instead thought that ground sloths had extensive nasal cartilage and prominent lip tissues (Bargo et al. 2006), but they lack features indicative of trunks or proboscides.

It's hard to see the Megatherium sculpture in full, especially from anterior view, so I've borrowed this photo from the Friends of Crystal Palace Dinosaurs website. Note the excellent rendering of the crouching, pedolateral hindlimbs, and also the strangely small left hand (the right is missing). You can see the colour mismatch between the left forearm and elbow - this marks a site of repair where the tree outgrew the grip of the sculpture. But what's with that tiny hand? Its detailing and grafting onto the forelimb is weird and doesn't match illustrations of the Megatherium sculpture from the 1850s (above). It's surely a crudely-sculpted replacement, not a replica of the original.
The hands (or rather, the hand - the right seems to be missing at present) are the second anomaly, and are far less explicable. As represented today, the surviving hand is curiously undersized, lacks claws and is so poorly shaped that I initially assumed it was a post-Hawkins replacement, akin to the replaced Palaeotherium head. The left hand was replaced after the growing girth of the tree broke the sculpture’s left forearm, and I presumed a diminutive hand was added due to lack of space. But no, at least according to McCarthy and Gilbert (1994), the Megatherium hand currently on the model is a replica of the real deal. This seems peculiar, and I'm not sure I buy it. Hawkins’ illustrations of Megatherium from the 1850s and 60s (including artwork associated with Crystal Palace) show appropriately large, clawed hands, and an illustration of the model being constructed in 1853 (above) shows it grasping a tree with sizeable, robust extremities. The rest of the model is so exact to Megatherium form that the embarrassingly inaccurate hands are entirely out of place - they look like they were made by someone who had no idea about Megatherium anatomy, which is patently not the case for the rest of the model. I strongly suspect that the hands of the model are crude replacements of lost originals, and that there’s a missing chapter in the history of this model.

Megaloceros, the Irish elk

The Crystal Palace Megaloceros bucks and doe on display in 2018. The attention to detail on these models is superb and, today, their situation close to pathways around the Geological Court allows visitors to get extremely close.
Probably the most spectacular mammal sculptures at Crystal Palace Park are the four Megaloceros giganteus situated in at the northeastern extent of the Geological Court. A reposed doe and fawn feature alongside two large bucks, each standing in a classic ‘regal’ pose with antlers aloft. So imposing are these sculptures that they would not look out of place situated among grand governmental buildings, or atop an enormous plinth in a city square. It seems strange that no iterations of the 19th century Crystal Palace Park Guides suggested starting tours of the Geological Court with this display. The Megaloceros, after all, slowly leads us into the strangeness of extinct animals ("they're deer, Jim, but not as we know them") as well as demonstrates Hawkins’ ability to create believable animals (a fact far easier to deduce with a deer than a dicynodont). Beginning a tour from the other end of the court, with the far less impactful and more distant Dicynodon, robs us of this effect. I also feel that the grandeur and strangeness of Victorian dinosaurs, marine reptiles and giant sloths overshadows Megaloceros somewhat. Sure, it's big and the antlers are impressive, but its "wow" factor is diminished after meeting the stranger, larger reptiles situated a few hundred metres away. I can understand why Hawkins was pushing for additional, less-familiar species - mammoths, dodos, moas - to place around this end of the Court before his funding was pulled.

Hawkins had no concern for missing anatomy or predicting proportions with Megaloceros. Decades before the Crystal Palace project was even conceived, Megaloceros osteology was extensively described and illustrated. From Cuvier (1827).
Megaloceros was a historic fossil species even to Hawkins and Owen. Remains of this animal were found in the late 1600s and, by the early 1800s, enough material was known to reconstruct the entire skeleton. Cuvier (1827) published several such illustrations, including two skeletal reconstructions that were widely reproduced in later texts on fossil mammals. This, and the glut of Megaloceros material held by British museums, would have given Hawkins an excellent insight into its anatomy. Visitors to Crystal Palace would not have known the Irish Elk as Megaloceros giganteus, however, but under Owen’s 1844 name for the species, Cervus (Megaceros) hibernicus. The nomenclatural history of M. giganteus is confused by several names being coined for this species in the 18th and 19th centuries. Owen’s subgenus (eventually promoted to a ‘true’ genus) Megaceros was the first to enter widespread use and almost became the accepted generic name for giganteus. However, Megaloceros was resurrected in 1945 (albeit somewhat corrupted from its original spelling, Megalocerus) and both names were applied to the Irish Elk until the 1980s. Adrian Lister (1987) finally brought an end to decades of confusion by establishing Megaloceros giganteus as the most appropriate name on grounds of both nomenclatural priority and usage.

Hawkins' Megaloceros doe and fawn, as seen in 2013. I really enjoy the detailing on their feet - if there was any doubt that Hawkins could make realistic-looking familiar creatures as well as weird fossil ones, these models dispel it.
Megaloceros was surely the least demanding of Hawkins' reconstruction assignments because of its close relationship to living deer. As did Cuvier, Owen realised that early interpretations of Megaloceros as a giant moose-like cervid were incorrect, and he placed it among Cervus, a genus that includes several other large, Old World deer species (e.g. Owen 1844). Hawkins appears to have referenced several Cervus anatomies in his reconstructions, especially the thick neck manes, deep fur over the withers, and a line of long, shaggy fur along their bellies. These are especially obvious on the bucks, but also present on the reclined doe. Manes are not common to many deer females, and I suspect Hawkins was referencing the winter appearance of certain elk subspecies (‘elk’ as in the wapiti Cervus canadensis, not the Eurasian elk/moose). The short, blunt tails seem to agree with this interpretation too. It would later be traditional to reconstruct Megaloceros like the red deer Cervus elaphus, but I wonder if Hawkins thought the shaggy appearance of winter elk was more apt for an Ice Age animal, or else if he thought the longer fur would look more obvious on his sculptures.

Hawkins' reconstructions of Megaloceros are, of course, some of the most scientifically credible of all his Crystal Palace artworks. But I have to admit that, on grounds that their restoration was nowhere near as complex as the other sculptures, I don’t think they’re the best examples of his palaeoartistic abilities. They certainly leave little doubt that Hawkins could produce convincing portrayals of semi-recognisable animals, but Megaloceros probably wasn’t much of a stretch for him. His artistic expertise included illustrating living mammals and he eventually wrote a series of books on this very topic (including one featuring deer, in 1876). For me, it’s his deductions about lesser-known and wholly unfamiliar fossil species that place him among the old masters of palaeoart, even though these insights are associated with sculptures that are scientifically more dated.

Today, we imagine Megaloceros almost as Hawkins did, excepting some different ideas about their colouration and soft-tissue anatomy. These have been provided by cave art and revelations about their relationships with modern deer.
While Hawkins’ Megaloceros are impressive reconstructions, they differ from our considerations of this animal today. It seems that Megaloceros was more closely related to fallow deer Dama than Cervus, and this implies some differences in facial anatomy and colouration, as well as some particulars of fur and soft-tissue distribution (e.g. a bulging laryngeal region and brush-like genital sheath). Some of these anatomies are confirmed in Megaloceros cave art (Geist 1999; Guthrie 2005), which also records other details unknown from fossils. These include a shoulder hump (presumably long hairs, fat or both) and a series of dark stripes: one at the base of the neck, one running from the shoulder towards the knee, and another surrounding a pale rump. Cave art also suggests, though not conclusively, that the head and neck were pale or white, while the body was darker, perhaps light brown. (There's a terrific summary of Megaloceros life appearance over at Tetrapod Zoology - check that out for additional details and discussion). This information was entirely unknowable to Hawkins, however, as the discovery of ancient European cave art post-dated the Crystal Palace project by over a decade, and its acceptance as the work of authentic Palaeolithic humans, and not modern vandals, was even longer coming. Moreover, even once ancient cave art was accepted as a genuine part of European history in the early 20th century, it would take decades to discover enough Megaloceros cave paintings to deduce meaningful details of its anatomy and colouration.

So, about those Crystal Palace Dinosaurs...

That brings our palaeoartistic review of the Crystal Palace palaeoartworks to a close, kudos to anyone who's read the entire series. This was meant to be little more than a series of brief notes and it's ended up being a number of long articles. Having already been interested in the Crystal Palace sculptures before writing this, I must admit to having a true fascination with them now. Writing these pieces has revealed so many gaps in our understanding of their history and development, allowed me to appreciate just what a good artist Hawkins was, and driven home the fact that the Geological Court sculptures were really not an Owen-Hawkins collaboration, but almost a solo Hawkins project. The latter point is already well made by records of correspondence between Owen and those involved in the Crystal Palace project, but trying to 'reverse engineer' Hawkins' artwork further demonstrates the token involvement Owen must have had. In this light, Hawkins really needs to be discussed more widely as one of the all-time greats of palaeoart. Despite relatively little scientific assistance, he produced spectacular, realistic and charismatic takes on fossil animals at a time when our understanding of animal anatomy - both fossil and modern - was a fraction of what it is today. Yes, he got many things 'wrong' with respect to our modern understanding and he perhaps leaned on living animals more than we would nowadays, but to focus on this, and not his achievements in anatomical prediction, his knowledge of contemporary science and attention to anatomical detail, does him a disservice. We have to evaluate the Geological Court models in light of what was known at the time, and in this respect they are truly first rate. Far superior in science and art, in fact, to the vast majority of palaeontological sculptures exhibited today.

Furthermore, I feel more committed than ever to the fact that these models should, no, must be conserved for future generations. As globally unique monuments to Victorian science and culture, we should regard them with pride, reverence and admiration, and not allow them to deteriorate through neglect, underfunding and (sad to say) deliberate vandalism. The ongoing work by the Friends of Crystal Palace Dinosaurs is essential to this mission and I salute them for pushing the value of these sculptures against the odds, and for their successes so far. As I've said repeatedly throughout this series, if you share my interest and concern for Hawkins' Crystal Palace palaeoartworks then check out the FOCPD website you see how you can help: chip in some money to help conserve the models or provide some elbow grease to help maintain the Geological Court. If we let these models slide too far into disrepair there's no coming back for them: all their artistry, history and scientific significance will be gone for good. Please take interest and help out if you can.

Enjoy monthly insights into palaeoart, fossil animal biology and occasional reviews of palaeo media? Support this blog for $1 a month and get free stuff!

This blog is sponsored through Patreon, the site where you can help online content creators make a living. If you enjoy my content, please consider donating $1 a month to help fund my work. $1 might seem a meaningless amount, but if every reader pitched that amount I could work on these articles and their artwork full time. In return, you'll get access to my exclusive Patreon content: regular updates on upcoming books, papers, painting and exhibitions. Plus, you get free stuff - prints, high-quality images for printing, books, competitions - as my way of thanking you for your support. As always, huge thanks to everyone who already sponsors my work!


  • Anonymous. (1871). Crystal Palace: Guide to the Palace and Park. R.K. Burt, London.
  • Argot, C. (2008). Changing views in paleontology: the story of a giant (Megatherium, Xenarthra). In Mammalian Evolutionary Morphology (pp. 37-50). Springer, Dordrecht.
  • Bargo, M. S., Toledo, N., & Vizcaíno, S. F. (2006). Muzzle of South American Pleistocene ground sloths (Xenarthra, Tardigrada). Journal of Morphology, 267(2), 248-263.
  • Cuvier, G. (1827). Essay on the Theory of the Earth. W. Blackwood.
  • Dawson, G. (2016). Show me the bone: Reconstructing prehistoric monsters in nineteenth-century Britain and America. University of Chicago Press.
  • Doyle, P. (2008). A vision of ‘deep time’: the ‘Geological Illustrations’ of Crystal Palace Park, London. Geological Society, London, Special Publications, 300(1), 197-205.
  • Doyle, P., & Robinson, E. (1993). The Victorian ‘Geological Illustrations’ of Crystal Palace Park. Proceedings of the Geologists' Association, 104(3), 181-194.Hooker, J. J. (2007). Bipedal browsing adaptations of the unusual Late Eocene–earliest Oligocene tylopod Anoplotherium (Artiodactyla, Mammalia). Zoological Journal of the Linnean Society, 151, 609-659.
  • Geist, V. (1999). Deer of the World. Swan Hill Press, Shrewsbury.
  • Guthrie, R. D. (2005). The Nature of Paleolithic Art. The University of Chicago Press.
  • Lister, A. M. (1987). Megaceros or Megaloceros? The nomenclature of the giant deer. Quaternary Newsletter, 52, 14-16.
  • McCarthy, S., & Gilbert, M. (1994). The Crystal Palace dinosaurs: The story of the world's first prehistoric sculptures. Crystal Palace Foundation.
  • Woodward, A. S., & Moreno, F. P. (1899). On a Portion of Mammalian Skin, named Neomylodon listai, from a Cavern near Consuelo Cove, Last Hope Inlet, Patagonia. Proceedings of the Zoological Society of London, 67, 144-156.
  • Owen, R. (1842). Description of the Skeleton of an Extinct Gigantic Sloth: Mylodon Robustus, Owen, with Observations on the Osteology, Natural Affinities, and Probable Habits of the Megatherioid Quadrupeds. J. van Voorst.
  • Owen, R. (1846). A History of British Fossil Mammals, and Birds. J. Van Voorst.
  • Owen, R. (1854). Geology and inhabitants of the ancient world (Vol. 8). Crystal Palace Library.
  • Owen, R. (1856). On the Affinities of the Stereognathus ooliticus (Charlesworth), a Mammal from the Oolitic Slate of Stonesfield. Quarterly Journal of the Geological Society, 13, 1-11.
  • Owen, R. (1857). Description of the Lower Jaw and Teeth of an Anoplotherioid quadruped (Dichobune ovina, Ow.) of the size of the Xiphodon gracilis, Cuv., from the Upper Eocene Marl, Isle of Wight. Quarterly Journal of the Geological Society, 13 254-260.
  • Phillips, S. (1854). Guide to the Crystal Palace and park. Crystal Palace Library.
  • Prothero, D. R. (2016). The Princeton field guide to prehistoric mammals. Princeton University Press.
  • Rémy, J. A. (1992). Observations sur l'anatomie crânienne du genre Palaeotherium (Perissodactyla, Mammalia); mise en évidence d'un nouveau sous-genre, Franzenitherium. Palaeovertebrata.
  • Rudwick, M. J. (1992). Scenes from deep time: early pictorial representations of the prehistoric world. University of Chicago Press.
  • Rudwick, M. J. (1997). Fossil Bones and Geological Catastrophes. New Translations & Interpretations of the Primary Texts. University of Chicago Press.
  • Rudwick, M. J. (2005). Bursting the limits of time: the reconstruction of geohistory in the age of revolution. University of Chicago Press.
  • Wall, W. P. (1980). Cranial evidence for a proboscis in Cadurcodon and a review of snout structure in the family Amynodontidae (Perissodactyla, Rhinocerotoidea). Journal of Paleontology, 54, 968-977.