Dinosaur Census Reveals Abundant Tyrannosaurus and Rare Ontogenetic Stages in the Upper Cretaceous Hell Creek Formation (Maastrichtian), Montana, USA Author John R. Horner Museum of the Rockies, Montana State University, Bozeman, Montana, United States of America jhorner@montana.edu Author Mark B. Goodwin Museum of Paleontology, University of California, Berkeley, California, United States of America Author Nathan Myhrvold Intellectual Ventures, Bellevue, Washington, United States of America text PLoS ONE 2011 e 16574 2011-02-09 6 2 1 9 journal article 10.1371/journal.pone.0016574 5a25e523-7350-43a2-b5cb-6cafa6259b25 1932-6203 PMC3036655 21347420 3744916 Results Geological Results Edmontosaurus , Ornithomimus and Ankylosaurus are found in siltstones or sandstones, and Thescelosaurus is found exclusively in mudstones, but the relative number of specimens is small and subsequently questionable as a real pattern of sediment preference or taphonomic artifact ( Tables S1 , S4). Other taxa are found in both channel and overbank sediments, but the majority of Triceratops , and in particular juvenile specimens, come primarily from mudstones [18]. There was no apparent sediment preference for preserving articulation. The basal sand unit (L3.lBS) produced both an articulated specimen of Edmontosaurus (‘‘X-rex’’/MOR 1142) and a disarticulated Tyrannosaurus ( ‘‘B-rex’’/MOR 1125 ). An articulated Tyrannosaurus ( ‘‘N-rex’’/Smithsonian Institution ) and a disarticulated Tyrannosaurus ( ‘‘G-rex’’/MOR 1128 ) were found in the lower mudstone unit (L3.lMS). Census Results The dinosaur census results are summarized in Table 1 by taxon with percentage of the fauna and absolute numbers given. Additional sedimentological details, more precise stratigraphic interval, preservation and ontogenetic designations are provided in Tables S1 , S2, S3, S4, S5, S6. The isolated, uncollected Triceratops skulls listed in Table S3 are not included in the census of skeletons from the lower Hell Creek Formation ( Table S1 ) at present because there is no way to know if they consist of three or more disarticulated pieces until they are collected. Thirty-nine skeletons (not counting the isolated Triceratops skulls) were recorded from the L3 strata. All but three of these skeletons were collected. The uncollected specimens were represented by at least three elements but were too severely eroded to yield data other than for this census. In addition, seven specimens (superscript 3 numbers in Table S1 ) consisted of only three elements each. Limited excavation around the elements failed to yield more material, and the sites were abandoned. Five specimens were found with some articulation, and of these, only one ( Edmontosaurus , ‘‘X-rex,’’ MOR 1142) was found with skin impressions. The most interesting census result in the L3 is the high number of Tyrannosaurus skeletons (n = 11) that is nearly double the number of Edmontosaurus skeletons (n = 6) and equals Triceratops (n = 11) (see Figure 1 and Table 1 ). However, as explained in the previous paragraph, it is likely the number of Triceratops skeletons will increase as these sites in L3 are excavated. Tyrannosaurus contributes to 28% of the dinosaur skeletons recorded in L3 while Edmontosaurus makes up only 15%. Considering the fact that Tyrannosaurus , Triceratops , and Edmontosaurus are all relatively similar in size as full-grown adults, we presume that there are few taphonomic biases that would amplify the Tyrannosaurus numbers to be greater than Edmontosaurus and this likely reflects a correct ratio of approximately 2:1. Thirty-two skeletons were collected from the U3 unit, four of which were collected prior to the Hell Creek Project by the Museum of the Rockies ( MOR 009 , Tyrannosaurus ; MOR 004, Triceratops ; MOR 555 , Tyrannosaurus ; MOR 622, Triceratops ; MOR 007, Edmontosaurus ). These were included in the census because they were found in the study area with documented stratigraphic and locality information, and they are cataloged into MOR. Triceratops skeletons (n = 22) greatly outnumbered other taxa ( Figure 2 , Table 1 ) and contribute to 69% of the total dinosaur skeletal fauna in U3. Specimens of Ornithomimus , Thescelosaurus , Ankylosaurus or Pachycephalosaurus were conspicuously absent, although isolated bones of Thescelosaurus , Ornithomimus , and Pachycephalosaurus were present in the Doldrum’s Lag deposit (MOR loc. HC-530) at the base of the Apex sandstone. Edmontosaurus and Tyrannosaurus skeletons were equal in number (n = 5) in U3 and comprise 16% each of the large dinosaur taxa. The pie charts in Figure 2 illustrate the similarity in overall percent composition between the large dinosaur fauna recorded in L3 and the two overlying lag deposits. The greatest contrast occurs within the upper Hell Creek (U3) record of dinosaur skeletons where Triceratops dominates (69%; n = 22), followed by Tyrannosaurus (16%; n = 5) and Edmontosaurus (16%; n = 5). Table 1. Hell Creek Formation dinosaur census.

Taxon
Stratigraphic level	Tric	Tyrn	Edmn	Thes	Orni	Pachy	Anky
Upper Hell Creek Fm (U3) skeletons	n = %	23 69%	5 16%	5 16%
Pie chart Iı Fig. 2 Table S4
"Doldrum’s" lag deposit at base of Apex sandstone (MOR locality HC-530)	n = %	16 41%	9 23%	7 18%	4 10%	2 5%	1 3%
Pie chart IIı Fig. 2 Table S6
"3B-1" lag deposit at base of Jen-rex sand (MOR locality HC-312)	n = %	23 33%	19 27%	18 26%	7 10%	2 3%	1 1%
Pie chart IIIı Fig. 2 Table S3
Lower Hell Creek Fm (L3) skeletons	n = %	11 28%	11 28%	6 15%	4 10%	5 13%	2 5%
Pie chart IVı Fig. 2 Table S1
Totals for the entire Hell Creek Formation (see Figure 4)	n = %	73 40%	44 24%	36 20%	15 8%	9 5%	2 1%	2 1%

Values determined from the dinosaur census tables (Tables S1ı S2ı S3ı S4ı S5ı S6). Empty cell indicates no record for that taxon. See Figure 2 for detailed stratigraphic section of the Hell Creek Formation and corresponding pie chart showing relative abundance of dinosaur genera. Abbreviations: Tricı Triceratops ; Tyrnı Tyrannosaurus ; Edmnı Edmontosaurus ; Thesı Thescelosaurus ; Orniı Ornithomimus ; Pachyı Pachycephalosaurus ; Ankyı Ankylosaurus . doi:10.1371/journal.pone.0016574.t001 Teeth were not collected or annotated because of the difficulties in using them for ontogenetic assessment with the exception of two large Tyrannosaurus teeth from the ‘‘3B-1 Lag’’ at the base of the Jen-rex sand. Tooth size varies as much as 300% in a single jaw, particularly in hadrosaurids (MOR 1609, Becky’s Giant), ceratopsids (MOR 2574, Quittin Time) and tyrannosaurids ( MOR 1125, B-rex ). This is one reason the assignment of some dinosaur teeth to ‘‘babies’’ [19] may be incorrect. These teeth are more accurately interpreted as being derived from the anterior or posterior portions of jaws from older individuals ( Figure 3 ). Only the largest and most robust tyrannosaurid teeth are reliable indicators of adults. The Triceratops specimens recorded in Table S6 represent specimens that were collected, but remain unprepared, uncataloged and consist of an unknown number of disarticulated elements. Ontogenetic Results In this census, growth stages at either end of the dinosaurian ontogenetic spectrum are least represented. Specimens of both the smallest and presumably youngest juveniles and the largest, and presumably oldest adults are the most rare dinosaurs recorded. The smallest specimen of Triceratops found during this project is a partially complete skull that is half again the length of the smallest previously known skull [20]. None of the Triceratops specimens found in the census area could be positively identified as ‘‘ Torosaurus ’’ size, although the specimen collected from the ‘‘BAB’’ locality has elongated squamosals characteristic of the ‘‘ Torosaurus ’’ morph. Two specimens of Edmontosaurus are in the ‘‘XL’’ size range: ‘‘Becky’s Giant’’ (MOR 1609) is a maxilla with a tooth-row length of 570 mm and the tail of ‘‘X-rex’’ (MOR 1142) is 7.5 meters in length from the posterior end of the sacrum. Both these specimens are indicative of greater size ranges then previously attributed to Edmontosaurus . Discussion Census The dinosaur collections made over the past decade during the Hell Creek Project yielded new information from an improved genus-level collecting schema and robust data set that revealed relative dinosaur abundances that were unexpected, and ontogenetic age classes previously considered rare. We recognize a much higher percentage of Tyrannosaurus ( Table 1 ) than previous surveys [3,4,21]. Tyrannosaurus equals Edmontosaurus in U3 and in L3 comprises a greater percentage of the large dinosaur fauna as the second most abundant taxon after Triceratops , followed by Edmontosaurus . This is surprisingly consistent in (1) the two major lag deposits (MOR loc. HC-530 and HC-312) in the Apex sandstone and Jen-rex sand ( Figure 2 ) where individual bones were counted and (2) in two-thirds of the formation reflected in L3 and U3 records of dinosaur skeletons only. Measured throughout the entire formation, Triceratops is by far the most common dinosaur at 40% (n = 72), Tyrannosaurus is second at 24% (n = 44), Edmontosaurus is third at 20% (n = 36), followed by Thescelosaurus at 8% (n = 15), Ornithomimus at 5% (n = 9), and Pachycephalosaurus and Ankylosaurus both at 1% (n = 2) are relatively rare (see Figure 4 ). Figure 3. Tyrannosaurus ( MOR 1125, ‘‘B-rex’ ’) teeth from the lower jaw of this medium-sized skeleton illustrate the extreme range in overall tooth size within one individual. A. A smaller posterior tooth from position #14 from the front of the jaw. B. A larger tooth from position #4 in the same jaw. This demonstrates why shed dinosaur teeth are not a reliable indicator of relative skeletal size and ontogenetic age. doi:10.1371/journal.pone.0016574.g003 Even though Triceratops dominates this census, associated specimens of Triceratops consisting of both cranial and postcranial elements remain relatively rare (see Tables S1 , S2). This contrasts with the record of isolated skulls that contribute to a significant portion of this census. We propose that this inconsistency may be explained by a historical collecting bias influenced by taphonomic controls. This is documented in museum collections [18]. Alternatively, predation, scavenging, or some as yet unknown vital effect of rapid deterioration of Triceratops limb elements may limit their preservation in the fossil record. We observed that postcranial elements are often located at some distance from the associated skull, particularly in the preservation of Triceratops . Thus, the limited discovery of postcranial elements may, in some circumstances, simply depend on how extensive a quarry is expanded after a skull is collected. Ontogenetic Stages When ontogenetic stages are considered, we observe a low number of both ‘‘A’’ and ‘‘F’’ class (see ontogeny column in Tables S1 , S3, S4, S6) of Triceratops individuals and ‘‘S’’ and ‘‘XL’’ individuals of other taxa. Overall, the dinosaur assemblages represented in the Hell Creek Formation consist primarily of subadult or small adult size individuals (based on comparisons with the largest specimens of known taxa). Small juveniles and large adults are both extremely rare, whereas subadult individuals (M & L and D & E) are relatively common. The paucity of juveniles seen in the Hell Creek Formation and contemporaneous sediments puzzled earlier researchers [22]. This can likely be explained by a combination of: (1) extended parental care [23–25]; (2) rapid juvenile growth [26,27]; and (3) colonial nesting in select geographic environments [19,28]. This pattern likely reflects either a preservational (taphonomic) or life history consequence acting on the dinosaur population. The uncommonness of apparently fully mature adults is more mysterious and not easily explained. What is now apparent, however, is this pattern contributed to an historical increase in the naming of new dinosaur species from the Hell Creek Formation. For example, over many decades it was presumed that the taxon ‘‘ Torosaurus ’’ represented a horned dinosaur that reached enormous proportions, even though there were no reported juveniles in the literature. The relatively expanded and fenestrated parietosquamosal frill exhibited by ‘‘ Torosaurus ’’ was among its most significant features [29]. With the advent of studies employing ontogenetic osteohistology, the alternative hypothesis that these giant dinosaurs were more likely mature individuals of existing taxa, rather than distinct taxa, became evident. This hypothesis is exemplified in recent studies of Triceratops ontogeny [12,30] that reinterpret ‘‘ Torosaurus ’’ as an adult Triceratops . Nonetheless, this hypothesis fails to explain why these giant, mature individuals are so rare, or more explicitly, why most Triceratops specimens are subadult sized. We propose that mature individuals of at least some dinosaur taxa either lived in a separate geographic locale analogous to younger individuals inhabiting an upland fauna, or these taxa experienced high mortality rates before reaching terminal size where late stage and often extreme cranial morphology is expressed. Figure 4. Pie chart of the time averaged census for largebodied dinosaurs from the entire Hell Creek Formation in the study area. Triceratops is the most common dinosaur at 40% (n = 72); Tyrannosaurus is second at 24% (n = 44); Edmontosaurus is third at 20% (n = 36) followed by Thescelosaurus at 8% (n = 15)ı Ornithomimus at 5% (n = 9)ı and Pachycephalosaurus and Ankylosaurus at 1% (n = 2). doi:10.1371/journal.pone.0016574.g004 Reproductive maturity in some dinosaurs was achieved during subadulthood (e.g., Tyrannosaurus , Allosaurus and Tenontosaurus ) and this event led to high adult mortality [31]. Interestingly however, our census data indicate the highest mortality occurred when Triceratops was about 2/3 grown (= skulls approximately 2.0 m in length compared to adults with 3.0 m long skulls) prior to the final ontogenetic stage of frill expansion and fenestration in Triceratops (= ‘‘ Torosaurus ’’). Edmontosaurus conforms to a similar scenario where the ‘‘XL’’ size individuals are the most rare, and the mid-size (‘‘M’’ and ‘‘L’’) individuals are the most common. This pattern is difficult to evaluate in Tyrannosaurus because of apparent variations in age relative to size [17]. Nonetheless, we predict a larger specimen of Tyrannosaurus than currently known will likely be discovered in future field studies. Although the lines of arrested growth (LAGs) observed in the largest yet known Tyrannosaurus specimens [17] suggest slowed growth, and therefore a presumed nearing of maturity, the cortex tissues of the femora and tibia of these individuals remain mostly primary. This contrasts with the femoral and tibial cortex tissues of the largest individuals of Triceratops and Edmontosaurus that are mostly secondary (dense Haversian), which is a much more mature form of cortical tissue. This suggests that Tyrannosaurus growth would have continued, resulting in a bulking-up of the skeleton by continued additions of periosteal bone tissues, possibly to the external fundamental system (EFS), which signifies maturity in other taxa [26]. Tyrannosaurus Abundance The abundance of Tyrannosaurus specimens both as skeletons and as isolated elements in the LAG deposits contradicts hypotheses concerning predator-prey ratios expected for large, predatory terrestrial animals such as tyrannosaurids [32,33]. Although constant ratios are suspect in modern ecosystems [34,35], there are always at least 75% more non-predators than predators, and in mammal populations the ratio is>90% [32 and references therein]. What is particularly interesting in this census is the indication that Tyrannosaurus is at least as abundant in the upper Hell Creek Formation as Edmontosaurus , an herbivore, previously suggested to be the primary food source of Tyrannosaurus [36] ( Figure 2 ). In the remaining two-thirds of the formation, Tyrannosaurus is more plentiful than Edmontosaurus ( Table 1 ). Because the smaller, predatory dinosaur taxa Troödon and dromaeosaurids (known from teeth found in microsites) are extremely rare (no skeletons or identifiable lag specimens), it stands to reason that Tyrannosaurus was not a typical predator [37]. In fact, the large numbers of Tyrannosaurus compared to the smaller theropods suggest that Tyrannosaurus benefited from much wider food choice opportunities than exclusively live prey and specific taxa such as Edmontosaurus [36]. A similar comparison can be made with mammal census numbers from the Serengeti plains where the hyena population is twice that of the combined population of lion, leopard and cheetah [38,39]. Tyrannosaurus may have acquired a larger percentage of meat from carrion sources than did smaller theropods, therefore filling the role of a more generalized, carnivorous opportunist such as a hyena. Based on energetic arguments [40], a Serengeti type ecosystem would have provided ample carrion to feed a Tyrannosaurus sized scavenger, particularly if Tyrannosaurus did not have to compete with avian scavengers. In addition, Tyrannosaurus adults may not have competed with Tyrannosaurus juveniles if the potential proclivity for carrion increased with size during ontogeny [41,42]. Such a situation might well explain why Tyrannosaurus teeth increase in overall robustness while the total number of teeth in the lower jaws decrease during late stages of ontogeny [15].