Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Dinosaur Census Reveals Abundant Tyrannosaurus and Rare Ontogenetic Stages in the Upper Cretaceous Hell Creek Formation (Maastrichtian), Montana, USA



A dinosaur census recorded during the Hell Creek Project (1999–2009) incorporates multiple lines of evidence from geography, taphohistory, stratigraphy, phylogeny and ontogeny to investigate the relative abundance of large dinosaurs preserved in the Upper Cretaceous Hell Creek Formation of northeastern Montana, USA. Overall, the dinosaur skeletal assemblages in the Hell Creek Formation (excluding lag-influenced records) consist primarily of subadult or small adult size individuals. Small juveniles and large adults are both extremely rare, whereas subadult individuals are relatively common. 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 environment where sedimentation rates were relatively less, or these taxa experienced high mortality before reaching terminal size where late stage and often extreme cranial morphology is expressed.

Methodology/Principal Findings

Tyrannosaurus skeletons are as abundant as Edmontosaurus, an herbivore, in the upper Hell Creek Formation and nearly twice as common in the lower third of the formation. Smaller, predatory dinosaurs (e.g., Troodon and dromaeosaurids) are primarily represented by teeth found in microvertebrate localities and their skeletons or identifiable lag specimens were conspicuously absent. This relative abundance suggests Tyrannosaurus was not a typical predator and likely benefited from much wider food choice opportunities than exclusively live prey and/or specific taxa. Tyrannosaurus adults may not have competed with Tyrannosaurus juveniles if the potential for selecting carrion increased with size during ontogeny.


Triceratops is the most common dinosaur and isolated skulls contribute to a significant portion of this census. Associated specimens of Triceratops consisting of both cranial and postcranial elements remain relatively rare. This rarity may be explained by a historical collecting bias influenced by facies and taphonomic factors. The limited discovery of postcranial elements may also depend on how extensive a fossil quarry is expanded after a skull is collected.


The Hell Creek Project (1999–2009), a collaborative, multi-institutional field study of the Upper Cretaceous Hell Creek Formation, northeastern Montana, produced this dinosaur census from a well-documented collection of all taxa. The overall goal of the project was to create a comprehensive biotic foundation from which paleobiological and geological hypotheses could be tested. One of the many projects included the collection of dinosaur specimens to test hypotheses focusing on the relative abundances and the presence or absence of various dinosaurian ontogenetic stages. Previous Hell Creek Formation surveys [1], [2] attempted to statistically support particular extinction hypotheses, but offered minimal information on the actual composition of the stratigraphically dispersed assemblages through the entire section of the Hell Creek Formation. A dataset by Sheehan et al. [1] was used by White et al. [3] to examine the structure of the dinosaurian assemblage with regard to taphonomy only. Russell and Manabe [4] provided a clearer picture of relative abundances within the dinosaur assemblage, but failed to subdivide the Hell Creek Formation stratigraphically or to include ontogenetic perspectives in their analysis of previous surveys and collections. We think it is essential that all temporal and spatial points of reference be considered synthetically when analyzing taxa from the fossil record – a type of unified field theory for paleontological specimens. Unified frames of reference (UFR) include geography, taphohistory (defined here as the history of the specimen from death to final disposition within space and time), stratigraphy, phylogeny and ontogeny.

In this current survey, we focused on the Upper Cretaceous Hell Creek Formation exposed around Fort Peck Reservoir in northeastern Montana. Here, contiguous outcrops are traceable over an area of about 1000 sq. km (Figure 1). Facies changes were tracked within the formation's 90 to 100 meter thickness. Each confirmed skeleton was recorded and evaluated with regard to its UFR. This dataset is archived in the Museum of the Rockies (MOR).

Figure 1. Index map of the Upper Cretaceous Hell Creek Formation along Ft. Peck Lake in northeastern Montana, USA.

Contiguous outcrops are traceable over an area of about 1000 square km adjacent to Fort Peck Lake shown in blue. Dark orange represents the lower Hell Creek Formation and light orange represents the middle and upper Hell Creek Formation. The enlarged view of the study area is indicated by the rectangle in the northeast quarter of the map of Montana.

Here we present the relative abundance of large dinosaur taxa, their general ontogenetic stage within a stratigraphic and sedimentological context and offer generalized hypotheses to explain several of the more interesting patterns that have emerged. This dinosaur census sets the stage for future statistical analysis and evaluation of the end Cretaceous faunal record and will add to recent paleoecological studies of diverse dinosaur faunas and clades that focused on body size, habitat partitioning and living space requirements [5] or abundance modeling [6].

The Hell Creek Project spanned 11 field seasons from 1999–2009. The first five years focused on the collection of specimens from the lower third of the formation (See Figure 1, dark orange area), while the remaining six years were aimed at specimen collection from the middle and upper strata (Figure 1, light orange area). Since several of the dinosaur specimens from the most recent field seasons (2008–2010) remain unprepared, the data presented here for the upper two-thirds of the formation will be updated as Hell Creek Project specimens are prepared and available from the corresponding author and on


Geological Methods

In order to evaluate possible changes in dinosaur taxa and their relative abundance through time, the Hell Creek Formation was divided into three stratigraphic units [7], [8] designated L3 (lowest), M3 (middle) and U3 (upper) (see Figure 2). Only specimens from the L3 and U3 are included in this census because these units show the greatest faunal contrast and have the highest resolution from sedimentological and stratigraphic controls in contrast to M3. A continuing evaluation of the geology and paleontology of M3 will be reported on in a follow-up study by the senior author and the Hell Creek Project team.

Figure 2. Stratigraphic divisions of the Hell Creek Formation with stratigraphic details of the upper third (U3) and lower third (L3) sequences referenced in the text and the associated pie chart showing dinosaur abundance.

Pie charts I – IV reflect the relative abundance of dinosaur genera based on skeletons (charts I and IV) and individual bones (charts II and III). Tyrannosaurus skeletons are as abundant as the hadrosaurid Edmontosaurus, an herbivore, in the upper Hell Creek Formation and nearly twice as common as Edmontosaurus in the lower Hell Creek Formation. Individual bones of Tyrannosaurus and Edmontosaurus are found approximately in the same order of magnitude within the lag deposits. Key: Pie chart I, dinosaur skeletons in the upper Hell Creek Formation (U3); Pie chart II, dinosaur bones from the “Doldrum’s” lag deposit at the base of the Apex Sandstone (AS); Pie chart III, dinosaur bones from the “3B-1” lag deposit at the base of the Jen-rex Sand; and Pie chart IV, dinosaur skeletons from the lower Hell Creek Formation (L3). Abbreviations: L3lBS, lower basal sandstone; L3.mBS, middle basal sandstone; L3.uBS, upper basal sandstone; L3.lMS, lower mudstone; L3.mMS, middle mudstone; L3.uMS, upper mudstone; U3.AS, Apex sandstone; U3.ibMS, interbedded mudstone; U3.10mS, 10 meter sandstone; U3.uMS, upper mudstone; U3.CZ, coal zone.

In this study, the L3 and U3 are subdivided based on sedimentological and stratigraphic details confirmed in the study area (Figure 1). The L3 basal sandstone (L3.BS) is further subdivided into three stratigraphic horizons: lower, L3.lBS; middle, L3.mBS; and upper, L3.uBS. An overlying variegated mudstone (L3.MS) is also subdivided into three traceable units: lower, L3.lMS; middle, L3.mMS; and upper, L3.uMS. Because a dinosaur – bearing lag deposit was not identified in the lower third (L3) of the Hell Creek Formation, we use the Jen-rex sand [7] at the base of M3 as a proxy for this facies in L3 because of its stratigraphic position along the M3–L3 contact (see Figure 2).

The U3 is comprised of two sandstone units: the Apex sandstone (U3.AS) and the 10-meter sandstone (U3.10mS). An interbedded variegated mudstone (U3.ibMS) separates these two sandstones and an upper variegated mudstone (U3.uMS) overlies the 10-meter sandstone. U3 is capped by an overlying uppermost coal zone (U3.CZ).

High-resolution digital video (remote control camera mounted on a Bell 407 helicopter) was utilized to track stratigraphic horizons and facies over large distances to assure stratigraphic resolution and supplement standard stratigraphic sections and mapping within the study area.

Census Methods

Census figures for the dinosaurs were determined using two different methods: (1) recording numbers of articulated or associated skeletons and (2) recording numbers of individuals based on isolated elements collected from lag deposits in the L3 ( = Jen-rex sand) and U3 ( = Apex sandstone). Although more than 150 microsites were recorded and sampled during the course of the Hell Creek Project, microsite census data are not included here because we think these sites impart biased dinosaur census data (contra [2]) because of the limited size of microsite specimens due to facies and size sorting dynamics of the particular hydrologic regime. While the shed teeth of adult dinosaurs such as Triceratops, Edmontosaurus, Thescelosaurus, Pachycephalosaurus, and Ankylosaurus are commonly found in microsites, the adult shed teeth of Tyrannosaurus are very rare, likely because they are much larger and heavier than the specimens commonly recovered from these deposits. In addition, the microsites sampled by the MOR crews were surface collected rather than screened, and could therefore potentially impart a collector-biased faunal record (e.g., we observed that small theropod teeth are often surface collected in greater numbers due to either a taphonomic or collector bias, or both, compared to ornithischian teeth). Nonetheless, these stratigraphically constrained microsites provide a basis for future comparisons with similar Mesozoic ecosystems [9][11].

Countable dinosaur specimens are herein defined as any group of three or more bones most likely belonging to one individual based on size, association and taphonomic details (color, wear, etc.). We assume that three bones represent the minimum number of elements of a skeleton or skull that has not been transported and redeposited after initial burial. As for bones collected from lag deposits, it is assumed that these sites contain a potentially time-averaged, cross section of the dinosaur fauna, as long as the skeletal elements vary in size from small teeth to large limb elements (≥0.50 m). Census results from two such deposits are included in this study: (1) the Jen-rex sand (3B-1 Lag: MOR locality no. HC-312) and (2) the U3 unit at the base of the Apex sandstone (Doldrum's Lag: MOR locality no. HC-530). Lag specimens of dinosaurs were collected and logged only if they could be reasonably identified to genus. Dinosaur genera recognized for this study include Triceratops (Torosaurus and Nedoceratops have been synonymized with Triceratops [12]), Edmontosaurus (“Anatotitan” is not recognized in this area), Thescelosaurus (Bugenosaura has been synonymized with Thescelosaurus [13]), Pachycephalosaurus (Stygimoloch and Dracorex have been synonymized with Pachycephalosaurus [14]), Ankylosaurus, Ornithomimus, and Tyrannosaurus (Nanotyrannus is considered a juvenile Tyrannosaurus [15]). No other dinosaurian material represented taxa we could identify to genus.

In addition to the census of skeletons and lag specimens (Tables S1,S3, S4, S5, S6) we also provide a list of Triceratops skulls (Table S2) that were located in the L3 unit, recorded, but not collected. Here, specimen collection varied according to the year of discovery. Triceratops specimens located in the early years (1999–2003) of the Hell Creek Project were not collected unless it was obvious that associated, disarticulated elements were present. If we could not determine the degree of articulation to indicate a skeleton by simple surface observation, the specimens were recorded and left in-situ. We later learned that extensive excavation was required to make this evaluation. This “learning process” probably explains the majority of the differences in Triceratops abundance between the census numbers from the L3 (n = 11) and the U3 (n = 22) units. When weathered Triceratops skulls were located during the first five-year period, their positions were recorded without excavation, even though it was clear that each of the skulls consisted of at least three distinct cranial elements (e.g., postorbital horn, squamosal and jugal). During the latter years of the project following 2003, every Triceratops specimen found was excavated regardless of how eroded or poorly preserved the exposed bones first appeared to be. Upon excavation, most specimens were found disarticulated with fair to very good preservation at depth. Those occurrences representing a countable skeleton were logged into the skeletal census while specimens that did not meet these criteria were not.

With the exception of one specimen of Tyrannosaurus rex (designated “N-rex”) curated into the Smithsonian Institution's National Museum of Natural History, all the dinosaur specimens in this study are curated into the paleontology collections of the Museum of the Rockies, Montana State University, Bozeman, MT, USA.

Ontogenetic Methods

In addition to producing a census of individuals of each taxon in the L3 and U3 units, we also determined the relative ontogenetic stages of the counted individuals. These general ontogenetic stages were determined by morphological ontogenetic markers in some taxa, such as Triceratops [16], using ontogenetic histological methods for others, such as Tyrannosaurus [17], or using a combination of these techniques for taxa such as Pachycephalosaurus [14].

Six ontogenetic stages identified with a letter ascending in size from “A” to “F” are employed for the Triceratops skeletal material (Tables S1). Triceratops has the widest published ontogenetic range [12], [16] and therefore allows for higher ontogenetic resolution than other taxa. Each letter corresponds to a relative age class: “A,” represents small juveniles; “B,” large juveniles; “C,” small subadults; “D,” large subadults; “E,” small adults; and “F,” large adults. These stages are simply determined relative to one another based on the smallest and largest end member skulls. Within specimens of Triceratops, for example, “A” individuals have skulls approximately 0.3 m in length; F-size skulls are approximately 3.0 m in length. Letters in-between “A” and “F” provide a general sense of intervening sizes. These stages are useful for the purpose of demonstrating the rarity or abundance of certain size classes of individuals and communicating these occurrences. Four size ranges are utilized for taxa other than Triceratops and for all the isolated fossils from the lag deposits (Tables S3, S4, S6): small “S”, medium “M,” large “L,” and extra-large “XL”.


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 (superscript3 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).

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.

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.

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.



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 4. Pie chart of the time averaged census for large-bodied 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), Ornithomimus at 5% (n = 9), and Pachycephalosaurus and Ankylosaurus at 1% (n = 2).

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.

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].


The Hell Creek Project generated eleven years (1999–2009) of collecting and field studies in the Upper Cretaceous Hell Creek Formation, eastern Montana, and resulted in the discovery of a wide variety of new dinosaur specimens, many of which revealed for the first time the ontogeny of well-known dinosaur taxa such as Triceratops. More than 240 associated or articulated dinosaur specimens were collected or recorded, and of these, smaller juveniles or larger adults were underrepresented, suggesting that these size individuals were not as common in the Hell Creek ecosystem. Nesting horizons in broadly contemporaneous formations suggest juveniles probably lived in other locations and the largest adults may have simply been relatively rare within their populations. Mortality rates appear to be higher among individuals that were not yet fully mature. This may have resulted in positive feedback acting on earlier maturation rates during ontogeny in some dinosaurs.

The relatively high abundance of Tyrannosaurus contradicts earlier suggestions that it was a very rare taxon in the Hell Creek Formation [21]. This census suggests that Tyrannosaurus was not strictly a predator, but instead more of an opportunistic feeder, possibly selecting similar food choices under circumstances comparable to that of hyenas in extant ecosystems, a trend unrecognized in earlier census studies.

Supporting Information

Table S1.

Lower Hell Creek Formation (L3) dinosaur skeletons in order of abundance.


Table S2.

Upper Hell Creek Formation (U3) Triceratops skulls collected. These specimens are either whole skulls or isolated elements and do not qualify to be counted under the “three bone rule.”


Table S3.

Dinosaur census from 3B1 lag deposit (MOR loc. HC-312) at the base of the Jen-rex sand in order of abundance.


Table S4.

Upper Hell Creek Formation (U3) skeletons in order of abundance.


Table S5.

Upper Hell Creek Formation (U3) Triceratops skulls recorded but not collected.


Table S6.

Census from the Doldrum's lag deposit (MOR loc. HC-530) at the base of the Apex sandstone in the upper Hell Creek Formation (U3.AS) in order of abundance.



We are grateful to Judd and Jay Twitchell for allowing us to camp on their land, and to the Twitchell families and Clay Taylor family for access to federal and state lands. We also thank Terry and Mary Kohler and the Windway Foundation for summer donations of their helicopter (with fuel, mechanic and pilot). Bill Berg and the staff of the Charles M. Russell Wildlife Refuge (USFW) in Lewistown, Gary Smith and Doug Melton and other staff of the Bureau of Land Management in Billings and Miles City, and Patrick Renne and staff at Montana State Lands (DNRC) in Helena facilitated access to land under their management and we are grateful for their cooperation and support. We also thank the staff of Hell Creek State Park and campground for hosting our field camp in the early years of the project. We thank Bob Harmon and his staff of field crew chiefs, and all the volunteers on the Hell Creek Project for their discoveries, and Carrie Ancell and Jamie Jette for the preparation of the fossil specimens. Thanks also to Denver Fowler, John Scannella, Jennifer Flight, Greg Wilson, Laura Wilson, and Ben Shoup for discussions concerning their thesis and dissertation data. David Polly and an anonymous reviewer provided thoughtful comments and reviews of an earlier draft of this manuscript. We thank Peter Roopnarine, PLoS One Academic Editor, for his editorial guidance and attention, and William Clemens and Patricia Holroyd for their comments and interest in this study. David Smith provided expert graphics assistance of the final figures. Pat Leiggi and Shelley McKamey were generous hosts on repeated visits to the Museum of the Rockies and Bozeman, Montana by MBG, and their support and friendship facilitated this research. This is University of California Museum of Paleontology contribution no. 2022.

Author Contributions

Conceived and designed the experiments: JH NM. Performed the experiments: JH MG. Analyzed the data: JH MG. Contributed reagents/materials/analysis tools: JH MG. Wrote the paper: JH MG.


  1. 1. Sheehan PM, Fastovsky DE, Hoffmann RG, Berghaus CB, Gabriel DL (1991) Sudden extinction of the dinosaurs: Latest Cretaceous, Upper Great Plains, USA. Science 254: 835–839.
  2. 2. Pearson DA, Schaefer T, Johnson KR, Nichols DJ, Hunter JP (2002) Vertebrate biostratigraphy of the Hell Creek Formation in southwestern North Dakota and northwestern South Dakota. In: Hartman JH, Johnson KR, Nichols DJ, editors. The Hell Creek Formation and the Cretaceous-Tertiary boundary in the northern Great Plains: An integrated continental record of the end of the Cretaceous. Boulder, Colorado: Geol Soc Am Special Paper 361. pp. 145–167.
  3. 3. White PD, Fastovsky DE, Sheehan PM (1998) Taphonomy and suggested structure of the dinosaurian assemblage of the Hell Creek Formation (Maastrichtian), eastern Montana and western North Dakota. Palaios 13: 41–51.
  4. 4. Russell DA, Manabe M (2002) Synopsis of the Hell Creek (uppermost Cretaceous) dinosaur assemblage. In: Hartman JH, Johnson KR, Nichols DJ, editors. The Hell Creek Formation and the Cretaceous-Tertiary boundary in the northern Great Plains: An integrated continental record of the end of the Cretaceous. Boulder, Colorado: Geol Soc Am Special Paper 361. pp. 169–176.
  5. 5. Farlow JO, Planka ER (2002) Body size overlap, habitat partitioning and living space requirements of terrestrial vertebrate predators: implications for the paleoecology of large theropod dinosaurs. Hist Biol 16: 21–40.
  6. 6. Farlow JO, Coroian ID, Foster JR (2010) Giants on the landscape: modeling the abundance of megaherbivorous dinosaurs of the Morrison Formation (Late Jurassic, western USA). Hist Biol 1–27. DOI:
  7. 7. Scannella JB, Fowler DW (2009) Anagenesis in Triceratops: evidence from a newly resolved stratigraphic framework for the Hell Creek Formation. 9th North American Paleontological Convention Abstracts. Cincinnati Museum Center Scientific Contributions 3. pp. 148–149.
  8. 8. Fowler DW (2009) A sequence stratigraphic subdivision of the Hell Creek Formation: beginnings of a high-resolution regional chronostratigraphic framework for the terminal Cretaceous. 9th North American Paleontological Convention Abstracts. Cincinnati Museum Center Scientific Contributions 3. 136 p.
  9. 9. Dodson P (1983) A faunal review of the Judith River (Oldman) Formation, Dinosaur Provincial Park, Alberta. Mosasaur 1: 89–118.
  10. 10. Dodson P (1987) Microfaunal studies of dinosaur paleoecology: Judith River Formation of southern Alberta. In: Currie PJ, Koster EH, editors. Fourth Symposium on Mesozoic Terrestrial Ecosystems. Drumheller, Alberta. Occasional Paper of the Tyrrell Museum of Palaeontology. 3. : 70–75.
  11. 11. Fowler DW (2009) A sequence stratigraphic subdivision of the Hell Creek Formation: beginnings of a high-resolution regional chronostratigraphic framework for the terminal Cretaceous. 9th North American Paleontological Convention Abstracts. Cincinnati Museum Center Scientific Contributions 3. 136 p.
  12. 12. Scannella JB, Horner JR (2010) Torosaurus Marsh, 1891 is Triceratops Marsh, 1889 (Ceratopsidae: Chasmosaurinae): synonymy through ontogeny. J Vert Paleont 30: 1157–1168.
  13. 13. Boyd CA, Brown CM, Scheetz RD, Clarke JA (2009) Taxonomic revision of the basal neornithischian taxa Thescelosaurus and Bugenasaura. J Vert Paleont 29: 758–770.
  14. 14. Horner JR, Goodwin MB (2009) Extreme cranial ontogeny in the Upper Cretaceous dinosaur Pachycephalosaurus. PLoS ONE 4(10): e7626.
  15. 15. Carr TD (1999) Craniofacial ontogeny in Tyrannosauridae (Dinosauria, Coelurosauria). J Vert Paleont 19: 497–520.
  16. 16. Horner JR, Goodwin MB (2006) Major cranial changes during Triceratops ontogeny. Proc R Soc Lond Biol 273: 2757–2761.
  17. 17. Horner JR, Padian K (2004) Age and growth dynamics of Tyrannosaurus rex. Proc R Soc Lond Biol 271: 1875–1880.
  18. 18. Goodwin MB, Horner JR (2010) Historical collecting bias and the fossil record of Triceratops in Montana. In: Ryan MJ, Chinnery-Allgeier BJ, Eberth DA, editors. New Perspectives on Horned Dinosaurs: The Royal Tyrrell Museum Ceratopsian Symposium. Bloomington: Indiana University Press. pp. 551–563.
  19. 19. Carpenter K (1982) Baby dinosaurs from the Late Cretaceous Lance and Hell Creek formations and a description of a new species of theropod. Contributions to Geology. University of Wyoming 20: 123–134.
  20. 20. Goodwin MB, Clemens WA, Horner JR, Padian K (2006) The smallest known Triceratops skull: new observations on ceratopsid cranial anatomy and ontogeny. J Vert Paleont 26: 103–112.
  21. 21. Brown B (1907) The Hell Creek Beds of the Upper Cretaceous of Montana. Bulletin of the American Museum of Natural History 23: 823–845.
  22. 22. Sternberg CM (1955) A juvenile hadrosaur from the Old Man Formation of Alberta. Annual Report of the National Museum of Canada 136: 120–122.
  23. 23. Horner JR, Makela R (1979) Nest of juveniles provides evidence of family structure among dinosaurs. Nature 282: 296–298.
  24. 24. Horner JR (2000) Dinosaur reproduction and parenting. Annual Review of Earth and Planetary Sciences 28: 19–45.
  25. 25. Meng Q, Liu J, Varricchio DJ, Huang T, Gao C (2004) Parental care in an ornithischian dinosaur. Nature 431: 145–146.
  26. 26. Horner JR, de Ricqlès AJ, Padian K (2000) Long bone histology of the hadrosaurid dinosaur Maiasaura peeblesorum: growth dynamics and physiology based on an ontogenetic series of skeletal elements. J Vert Paleont 20: 115–129.
  27. 27. Horner JR, Padian K, de Ricqlès AJ (2001) Comparative osteohistology of some embryonic and neonatal archosaurs: phylogenetic and behavioral implications for dinosaurs. Paleobiol 27: 39–58.
  28. 28. Horner JR (1982) Evidence of colonial nesting and ‘site fidelity’ among ornithischian dinosaurs. Nature 297: 675–676.
  29. 29. Marsh OC (1891) Notice of new vertebrate fossils. American Journal of Sciences series 3: 265–269.
  30. 30. Scannella JB (2009) And then there was one: synonymy consequences of Triceratops cranial ontogeny. J Vert Paleont 29: 128(A).
  31. 31. Lee AH, Werning S (2008) Sexual maturity in growing dinosaurs does not fit reptilian growth models. P Natl Acad Sci USA 105: 582–587.
  32. 32. Bakker RT (1972) Anatomical and ecological evidence of endothermy in dinosaurs: Nature 238: 81–85.
  33. 33. Farlow JO, Holtz TR Jr (2002) The fossil record of predation in dinosaurs. Paleontological Society Papers 8: 251–265.
  34. 34. Closs G, Watterson GA, Donnelly PJ (1993) Constant predator-prey ratios: an arithmetical artifact? Ecology 74: 238–243.
  35. 35. Wilson JB (1996) The myth of constant predator: prey ratios. Oecologia 106: 272–276.
  36. 36. Paul GS (2008) The extreme lifestyles and habits of the gigantic tyrannosaurid superpredators of the Late Cretaceous of North America and Asia. In: Larson P, Carpenter K, editors. Tyrannosaurus Rex. Bloomington: Indiana University Press. pp. 307–352.
  37. 37. Farlow JO (1994) Speculations about the carrion-locating ability of tyrannosaurs. Hist Biol 7: 159–165.
  38. 38. Mills MGL (2003) Kalahari Hyenas, Comparative Behavioral Ecology of Two Species. The Blackburn Press. 298 p.
  39. 39. Sinclair ARE, Arcese P, editors. (1995) Serengeti II: dynamics, management, and conservation of an ecosystem. University of Chicago Press. 673 p.
  40. 40. Ruxton GD, Houston DC (2003) Could Tyrannosaurus rex have been a scavenger rather than a predator? An energetics approach. Proc R Soc Lond Biol 270: 731–733.
  41. 41. Russell DA (1977) A Vanished World: The Dinosaurs of Western Canada. National Museum of Natural Sciences. Natural History Series No. 4. National Museum of Canada, Ottawa. 142 p.
  42. 42. Holtz TR Jr (2008) A critical reappraisal of the obligate scavenging hypothesis for Tyrannosaurus rex and other tyrant dinosaurs. In: Larson P, Carpenter K, editors. Tyrannosaurus Rex. Bloomington: Indiana University Press. pp. 371–396.