Edmontosaurus

Edmontosaurus has had a long and complicated history in paleontology, having spent decades with various species classified in other genera. Its taxonomic history intertwines at various points with the genera Agathaumas, Anatosaurus, Anatotitan, Claosaurus, Hadrosaurus, Thespesius, and Trachodon,[4][5] and references predating the 1980s typically use Anatosaurus, Claosaurus, Thespesius, or Trachodon for edmontosaur fossils (excluding those assigned to E. regalis), depending on author and date. Although Edmontosaurus was only named in 1917, its oldest well-supported species (E. annectens) was named in 1892 as a species of Claosaurus.

Skeletal restoration of the E. annectens (then Claosaurus) holotype, by Othniel Charles Marsh

The first well-supported species of Edmontosaurus was named in 1892 as Claosaurus annectens by Othniel Charles Marsh. This species is based on USNM 2414, a partial skull-roof and skeleton, with a second skull and skeleton, YPM 2182, designated the paratype. Both were collected in 1891 by John Bell Hatcher from the late Maastrichtian-age Upper Cretaceous Lance Formation of Niobrara County (then part of Converse County), Wyoming.[6] This species has some historical footnotes attached: it is among the first dinosaurs to receive a skeletal restoration, and is the first hadrosaurid so restored;[5][7] and YPM 2182 and UNSM 2414 are, respectively, the first and second essentially complete mounted dinosaur skeletons in the United States.[8] YPM 2182 was put on display in 1901,[5] and USNM 2414 in 1904.[8]

E. annectens paratype YPM 2182 at the Yale University Museum, the first nearly complete dinosaur skeleton mounted in the United States.[8]

Because of the incomplete understanding of hadrosaurids at the time, following Marsh's death in 1897 Claosaurus annectens was variously classified as a species of Claosaurus, Thespesius or Trachodon. Opinions varied greatly; textbooks and encyclopedias drew a distinction between the "Iguanodon-like" Claosaurus annectens and the "duck-billed" Hadrosaurus (based on remains now known as adult Edmontosaurus annectens), while Hatcher explicitly identified C. annectens as synonymous with the hadrosaurid represented by those same duck-billed skulls.[5] Hatcher's revision, published in 1902, was sweeping: he considered almost all hadrosaurid genera then known as synonyms of Trachodon. This included Cionodon, Diclonius, Hadrosaurus, Ornithotarsus, Pteropelyx, and Thespesius, as well as Claorhynchus and Polyonax, fragmentary genera now thought to be horned dinosaurs.[9] Hatcher's work led to a brief consensus, until after 1910 new material from Canada and Montana showed a greater diversity of hadrosaurids than previously suspected.[5] Charles W. Gilmore in 1915 reassessed hadrosaurids and recommended that Thespesius be reintroduced for hadrosaurids from the Lance Formation and rock units of equivalent age, and that Trachodon, based on inadequate material, should be restricted to a hadrosaurid from the older Judith River Formation and its equivalents. In regards to Claosaurus annectens, he recommended that it be considered the same as Thespesius occidentalis.[10] His reinstatement of Thespesius for Lance-age hadrosaurids would have other consequences for the taxonomy of Edmontosaurus in the following decades.

During this time frame (1902–1915), two additional important specimens of C. annectens were recovered. The first, the "mummy" specimen AMNH 5060, was discovered in 1908 by Charles Hazelius Sternberg and his sons in Lance Formation rocks near Lusk, Wyoming. Sternberg was working for the British Museum of Natural History, but Henry Fairfield Osborn of the American Museum of Natural History was able to purchase the specimen for $2,000.[11] The Sternbergs recovered a second similar specimen from the same area in 1910,[12] not as well preserved but also found with skin impressions. They sold this specimen (SM 4036) to the Senckenberg Museum in Germany.[11]

As a side note, Trachodon selwyni, described by Lawrence Lambe in 1902 for a lower jaw from what is now known as the Dinosaur Park Formation of Alberta,[13] was erroneously described by Glut (1997) as having been assigned to Edmontosaurus regalis by Lull and Wright.[14] It was not, instead being designated "of very doubtful validity."[15] More recent reviews of hadrosaurids have concurred.[4][16]

Edmontosaurus itself was coined in 1917 by Lawrence Lambe for two partial skeletons found in the Horseshoe Canyon Formation (formerly the lower Edmonton Formation) along the Red Deer River of southern Alberta, Canada.[17] These rocks are older than the rocks in which Claosaurus annectens was found.[18] The Edmonton Formation lends Edmontosaurus its name.[5] The type species, E. regalis ("regal", or, more loosely, "king-sized"),[5] is based on NMC 2288, consisting of a skull, articulated vertebrae up to the sixth tail vertebra, ribs, partial hips, an upper arm bone, and most of a hind limb. It was discovered in 1912 by Levi Sternberg. The second specimen, paratype NMC 2289, consists of a skull and skeleton lacking the beak, most of the tail, and part of the feet. It was discovered in 1916 by George F. Sternberg. Lambe found that his new dinosaur compared best to Diclonius mirabilis (specimens now assigned to Edmontosaurus annectens), and drew attention to the size and robustness of Edmontosaurus.[17] Initially, Lambe only described the skulls of the two skeletons, but returned to the genus in 1920 to describe the skeleton of NMC 2289.[19] The postcrania of the type specimen remains undescribed, still in its plaster jackets.[14]

Specimen CMNFV 8399, holotype of E. edmontoni, now thought to be a young E. regalis.

Two more species that would come to be included with Edmontosaurus were named from Canadian remains in the 1920s, but both would initially be assigned to Thespesius. Gilmore named the first, Thespesius edmontoni, in 1924. T. edmontoni also came from the Horseshoe Canyon Formation. It was based on NMC 8399, another nearly complete skeleton lacking most of the tail. NMC 8399 was discovered on the Red Deer River in 1912 by a Sternberg party.[20] Its forelimbs, ossified tendons, and skin impressions were briefly described in 1913 and 1914 by Lambe, who at first thought it was an example of a species he had named Trachodon marginatus,[21] but then changed his mind.[22] The specimen became the first dinosaur skeleton to be mounted for exhibition in a Canadian museum. Gilmore found that his new species compared closely to what he called Thespesius annectens, but left the two apart because of details of the arms and hands. He also noted that his species had more vertebrae than Marsh's in the back and neck, but proposed that Marsh was mistaken in assuming that the annectens specimens were complete in those regions.[20]

In 1926, Charles Mortram Sternberg named Thespesius saskatchewanensis for NMC 8509, a skull and partial skeleton from the Wood Mountain plateau of southern Saskatchewan. He had collected this specimen in 1921, from rocks that were assigned to the Lance Formation,[23] now the Frenchman Formation.[4] NMC 8509 included an almost complete skull, numerous vertebrae, partial shoulder and hip girdles, and partial hind limbs, representing the first substantial dinosaur specimen recovered from Saskatchewan. Sternberg opted to assign it to Thespesius because that was the only hadrosaurid genus known from the Lance Formation at the time.[23] At the time, T. saskatchewanensis was unusual because of its small size, estimated at 7 to 7.3 metres (23 to 24 ft) in length.[24]

Outdated E. annectens skeleton mounts, formerly referred to Anatosaurus

In 1942, Lull and Wright attempted to resolve the complicated taxonomy of crestless hadrosaurids by naming a new genus, Anatosaurus, to take in several species that did not fit well under their previous genera. Anatosaurus, meaning "duck lizard", because of its wide, duck-like beak (Latin anas = duck + Greek sauros = lizard), had as its type species Marsh's old Claosaurus annectens. Also assigned to this genus were Thespesius edmontoni, T. saskatchewanensis, a large lower jaw that Marsh had named Trachodon longiceps in 1890, and a new species, Anatosaurus copei, for two skeletons on display at the American Museum of Natural History that had long been known as Diclonius mirabilis (or variations thereof). Thus, the various species became Anatosaurus annectens, A. copei, A. edmontoni, A. longiceps, and A. saskatchewanensis.[24] Anatosaurus would come to be called the "classic duck-billed dinosaur."[25]

Skin impression of the specimen nicknamed "Dakota", which was found in 1999

This state of affairs persisted for several decades, until Michael K. Brett-Surman reexamined the pertinent material for his graduate studies in the 1970s and 1980s. He concluded that the type species of Anatosaurus, A. annectens, was actually a species of Edmontosaurus and that A. copei was different enough to warrant its own genus.[26][27][28] Although theses and dissertations are not regarded as official publications by the International Commission on Zoological Nomenclature, which regulates the naming of organisms, his conclusions were known to other paleontologists, and were adopted by several popular works of the time.[29][30] Brett-Surman and Ralph Chapman designated a new genus for A. copei (Anatotitan) in 1990.[31] Of the remaining species, A. saskatchewanensis and A. edmontoni were assigned to Edmontosaurus as well,[16] and A. longiceps went to Anatotitan, as either a second species[32] or as a synonym of A. copei.[16] Because the type species of Anatosaurus (A. annectens) was sunk into Edmontosaurus, the name Anatosaurus is abandoned as a junior synonym of Edmontosaurus.

The conception of Edmontosaurus that emerged included three valid species: the type E. regalis, E. annectens (including Anatosaurus edmontoni, amended to edmontonensis), and E. saskatchewanensis.[16] The debate about the proper taxonomy of the A. copei specimens continues to the present: returning to Hatcher's argument of 1902, Jack Horner, David B. Weishampel, and Catherine Forster regarded Anatotitan copei as representing specimens of Edmontosaurus annectens with crushed skulls.[4] In 2007 another "mummy" was announced; nicknamed "Dakota", it was discovered in 1999 by Tyler Lyson, and came from the Hell Creek Formation of North Dakota.[33][34]

In a 2011 study by Nicolás Campione and David Evans, the authors conducted the first ever morphometric analysis to compare the various specimens assigned to Edmontosaurus. They concluded that only two species are valid: E. regalis, from the late Campanian, and E. annectens, from the late Maastrichtian. Their study provided further evidence that Anatotitan copei is a synonym of E. annectens; specifically, that the long, low skull of A. copei is the result of ontogenetic change and represents mature E. annectens individuals.[18]

Skull of E. annectens, showing duck-bill and dentition – Oxford University Museum of Natural History

The skull of a fully grown Edmontosaurus could be over a metre long. One skull of E. annectens (formerly Anatotitan) measures 3.87 feet (1.18 m) long.[48] The skull was roughly triangular in profile,[19] with no bony cranial crest.[24] Viewed from above, the front and rear of the skull were expanded, with the broad front forming a duck-bill or spoon-bill shape. The beak was toothless, and both the upper and lower beaks were extended by keratinous material.[4] Substantial remains of the keratinous upper beak are known from the "mummy" kept at the Senckenberg Museum.[14] In this specimen, the preserved nonbony part of the beak extended for at least 8 centimetres (3.1 in) beyond the bone, projecting down vertically.[49] The nasal openings of Edmontosaurus were elongate and housed in deep depressions surrounded by distinct bony rims above, behind, and below.[35] In at least one case (the Senckenberg specimen), rarely preserved sclerotic rings were preserved in the eye sockets.[50] Another rarely seen bone, the stapes (the reptilian ear bone), has also been seen in a specimen of Edmontosaurus.[4]

Teeth were present only in the maxillae (upper cheeks) and dentaries (main bone of the lower jaw). The teeth were continually replaced, taking about half a year to form.[51] They were composed of six types of tissues, rivaling the complexity of mammal teeth.[52] They grew in columns, with an observed maximum of six in each, and the number of columns varied based on the animal's size.[44] Known column counts for the two species are: 51 to 53 columns per maxilla and 48 to 49 per dentary (teeth of the upper jaw being slightly narrower than those in the lower jaw) for E. regalis; and 52 columns per maxilla and 44 per dentary for E. annectens (an E. saskatchewanensis specimen).[24]

Hip bones of E. annectens

The number of vertebrae differs between specimens. E. regalis had thirteen neck vertebrae, eighteen back vertebrae, nine hip vertebrae, and an unknown number of tail vertebrae.[24] A specimen once identified as belonging to Anatosaurus edmontoni (now considered to be the same as E. regalis) is reported as having an additional back vertebra and 85 tail vertebrae, with an undisclosed amount of restoration.[24] Other hadrosaurids are only reported as having 50 to 70 tail vertebrae,[4] so this appears to have been an overestimate. The anterior back was curved toward the ground, with the neck flexed upward and the rest of the back and tail held horizontally.[4] Most of the back and tail were lined by ossified tendons arranged in a latticework along the neural spines of the vertebrae. This condition has been described as making the back and at least part of the tail "ramrod" straight.[53][54] The ossified tendons are interpreted as having strengthened the vertebral column against gravitational stress, incurred through being a large animal with a horizontal vertebral column otherwise supported mostly by the hind legs and hips.[53]

Reconstruction of E. regalis

The shoulder blades were long flat blade-like bones, held roughly parallel to the vertebral column. The hips were composed of three elements each: an elongate ilium above the articulation with the leg, an ischium below and behind with a long thin rod, and a pubis in front that flared into a plate-like structure. The structure of the hip hindered the animal from standing with its back erect, because in such a position the thigh bone would have pushed against the joint of the ilium and pubis, instead of pushing only against the solid ilium. The nine fused hip vertebrae provided support for the hip.[44]

The fore legs were shorter and less heavily built than the hind legs. The upper arm had a large deltopectoral crest for muscle attachment, while the ulna and radius were slim. The upper arm and forearm were similar in length. The wrist was simple, with only two small bones. Each hand had four fingers, with no thumb (first finger). The index (second), third, and fourth fingers were approximately the same length and were united in life within a fleshy covering. Although the second and third finger had hoof-like unguals, these bones were also within the skin and not apparent from the outside. The little finger diverged from the other three and was much shorter. The thigh bone was robust and straight, with a prominent flange about halfway down the posterior side.[44] This ridge was for the attachment of powerful muscles attached to the hips and tail that pulled the thighs (and thus the hind legs) backward and helped maintain the use of the tail as a balancing organ.[55] Each foot had three toes, with no big toe or little toe. The toes had hoof-like tips.[44]

AMNH 5060: an E. annectens with skin impressions

Multiple specimens of Edmontosaurus annectens have been found with preserved skin impressions. Several have been well-publicized, such as the "Trachodon mummy" of the early 20th century,[56][57] and the specimen nicknamed "Dakota",[33][34][58] the latter apparently including remnant organic compounds from the skin.[58] Because of these finds, the scalation of Edmontosaurus annectens is known for most areas of the body. Skin impressions are less well known for E. regalis, but some well-preserved examples have been studied, including one which preserves a soft tissue crest or wattle on the head. It is unknown whether such a crest was present on E. annectens, and whether it was an indicator of sexual dimorphism.[3]

A preserved rhamphotheca present in the E. annectens specimen LACM 23502, housed in the Los Angeles County Museum, indicates the beak of Edmontosaurus was more hook-shaped and extensive than many illustrations in scientific and public media have previously depicted. Whether or not the specimen in question preserved the true rhamphotheca or just a cast of the inner structure attached to the bone is not known at present.[59][60][61]

In a 2011 study, Campione and Evans recorded data from all known "edmontosaur" skulls from the Campanian and Maastrichtian and used it to plot a morphometric graph, comparing variable features of the skull with skull size. Their results showed that within both recognized Edmontosaurus species, many features previously used to classify additional species or genera were directly correlated with skull size. Campione and Evans interpreted these results as strongly suggesting that the shape of Edmontosaurus skulls changed dramatically as they grew. This has led to several apparent mistakes in classification in the past. The Campanian species Thespesius edmontoni, previously considered a synonym of E. annectens due to its small size and skull shape, is more likely a subadult specimen of the contemporary E. regalis. Similarly, the three previously recognized Maastrichtian edmontosaur species likely represent growth stages of a single species, with E. saskatchewanensis representing juveniles, E. annectens subadults, and Anatotitan copei fully mature adults. The skulls became longer and flatter as the animals grew.[18]

A 1905 chart showing the relatively small brains of a Triceratops (top) and E. annectens

The brain of Edmontosaurus has been described in several papers and abstracts through the use of endocasts of the cavity where the brain had been. E. annectens[64][65] and E. regalis,[19] as well as specimens not identified to species,[66][67][68] have been studied in this way. The brain was not particularly large for an animal the size of Edmontosaurus. The space holding it was only about a quarter of the length of the skull,[19] and various endocasts have been measured as displacing 374 millilitres (13 US fl oz)[68] to 450 millilitres (15 US fl oz),[67] which does not take into account that the brain may have occupied as little as 50% of the space of the endocast, the rest of the space being taken up by the dura mater surrounding the brain.[67][68] For example, the brain of the specimen with the 374 millilitre endocast is estimated to have had a volume of 268 millilitres (9 US fl oz).[68] The brain was an elongate structure,[67] and as with other non-mammals, there would have been no neocortex.[68] Like Stegosaurus, the neural canal was expanded in the hips, but not to the same degree: the endosacral space of Stegosaurus had 20 times the volume of its endocranial cast, whereas the endosacral space of Edmontosaurus was only 2.59 times larger in volume.[67]

In 2003, evidence of tumors, including hemangiomas, desmoplastic fibroma, metastatic cancer, and osteoblastoma, was described in Edmontosaurus bones. Rothschild et al. tested dinosaur vertebrae for tumors using computerized tomography and fluoroscope screening. Several other hadrosaurids, including Brachylophosaurus, Gilmoreosaurus, and Bactrosaurus, also tested positive. Although more than 10,000 fossils were examined in this manner, the tumors were limited to Edmontosaurus and closely related genera. The tumors may have been caused by environmental factors or genetic propensity.[69]

Osteochondrosis, or surficial pits in bone at places where bones articulate, is also known in Edmontosaurus. This condition, resulting from cartilage failing to be replaced by bone during growth, was found to be present in 2.2% of 224 edmontosaur toe bones. The underlying cause of the condition is unknown. Genetic predisposition, trauma, feeding intensity, alterations in blood supply, excess thyroid hormones, and deficiencies in various growth factors have been suggested. Among dinosaurs, osteochondrosis (like tumors) is most commonly found in hadrosaurids.[70]

E. annectens in a quadrupedal pose

Like other hadrosaurids, Edmontosaurus is thought to have been a facultative biped, meaning that it mostly moved on four legs, but could adopt a bipedal stance when needed. It probably went on all fours when standing still or moving slowly, and switched to using the hind legs alone when moving more rapidly.[4] Research conducted by computer modeling in 2007 suggests that Edmontosaurus could run at high speeds, perhaps up to 45 kilometres per hour (28 mph).[33] Further simulations using a subadult specimen estimated as weighing 715 kilograms (1,576 lb) when alive produced a model that could run or hop bipedally, use a trot, pace, or single foot symmetric quadrupedal gait, or move at a gallop. The researchers found to their surprise that the fastest gait was kangaroo-like hopping (maximum simulated speed of 17.3 metres per second (62 km/h; 39 mph)), which they regarded as unlikely based on the size of the animal and lack of hopping footprints in the fossil record, and instead interpreted the result as indicative of an inaccuracy in their simulation. The fastest non-hopping gaits were galloping (maximum simulated speed of 15.7 metres per second (57 km/h; 35 mph)) and running bipedally (maximum simulated speed of 14.0 metres per second (50 km/h; 31 mph)). They found weak support for bipedal running as the most likely option for high-speed movement, but did not rule out high-speed quadrupedal movement.[71]

While long thought to have been aquatic or semiaquatic, hadrosaurids were not as well-suited for swimming as other dinosaurs (particularly theropods, who were once thought to have been unable to pursue hadrosaurids into water). Hadrosaurids had slim hands with short fingers, making their forelimbs ineffective for propulsion, and the tail was also not useful for propulsion because of the ossified tendons that increased its rigidity, and the poorly developed attachment points for muscles that would have moved the tail from side to side.[72][73]

Extensive bone beds are known for Edmontosaurus, and such groupings of hadrosaurids are used to suggest that they were gregarious, living in groups.[4] Three quarries containing Edmontosaurus remains are identified in a 2007 database of fossil bone beds, from Alberta (Horseshoe Canyon Formation), South Dakota (Hell Creek Formation), and Wyoming (Lance Formation). One edmontosaur bone bed, from claystone and mudstone of the Lance Formation in eastern Wyoming, covers more than a square kilometre, although Edmontosaurus bones are most concentrated in a 40 hectares (0.15 sq mi) subsection of this site. It is estimated that disassociated remains pertaining to 10,000 to 25,000 edmontosaurs are present here. [74]

Unlike many other hadrosaurids, Edmontosaurus lacked a bony crest. It may have had soft-tissue display structures in the skull, though: the bones around the nasal openings had deep indentations surrounding the openings, and this pair of recesses are postulated to have held inflatable air sacs, perhaps allowing for both visual and auditory signaling.[35] Edmontosaurus may have been dimorphic, with more robust and more lightly built forms, but it has not been established if this is related to sexual dimorphism.[75]

Edmontosaurus has been considered a possibly migratory hadrosaurid by some authors. A 2008 review of dinosaur migration studies by Phil R. Bell and Eric Snively proposed that E. regalis was capable of an annual 2,600 kilometres (1,600 mi) round-trip journey, provided it had the requisite metabolism and fat deposition rates. Such a trip would have required speeds of about 2 to 10 kilometres per hour (1 to 6 mph), and could have brought it from Alaska to Alberta.[76][77] In contrast to Bell and Snively, Anusuya Chinsamy and colleagues concluded from a study of bone microstructure that polar Edmontosaurus overwintered.[78]

E. annectens skull preserving the keratinous beak (partially removed on the right side by accident), Natural History Museum of Los Angeles

Between the mid-1980s and the first decade of the 2000s, the prevailing interpretation of how hadrosaurids processed their food followed the model put forward in 1984 by David B. Weishampel. He proposed that the structure of the skull permitted motion between bones that resulted in backward and forward motion of the lower jaw, and outward bowing of the tooth-bearing bones of the upper jaw when the mouth was closed. The teeth of the upper jaw would grind against the teeth of the lower jaw like rasps, processing plant material trapped between them.[4][79] Such a motion would parallel the effects of mastication in mammals, although accomplishing the effects in a completely different way.[80] Work in the early 2000s has challenged the Weishampel model. A study published in 2008 by Casey Holliday and Lawrence Witmer found that ornithopods like Edmontosaurus lacked the types of skull joints seen in those modern animals that are known to have kinetic skulls (skulls that permit motion between their constituent bones), such as squamates and birds. They proposed that joints that had been interpreted as permitting movement in dinosaur skulls were actually cartilaginous growth zones.[81] An important piece of evidence for Weishampel's model is the orientation of scratches on the teeth, showing the direction of jaw action. Other movements could produce similar scratches though, such as movement of the bones of the two halves of the lower jaw. Not all models have been scrutinized under present techniques.[81] Vincent Williams and colleagues (2009) published additional work on hadrosaurid tooth microwear. They found four classes of scratches on Edmontosaurus teeth. The most common class was interpreted as resulting from an oblique motion, not a simple up-down or front-back motion, which is consistent with the Weishampel model. This motion is thought to have been the primary motion for grinding food. Two scratch classes were interpreted as resulting from forward or backward movement of the jaws. The other class was variable and probably resulted from opening the jaws. The combination of movements is more complex than had been previously predicted.[82]

Close up of teeth in the lower jaw

Weishampel developed his model with the aid of a computer simulation. Natalia Rybczynski and colleagues have updated this work with a much more sophisticated three-dimensional animation model, scanning a skull of E. regalis with lasers. They were able to replicate the proposed motion with their model, although they found that additional secondary movements between other bones were required, with maximum separations of 1.3 to 1.4 centimetres (0.51 to 0.55 in) between some bones during the chewing cycle. Rybczynski and colleagues were not convinced that the Weishampel model is viable, but noted that they have several improvements to implement to their animation. Planned improvements include incorporating soft tissue and tooth wear marks and scratches, which should better constrain movements. They note that there are several other hypotheses to test as well.[80] Further research published in 2012 by Robin Cuthbertson and colleagues found the motions required for Weishampel's model to be unlikely, and favored a model in which movements of the lower jaw produced grinding action. The lower jaw's joint with the upper jaw would permit anterior–posterior motion along with the usual rotation, and the anterior joint of the two halves of the lower jaw would also permit motion; in combination, the two halves of the lower jaw could move slightly back and forth as well as rotating slightly along their long axes. These motions would account for the observed tooth wear and a more solidly constructed skull than modeled by Weishampel.[83]

Edmontosaurus was a wide-ranging genus in both time and space. The rock units from which it is known can be divided into two groups by age: the older Horseshoe Canyon and St. Mary River formations, and the younger Frenchman, Hell Creek, and Lance formations. The time span covered by the Horseshoe Canyon Formation and equivalents is also known as Edmontonian, and the time span covered by the younger units is also known as Lancian. The Edmontonian and Lancian time intervals had distinct dinosaur faunas.[84]

Horseshoe Canyon Formation near Drumheller. The dark bands are coal seams.

The Edmontonian land vertebrate age is defined by the first appearance of Edmontosaurus regalis in the fossil record.[85] Although sometimes reported as of exclusively early Maastrichtian age,[41] the Horseshoe Canyon Formation was of somewhat longer duration. Deposition began approximately 73 million years ago, in the late Campanian, and ended between 68.0 and 67.6 million years ago.[86] Edmontosaurus regalis is known from the lowest of five units within the Horseshoe Canyon Formation, but is absent from at least the second to the top.[87] As many as three quarters of the dinosaur specimens from badlands near Drumheller, Alberta may pertain to Edmontosaurus.[88]

The Lancian time interval was the last interval before the Cretaceous–Paleogene extinction event that eliminated non-avian dinosaurs. Edmontosaurus was one of the more common dinosaurs of the interval. Robert Bakker reports that it made up one-seventh of the large dinosaur sample, with most of the rest (five-sixths) made up of the horned dinosaur Triceratops.[89] The coastal plain Triceratops–Edmontosaurus association, dominated by Triceratops, extended from Colorado to Saskatchewan.[90]

Fossils of Edmontosaurus discovered in eastern Montana (Hell Creek Formation).

The Lance Formation, as typified by exposures approximately 100 kilometres (62 mi) north of Fort Laramie in eastern Wyoming, has been interpreted as a bayou setting similar to the Louisiana coastal plain. It was closer to a large delta than the Hell Creek Formation depositional setting to the north and received much more sediment. Tropical araucarian conifers and palm trees dotted the hardwood forests, differentiating the flora from the northern coastal plain.[91] The climate was humid and subtropical, with conifers, palmettos, and ferns in the swamps, and conifers, ash, live oak, and shrubs in the forests.[37] Freshwater fish, salamanders, turtles, diverse lizards, snakes, shorebirds, and small mammals lived alongside the dinosaurs. Small dinosaurs are not known in as great of abundance here as in the Hell Creek rocks, but Thescelosaurus once again seems to have been relatively common. Triceratops is known from many skulls, which tend to be somewhat smaller than those of more northern individuals. The Lance Formation is the setting of two edmontosaur "mummies".[91]

1897 restoration by Charles R. Knight of E. annectens as semi-aquatic animals that could only chew soft water plants, a popular idea at the time which is now outdated

As a hadrosaurid, Edmontosaurus was a large terrestrial herbivore. Its teeth were continually replaced and packed into dental batteries that contained hundreds of teeth, only a relative handful of which were in use at any time.[4] It used its broad beak to cut loose food, perhaps by cropping,[4] or by closing the jaws in a clamshell-like manner over twigs and branches and then stripping off the more nutritious leaves and shoots.[53] Because the tooth rows are deeply indented from the outside of the jaws, and because of other anatomical details, it is inferred that Edmontosaurus and most other ornithischians had cheek-like structures, muscular or non-muscular. The function of the cheeks was to retain food in the mouth.[92][93] The animal's feeding range would have been from ground level to around 4 metres (13 ft) above.[4]

Before the 1960s and 1970s, the prevailing interpretation of hadrosaurids like Edmontosaurus was that they were aquatic and fed on aquatic plants.[94] An example of this is William Morris's 1970 interpretation of an edmontosaur skull with nonbony beak remnants. He proposed that the animal had a diet much like that of some modern ducks, filtering plants and aquatic invertebrates like mollusks and crustaceans from the water and discharging water via V-shaped furrows along the inner face of the upper beak.[47] This interpretation of the beak has been rejected, as the furrows and ridges are more like those of herbivorous turtle beaks than the flexible structures seen in filter-feeding birds.[94]

Because scratches dominate the microwear texture of the teeth, Williams et al. suggested Edmontosaurus was a grazer instead of a browser, which would be predicted to have fewer scratches due to eating less abrasive materials. Candidates for ingested abrasives include silica-rich plants like horsetails and soil that was accidentally ingested due to feeding at ground level.[82] The tooth structure indicates combined slicing and grinding capabilities.[52]

Reports of gastroliths, or stomach stones, in the hadrosaurid Claosaurus are actually based on a probable double misidentification. First, the specimen is actually of Edmontosaurus annectens. Barnum Brown, who discovered the specimen in 1900, referred to it as Claosaurus because E. annectens was thought to be a species of Claosaurus at the time. Additionally, it is more likely that the supposed gastroliths represent gravel washed in during burial.[5]

Possible gut contents were reported from the "Trachodon mummy" at the American Museum of Natural History, but were never described.

Both of the "mummy" specimens collected by the Sternbergs were reported to have had possible gut contents. Charles H. Sternberg reported the presence of carbonized gut contents in the American Museum of Natural History specimen,[95] but this material has not been described.[96] The plant remains in the Senckenberg Museum specimen have been described, but have proven difficult to interpret. The plants found in the carcass included needles of the conifer Cunninghamites elegans, twigs from conifer and broadleaf trees, and numerous small seeds or fruits.[97] Upon their description in 1922, they were the subject of a debate in the German-language journal Paläontologische Zeitschrift. Kräusel, who described the material, interpreted it as the gut contents of the animal,[97] while Abel could not rule out that the plants had been washed into the carcass after death.[98]

At the time, hadrosaurids were thought to have been aquatic animals, and Kräusel made a point of stating that the specimen did not rule out hadrosaurids eating water plants.[53][97] The discovery of possible gut contents made little impact in English-speaking circles, except for another brief mention of the aquatic-terrestrial dichotomy,[99] until it was brought up by John Ostrom in the course of an article reassessing the old interpretation of hadrosaurids as water-bound. Instead of trying to adapt the discovery to the aquatic model, he used it as a line of evidence that hadrosaurids were terrestrial herbivores.[53] While his interpretation of hadrosaurids as terrestrial animals has been generally accepted,[4] the Senckenberg plant fossils remain equivocal. Kenneth Carpenter has suggested that they may actually represent the gut contents of a starving animal, instead of a typical diet.[100][101] Other authors have noted that because the plant fossils were removed from their original context in the specimen and were heavily prepared, it is no longer possible to follow up on the original work, leaving open the possibility that the plants were washed-in debris.[96][102]

The Senckenberg Museum specimen

The diet and physiology of Edmontosaurus have been probed by using stable isotopes of carbon and oxygen as recorded in tooth enamel. When feeding, drinking, and breathing, animals take in carbon and oxygen, which become incorporated into bone. The isotopes of these two elements are determined by various internal and external factors, such as the type of plants being eaten, the physiology of the animal, salinity, and climate. If isotope ratios in fossils are not altered by fossilization and later changes, they can be studied for information about the original factors; warmblooded animals will have certain isotopic compositions compared to their surroundings, animals that eat certain types of plants or use certain digestive processes will have distinct isotopic compositions, and so on. Enamel is typically used because the structure of the mineral that forms enamel makes it the most resistant material to chemical change in the skeleton.[51]

A 2004 study by Kathryn Thomas and Sandra Carlson used teeth from the upper jaw of three individuals interpreted as a juvenile, a subadult, and an adult, recovered from a bone bed in the Hell Creek Formation of Corson County, South Dakota. In this study, successive teeth in columns in the edmontosaurs' dental batteries were sampled from multiple locations along each tooth using a microdrilling system. This sampling method takes advantage of the organization of hadrosaurid dental batteries to find variation in tooth isotopes over a period of time. From their work, it appears that edmontosaur teeth took less than about 0.65 years to form, slightly faster in younger edmontosaurs. The teeth of all three individuals appeared to show variation in oxygen isotope ratios that could correspond to warm/dry and cool/wet periods; Thomas and Carlson considered the possibility that the animals were migrating instead, but favored local seasonal variations because migration would have more likely led to ratio homogenization, as many animals migrate to stay within specific temperature ranges or near particular food sources.[51]

The edmontosaurs also showed enriched carbon isotope values, which for modern mammals would be interpreted as a mixed diet of C3 plants (most plants) and C4 plants (grasses); however, C4 plants were extremely rare in the Late Cretaceous if present at all. Thomas and Carlson put forward several factors that may have been operating, and found the most likely to include a diet heavy in gymnosperms, consuming salt-stressed plants from coastal areas adjacent to the Western Interior Seaway, and a physiological difference between dinosaurs and mammals that caused dinosaurs to form tissue with different carbon ratios than would be expected for mammals. A combination of factors is also possible.[51]

Edmontosaurus annectens mounted as being hunted by a Tyrannosaurus, which would have preyed on the species in life

The time span and geographic range of Edmontosaurus overlapped with Tyrannosaurus, and an adult specimen of E. annectens on display in the Denver Museum of Nature and Science shows evidence of a theropod bite in the tail. Counting back from the hip, the thirteenth to seventeenth vertebrae have damaged spines consistent with an attack from the right rear of the animal. One spine has a portion sheared away, and the others are kinked; three have apparent tooth puncture marks. The top of the tail was at least 2.9 metres (9.5 ft) high, and the only theropod species known from the same rock formation that was tall enough to make such an attack is T. rex. The bones are partially healed, but the edmontosaur died before the traces of damage were completely obliterated. The damage also shows signs of bone infection. Kenneth Carpenter, who studied the specimen, noted that there also seems to be a healed fracture in the left hip which predated the attack because it was more fully healed. He suggested that the edmontosaur was a target because it may have been limping from this earlier injury. Because it survived the attack, Carpenter suggested that it may have outmaneuvered or outrun its attacker, or that the damage to its tail was incurred by the hadrosaurid using it as a weapon against the tyrannosaur.[103] However, more modern studies dispute the idea of an attack but rather other factors unrealated to an attack from a tyrannosaur.[104]

Another specimen of E. annectens, pertaining to a 7.6 metres (25 ft) long individual from South Dakota, shows evidence of tooth marks from small theropods on its lower jaws. Some of the marks are partially healed. Michael Triebold, informally reporting on the specimen, suggested a scenario where small theropods attacked the throat of the edmontosaur; the animal survived the initial attack but succumbed to its injuries shortly thereafter.[105] Some edmontosaur bone beds were sites of scavenging. Albertosaurus and Saurornitholestes tooth marks are common at one Alberta bone bed,[106] and Daspletosaurus fed on Edmontosaurus and fellow hadrosaurid Saurolophus at another Alberta site.[107]