Introduction

The record of Early Cretaceous sauropod dinosaurs in North America was for many years poorly represented, with the only generic names in use being Astrodon Leidy, 1865 and Pleurocoelus Marsh, 1888, the former represented only by teeth and often considered synonymous with the latter (e.g., Hatcher 1903a; Gilmore 1921; Carpenter and Tidwell 2005). In recent years, this record has been greatly expanded and clarified by the discovery and description of Sonorasaurus Ratkevich, 1998; Cedarosaurus Tidwell, Carpenter, and Brooks, 1999; Sauroposeidon Wedel, Cifelli, and Sanders, 2000a; Venenosaurus Tidwell, Carpenter, and Meyer, 2001; Paluxysaurus Rose, 2007; and Abydosaurus Chure, Britt, Whitlock, and Wilson, 2010. Further material, representing yet more new sauropod taxa, is known and awaits description: for example, two new taxa from the Dalton Wells Quarry, a camarasaurid and a titanosaurian (Eberth et al. 2006: 220); and a titanosaurian from the Yellow Cat Member of the Cedar Mountain Formation represented by an articulated sequence of five presacral vertebrae (Tidwell and Carpenter 2007).

Here we describe a new sauropod taxon which further extends the record of Early Cretaceous North American sauropods. The ilium and scapula of this sauropod were figured, but not described, by Kirkland et al. (1997: 93), who considered them as “comparable to Pleurocoelus” (at that time the only known Early Cretaceous sauropod from North America).

Institutional abbreviations.—AMNH, American Museum of Natural History, New York, USA; BMNH, the Natural History Museum, London, UK; CCG, Chengdu College of Geology, Chengdu, China; CM, Carnegie Museum of Natural History, Pittsburgh, USA; DMNH, Denver Museum of Natural History, Denver, USA; DNM, Dinosaur National Monument, Dinosaur, USA; FMNH, Field Museum of Natural History, Chicago, USA; FWMSH, Fort Worth Museum of Science and History, Fort Worth, USA; HMN, Humboldt Museum für Naturkunde, Berlin, Germany; OMNH, Oklahoma Museum of Natural History, Norman, USA; ZDM, Zigong Dinosaur Museum, Zigong, China.

Anatomical nomenclature.—We follow Upchurch et al. (2004a) in describing scapulae as though oriented horizontally: the coracoid articular surface is designated anterior and the end commonly called “distal” is designated posterior. See Taylor (2009: 787) for a summary of other schemes that have been used. Names of clades are used as summarized in Table 1.

Geological and faunal context

The holotype and referred specimens described herein were collected at OMNH locality V857, known as the Hotel Mesa Quarry (Kirkland et al. 1997: 96–97, fig. 28). One of us (RLC), who had already been working in Lower Cretaceous rocks of the region, was notified about the site through the courtesy of James I. Kirkland, and was guided to it by Bill Hawes of Grand Junction, Colorado, in September 1994. Additional collecting at the site for OMNH was conducted by Randall L. Nydam and James I. Kirkland in March 1995.

OMNH V857 lies on the southwest flank of Hotel Mesa in easternmost Grand County, Utah (see Kirkland and Madsen 2007: fig. 2), about 1.3 km north-northeast of the junction of the Colorado and Dolores rivers (sec. 4, T23S R24E, Dewey 7.5' quadrangle, USGS provisional edition 1985; precise locality data are on file at OMNH and are available to qualified investigators upon request). This site had previously been worked by private collectors; the number, identity, condition, and disposition of the privately collected fossils cannot now be determined. However, given the density of bone still present and exposed, and the fact that the existing quarry was already some 5–6 m long and 3 m deep, it appears that a considerable number of elements were removed from the quarry and that the loss of valuable scientific information has unfortunately been considerable. Bones left exposed by these previous collectors were in various states of disrepair: some had been broken and their pieces used to hold down the remnants of a plastic tarpaulin. The material reported herein was recovered in the course of the two brief OMNH survey/salvage visits to this site.

Table 1.

Clade names used in this study and the definitions used. For simplicity, specifiers are indicated by genus rather than species; in each case, the type species of the genus is intended. Node-based clades are indicated by “+”, branch-based clades by “not”.

Stratigraphically, OMNH V857 lies in a sequence of Lower Cretaceous rocks interposed between the Morrison Formation (Kimmeridgian) below and the Dakota Formation (Cenomanian) above. Westward, these rocks are recognized as the Cedar Mountain Formation; eastward, the Burro Canyon Formation. The arbitrary dividing line between these entities is generally placed at the Colorado River (Stokes 1952; Tschudy et al. 1984) which technically places OMNH V857 within the Burro Canyon Formation. However, we will refer to the locality as belonging to the more widely recognized Cedar Mountain Formation, as it is in this formation that comparable specimens are known, and the stratigraphy and sedimentology do not change across the arbitrary border. The most extensive recent applicable stratigraphic work is that of Kirkland et al. (1997, 1999), who recognized five members within the Cedar Mountain Formation (in ascending order): the Buckhorn Conglomerate, Yellow Cat, Poison Strip Sandstone, Ruby Ranch, and Mussentuchit members. (The Buckhorn Conglomerate Member has a non-overlapping geographic distribution with respect to the Yellow Cat and Poison Strip Sandstone members and may be a lateral equivalent of one or both; see Kirkland et al. 1999: figs. 1, 2) The existing paleontological record from these units is patchy, but there is evidence of three distinct dinosaur faunas from the Yellow Cat (Barremian or older), Poison Strip Sandstone plus Ruby Ranch (Aptian—Albian), and Mussentuchit (Albian—Cenomanian) members, respectively (Kirkland et al. 1997). While spanning a total of some 30 Ma, the faunal record from the Cedar Mountain Formation is punctuated by significant hiatuses (Sames et al. 2010), and fossil horizons that are stratigraphically close may therefore be widely separated in time.

Like most terrigenous Mesozoic units of the Western Interior (see for example Ostrom 1970: 14), the laterally equivalent Cedar Mountain and Burro Canyon formations are enormously variable. At Hotel Mesa, the lowest two (Buckhorn Conglomerate, Yellow Cat) and uppermost (Mussentuchit) members are absent, and only two members of the Cedar Mountain Formation may be recognized: the Poison Strip Sandstone (below) and Ruby Ranch (above) members. Stratigraphically, OMNH locality V857 lies within the Ruby Ranch Member, a few meters below the local contact with the Dakota Formation (see Kirkland et al. 1999: fig. 8). As a working hypothesis, we regard the fossils from the Hotel Mesa site as being Aptian—Albian, and therefore approximately equivalent in age to assemblages from the Cloverly Formation of Wyoming and Montana and the Trinity Group of Texas and Oklahoma (e.g., Ostrom 1970; Winkler et al. 1990; Jacobs et al. 1991; Brinkman et al. 1998; Davis et al. 2008). As elsewhere, the Ruby Ranch Member in this area is comprised predominantly of drab, pale green, and mauve mudstones with abundant carbonate nodules. The fossil horizon, which is mainly sandstone with some pebbles and intermittent, irregular mudstone stringers, is about 40–50 cm thick. Locally, these mudstones contain sparse microvertebrate fossils, often rounded or broken. About 100 kg of this matrix was collected and transported back to the OMNH for fossil extraction using underwater screen washing and associated concentration techniques (Cifelli et al. 1996).

Although the microvertebrate assemblage from Hotel Mesa is neither very diverse nor particularly well represented, it does merit attention as being the only such site of Aptian—Albian age within the Colorado Plateau. A faunal list is presented in Table 2. As is generally the case for such assemblages, a number of aquatic taxa are represented though no remains of Testudines have been recovered yet. Probably two chondrichthyans are present, a hybodontid (possibly Hybodus, which was listed in the Ruby Ranch fauna by Kirkland et al. 1999: table 2) and a polyacrodontid, perhaps Polyacrodus. The neoceratodontid form genus Ceratodus, widespread in Aptian—Albian faunas of North America (see review by Kirkland 1987), is represented by a poorly preserved tooth plate; two actinopterygians are present, one represented by teeth, and ganoid scales probably referable to Lepisosteidae. No mammalian or lissamphibian fossils have been recovered, and Reptilia is represented solely by archosaurians, of which crocodyliforms (three or four taxa, including a species of the widespread taxon Bernissartia and unidentified species of Goniopholididae and Atoposauridae) are most diverse. Megafossils of the Hotel Mesa Quarry mainly represent Sauropoda (including many unidentified fragmentary specimens in addition to those described herein); theropods are known by isolated teeth and rare, incomplete postcranial elements. The fauna includes an ornithopod represented by a single non-diagnostic tooth. Kirkland et al. (1999) mentioned the presence in the Ruby Ranch fauna of Tenontosaurus, otherwise known from the Cloverly Formation of Wyoming and Montana (Ostrom 1970) and the Trinity Group of Texas and Oklahoma (Langsten 1974; Winkler et al. 1997b; Brinkman et al. 1998), based on specimens from an unidentified site on the western side of the San Rafael Swell, Utah. The ornithopod specimen from the Hotel Mesa Quarry, OMNH 27824, probably does not represent Tenontosaurus, but further identification cannot be firmly established at present.

Table 2.

Vertebrate fauna of the Hotel Mesa Quarry (OMNH V857), Ruby Ranch Member, Cedar Mountain Formation (?Aptian—Albian), Grand County, Utah.

Although the quarry has not been worked in several years, it is not exhausted and may contain additional relevant material (James Kirkland, personal communication, October 2007).

Systematic paleontology

Table 3.

Material referred to the camarasauromorph sauropod Brontomerus mcintoshi from the Lower Cretaceous Cedar Mountain Formation of Utah.

Fig. 1.

Skeletal inventory of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, in left lateral view. Preserved elements are white, missing elements are reconstructed in gray. After a Camarasaurus grandis reconstruction kindly provided by Scott Hartman.

Fig. 2.

Left ilium of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, type specimen OMNH 66430 in lateral view reconstructed from the three fragments (A), and ventral view (B).

Fig. 3.

Ilia of sauropod dinosaurs, scaled to same total length. A. Mamenchisaurus hochuanensis Young and Zhao, 1972, holotype CCG V 20401, right ilium reversed, modified from Young and Zhao (1972: pl. 6: 1a). B. Diplodocus carnegii Hatcher, 1901, CM 94, right ilium reversed, modified from Hatcher (1901: pl. 10: 1). C. Camarasaurus supremus Cope, 1877, AMNH 5761 Il. 1, left ilium, modified from Osborn and Mook (1921: fig. 87). D. Giraffatitan brancai (Janensch, 1914), HMN J1, left ilium, modified from Janensch (1961: pl. E: 2). E. Rapetosaurus krausi Curry Rogers and Forster, 2001, holotype FMNH PR 2209, left ilium, modified from Curry Rogers (2009: fig. 39B). F. Brontomerus mcintoshi gen. et sp. nov. holotype OMNH 66430, left ilium.

Fig. 4.

Measurement protocol for sauropod ilia as illustrated in Fig. 3 and shown in Table 4. Total length is measured along the longest axis of the ilium; lengths of preacetabular and postacetabular lobes are measured parallel to this axis, and extend from the extremity of the lobe to the anterior margin of the pubic peduncle and posterior margin of the ischiadic peduncle respectively. Supracetabular height is measured perpendicular to the longest axis, and extends from the highest point of the acetabulum to the point level with the highest part of the ilium.

Table 4.

Relative measurements of ilia in sauropods. Measurement protocol is illustrated in Fig. 4.

Fig. 5.

Damaged presacral vertebra of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, OMNH 66429, in dorsal view, as photograph (A) and interpretive drawing (B). Shading indicates air spaces.

Fig. 6.

Mid-caudal vertebra of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, OMNH 61248 in dorsal (A), anterior (B), left lateral (C), posterior (D), and ventral (E) views.

Fig. 7.

First right dorsal rib of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, OMNH 27766 in posterior view: head of rib, showing pneumatic invasion of shaft (A) and complete rib, showing laterally directed curvature of shaft (B).

Fig. 8.

Partial left scapula of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, OMNH 27761, in lateral view, tentatively reconstructed after Giraffatitan brancai HMN Sa 9 (Janensch 1961: pl. 15: 1).

Fig. 9.

Scapulocoracoids and scapulae of sauropod dinosaurs, scaled to same length of scapular blade from posterior point of glenoid to posterior margin of blade. A. Mamenchisaurus youngi Young and Zhao, 1972, holotype ZDM0083, left scapulocoracoid, modified from Ouyang and Ye (2002: fig. 22). B. Diplodocus longus Hatcher, 1901, USNM 10865, right scapulocoracoid reversed, photograph by MPT. C. Camarasaurus supremus Cope, 1877, AMNH 5761 Sc. 1, left scapula, and AMNH 5761 Cor. 1, left coracoid, probably associated, modified from Osborn and Mook (1921: figs. 75, 81a). D. Giraffatitan brancai (Janensch, 1914), HMN Sa 9, left scapula, modified from Janensch (1961: pl. 15: 1). E. Rapetosaurus krausi Curry Rogers and Forster, 2001, holotype FMNH PR 2209, right scapula reversed, modified from Curry Rogers (2009: fig. 32). F. Brontomerus mcintoshi gen. et sp. nov. OMNH 27761, left scapula, tentatively reconstructed after Giraffatitan brancai.

Fig. 10.

Partial paired sternal plates of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, OMNH 66431 and 66432, in ?ventral view.

Comparison with other Early Cretaceous North American sauropods

Brontomerus cannot be directly compared with Astrodonl Pleurocoelus, Sonorasaurus, or Sauroposeidon due to the absence of overlapping material between these genera, nor with Abydosaurus as the overlapping elements have not yet been adequately described (Carpenter and Tidwell 2005; Ratkevich 1998; Wedel et al. 2000b; Chure et al. 2010). Two other sauropod dinosaurs are already known from the Cedar Mountain Formation: Cedarosaurus from the Yellow Cat Member and Venenosaurus from the Poison Strip Member. Brontomerus is from the stratigraphically higher Ruby Ranch Member, and can be distinguished from both of these taxa as discussed below. Since the Yellow Cat Member is Barremian in age, the Poison Strip Member is Aptian, and the upper part of the Ruby Ranch Member (where the Hotel Mesa Quarry is located) is Aptian—Albian, these three sauropods together probably span the last three ages of the Early Cretaceous. Brontomerus is also distinct from Paluxysaurus, as shown below.

Cedarosaurus .—Cedarosaurus is known from a single partial, semi-disarticulated skeleton, DMNH 39045, described by Tidwell et al. (1999). Although much of the skeleton is preserved, relatively few elements overlap with the material of Brontomerus described above: only dorsal vertebrae, mid-to-posterior caudal vertebrae, dorsal ribs, partial scapulae, and sternal plates are in common. The single crushed presacral centrum of Brontomerus cannot be usefully compared with the dorsal vertebrae of Cedarosaurus beyond the observation that the presacral bone texture described by Tidwell et al. (1999: 23) as “numerous matrix filled chambers which are separated by thin walls of bone” is a good match. Tidwell et al. (1999) did not figure either the ribs or sternal plates of Cedarosaurus, but photographs supplied by Virginia Tidwell show that its sternals are generally similar in shape to those of Brontomerus, though much larger and somewhat less elongate. Tidwell et al. (1999: 25) noted the absence of pneumatic foramina in the two preserved rib heads while recognizing the possibility that anterior ribs might be pneumatic while posterior ribs of the same individual lack this feature. Photographs of a rib head were supplied by Virginia Tidwell and show little resemblance to that of Brontomerus, but damage to both articular facets hinders comparison. The preserved portions of Cedarosaurus scapulae are from the anterior end and therefore do not greatly overlap with the more posterior preserved portion of the Brontomerus scapula; however, the mid scapular region of Cedarosaurus differs in the possession of a more pronounced acromion process, less straight ventral border and relatively narrower scapular shaft. Finally, the mid-to-posterior caudal vertebra of Brontomerus lacks the distinctive sharp ridge extending along the edge of the neural arch described by Tidwell et al. (1999: 25, fig. 5); but other differences such as its greater elongation and greatly reduced neural spine are not inconsistent with the caudals of Cedarosaurus, taking into account that the Brontomerus caudal is from a more distal position in the caudal sequence than any of those figured by Tidwell et al. (1999). In conclusion, the preponderance of the scant morphological evidence supports the generic separation of Brontomerus from Cedarosaurus. Furthermore, the Yellow Cat Member is no younger than Barremian in age, and may be Berriasian—Valanginian (Sames and Madsen 2007), giving Cedarosaurus a minimum age of 121 My a and a maximum age of∼140 My a, while the Aptian—Albian position of the Hotel Mesa Quarry at the top of the Ruby Ranch Member suggests a much younger age. While this gap does not in itself prove generic separation, it corroborates this conclusion.

Venenosaurus .—Venenosaurus was originally described from a single small adult specimen, DMNH 40932, although elements from one or more juveniles were also present in the quarry (Tidwell et al. 2001: 140). Some of the juvenile material was subsequently described by Tidwell and Wilhite (2005), but none of this material overlaps with that of Brontomerus, so comparisons with Venenosaurus must be made on the basis of the type specimen alone. The overlapping material consists of caudal vertebrae, dorsal ribs, and a left scapula. The “distal” caudal of Venenosaurus figured by Tidwell et al. (2001: fig. 11.4C) is similar to the mid-to-posterior caudal of Brontomerus in the proportions of the centrum and in the elongation of the posteriorly directed postzygapophyseal ramus. However, the Venenosaurus distal caudal has very much shorter prezygapophyses, and a much less tall neural arch which is set forward almost to the margin of the centrum rather than set back 10% of the centrum's length. It also lacks the characteristic ventral process that hangs from the postzygapophyseal ramus in Brontomerus and bears the postzygapophyseal facets. The scapula of Venenosaurus figured by Tidwell et al. (2001: fig. 11.5A) does not resemble that of Brontomerus, having a more curved ventral border, a much less steep ascent of the dorsal border towards the anterior expansion, a less expanded posterior blade, and no sign of the “steps” apparent on both borders of the blade of Brontomerus. The illustrated dorsal rib head of Venenosaurus (Tidwell et al. 2001: fig. 11.9) differs from that of Brontomerus, having a very short tuberculum, a capitulum no broader than the ramus that supports it, and a very different pneumatic excavation which invades the bone in a dorsal direction, penetrating the capitulum, rather than ventrally, penetrating the shaft. These differences of the ribs, however, may be less significant than they appear: the tuberculum of the Venenosaurus rib is “somewhat eroded” (Tidwell et al. 2001: 153) which may explain its shortness; the degree of expansion of the capitular head may vary serially, with the Venenosaurus rib being from a more posterior position than the Brontomerus rib; and pneumatic features tend to vary both serially and between individuals, and even on occasion between the two sides of a single element (e.g., in Xenoposeidon: see Taylor and Naish 2007: 1552–1553). Nevertheless, the balance of evidence strongly indicates that Brontomerus is distinct from Venenosaurus.

Paluxysaurus .—The internal structure of the presacral vertebrae of Paluxysaurus seems to be camellate, like that of Brontomerus, based on the referred isolated dorsal centrum FWMSH 93B-10-48 (Rose 2007: 17, 44–45). Some of the “distal” caudal vertebrae of Paluxysaurus (Rose 2007: fig. 17) somewhat resemble the Brontomerus mid-to-posterior caudal, but none has the very elongate prezygapophyses or ventral process of the postzygapophyseal ramus that characterize the latter, and the Paluxysaurus caudals have distinct, dorsally projecting neural spines. The figured ribs of Paluxysaurus (Rose 2007: fig. 15) differ from that of Brontomerus in every respect save the pneumatic invasion of the rib-head in the direction of the shaft, but these ribs are too badly damaged for useful comparison and in any case are probably from a more posterior position. The prepared scapulae of Paluxysaurus (Rose 2007: fig. 20) differ from that of Brontomerus in their more concave ventral border, narrower blade, less expanded posterior extremity, and lack of “steps” on the anterior and posterior borders. The sternal plates of Paluxysaurus are much less proportionally elongate than those of Brontomerus. The ilium of Paluxysaurus is not figured, but according to the description it differs in almost every respect from that of Brontomerus. In conclusion, significant differences in the mid-caudal vertebrae, scapulae and sternal plates and probably in the ribs demonstrate that these two genera are separate.

Table 5.

Character scores for the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah in the matrix used for the phylogenetic analysis of this paper. Apart from the addition of Brontomerus and the rescoring of character 261 for Rapetosaurus, the matrix is identical to that of Harris (2006). Brontomerus is scored only for the listed characters. Conventional anatomical nomenclature is here used in place of the avian nomenclature of Harris (2006).

Phylogenetic analysis

On the assumption that the referred material belongs to Brontomerus, we attempted to establish the phylogenetic position of the new taxon by means of a phylogenetic analysis. We used the matrix of Harris (2006) as a basis, adding the single new taxon to yield a matrix of 31 taxa (29 ingroups and two outgroups) and 331 characters. The only other change made was the rescoring of character 261 for Rapetosaurus (“in lateral view, the anteroventralmost point on the iliac preacetabular process”) changing it from state 1 (“is posterior to the anteriormost part of the process (process is semicircular with posteroventral excursion of cartilage cap”) to state 0 (“is also the anteriormost point (preacetabular process is pointed”). Brontomerus could be scored for 20 of the 331 characters, 6% of the total (Table 5).

Following Harris (2006), PAUP* 4.0b10 (Swofford 2002) was used to perform a heuristic search using random stepwise addition with 50 replicates and with maximum trees = 500,000. The analysis yielded 180 equally parsimonious trees with length = 788, consistency index (CI) = 0.5228, retention index (RI) = 0.6848, and rescaled consistency index (RC) = 0.3581.

Fig. 11.

Phylogenetic relationships of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah, produced using PAUP* 4.0b10 on the matrix of Harris (2006) augmented by Brontomerus, having 31 taxa and 331 characters. Left side, strict consensus of 180 most parsimonious trees (length = 788; CI = 0.5228; RI = 0.6848; RC = 0.3581); right side, 50% majority rule consensus.

The strict consensus tree (Fig. 11A) is poorly resolved, with Titanosauriformes collapsing into a broad polytomy within which only Saltasauridae is differentiated. This represents a dramatic loss of resolution compared to the results without Brontomerus (Harris 2006: fig. 5A). A posteriori deletion of Brontomerus, however, yields a well resolved Macronaria similar to that of Harris's (2006) analysis, with Camarasaurus; Brachiosaurus Riggs, 1903; Euhelopus Romer, 1956; and Malawisaurus Jacobs, Winkler, Downs, and Gomani, 1993 as successive singleton outgroups to a group of more derived titanosaurians. This demonstrates that the addition of Brontomerus to the matrix does not cause instability in the relationships between these more fully represented taxa, and that it is only the position of Brontomerus itself that is unstable. Among the equally most parsimonious positions of Brontomerus are as a non-titanosauriform camarasauromorph, a basal titanosauriform, a basal somphospondyl, the sister taxon to Euhelopus, a basal titanosaurian, a basal lithostrotian and a derived non-saltasaurid lithostrotian. One further step is sufficient to place Brontomerus as a brachiosaurid, a basal (non-camarasauromorph) macronarian, a basal (non-diplodocid) diplodocoid or even a non-neosauropod. Three further steps are required for Brontomerus to be recovered as a saltasaurid, specifically an opisthocoelicaudiine.

In the 50% majority rule tree (Fig. 11B) all the standard sauropod clades are recovered. This tree shows the most likely position of Brontomerus as a basal somphospondyl, in a trichotomy with Euhelopus and Titanosauria.

In order to investigate a possible source of instability, we also re-ran the analysis with the Rapetosaurus ilium character restored to the state given by Harris (2006), and found that the value of this character made no difference to the results: all trees are one step longer with the new value, but the topology of all consensus trees (strict, semistrict, and majority rule) is unaffected by the changed scoring.

When Brontomerus is scored for the holotype ilium only, and the referred elements are ignored, the strict consensus constrains Brontomerus only to be a eusauropod more derived than Shunosaurus Dong, Zhou, and Zhang, 1983.

Discussion

Functional anatomy.—The functional significance of the unusual ilium of Brontomerus is difficult to interpret due to the absence of functionally related skeletal elements such as the pubis and ischium, posterior dorsal and anterior caudal vertebrae, and femur. In life, these elements work together as a functional complex, each affecting the function of the others. Some speculation is warranted, however.

Fig. 12.

Speculative life restoration of the camarasauromorph sauropod Brontomerus mcintoshi gen. et sp. nov. from the Lower Cretaceous Cedar Mountain Formation of Utah. Adult individual (sized according to the referred scapula) protects juvenile (sized according to the holotype ilium) from a Utahraptor. the enlarged femoral protractors may have enabled a powerful kick. Executed by Francisco Gascó under direction from MPT and MJW, reproduced with permission.

The large preacetabular blade of the ilium provides an anchor for large protraction muscles, which would have allowed the leg to be moved forwards powerfully; this is surprising as femoral retraction is required for forward locomotion, requiring large muscles to pull the femur backwards, and the ilium of Brontomerus offers almost no attachment area for posteriorly located muscles. The caudofemoralis muscle is the main power generator in locomotion for extant reptiles, and osteological correlates such as the fourth trochanter of the femur indicate that this was also true of non-avian dinosaurs. This muscle connects the femur to the tail, so in the absence of proximal caudal vertebrae and chevrons of Brontomerus it is not possible to determine whether the femoral retractors were enlarged in proportion with the protractors. If so, then this increase in musculature would indicate that Brontomerus may have been unusually athletic for a sauropod. Another possibility is that powerful femoral protraction was useful for delivering a strong kick (Fig. 12).

As well as protractors, the preacetabular blade of the ilium also anchors abductors (i.e., muscles which draw the leg laterally away from the median plane). These muscles are important for creating abduction torque when standing, and may have facilitated occasional bipedal stance or even limited bipedal walking.

A third possibility is that the proportionally large leg muscles were required to drive unusually long legs. The large anterior expansion of the scapula provides weak additional support for this hypothesis. If this interpretation were correct, Brontomerus might have resembled a giraffe in gross morphology.

Did sauropods really decline in the Early Cretaceous of North America?—Apart from Cetiosaurus oxoniensis Phillips, 1871, the first reasonably complete remains of sauropods were discovered in the Morrison Formation of the American West in the 1870s and 1880s, and these remains provided the first accurate understanding of the size and bauplan of the clade (Cope 1877, 1878; Marsh 1877). Sauropod remains had previously been assumed to represent gigantic crocodiles (Owen 1859) or pterosaur-bird intermediates (Seeley 1870). Nineteenth century workers also quickly recognized that sauropods were common in the Morrison Formation and essentially absent in the Early Cretaceous “Dakota” Formations, which at the time included the Cloverly and Cedar Mountain Formations in addition to the Dakota Formation itself (Witzke and Ludvigson 1994). In fact, the absence of sauropods from Early Cretaceous rocks was so well-established that Marsh (1888) erroneously identified the age of the Potomac Formation (now the Potomac Group) as Late Jurassic based on the presence of the sauropod Pleurocoelus (now synonymized with Astrodon; Carpenter and Tidwell 2005).

Table 6.

Early Cretaceous North American sauropod diversity. MNI = Minimum Number of Individuals.

In the following century, the infrequent recovery of sauropod material from Early Cretaceous deposits across North America (e.g., Larkin 1910; Ostrom 1970; Langston 1974) reinforced the perception that the Late Jurassic was the apex of sauropod success, and the Early Cretaceous its nadir. This apparent dichotomy became entrenched in both scientific (Bakker 1978; Behrensmeyer et al. 1992) and popular (Bakker 1986) accounts of the evolutionary history and paleoecology of dinosaurs, including previous publications by the junior authors (Wedel et al. 2000a, b; Wedel and Cifelli 2005). However, a wave of discovery of new material of Early Cretaceous sauropods (Britt and Stadtman 1996, 1997; Britt et al. 1997, 1998, 2004; Winkler et al. 1997a; Gomani et al. 1999; Burge et al. 2000; Maxwell and Cifelli 2000; Burge and Bird 2001; Coulson et al. 2004; Bird 2005; Tidwell and Carpenter 2007) and the description of many new taxa in the past decade (reviewed below and in Table 6) prompts us to reconsider the question of whether sauropods really did decline in the Early Cretaceous of North America.

The diversity of Early Cretaceous North American sauropods.—To date, eight sauropod genera have been described from the Early Cretaceous of North America: Astrodon, including Pleurocoelus (Leidy 1865; Marsh 1888), Sonorasaurus (Ratkevich 1998), Cedarosaurus (Tidwell et al. 1999), Sauroposeidon (Wedel et al. 2000a), Venenosaurus (Tidwell et al. 2001), Paluxysaurus (Rose 2007), Abydosaurus (Chure et al. 2010), and now Brontomerus (this paper).

The pool of named taxa will always be only a subset of the taxa that have been or are being studied, and the taxa under study are in turn only a subset of the pool of potentially new taxa that have been discovered in the field, excavated, and prepared. Almost all of the Early Cretaceous North American sauropods that have been recently described or are currently under study were discovered and excavated in the 1990s or 2000s. Sauropod fossils are often logistically challenging to excavate, prepare, manipulate, photograph, and even measure, simply because they are so large and heavy. In our experience the process of preparation and description of a new sauropod can be very protracted compared to the preparation and description of other, smaller vertebrates. The recently named taxa represent the first “graduates” of the arduous descriptive process, and many more new taxa will probably be described in the coming decade. These include:

Additional sauropod remains that are extremely incomplete or not diagnostic, and therefore unfit for naming, are known from several formations. Some of these may eventually be referred to existing or forthcoming named taxa, or might attain holotype status following the discovery of more complete material or additional study. This material includes:

If all of the apparently diagnostic sets of material (first list, above) were named as new genera, the generic diversity of Early Cretaceous North American sauropods could rise from the current total of 8 to as many as 15. On the other hand, some of this undescribed material may eventually be referred to existing genera, and even some of the eight existing genera might eventually be synonymized. For example, the Sauroposeidon type material might represent the neck of Cedarosaurus or of Venenosaurus; in the former case Sauroposeidon would become a junior synonym of Cedarosaurus and in the latter Venenosaurus would become a junior synonym of Sauroposeidon. Note however, that no existing evidence supports these hypothetical synonymizations and they seem unlikely given the temporal and geographic distance separating the taxa concerned.

The statistical analysis of Mannion et al. (2010: 17–18) reported that, when apparent diversity is corrected for availability of relevantly-aged rocks, the Early Cretaceous represents a time of low diversity for sauropods. However, their analysis was based only on validly named genera, and as shown here there is evidence for many additional genera that are yet to be formally recognized.

The diversity of Early Cretaceous North American sauropods has traditionally been contrasted with that of the sauropod taxa of the Late Jurassic Morrison Formation, which is reviewed below.

The diversity of Morrison sauropods.—Six well-known genera have been recognized from the Morrison Formation for over a century: Apatosaurus, Barosaurus, Brachiosaurus, Camarasaurus, Diplodocus, and Haplocanthosaurus. Several other taxa that have long been recognized are poorly represented but nevertheless probably valid: Dystrophaeus, Amphicoelias, and “Apatosaurus” minimus, which is not referable to Apatosaurus and is probably not even a diplodocoid (McIntosh 1990: 398; Upchurch et al. 2004a: 298). In addition, several new taxa have been described in recent decades or are currently under study: Supersaurus, Eobrontosaurus, Suuwassea, and an apparently new diplodocoid announced by Vietti and Hartman (2004). Upchurch et al. (2004b: 307) suggested that Eobrontosaurus is referrable to Camarasaurus but this assertion was not based on the identification of any shared apomorphies. The holotype is distinct from both Camarasaurus and Apatosaurus and probably represents a valid taxon (Ray Wilhite, personal communication, May 2008). Finally, at least five Morrison Formation genera have been synonymized in recent years: Cathetosaurus is now classified as a distinct species of Camarasaurus (C. lewisi; McIntosh et al. 1996), Ultrasauros and Dystylosaurus are both junior synonyms of Supersaurus (Curtice et al. 1996; Curtice and Stadtman 2001), Elosaurus is a juvenile Apatosaurus (Upchurch et al. 2004b), and Seismosaurus has been synonymized with Diplodocus: as a distinct species, D. hallorum, by Lucas et al. (2006), and as the existing species D. longus by Lovelace et al. (2008). This leaves a current total of 12 named genera, plus one potentially new genus under study.

Early Cretaceous and Morrison sauropods compared.—If we simply count existing genera, then Early Cretaceous North American sauropods are only two thirds as diverse as their Morrison forebears (8 genera compared to 12). If we include all of the potentially new taxa that are currently under study, the diversity of sauropods in the Early Cretaceous of North America may eventually exceed that of the Morrison (16 genera to 13). However, so simple a comparison of the sauropod faunas of the North American Late Jurassic and Early Cretaceous is misleading, for the following reasons.

First, the North American Early Cretaceous rock units, separately or together, are simply dwarfed by the Morrison Formation. Foster (2003) listed more than 270 vertebrate localities in the Morrison Formation, in Arizona, Colorado, Idaho, Kansas, Montana, New Mexico, Oklahoma, South Dakota, Utah, and Wyoming. Based on the sheer volume and outcrop area of the Morrison Formation, we might expect its vertebrate fauna to appear more diverse than those of the Early Cretaceous units.

Second, the Morrison Formation is exposed along the Rocky Mountains and in part of the Great Basin, whereas the North American Lower Cretaceous units are scattered across the continent, from the Arundel Formation on the eastern seaboard to the Turney Ranch Formation in the desert southwest, and from the Cloverly Formation in southern Montana to the Trinity Group in central Texas. Although there is some evidence for biogeographic differentiation within the Morrison Formation (Harris and Dodson 2004), the Lower Cretaceous units span much more of the continent and much more time. Temporal differences among the Lower Cretaceous dinosaur faunas are well established (Kirkland et al. 1997), and there may have been biogeographic differentiation as well (Jacobs and Winkler 1998; Nydam and Cifelli 2002). Most relevant in this context are the Aptian—Albian assemblages of the Trinity Group, Texas (Winkler et al. 1990) and Oklahoma (Brinkman et al. 1998); and the Arundel Clay facies of the Potomac Group, Maryland (Kranz 1998). On this basis, we might therefore expect the combined fauna of the Lower Cretaceous units to be more diverse than that of the Morrison Formation.

Third, the Morrison Formation and the various Lower Cretaceous terrestrial units do not represent equivalent amounts of time. The Morrison Formation represents about seven million years of deposition, from 155–148 Mya (Kowallis et al. 1998). The most temporally extensive North American Lower Cretaceous unit is the Cedar Mountain Formation, the dinosaur-bearing parts of which span at least 28 million years from the Barremian (126 ± 2.5 Mya; Kirkland and Madsen 2007) to the earliest Cenomanian (98.37 ± 0.07 Mya; Cifelli et al. 1997b). If the Yellow Cat Member of the Cedar Mountain Formation is indeed Berriasian±Valanginian (Sames and Madsen 2007), the Cedar Mountain Formation may span 40 million years or more (albeit with large depositional hiatuses). All of the other Lower Cretaceous units fall into the span of time represented by the Cedar Mountain Formation, which is four to six times as long as that represented by the Morrison Formation. On this basis, we might also expect the faunas of the Lower Creta- ceous units, or even that of the Cedar Mountain Formation by itself, to be more diverse than that of the Morrison Formation.

Fourth, the Morrison Formation has been intensively sampled for more than 130 years, and numerous expeditions have gone to the Morrison Formation in the express hope of finding complete skeletons of sauropods. The less spectacular material of the Lower Cretaceous units has drawn much less attention from both paleontologists and the public. The Cloverly Formation has only been intensively sampled since the 1960s, and the Antlers and Cedar Mountain Formations since the 1990s (although earlier collections had been made from all of these formations). The explosive growth of our knowledge of Early Cretaceous sauropods in North America is a direct result of more intensive sampling of all of the North American Lower Cretaceous units in the past two to three decades. This new pulse of collecting is mitigating the collector bias that favored the Morrison Formation, but a serious imbalance still persists and may never be completely alleviated. This is partly because fieldwork in the Morrison Formation has not stopped and probably will not in the foreseeable future, and new sauropods continue to be discovered (e.g., Vietti and Hartman 2004) and described (e.g., Suuwassea, Harris and Dodson 2004) from the Morrison Formation.

In summary, four factors bias our perception of sauropod diversity in the Late Jurassic and Early Cretaceous of North America: the Morrison Formation is larger than all of the Lower Cretaceous units combined and has been more intensively sampled and for much longer, but the Lower Cretaceous units span more space and time. It is not clear how these various biases should be weighted. Nevertheless, three major conclusions can be drawn.

First, the North American sauropod fauna shifted from being diplodocoid-dominated during the Late Jurassic, with a low diversity of macronarians, to being exclusively composed of macronarians during the Early Cretaceous. In particular, the Morrison Formation is dominated by diplodocids, which seem to have hit their global peak of diversity during the Kimmeridgian and Tithonian (including Tornieria and Australodocus in Africa; Remes 2006, 2007) and then swiftly disappeared. No definitive Cretaceous diplodocid is known from anywhere in the world, although other diplodocoid clades (Rebbachi-sauridae and Dicraeosauridae) persisted on southern continents (Salgado and Bonaparte 1991; Sereno et al. 1999, 2007; Rauhut et al. 2005). Upchurch and Mannion (2009) referred an isolated partial anterior caudal vertebra from the Early Cretaceous of China to Diplodocidae, but this referral has been questioned (Whitlock et al. 2011). Thus the clade Diplodocidae, generally considered successful, seems to have been limited in both time and space (Taylor 2006).

Second, per fauna and per quarry diversity of sauropods is much lower in the Lower Cretaceous than in the Morrison Formation. The Cedar Mountain Formation may eventually have as many as 10 named genera of sauropods, but these are spread across at least three members, each with their own distinct dinosaurian faunas, and no single fauna is yet known to contain more than three sauropod genera. Large quarries in the Morrison Formation often have more than one sauropod genus present, and the Marsh-Felch and Dry Mesa quarries each contain five sauropod genera±until recently, more than were known from the entire Lower Cretaceous of North America.

Third, whereas sauropods are the most abundant large vertebrates in the Morrison Formation, they are among the rarest elements in their respective faunas in the North American Lower Cretaceous. The “big six” Morrison Formation genera±Apatosaurus, Barosaurus, Brachiosaurus, Camarasaurus, Diplodocus, and Haplocanthosaurus—are all represented by more than 10 individuals, and Camarasaurus, Diplodocus, and Apatosaurus are represented by close to 100 or more (Foster 2003: table 5). In contrast, the Early Cretaceous sauropods known from the most specimens are the Dalton Wells titanosaurian (14), the CEU brachiosaur (6), Paluxysaurus (4), and Abydosaurus (4), of which the first two are currently undescribed. Most of the named taxa are known from only one or two individuals.

Summary.—Sauropods are both diverse and abundant in the Morrison Formation, multiple sauropod genera are often found together in a quarry, and these sauropods are mostly diplodocids and Camarasaurus. Representatives of other sauropod clades are sparse. In particular, macronarians are not diverse, although Camarasaurus is abundant, and titanosauriforms are represented only by Brachiosaurus. (Haplocanthosaurus may also be a macronarian, having been recovered within this clade by the phylogenetic analyses of Wilson and Sereno [1998] and Upchurch et al. [2004a], but as a non-diplodocimorph diplodocoid by Wilson [2002]. However, it is among the rarest of Morrison Formation sauropods.)

In contrast, during the Early Cretaceous of North America sauropods were relatively rare (at least compared to their Morrison Formation abundance). Although the total diversity of sauropods across all North American Early Cretaceous formations is high, the diversity within each fauna is low, and co-occurrence of multiple taxa at a single locality is the exception rather than the rule. All known Early Cretaceous North American sauropods are macronarians, and most are brachiosaurids or other basal titanosauriforms, although a few titanosaurians and one probable camarasaurid are also present. The implications of these differences for Mesozoic paleoecology, especially energy flow and nutrient cycling in terrestrial ecosystems, are only beginning to be explored (e.g., Farlow et al. 2010).

Earliest Cretaceous sauropods and the fate of Diplodocidae.—Understanding of the history and evolution of sauropods in the mid-Mesozoic has been impaired by the unavailability of rocks from the earliest Cretaceous in many parts of the world. Possibly for this reason, the fossil record of diplodocids is extremely limited, with all known definitive diplodocid genera worldwide having arisen in the Kimmeridgian—Tithonian (Taylor 2006: 137)—although non-diplodocid diplodocoids, notably rebbachisaurids, are known to have persisted well into the Cretaceous. However, it is also possible that, rather than becoming extinct at the end of the Jurassic, diplodocids persisted into the earliest Cretaceous and were only gradually replaced by the macronarian sauropod fauna that characterizes the Cedar Mountain Formation and other North American Lower Cretaceous units.

This idea can be investigated by searching for late-surviving diplodocids in the lowest parts of the Cedar Mountain Formation and in earliest Cretaceous strata outside North America. Until the recognition of the possible diplodocid Dinheirosaurus Bonaparte and Mateus, 1999 from the Late Jurassic of Portugal, no diplodocid genus had been named from outside North America, although the type species of Lourinhasaurus Dantas, Sanz, Silva, Ortega, Santos, and Cachão, 1998, “Apatosaurus” alenquerensis Lapparent and Zbyszewski, 1957, was considered by its describers to represent a diplodocid and the referred species “Barosaurus” africanus Fraas, 1908 was known from Tendaguru in Tanzania. The African “Barosaurus” material is now recognised as comprising two distinct new diplodocid genera, Tornieria Sternfeld, 1911 (Remes 2006) and Australodocus Remes, 2007, both in fact belonging to Diplodocinae, so the existence of Late Jurassic diplodocids is now well established outside North America, with representatives in both Europe and Africa. Both the Portuguese Lourinh Formation and the African Tendaguru Formation end at the Jurassic/Cretaceous boundary, but other latest-Jurassic formations in Portugal are conformably overlain by Lower Cretaceous strata correlative with the Wealden Supergroup of England. It is in these strata that Early Cretaceous diplodocids may most usefully be sought, and there are signs that diplodocids may indeed have been present in the Wealden: Taylor and Naish (2007: 1560) reported the presence of a large sauropod metacarpal from the Hastings Group of the Wealden which has been identified as diplodocid, and Naish and Martill (2001: 232–234) discussed other putative, though not definitive, Wealden diplodocid material. Thus it seems likely that diplodocids did indeed survive into the Cretaceous, at least in Europe and possibly also in North America, and that their apparent end-Jurassic extinction is actually an artifact produced by the lack of representative strata from the earliest Cretaceous.

Conclusions

Brontomerus mcintoshi is a new genus and species of macronarian sauropod from the Ruby Ranch Member of the Cedar Mountain Formation of Utah. The new taxon is represented by at least two individuals of different sizes, probably a juvenile and an adult. It is clearly separate from all previously known Cedar Mountain Formation sauropods, and is distinguished from all other sauropods by several unique characters of the ilium and the scapula. The new taxon is probably a fairly basal camarasauromorph, although resolution is poor due to the incompleteness of the material. The distinctive characters of the ilium (e.g., huge preacetabular blade, no postacetabular blade, very tall overall, transversely thin) probably have functional significance, but this is difficult to assess in the absence of other pelvic elements, femora and proximal caudals.

The improving record of Early Cretaceous sauropods in North America is extended by the new genus, so that generic-level diversity of sauropods in this epoch now approaches that of the Late Jurassic. The most striking differences between Late Jurassic and Early Cretaceous sauropods in North America is that the former are abundant and dominated by diplodocids, whereas the latter are comparatively scarce and dominated by macronarians. It is currently impossible to determine whether this shift happened suddenly, or gradually over many millions of years in the earliest Cretaceous. It is natural to assume that if the shift was sudden, it happened at the end of the Jurassic, but that is not necessarily the case. The timing and tempo of this faunal shift remain uncertain; future inferences will have to be based on improved understanding of global changes in conditions in the earliest Cretaceous, and careful analysis of faunal changes on neighbouring continents, especially Europe.