May 17, 2016

Today, the largest turtle alive is easily the leatherback turtle (Dermochelys coriacea), at a whopping 916 kg (2,015 lbs, Eckert & Luginbuhl 1988). On land, the largest turtle goes to Chelonoidis nigra (Galápagos tortoise) which has been reliably recorded as reaching up to 417 kg (919 lbs) in weight (Guinness World Records). However, both extant turtles are dwarfed in size by an immense land tortoise that lived as little as 1.7 million years ago, in the Pleistocene.

My name is…who?

Figuring out the name for this giant tortoise has not been easy. The naming of this immense beast has taken a remarkably complicated journey. When it was originally described, the authors—Hugh Falconer and Captain Proby Thomas Cautley (1837)—christened it: Megalochelys sivalensis, “The mega turtle from the Siwalik Hills”. However, this name was fairly quickly withdrawn in favour of the even more impressive name: Colossochelys atlas, “The colossal turtle that held up the world” (Falconer and Cautley 1844). Indeed, aesthetics were the sole reason for the new name, as quoted by Murchison (1868):

‘…the term “Megalochelys” was thought not to convey a sufficiently expressive idea of the size.’

Unfortunately, the timing of this turtle’s discovery was not the best time for reptile paleontology. Subsequent authors argued that Colossochelys atlas was so similar to modern-day tortoises as to be indistinguishable at the generic level, and thus they sunk Colossochelys atlas into the bloated genus: Testudo (Murchison 1868; Lydekker 1880, 1889). Still later authors seemed unhappy with plopping the animal in Testudo and instead opted to stick it in the equally bloated genus: Geochelone (Hooijer 1971; Auffenberg 1974). Thankfully, during the past ten years many herpetologists have started to take more of a “splitter” mentality to reptilian taxonomy, and many of these formally bloated genera have since been split. This has largely been done by simply elevating the subgenera that often accompanied these bloated genera (in a sense, formally recognizing the diversity that many previous authors had already hinted at).

So both Testudo and Geochelone are now out of the running, which leaves us with our original two choices.

Well, due to the vagaries of taxonomic nomenclature, the name was still in a bit of a taxonomic quagmire. It turns out that the original description of the beast (Falconer and Cautley 1837) was not a description at all, so much as a brief communication.

Testudinata are represented by the Megalochelys Sivalensis (Nob), a tortoise of enormous dimensions which holds in its order the same rank that the Iguanodon and Megalosaurus do among the Saurians.This huge reptile (the Megalochelys)—certainly the most remarkable of all the animals which the Sewaliks have yielded—from its size carries the imagination back to the era of gigantic Saurians. We have leg bones derived from it, with corresponding fragments of the shell, larger than the bones in the Indian unicorned Rhinoceros! — Falconer & Cautley 1837

That’s it. That was all that was originally written. Basically, a quick blurb that focused mostly on “how awesome” the animal was. According to the rules of the International Coalition on Zoological Nomenclature (ICZN), this meant that the taxonomic name, Megalochelys sivalensis was a nomen nudum, or a “naked name.” There was no described material to associate with the specimen, so the name became invalid. This wasn’t that big a deal, initially, given that Falconer and Cautley did provide a description for their material, later in 1844 when they named it Colossochelys atlas. However, as with all bureaucracies, the devil was in the details. According to Rhodin et al. (2015), although the lack of description for the original name declaration: Megalochelys sivalensis , did invalidate the species name, it didn’t invalidate the genus name. According to the ICZN Article 11.4.1:

A published work containing family-group names or genus-group names without associated nominal species is accepted as consistent with the Principle of Binominal Nomenclature in the absence of evidence to the contrary.

So, the genus: Megalochelys remains valid even if its species epithet does not. This meant that Colossochelys is forever sunk as a junior synonym of Megalochelys, but the species name: “atlas” remains. Thus we are left with the current name for the species: Megalochelys atlas (Rhodin et al. 2015).
How mega was Megalochelys?

This has been another surprisingly difficult thing to nail down. Read a generic book on prehistoric animals, or even check the internet and one will find a maximum weight of 4000 kg bandied about. I’m not entirely sure where this number originally came from (but see below for a potential origin). Given that M. atlas is often compared to a Volkswagen beetle in size, one can see hints of how this over-exaggeration can come about.

What of the actual, scientific measurements?

Well, it turns out that they are actually pretty few and far between. The original, gigantic size for M. atlas can be traced back to its original description by Falconer and Cautley (as detailed in Murchison 1868). The authors repeatedly compared the size of their specimen to that of an Indian rhinoceros (Rhinoceros unicornis), which can reach masses of 2300 kg (Macdonald 2001; Clauss et al. 2005). This was not just hyperbole, though, as the authors wrote:

A specimen comprising the upper part of the humerus was exhibited, exceeding in size the same portion of the corresponding bone of the Indian Rhinoceros…

So, based on general size comparisons with an Indian rhino, it would appear that M. atlas may have reached slightly over two tonnes in mass. Of course, this is operating under the assumption that the rhino they looked at was a large male, and that they accounted for the differences in anatomical proportions between rhino and tortoise humeri.

Falconer and Cautley never gave a weight estimate to their M. atlas fossils, but Falconer did provide an estimate of overall length, based on a comparison of bony elements between M. atlas and a Chelonoidis nigra specimen available to him from the College of Surgeons. That specimen had a carapace length of 107 cm (42 inches), and a curved carapace length (CCL) of 137 cm (54 inches). According to the Galápagos Islands tourism site, that translates roughly to an animal of about 227 kg (500 lbs) in weight. In contrast, the estimated carapace length and CCL for M. atlas was estimated as: 373 cm (12 ft 3 inches) and 480 cm (15 ft 9 inches) respectively. That is 3.5 times longer than the modern-day Galápagos tortoise. If we assume that the overall anatomy was similar between both species, and if we assume that M. atlas scaled isometrically to C. nigra, then we can use known scaling relationships such as the fact that volume (and hence, mass) scales to the third power, to calculate the estimated mass for Falconer and Cautley’s large M. atlas specimen.

3.53 = 43 times heavier.

227 kg * 43 = 9,761 kg (9.7 tonnes)

This would have made Megalochelys atlas insanely huge, putting it on par with a very large bull African elephant. Of course, this is an isometric scaling attempt, which works for members of the same species (think Shaq vs. Jet Li), but doesn’t hold true across species. Modern-day giant tortoises are a lot lighter than they would appear based off a cursory glimpse of their skeletons. This is due to the evolution of weight saving features on their bodies, such as a honeycombing of the shell microstructure (Scheyer and Sánchez-Villagra 2007). Although there has yet to be any microstructural analyses of M. atlas shells, gross anatomical descriptions indicate that similar weight-saving features were likely present in this animal as well (see below). As such, the estimated weight based on isometry is likely going to be too much. Exactly how much is difficult to assess, but it is still unlikely to bring the weight down to the typical heavyweight estimate of 4 tonnes.

What else could explain these enormous estimates? For that we need to return to the original description of the animal.

The original dimensions of M. atlas, as estimated by Falconer and Cautley, were huge. Falconer calculated that Megalochelys atlas had a 3.73 meter (12 ft 3 inch) carapace, with a 1.8 meter (6 ft) neck, and a 0.6 meter (2 ft) head and tail. So, the initial total length of the animal was estimated at 6.8 meters (22 ft 3 inches). At this colossal size, a mass of 5 or 6 tonnes would not be out of the question. However, it turns out that the original estimates given by Falconer and Cautley (1844) were mismeasurements taken from a composite specimen (composite specimens will compound any errors in the original measurements). Lydekker (1889) first called this into question and suggested that the actual length of the carapace in the large M. atlas specimen found by Falconer and Cautley, was probably about 1.8 meters (6 ft). This result agreed well with a second, preserved shell discovered by Barnum Brown in 1931. Upon reconstruction, Brown found the curved carapace length to be 2.2 meters (7 ft 4 inches). According to Brown (1931), the specimen they recovered was approximately the same size as the one Falconer and Cautley had discovered nearly a century earlier. Brown’s reconstruction resulted in a tortoise that was 1.6 times longer than a modern-day Galápagos tortoise, which translates to a 4x estimated higher mass, or approximately 929 kg (2,046 lbs). Interestingly, Brown attempted to estimate the mass of Megalochelys atlas using a 1:1 scale model that was immersed in water to determine its volume via displacement. His results suggested that their M. atlas specimen would have weighed 955 kg (2,100 lbs), which isn’t too far off from the isometry estimate.

Mind you, the 1 tonne estimate for M. atlas, comes strictly from the carapace measurements. Falconer and Cautley’s original material was a composite specimen, meaning that large, rhinoceros-sized humerus, may have belonged to an animal that was hovering closer to 2–3 tonnes rather than one. Further, recent work by Hirayama et al. (2015, preprint, so be wary), has uncovered an animal with an estimated shell length of 2.7 meters (9 ft), which is closing in on the original length estimate that Falconer & Cautley proposed back in 1837.

It’s important to remember that It is very unlikely that the only preserved material we have found so far belonged to the largest individuals of the species. I would say that a maximum weight around 2–3 tonnes or so was probably not out of the question for this titanic turtle.

Most restorations of Megalochelys atlas restore the animal as a scaled up Galápagos tortoise, and sometimes a snapping turtle (!?). Rarely (if ever) have I seen a paleo artist reconstruct this animal based on the fossil material. It doesn’t help that the few descriptions available in the literature come from a time where paleontologists viewed all turtles as looking essentially the same. Further compounding the difficulty is that most described material includes bits of shell. Very little non-shell bones seem to be recorded for the species. To date, all that I have found on non-shell material for M. atlascome from the original description by Falconer and Cautley, and the more recent description of related material by Setiyabudi (2009). Even the best skeletal mount for M. atlas , found at the American Museum of Natural History, was a chimera based on a large, but not particularly well-preserved, shell collected by Brown in 1931, and enhanced with casts of the skull and limb bones originally housed in the British Museum of Natural History. Still, even with this rare and fragmentary material one can discern aspects of M. atlas that would have made it unique apart from just being a scaled up giant tortoise.
The skull

The only skull that seems to have ever been described for M. atlas, comes from the original work of Falconer and Cautley. Transcribed data from that time indicates that only a partial description of the skull was ever given. Thankfully, there is at least one good drawing of the skull (Murchison 1868). The skull was only partially preserved (and came from an animal believed to be smaller than the one that the main reconstruction was based on). As mentioned above, Falconer and Cautley estimated the length of the skull to be 0.61 m (2 ft) long. Preserved bits indicate that the head had the deep emarginations for jaw muscle attachments that are typical for tortoises. However, the supraoccipital crest was broken during fossilization, so it is hard to tell if it was long like in Galápagos tortoises, or truncated as in Aldabrachelys gigantea (Aldabra tortoises). Interestingly, the nasal passage seems to have been fairly elaborate, converging on (or shared with, depending on phylogenetic relatedness) the skull shape seen in Aldabra tortoises. Given the similarities in skull shape between M. atlas and A. gigantea, it’s possible that the original skull estimates of 0.61 meters (2 ft) may have been over-exaggerated. Extant A. gigantea are known for their almost comically small heads compared to their bodies. Perhaps M. atlas had something similar. Less known about Aldabra tortoises are their extremely elaborated nasal passages (Arnold 1979; Gerlach 2005). The nasal passages form a deeply ascending loop prior to entering the throat. Furthermore, this species has a fleshy, “flap” located near the opening into the olfactory recess. Arnold (1979) described the flap as being comprised of cavernous tissue, and directed anteriorly. He hypothesized (based on observations of tortoises apparently drinking, by I. R. Swingland) that the flap could be brought down to block the entrance into the olfactory recess, keeping water from entering this region of the nose during nasal drinking. The soft-tissue nostril in Aldabra tortoises is pointed, and the nasal elaboration found in this taxon, is associated with deep excavations of the nasal cavity. Similar excavations can be seen in both the illustrated skull and the reconstruction for M. atlas, suggesting that a similarly-shaped nose may have been present in this species too.

The shell

Most of what we know of M. atlas comes from its shell. Since turtle shells are comprised of numerous bones and these bones likely evolved for a protective function (selecting for strength) it makes sense that they would be the parts of the body that would most likely preserve. Of the numerous accounts of M. atlas in the literature, most are based on the shell.

As mentioned above, we have no complete shell material for M. atlas. The closest we have is a shell collected by Barnum Brown back in 1931, yet even that was not complete. In fact, Brown considered the material so scrappy that he was originally going to leave it behind in favour of better material, but he decided to take it with him after a year and a half of prospecting produced nothing better (Brown 1931). The material that now forms the mounted skeleton on exhibit at the AMNH (see image) consisted of over a thousand broken pieces that weighed in at 363 kg (800 lbs). It took over a year to reconstruct the specimen currently on exhibit. Compare that to Cretaceous baenid turtles, or Paleogene emydids and there is a noticeable difference. For these other turtles species, the shell tends to preserve nearly complete, or at least in large, obvious chunks. M. atlas, on the other hand, seemed to shatter during preservation.

The reason behind this is related to the relative thinness of the shell. Despite being the size of a small car, M. atlas appears to have been built remarkably light. As with extant large tortoises, M. atlas kept its shell material pretty thin. Falconer and Cautley commented on it back in 1844:

The thickness of bone in the convexity is almost in an inverse ratio to the size….There are no complete pieces in the fossil remains, and the reasons of this are the curvature and thinness.

Setiyabudi (2009) measured parts of the shell of specimen found in Indonesia that was referred to M. atlas. He found the costals to average about 1 cm (0.4 inches) in thickness, with the bridge attaching the carapace to plastron, being about 7 cm (2.8 inches) thick. For reference, the shell of a large Aldabra tortoise averages about 7.6 cm (3 inches). Interestingly, Aldabra tortoises are unique among extant giant tortoises in that they do have relatively thick shells (Gunther 1877).

The epiplastron, in many turtles, consists of the first pair of bony plates beyond the neck (see the image above). They tend to be fairly wide and blend into the hyoplastral and entoplastral bones. In giant tortoises, the epiplastra get constricted, leaving more room for the front limbs to swing forward and reach under the body. M. atlas took this a bit further, Not only are the epiplastra in M. atlas, constricted, but they had a keel running on the ventral side and the epiplastra themselves extended forward to flair laterally into two rounded spikes. These bones were thick, too. They were thicker than any other parts of the shell. The epiplastra had a base diameter of 20.3 cm (8 inches) and a thickness of 16.5 cm (6.5 inches, Murchison 1868). These were heavily built structures. In life, the epiplastra would have been covered by keratinous scutes known as gulars. This would have extended the rounded spikes on M. atlas even more forward.

In other words, Megalochelys atlas carried a tire iron on its shell.

Although both humeri (Falconer and Cautley 1844) and femora (Brown 1931, Setiyabudi 2009) are known, the only real description I could come across was for the humeri. According to Falconer and Cautley, the humeral head in M. atlas was extremely convex and nested deeply in the glenoid of the pectoral girdle. The humeral shaft exhibited a stronger curve than is seen in smaller tortoises (Murchison 1868). Taken together, Falconer and Cautley argued that this arrangement would have allowed M. atlas to bring its forelimbs under its body to a greater degree than is seen in smaller tortoises. The constriction of the epiplastra discussed above, would have further allowed the limbs to travel more parasagittally. All of these are structures that we would expect to see in a large animal looking to hold up its weight (Schmidt-Nielsen 1987). Indeed, we do see this kind of change in stance in the extant Galápagos and Aldabra tortoises. Whether the femora show similar signs is harder to tell. Setiyabudi (2009) does provide some images of an M. atlas femur, and reconstructions of the plastron indicate constrictions around the limb pockets that would have allowed the legs to travel further under the body.

Taken together it would seem that Megalochelys atlas had evolved limbs that allowed it to better swing its weight around.
Neck and tail

This remains an unknown aspect of Megalochelys atlas. Falconer and Cautley never recovered bits from this region, so important things such as neck length and mobility remain unknown for M. atlas, although it is suspected that the neck was probably fairly long. Exactly how long is unknown. Gunther (1877) described notable differences between neck length in Aldabra tortoises vs. Galápagos tortoises. The former were found to have shorter, thicker and less mobile necks than their Galápagos counterparts. If the head of M. atlas looked most similar to that of extant Aldabra tortoises, then perhaps it too had a relatively short neck. If M. atlas had a neck more like extant, saddle-backed Galápagos tortoises, or the Mascarene tortoises (Cylindraspis sp.) then it would have extended substantially past the shell. Regardless, the neck of M. atlas was probably fairly long as most turtles have necks that are at least as long as their carapaces.
Potential Paleobiology

This is always a difficult question to answer with extinct species. There are so many aspects of an animal that just don’t fossilize, forcing us to rely heavily on extant animals to give us some idea of potential behaviours.
Paleoenvironment

M. atlas is best known from the Siwaliks. The Siwaliks are the foothills, or outer mountain range of the Himalayas. The name may also be written as Sivalik or Shivalik, with the latter more accurately approaching the actual pronunciation (at least for English speakers). The name literally translates as: “Shiva’s tresses” (hair), in reference to the Hindu god of destruction. This is the youngest part of the Himalayas, at around 5–15 million years old (Gautam & Fujiwara 2000), and would have been well in place by the time M. atlas came on the scene. Paleofloral analysis indicates that the Pleistocene Siwaliks were a warm, humid time, with a strong monsoon season and lots of leafy, flowering vegetation for this giant tortoise to consume (Khan et al. 2014). Although the Siwalik hills are most commonly associated with the discovery of M. atlas, fossils of this gigantic chelonian indicate that it ranged across much of Southeast Asia, extending as far (to date) as Timor, although that may be a distinct species (Setiyabudi 2009) in the same genus (more of that below). Wikipedia mentions that M. atlas material has been found in Pakistan, but I haven’t been able to find any literature to corroborate this and I suspect it was based on the fact that M. atlas material was originally found in the state of Punjab, India, which happens to be next to Punjab, Pakistan (a similar situation exists with Kansas City, Kansas, and Kansas City, Missouri, in the United States).
Necks for sex…and feeding

Reconstructions of the shell in Megalochelys atlasindicate that the rostral carapace opening was modestly flared dorsally (Brown 1931, Badam 1981). So M. atlas probably wasn’t routinely extending its neck to reach the tops of small trees and/or compete with conspecifics, but it probably also wasn’t a low grazer all the time either. It was likely that the animal grazed across a variety of heights based on available vegetation, and that neck raising wasn’t a big part in intraspecific battles like in extant, saddle-backed Galápagos tortoises. If we assume a conservative neck length as long as the carapace, that gives a neck length of 1.8–2.1 meters (6–7 ft), which produces a fairly impressive feeding envelope. This would have made M. atlas a pretty darned efficient forager in its environment.

Speaking of competition, the skeleton of M. atlas does indicate a potential behaviour that males may have engaged in.

Turtles may be the only group of reptiles that show osteological evidence for sexual dimorphism. Male turtles have plastra that are concave. This lets the males more easily mount females during mating. Because of this morphology, Brown (1931) was able to deduce that his recovered shell belonged to a large male. This means that those enlarged epiplastron bones belonged to a male tortoise. Although M. atlas probably didn’t do neck raising competitions to handle intraspecific spats, as Galápagos tortoises do, the presence of this potential bony battering ram suggests that males may have employed a different method of handling arguments over mates and territory.

Many tortoise species engage in shoving contests where one animal pushes and attempts to flip the other animal. In some tortoise species (notably, ploughshare tortoises [Astrochelys yniphora]) the gular scales and underlying epiplastra bones, are used like a tire iron, giving the animals the leverage needed to shove and flip their rivals over. Although the epiplastron bones in M. atlas were not as excessive as those in ploughshare tortoises, they do appear to be fairly similar to the epiplastra of extant sulcata tortoises (Centrochelys sulcata). These large tortoises also engage in male on male shoving matches using their gular scales and epiplastra (for example). So it would appear that Megalochelys atlas males may have solved disputes over mates by using their enlarged gulars to shove the other opponents away, or even flipping them over (although given the immense weight of adult animals, that would probably have been pretty rare). Of course, this is technically all conjecture. We don’t have enough specimens to see evidence of shell damage caused by gular spikes, but we can at least say that M. atlas had the necessary tools to make this behaviour possible. One can only imagine what such a sight must have looked like. I’d imagine the sounds of two large males engaged in combat was probably similar to listening to an avalanche. Given the tenacity of turtles when they get their minds set on something, I suspect that getting in the way of a large Megalochelys atlas was like stepping in front of a bulldozer.

Now if we could just get some intrepid paleo-artists to draw these types of interactions…

Extant giant tortoises are remarkably slow (though not nearly as slow as media portrayals would have us think). Researchers studying Galápagos tortoise locomotion found that, unlike other animals, giant tortoises appear not to use pendular motion to recover locomotor energy when walking (Zani et al. 2005). The authors suspected that the extremely slow walking speed of the animals (0.16 m/s) was responsible for this lack of pendular motion, and that for the tortoises to take advantage of pendulum mechanics, they would need to walk about 2–3 times faster than their current walking speed, or approximately 0.48 m/s. Nonetheless, giant tortoises still revealed very efficient locomotion, suggesting that energy recovery is just not as important for tortoises (or reptiles in general) as it is for mammals. All that said, I would caution against assuming that the large size of the tortoises is responsible for their slow speed. If we look at the locomotion of baby Galápagos tortoises, and compare them to similar sized juvenile and adult tortoises of smaller (but still large) species (e.g., Newman the sulcata tortoise), it’s apparent that Galápagos tortoises walk slower at all stages of life. I suspect that island life is to blame for the slower pace (more below).

There does seem to be an exception to this island rule. The giant tortoises of the Mascarenes (Cylindraspis sp.) seem to have been built for speed. They had long legs, very open shells (cylinder-shaped, as their generic name implies), that were extremely thin (averaging 1mm thick). An early taxidermied specimen features the animal in a galloping pose, leading to the common name: “racing tortoise” (Gerlach 1998). Although this was likely an exaggerated posture chosen by the taxidermist, the length of the legs and lightness of the shell does suggest that they probably were pretty speedy animals. Exactly why this genus was seemingly more active than Galápagos and Aldabra species remains unknown, but then most of the ecology and biology of the Mascarenes tortoises remains unknown. One possible reason could be that the genus had speciated long before Mauritus and its neighbouring islands first emerged, thus making them potentially former continental species (Gerlach 2004).
Growth rate

The growth rate of M. atlas is currently unknown, and in some ways it is unknowable. Although numerous advances in paleohistology have allowed us to make estimates of growth rate, the actual rate (as measured in grams / day) is not something that we can currently assess. Still, even a look at the histology of the long bones would be a step forward.

One thing we can say for certain is that the original estimates of age for M. atlas were way off the mark. Barnum Brown (1931) made an estimate of the age for his large M. atlas specimen. Brown estimated that his specimen was “three or four hundred years old when he died.” Such an estimate played into the current held view of reptiles at the time (i.e., that they are slow and long-lived animals), as well as documented cases of extreme longevity in island giant tortoises (though even these are mostly apocryphal). The thinking being, I suppose, that if a 500 kg tortoise can reach 200 years of age, a tortoise of twice that size should live twice as long. Brown’s thinking was further coloured by a commonly held myth that still persists to this day. Namely that large reptiles get large by consistently growing and living a long time. Current knowledge of reptilian growth rates has shown that, just like mammals and birds, growth is fastest when young, and growth slows considerably once sexual maturity has been reached. Thus, large reptiles get that way not be living longer, but by growing faster in their “formative” years. To put it another way, a 20–30 year old Galápagos tortoise, will be just as big as a 150 year old individual (Daggett 1915; Garman 1917).

Another reason to question Brown’s estimate of the age for M. atlas is that this tortoise was a continental animal. Continental animals are faced with tougher hardships than island animals (mostly in regards to predators), which tends to lessen life expectancy considerably. A large M. atlas may have lived to a ripe old age of 150–200 years, but I suspect that even that would have been a rarity. Similarly, continental animals live faster lives than island animals. The limited resources on islands (both in quantity and quality of food) produces a strong selection pressure to slow down life histories. This has been documented in extant New Zealand Kiwis (maturity at 5–6 years, Bourdon et al. 2009), kakapos (reproduction every 2–7 years, sexual maturity 5–11 years, Eason et al. 2006), and Belearic island cave goats. Data on growth rate and age at sexual maturity for giant tortoises are harder to come by despite decades of conservation management on these species. Gaymer (1968) and Grubb (1971a) used carapace scute widths to estimate growth rate in Aldabra tortoises. They estimated an age of around 10–20 years for adult size to be reached, with captive-raised individuals showing a much higher spike in growth rate (Gaymer 1968). Studies on growth in juvenile Galápagos tortoise species (Chelonoidis nigra hoodensis) found that animals fed a diet of native plants, three times a week, grew a tenth as fast as captive-raised animals (Furrer et al. 2004). This reveals some of the environmental pressures that are in place to keep animals from maturing quickly on the islands (though why the animals were only fed three times a week instead of ad libitum, like wild animals, is beyond me). In contrast, smaller continental sulcata tortoises are the third largest tortoises in the world, reaching body masses around 90 kg (198 lbs, Grubb 1971b) in 6–9 years (Ritz et al. 2010). As a continental animal with higher selective pressure to grow out of hunting size quickly, Megalochelys atlas probably grew remarkably fast.
Cleaning stance

That said, it’s always important to keep in mind the limitations of our inferences when assigning behaviours to extinct animals. In this case, the cleaning response is only seen in island tortoises. Large sulcatas and leopard tortoises don’t exhibit this behaviour. Predation risks may make this stance a little too vulnerable for continental animals. Alternatively, this mutualistic relationship may not develop until turtles reach a certain threshold size that modern-day, continental tortoises just don’t reach.

So it’s a cool behaviour to hypothesize, but one that should be done cautiously.
Multitudinous Megalochelys

~ Jura
References
Arnold, E.N. 1979. Indian Ocean Giant Tortoises: Their Systematics and Island Adaptations. Phil. Trans. R. Soc. Lond. B. Biol. Sci. 286(1011):127–145.
Auffenberg, W. 1974. Checklist of Fossil Land Tortoises (Testudinidae). Bull. FL. State Mus. Biol. Sci. 18:121–251.
Badam, G.L. 1981. Colossochelys atlas, A giant Tortoise from the Upper Siwaliks of North India. Bull. Deccan. College. Res. Inst. 40:149–153.
Bourdon, E., Castanet, J., de Ricqules, A., Scofield, P., Tennyson, A., Lamrous, H., Cubo, J. 2009. Bone Growth Marks Reveal Protracted Growth in New Zealand Kiwi (Aves, Apterygidae). Biol. Lett. 5:639–642.
Brown, B. 1931. The Largest Known Land Tortoise. Nat. Hist. 31(2):184–187.
Chiari, Y., Claude, J. 2011. Study of the Carapace Shape and Growth in Two Galapagos Tortoise Lineages. J. Morph. 272:379–386.
Christian, A. 2002. Neck Posture and Overall Body Design in Sauropods. Mitt. Mus. Nat.kd. Berl. Geowiss. Reiche. 5:271–281.
Clauss, M., Polster, C., Kienzle, E., Wiesner, H., Baumgartner, K,. von Houwald, F., Ortmann, S., Streich, W.J., Dierenfeld, E.S. 2005. Studies on Digestive Physiology and Feed Digestiblities in Captive Indian Rhinoceros (Rhinoceros unicornis). J. Animal. Physiol. Animal. Nut. 89:229–237.
Daggett, F.S. 1915. A Galapagos Tortoise. Science. 42(1096):933–934.
Davis, A.C. 2006. Target Training and Voluntary Blood Drawing of the Aldabra Tortoise (Geochelone gigantea). AAZK Conference Proceedings:156–164.
Eason, D.K., Elliot, G.P., Merton, D.V., Jansen, P.W., Harper, G.A., Moorhouse, R.J. 2006. Breeding Biology of Kakapo (Strigops habroptilus) on Offshore Island Sanctuaries, 1990–2002. Notornis. 53(1):27–36.
Eckert, K.L., Luginbuhl, C. 1988. Death of a Giant. Mar.Turtle. Newsl. 43:2-3.
Falconer, H., Cautley, P.T. 1837. On Additional Fossil Species of the Order Quadrumana from the Siwalik Hills. J. Asiatic Soc. Bengal. Vol. 6:354–360.
Falconer, H., Cautley, P.T. 1844. Communication on the Colossochelys atlas, A Fossil Tortoise of Enormous Size from the Tertiary Strata of the Siwalk Hills in the North of India. Proc. Zool. Soc. Lond. 12:54–84.
Furrer, S.C., Hatt, J.M., Snell, H., Marquez, C., Honegger, R.E., Rubel, A. 2004. Comparative Study on the Growth of Juvenile Galapagos Giant Tortoises (Geochelone nigra) at the Charles Darwin Research Station (Galapagos Islands, Ecuador) and Zoo Zurich (Zurich, Switzerland). Zoo. Biol. 23:177–183.
Garman, S. 1917. The Galapagos Totoises. Mem. Mus. Comp. Zool. 30(4):261–290 with 42 plates.
Gautam, P., Fujiwara, Y. 2000. Magnetic Polarity Stratigraphy of Siwalik Group Sediments of Karnali River Section in Western Nepal. Geophys. J. Int. 142:812–824.
Gaymer, R. 1968. The Indian Ocean Giant Tortoise Testudo gigantea on Aldabra. J. Zool. Lond. 154:341–363.
Gerlach, J. 1998. “The Racing Tortoise” in Gerlach, J. (ed). Famous Tortoises. Privately Published. pps: unlabeled,
but available here
Gerlach, J. 2004. Giant Tortoises of the Indian Ocean, The Genus ‘Dipsochelys’ inhabiting the Seychelles Islands and the extinct giants of Madagascar and the Mascarenes. Chimaira/Serpents Tale NHBD.
Gerlach, J. 2005. Interpreting Morphological and Molecular Data on Indian Ocean Giant Tortoises. in Hubert, B.A., Sinclair, B.J., Lampe, K-H. (eds). African Biodiversity: Molecules, Organisms, Ecosystems. Springer, New York. pp: 213–219.
Grubb, P. 1971a. The Growth, Ecology and Population Structure of Giant Tortoises on Aldabra. Phil. Trans. R. Soc. Lond. B. Biol. Sci. 260(836):327–372.
Grubb, P. 1971b. Comparative Notes on the Behavior of Geochelone sulcata. Herpetologica. 27(3):328–333.
Gunther, A.C.L.G. 1877. The Gigantic Land-Tortoises (Living and Extinct) in the Collection of the British Museum. London. Order of the Trustees.
Hirayama, R., Sonoda, T., Takai, M., Htike, T., Thein, Z.M.M., Takahashi, A. 2015. Megalochelys: Gigantic Tortoise from the Neogene of Myanmar. PeerJ
(preprint)
Hooijer, D.A. 1971. A Giant Land Tortoise, Geochelone atlas (Falconer
and Cautley), from the Pleistocene of Timor. Proc. Koninklijke Nederlandsche Akademie van Wetenschappen, Ser. B. Phys. Sci. 74(5):504–525.
Hughes, S., Bary, J., Russell, J., Bell, R., Gurung, S. 2016. Neck Length and Mean Arterial Pressure in the Sauropod Dinosaurs. J. Exp. Biol. 219:1154–1161.
Khan, M.A., Spicer, R.A., Bera, S., Ghosh, R., Yang, J., Spicer, T.E.V., Guo, S-X., Su, T., Jacques, F., Grote, P.J. 2014. Miocene to Pleistocene Floras and Climate of the Eastern Himalayan Siwaliks, and New Palaeoelevation Estimates for the Namling-Oiyung Basin, Tibet. Global. Planet. Change. 113:1–10.
Lydekker, R. 1880. A Sketch of the Land Tortoises of the Siwaliks. Records of the Geological Survey of India. 22:209―212.
Lydekker, R. 1889. Catalogue of the Fossil Reptilia and Amphibia in the British Museum. Part III. Chelonia. London: Brit. Mus. Nat. Hist. 3:74
Macdonald, D. 2001. The New Encyclopedia of Mammals. Oxford. Oxford U. Press.
Murchison, C.D. 1868. Palaeontological Memoirs and Notes of the late Hugh Falconer: With a Biographical Sketch of the Author Compiled and Edited by Charles Murchison. Rob. Hardwicke, 1868.
Rhodin, A.G.J., Thomson, S., Georgalis, G.L., Hans-Volker, K., Danilov, I.G., Takahashi, A., de la Fuente, M.S., Bourque, J.R., Delfino, M., Bour, R., Iverson, J.B., Shaffer, H.B., van Dijk, P.P. 2015. Turtles and Tortoises of the World during the Rise and Global Spread of Humanity: First Checklist and Review of Extinct Pleistocene and Holocene Chelonians: A Compilation Project of the IUCN/SSC Tortoise and Freshwater Turtle Specialist Group. Chelonian Res. Monograph. 5:1-66.
Ritz, J., Griebeler, E.M., Huber, R., Clauss, M. 2010. Bdy Size Development of Captive and Free-Ranging African Spurred Tortoises (Geochelone sulcata): High Plasticity in Reptilian Growth Rates. Herp. J. 20:213–216.
Schafer, S.F., Krekorian, C.O’N. 1983. Agonistic Behavior of the Galapagos Tortoise, Geochelone elephantopus, with Emphasis on Its Relationship to Saddle-Backed Shell Shape. Herpetologica. 39(4):448–456.
Scheyer, T.M., Sánchez-Villagra M.R. 2007. Carapace Bone Histology in the Giant Pleurodiran Turtle Stupendemys geographicus: Phylogeny and Function. Acta. Palaeontol. Pol. 52(1):137–154.
Schmidt-Nielsen, K. 1987. Animal Physiology. 5th Edition. Cambridge University Press.
Setiyabudi, E. 2009. An Early Pleistocene Giant Tortoise (Reptilia; Testudines; Testudinidae) from the Bumiayu Area, Central Java, Indonesia. J. Fossil. Res. 42(1):1–11.
Seymour, R.S. 2009. Raising the Sauropod Neck: It Costs More to Get Less. Biol. Lett. 5(3):317–319.
Seymour, R.S., Lilywhite, H.B. 2000. Hearts, Neck Posture and Metabolic Intensity of Sauropod Dinosaurs. Proc. R. Soc. Lond. B. 267:1883–1887.
Stevens, K., Parrish, J.M. 1999. Neck Posture and Feeding Habits of Two Jurassic Sauropod Dinosaurs. Science. 284:798–800.
Stevens, K.A. and Parrish, M.J. 2005a. Digital reconstructions of sauropod dinosaurs and implications for feeding. In: K.A. Curry Rogers and J.A.
Wilson (eds.), The Sauropods: Evolution and Paleobiology, 178–200. California University Press, Berkeley.
Stevens, K.A. and Parrish, M.J. 2005b. Neck posture, dentition, and feeding strategies in Jurassic sauropod dinosaurs. In: V. Tidwell and K. Carpenter (eds.), Thunder−lizards. The Sauropodomorph Dinosaurs, 212–232. Indiana University Press, Bloomington and Indianapolis.
Taylor, M.P. 2014. Quantifying the Effect of Intervertebral Cartilage on Neck Posture in the Necks of Sauropod Dinosaurs. PeerJ 2:e712.
Taylor, M.P., Wedel, M.J., Naish, D. 2009. Head and Neck Posture in Sauropod Dinosaurs Inferred from Extant Animals. Acta. Palaeontol. Pol. 54(2):213–220.
Zangerl, R. 1969. The Turtle Shell. in Gans, C., Bellairs, A. d’A., Parsons, T. (eds) Biology of the Reptilia. Volume 1.London and New York. Academic Press. pp:311–339.
Zani, P.A., Gottschall, J.S., Kram, R. 2005. Giant Galapagos Tortoises Walk Without Inverted Pendulum Mechanical-Energy Exchange. J. Exp. Biol. 208:1489–1494.