Anatomy, taphonomy, and phylogenetic implications of a new specimen of Eolambia caroljonesa (Dinosauria: Ornithopoda) from the Cedar Mountain Formation, Utah, USA

Background Eolambia caroljonesa is the most abundant dinosaur in the lower Cenomanian Mussentuchit Member of the Cedar Mountain Formation of Utah, and one of the most completely known non-hadrosaurid iguanodontians from North America. In addition to the large holotype and paratype partial skulls, copious remains of skeletally immature individuals, including three bonebeds, have been referred to E. caroljonesa. Nevertheless, aspects of the postcranial anatomy of this taxon, particularly the pelvic girdle, have remained ambiguous due to the lack of associated postcranial material of larger, more mature individuals. Methodology/Principal findings Here we describe a recently discovered associated partial postcranial skeleton of a large Eolambia caroljonesa. This specimen, FMNH PR 3847, provides new anatomical data regarding the vertebral column and pelvic girdle, supplementing previous diagnoses and descriptions of E. caroljonesa. A new phylogenetic analysis incorporating information from FMNH PR 3847 places E. caroljonesa as a basal hadrosauromorph closely related to Protohadros byrdi from the Cenomanian Woodbine Formation of Texas. Histological analysis of FMNH PR 3847 reveals that it represents a subadult individual eight to nine years of age. Taphonomic analysis indicates that FMNH PR 3847 was preserved in a crevasse splay deposit, along with an unusual abundance of small crocodylomorph material. Conclusions/Significance FMNH PR 3847 provides a wealth of new morphological data, adding to the anatomical and systematic characterization of Eolambia caroljonesa, and histological data, revealing new information on growth history in a basal hadrosauromorph. Taphonomic characterization of FMNH PR 3847 and associated vertebrate material will allow comparison with other vertebrate localities in the Mussentuchit Member of the Cedar Mountain Formation.


Methodology/Principal findings
Here we describe a recently discovered associated partial postcranial skeleton of a large Eolambia caroljonesa. This specimen, FMNH PR 3847, provides new anatomical data regarding the vertebral column and pelvic girdle, supplementing previous diagnoses and descriptions of E. caroljonesa. A new phylogenetic analysis incorporating information from FMNH PR 3847 places E. caroljonesa as a basal hadrosauromorph closely related to Protohadros byrdi from the Cenomanian Woodbine Formation of Texas. Histological analysis of FMNH PR 3847 reveals that it represents a subadult individual eight to nine years of age. Taphonomic analysis indicates that FMNH PR 3847 was preserved in a crevasse splay deposit, along with an unusual abundance of small crocodylomorph material.

Conclusions/Significance
FMNH PR 3847 provides a wealth of new morphological data, adding to the anatomical and systematic characterization of Eolambia caroljonesa, and histological data, revealing new information on growth history in a basal hadrosauromorph. Taphonomic characterization of PLOS

Histological analysis
Histological age determination on hadrosaurids is typically performed on limb bones [33,34,35,36] as these appear to preserve the largest number of growth markers in specimens for which multiple elements have been examined [33]. However, as no limb bones were preserved with FMNH PR 3847, we selected a rib for sectioning because ribs have been shown to preserve a good growth record in sauropod [37] and theropod dinosaurs [38]. A small section (~1 cm) was taken from an incomplete caudal dorsal rib from the left side, identified as such by having a tuberculum that is developed as a small saddle shaped depression rather than projecting from the rib shaft and by the nearly continuous arc between the capitulum and rib shaft. The section was taken from the proximal end of the shaft approximately 8 cm below the tuberculum, as Waskow and Sander [37] found this region to preserve more growth markers than more distal rib sections. After cutting, the section was embedded in polyester resin for thin sectioning. The section was adhered to Plexiglass slides using epoxy resin after which thin sections were cut at a thickness of~0.5 mm. All cutting was carried out on a Buehler Isomet 4000 Linear Precision Saw. Sections were then ground and polished on a Buehler Ecomet 5 using 600/1200 gauge grit paper until a desired thickness was achieved allowing for microscopic examination and photography. The sections were finally buffed using 0.3μm Alumina Microspolish II (Buehler, IL) after the desired thickness was achieved.
The sections were examined and photographed at 4X magnification using an Olympus Bx 60 microscope mounted with an Olympus DP25 digital camera. Images were captured under regular and polarized light using CellSans software run on a Dell Dimension 8400 Pentium IV desktop computer running the Microsoft XP operating system. Individual photos were stitched together in the PT Gui photo stitching software (New House Internet Services BV, Rotterdam) ver. 10.0.15 on an HP 7500 terminal running Windows 7. Lens settings were set to rectilinear lens with a near infinite focal length (100000 mm) as we assumed minimal to no parallax in the micrographs. The stitched panorama and individual polarized micrographs were opened in Adobe Photoshop CS 5 software for cropping and labeling. The polarized light micrograph was sharpened using the sharpening filter to overcome slight differences in focus due to uneven thickness between the middle and peripheral parts of the thin section.
Distribution and Horizon. All specimens of Eolambia caroljonesa have been found in Emery County, Utah, in the Mussentuchit Member, Cedar Mountain Formation (lower Cenomanian) [17,18,19,20]. More precise locality data are on file at CEUM, OMNH, and FMNH.

Description of FMNH PR 3847
Measurements of FMNH PR 3847 are given in the online Supporting Information (S1 Table). The approximate positions of the disarticulated but associated cervical, dorsal, and caudal vertebrae of FMNH PR 3847 were derived through comparison with iguanodontians for which complete, articulated vertebral columns are known, especially Iguanodon bernissartensis [61] and Mantellisaurus atherfieldensis [62]. Comparison also was made with the disarticulated Eolambia bonebed material described by McDonald et al. [20].
Cervical vertebrae. FMNH PR 3847 includes the neural arch of a middle cervical vertebra, the complete right half of a caudal cervical vertebra, the left half of the neural arch of another caudal cervical vertebra, and an eroded opisthocoelous centrum of uncertain position in the cervical column (Fig 1).  [62] as a template. Numbers correspond to table of measurements (S1 Table). Caudal cervical vertebra (4), caudal dorsal neural arch (7), right dorsal rib (13), two proximal caudal centra (17 and 18), left pubis (27, medial view), and right ischium (28) are reversed. Sacral ribs and more recently prepared dorsal vertebrae are not shown. Scale bar equals 1 meter.
The middle cervical neural arch is complete except for the neural spine. The neural canal is elliptical, with the long axis oriented mediolaterally, although there appears to be some distortion (Fig 2A and 2B). The prezygapophyses are pedestals that project dorsolaterally from the cranial margin of the neural arch, terminating in subcircular, dorsomedially-facing articular surfaces (Fig 2A, 2C and 2D). The blunt, rod-like diapophyses are caudoventral to the prezygapophyses, on the lateral surfaces of the neural arch, and project laterally (Fig 2A-2C). Caudal to the prezygapophyses, the dorsal surface of the neural arch slopes gently dorsally towards the subtriangular base of the neural spine (Fig 2C and 2D). Caudal to the base of the neural spine, the neural arch splits into two elongate postzygapophyses that project caudolaterally and curve slightly ventrally along their lengths; each postzygapophysis terminates in an ovoid, ventrolaterally-facing articular surface (Fig 2B-2E).
The centrum of the caudal cervical vertebra is strongly opisthocoelous, with the cranial surface forming a hemispherical bulge and the caudal surface a deep concavity (Fig 2F-2H); this morphology is present in many styracosternans, including Hippodraco [4], Lurdusaurus [79], Lanzhousaurus [80], Barilium [81], Hypselospinus [58], Iguanodon [61], Mantellisaurus [62], Ouranosaurus [59], Jinzhousaurus [82], Bolong [63], Equijubus [68], Probactrosaurus [65], Jeyawati [22], Bactrosaurus [72], Gilmoreosaurus [74], Tanius [77], Huehuecanauhtlus [23], Yunganglong [15], Claosaurus [83], and hadrosaurids [7]. The right lateral surface of the centrum is damaged and the left is missing, precluding description of the parapophyses. In contrast, the right side of the neural arch is complete and well preserved. As in the middle cervical neural arch described above, the neural canal appears to have been elliptical (Fig 2G and 2H). Rather than forming a distinct pedestal as in the middle cervical neural arch, the prezygapophysis of the caudal cervical is a cranially-convex ledge on the craniodorsal surface of a large transverse process (Fig 2F-2I). The rounded diapophysis is at the end of the transverse process (Fig 2F-2I). The postzygapophysis of the caudal cervical projects caudolaterally and terminates in a ventrolaterally-facing articular surface, similar to the postzygapophyses of the middle cervical neural arch (Fig 2F-2I). However, the postzygapophysis of the caudal cervical lacks the slight ventral curvature of those of the middle cervical, instead projecting dorsally for its entire length. Furthermore, although the left side of the neural arch is missing, the postzygapophyses of the caudal cervical clearly would have been more divergent, with a greater angle between them in dorsal view, than those of the middle cervical neural arch.
Dorsal vertebrae and ribs. FMNH PR 3847 includes the nearly intact first dorsal vertebra, two nearly complete cranial dorsal vertebrae missing only the distal ends of their neural spines, the complete neural arches of two additional cranial dorsal vertebrae, the complete neural arch of a caudal dorsal vertebra, and the centra of five caudal dorsal vertebrae (Fig 1). The specimen also includes 11 dorsal ribs in various states of preservation.
The first dorsal vertebra (D1) was identified as such based upon its combination of an opisthocoelous centrum; parapophyses located on the lateral surfaces of the neural arch, rather than on the centrum; and small neural spine, as indicated by the size of the broken base ( Fig  3A-3D). In this combination of features, the D1 of FMNH PR 3847 compares closely with those of Iguanodon [61] and Mantellisaurus [62], for which articulated vertebral series are known. D1 is similar to the caudal cervical vertebra described above in the aforementioned opisthocoelous centrum, as well as prezygapophyses that form ledges on the dorsal surfaces of the large, laterally-projecting transverse processes and elongate, caudolaterally-directed, and widely divergent postzygapophyses (Fig 3A-3D).
The neural arch of another cranial dorsal vertebra probably represents either D3 or D4 based upon comparison with the dorsal series of Iguanodon [61] and Mantellisaurus [62]. The prezygapophyses are oblong, tab-like, and project cranially, with the articular surfaces steeply inclined ventromedially (Fig 3E, 3G and 3H). Each prezygapophysis is bordered caudomedially by a small but well-defined depression, and a low ridge separates these depressions along the midline. This ridge is better developed in more caudally positioned vertebrae, and is observed in the mid-dorsal and caudal dorsal vertebrae of Edmontosaurus [84]. The parapophyses are shallow, rugose, and roughly elliptical depressions on the lateral surfaces of the neural arch, immediately caudal to the prezygapophyses and cranial to the bases of the transverse processes (Fig 3G). They extend ventrally nearly to the neurocentral suture as in the cranial dorsals of Edmontosaurus [84], and unlike the other preserved dorsal vertebrae that all have parapophyses situated farther from the neurocentral suture. The transverse processes sweep New skeleton of Eolambia dorsolaterally in cranial and caudal views, and caudolaterally in dorsal view (Fig 3E-3H), terminating in rounded, rugose diapophyses ( Fig 3G). Each transverse process is composed of three laminae, one extending from the prezygapophysis to the diapophysis (Fig 3E), another from the postzygapophysis to the diapophysis (Fig 3F), and a third from near the neurocentral suture to the diapophysis ( Fig 3G); the dorsal surfaces of the transverse processes are flat ( Fig 3H). The neural spine arises between the transverse processes and, though incomplete along its dorsal and caudal margins, appears to have been subrectangular (Fig 3E-3H). The postzygapophyses are ventral to the caudal margin of the neural spine and form a pair of ventrolaterally-facing ledges with a quadrangular depression between them (Fig 3F). A rod-like fragment of ossified tendon is adhered by matrix to the left lateral surface of the neural spine ( Fig 3F).
A nearly complete vertebra with the centrum and arch in articulation likely represents the next element (D5?) following the arch described above. The centrum has a flat cranial intercentral articulation and a faintly concave caudal one, and lacks a ventral keel. The neural canal is elliptical cranially but circular in caudal view. The parapophyses are elongate with a dorsoventral long axis, but do not reach the neurocentral suture ventrally. The prezygapophyses are situated close to the midline and separated by a distinct ridge below the base of the neural spine. The neural spine has a broad base in lateral view, and is inclined caudally. Although it is missing its distal section, the spine is taller than its corresponding centrum. The transverse processes are less backswept than in the preceding neural arch, and separated from the postzygapophyses by a sharp but shallow notch in dorsal view. A thick, tear-drop shaped septum projects ventrally from the midline where the postzygapophyses meet. A shallow depression occupies the base of the neural spine between the postzygapophyses in caudal view.
Another vertebra bears a neural arch that is similar in anatomy to the preceding element, but has a wider base to its neural spine, a circular neural canal opening in cranial view and parapophyses even farther removed from the neurocentral suture. The transverse processes are shorter and wider in dorsal view, and are separated from the postzygapophyses by a distinct notch. The midline septum projecting ventrally below the postzygapophyses is well developed and tapers caudally in ventral view. These traits suggest a more caudal position for this vertebra, perhaps as D6 or D7. A centrum was found disarticulated from, but adjacent to and touching this arch. The centrum has a slightly convex cranial intercentral articulation and a weak ventral keel typical of more cranial dorsal vertebrae [84], and may not belong with the arch despite their close proximity.
A more caudally positioned neural arch exhibits parapophyses that do not extend below the level of the prezygapophyses, which in turn have less inclined articulations than those of the preceding vertebrae. The transverse processes are slender and less backswept than in more cranial elements and are not separated caudally from the poszygapophyses by a notch. The postzygapophyseal septum is robust in ventral view. The neural spine is caudally inclined and its base overhangs the arch peduncles by a wide margin. Comparison to the vertebral column of Iguanodon [61] suggests a position as the 10 th dorsal vertebra.
An isolated dorsal centrum with a narrow neural canal and a well developed keel appears to be from a similar position along the vertebral column, but taphonomic distortion of both the centrum and arch render a fit between the two uncertain. The cranial intercentral articulation is taller than wide and flat, whereas the caudal face is markedly wider as in the caudal dorsal vertebrae of Iguanodon [61].
Two neural arches of caudal dorsal vertebrae probably represent either D12 and D13, or D13 and D14, bearing a close resemblance to those vertebrae of Iguanodon [61] and Mantellisaurus [62] in the horizontal, laterally-directed transverse processes and positions of the parapophyses at the cranial margins of the bases of the transverse processes (Fig 4A-4E). The parapophyses themselves are laterally-protruding, subcircular structures, each forming a cup-like depression (Fig 4A, 4C and 4E). Both differ from the preceding dorsal vertebrae in that the articular surfaces of the prezygapophyses are nearly horizontal (Fig 4A), the transverse processes are simple flat structures lacking distinct laminae (Fig 4A-4E), the neural spines are displaced caudally beyond the arch peduncles (Fig 4C), and the postzygapophyses face ventrally rather than ventrolaterally (Fig 4B and 4D). There is a short section of a dorsal rib attached to the neural arch near the left neurocentral suture of the more caudal of the two arches (Fig 4B), identified as such by its smaller prezygapophyses.
Based upon their shapes and comparisons with Iguanodon [61] and Mantellisaurus [62], the five dorsal centra might represent D12-D16. One of these appears to fit reasonably well with the arch we tentatively identify as D12. All five centra are amphiplatyan with subcircular articular faces (Fig 5A-5E). In all five centra, the caudal articular face is transversely wider than the cranial face, making the centra spool-shaped in lateral and ventral views (Fig 5A-5E). The ventral margins of the centra are concave in lateral view and there is an offset between the cranial and caudal faces. The centrum identified as D15 retains a small portion of the neural arch; the only recognizable features are the prezygapophyses, which closely resemble those of the caudal dorsal neural arch (D12 or D13) described above, except that they are craniocaudally shorter and face dorsally (Fig 5D).
The proximal ends of the dorsal ribs bifurcate into the ventral capitulum and the dorsal tuberculum. The subrectangular capitulum is craniocaudally compressed; the proximal end of the capitulum, where it would have articulated with the parapophysis of a dorsal vertebra, is rounded and rugose (Fig 6). The subtriangular tuberculum is located at the base of the capitulum and projects dorsally (Fig 6). The dorsal ribs of FMNH PR 3847 closely resemble those of juvenile Eolambia [20], and those of other iguanodontians, such as Iguanodon [61], Mantellisaurus [62], and Jinzhousaurus [82]. Sacral ribs. FMNH PR 3847 includes four disarticulated sacral ribs in various states of preservation. The most complete is tentatively identified as the fifth sacral rib from the left side, based upon comparisons with the articulated sacra of Iguanodon [61], Mantellisaurus [62], and Bactrosaurus [72]. The medial surface of the rib forms a rugose, ventrally-expanded facet for articulation with the fifth and sixth sacral vertebrae (Fig 7A and 7B). Lateral to the articular surface, the rib expands craniocaudally (Fig 7C and 7D), yet becomes markedly thinner dorsoventrally, especially along the caudal margin (Fig 7A and 7B). The lateral margin of the rib is dorsally expanded and rugose, forming a facet at which the rib would articulate with the corresponding facet on the medial surface of the ilium (Fig 7A and 7B).
Caudal vertebrae and chevrons. FMNH PR 3847 includes the centra of five proximal caudal vertebrae, the neural spines of two proximal caudal vertebrae, two nearly complete middle caudal vertebrae, the centra of two additional middle caudal vertebrae, and two complete chevrons (Fig 1).
The cranial and caudal articular faces of a proximal caudal centrum are subcircular and gently concave (Fig 8A and 8B). The ventral surface of the centrum bears two craniocaudallydirected ridges that are laterally offset from the midline of the centrum ( Fig 8C); caudally, these ridges terminate in a pair of prominent facets for the articulation of a proximal chevron (Fig 8C and 8D). Between the ridges is a deep elliptical depression. The caudal ribs are dorsoventrally compressed and project caudolaterally from the dorsolateral surface of the centrum (Fig 8A-8D). The prezygapophyses project craniodorsally and face dorsomedially (Fig 8A, 8D and 8E). The postzygapophyses are located on the caudal margin of the base of the neural spine and face ventrolaterally (Fig 8E and 8F). The neural spine is straight and inclined caudally, with a rectangular distal end (Fig 8F). The middle caudal vertebrae are broadly similar to the proximal caudals, with slightly concave articular faces and prominent chevron facets on the centra, dorsomedially-facing prezygapophyses, and ventrolaterally-facing postzygapophyses on the base of the neural spine ( Fig  8G-8I). However, the middle caudal vertebrae differ from the more proximal members of the series in having hexagonal, rather than subcircular, articular faces on the centra; caudal ribs reduced to small, rounded nubs; more elongate prezygapophyses; and neural spines that curve dorsally, with gently concave cranial margins and convex caudal margins (Fig 8G-8I).
The proximal end of the proximal chevron is mediolaterally and craniocaudally expanded into two articulatar facets proximal to the haemal canal (Fig 8J). The distal portion of the chevron curves caudally and is rounded at its distal end (Fig 8K). A second, smaller chevron is similar in its anatomy, but is about two thirds the length of the more proximal element.
Ilium. FMNH PR 3847 includes the intact, undistorted left ilium (Fig 1). The right ilium is also preserved, but exhibits a break near the base of the preacetabular process as well as poorer preservation of periosteal surfaces, so our description will focus on the better preserved left element. The preacetabular process projects almost straight cranially, with little change in slope between the dorsal margin of the process and that of the body of the ilium dorsal to the pubic peduncle (Fig 9A and 9B). The distal end of the preacetabular process is ventrally expanded into a horizontal 'boot', as in many other styracosternans, including Iguanacolossus [4], Cedrorestes [27], Planicoxa [25], Osmakasaurus [26,73], Barilium [81], Iguanodon [61], Mantellisaurus [62], Ouranosaurus [59], Xuwulong [69], and Probactrosaurus [65]. The medial surface of the preacetabular process exhibits a mediolaterally-thickened shelf that becomes dorsoventrally deeper caudally, reaching its greatest depth dorsal to the pubic peduncle. Immediately cranial to this deepest point, the shelf bears a well-demarcated facet for the articulation of the first sacral rib (Fig 9B); the other sacral rib facets are indistinct. The shelf extends farther caudally along the body of the ilium, terminating near the caudal end of the medial surface of the postacetabular process and forming a narrow brevis fossa.
The subtriangular pubic peduncle and oval, ventrolaterally-facing ischial peduncle define a broad, shallow acetabulum ( Fig 9A). The dorsal margin of the body of the ilium exhibits a dorsally-directed flange extending from above the approximate midpoint of the acetabulum to a point on the dorsal margin of the postacetabular process caudal to the ischial peduncle (Fig 9A  and 9B). A similar flange is also present on ilia of skeletally immature individuals from the Eolambia #2 quarry in the CEUM collection, albeit subtler, extending only as far cranially as the ischial peduncle (Fig 9C and 9D). A similar dorsal flange also is present on a hadrosauroid ilium from the Lewisville Member of the Woodbine Formation [78]. The postacetabular process tapers towards its caudal end with no break in the slope of its dorsal margin, as in Iguanodon [61], Bactrosaurus [72], Gilmoreosaurus [74], and Claosaurus [83]. The length of the postacetabular process relative to the body of the ilium appears to increase through ontogeny (Fig 9A, 9C and 9D).

Taphonomy of FMNH PR 3847
Geology. Several authors provide detailed geologic descriptions of the Mussentuchit Member [19,88,89]. FMNH locality UT080821-1 is positioned approximately 30 meters from the base of the Cedar Mountain Formation, which in this outcrop includes only the Buckhorn Conglomerate, Ruby Ranch, and Mussentuchit members.
The locality resides in an approximately two-meter coarsening upward sequence starting approximately one meter below the base of the bone-bearing horizon with a freshwater limestone abruptly changing to a grey fine-grained mudstone. This layer is truncated by a coarsegrained, white sandstone containing large rip-up clasts of the underlying grey mudstone (Fig  10). The sandstone lens fines upwards to a sharp contact with another whitish-grey silt/mudstone. As is common throughout much of the Mussentuchit Member, the FMNH Locality UT080821-1 sandstone contains a large amount of clay, presumably from altered volcanic ash.
Microvertebrate assemblage. In addition to the partial Eolambia skeleton, hundreds of microfossils have been recovered from the site, almost exclusively referable to neosuchian crocodylomorphs. Teeth dominate this microvertebrate assemblage, although some postcranial and possible cranial elements from small neosuchian individuals are present. Generic identification of the crocodylomorph fauna from the Mussentuchit Member is still tenuous. Prior authors [19,28,90] hypothesized at least three groups of crocodilians based on tooth morphology: "atoposaurid", "globidontid/bernissartid" (two species of Bernissartia [19]), and "pholidosaurid". Recently though, Irmis [91] clarified that the general dental traits used to identify each of these groups could not reliably be used for taxonomic purposes. Therefore, we classify the crocodilians strictly as Neosuchia indet. (Fig 11).
The largest teeth in the sample are what have been considered pholidosaurid, being broad and round at the base with straight longitudinal ridges extending from the base to the tip ( Fig  11A and 11B). Another "pholidosaurid" tooth morphology from our sample is smooth-sided apart from two carinae on the lateral portions of the lingual surface (Fig 11C). The "globidontid" morph consists of ovoid teeth with minor pitting and ridges on the edges. Both of the latter two tooth morphotypes have constricted bases unlike the "pholidosaurid" form (Fig 11C and 11D). Finally, the "atoposaurid"-type teeth are labio-lingually compressed, triangular, and highly decorated with irregularly-spaced and non-linear longitudinal ridges (Fig 11F-11H). In many cases, such as FMNH PR 3844, the tips curve lingually.
One important aspect of this quarry is that almost all of the neosuchian material is small. A small neosuchian femur, FMNH PR 3028, is the largest non-tooth element recovered, at 53 were taken with a Keyence VCX-5000 digital microscope. Image in C was processed through the Keyence VCX-5000 High Dynamic Range filter prior to photographing in order to reduce glare and illuminate specimen detail. Scale bars in A and C equal 5 mm. Scale bars in B and D-H equal 1 mm.
https://doi.org/10.1371/journal.pone.0176896.g011 mm maximum length. As such, it seems reasonable that these elements might be referable to the same species. If this supposition is true, then some of the "atoposaurid" and "globidontid"type teeth might actually belong to the same neosuchian taxon. This observation is important because no complete crocodylomorph jaw elements with dentition have been described from the Mussentuchit Member. If true, this follows the observation of Irmis [91] that in other species these two morphotypes co-occur.
Non-crocodylomorph microvertebrate fossils recovered from FMNH Locality UT080821-1 include a single gar tooth (FMNH PR 15907; Fig 12A-12C) and theropod teeth provisionally referred to? Siats meekerorum [30] based on provenance and size of teeth from other nearby localities with the same serration densities (Fig 12D). The gar tooth possesses a spatulate morphology similar to species of Atractosteus [92] and other Late Cretaceous examples from Utah [93], although it differs in being lingually concave. It is unclear if the distal-most point is blunted because of breakage or if this morphology is inherent to the species.
Importantly, FMNH PR 15907 is one of the earliest definitive gar specimens in North America. Garrison et al. [19] figured plausible gar scales from the Cifelli-2 site, a locality near FMNH Locality UT080821-1, but scales are less diagnostic than teeth.
Taphonomic interpretation. Based on current geological evidence, we interpret FMNH Locality UT080821-1 as a crevasse splay burial of a partial Eolambia skeleton (FMNH PR 3847) on a floodplain. Proximity to the river is unknown, but given the size of the rip-up clasts we assume that the levee breakage was relatively close to the skeleton.
The provenance of the crocodylomorph material remains unclear. The tiny size of the preserved elements might indicate that the floodwaters excavated a buried nest, but this would not explain the disarticulation of elements prior to burial. Nonetheless, the preponderance of crocodylomorph bones compared to the paucity of other species would suggest at least a gathering of small crocodylomorph skeletons prior to the flood event, and not a random sampling of the local microvertebrate fauna.

Histological description and age estimation of FMNH PR 3847
The histological cross section of the caudal dorsal rib exhibits extensive secondary remodeling of the bone matrix. The medullary region is marked by numerous, irregular erosional lacunae, and the lateral side of the rib shaft exhibits almost complete replacement of the bone interior by Haversian systems, with some secondary osteons crosscutting the boundaries of prior secondary osteonal growth. The inner surface of the rib shaft is less remodeled preserving zonal bone with primary osteons and growth marks, particularly in the caudomedial region (Fig 13), although scattered secondary osteons are present. Five Lines of Arrested Growth (hereafter LAGs) are clearly visible in the widest preserved band of zonal bones (white arrows in Fig  13B); all are truncated, but the outermost are truncated by remodeling toward the lateral side of the rib. Traces of a probable sixth growth line are observed in the Haversian bone in the medullary region (grey arrow in Fig 13B), but have very little lateral extent so identification of this feature as an annual marker is less certain. Spacing between definitive LAGs decreases noticeably toward the periphery indicating that growth in the rib was slowing, but presence of primary osteons in the outermost zone suggest that an External Fundamental System indicative of growth cessation was not yet being formed when the individual died.
Observed growth markers provide a minimum age of five years for this individual, or six years if the probable growth marker in the medullary region is counted. All of these growth marks are located in the outer 50% of the bone cortex as measured along a transect extending between the long axis of the cross section and perpendicular to the tangent to the external bone surface above the section of the rib preserving the best growth record (Fig 13), so these ages clearly represent an underestimate. Sophisticated mathematical methods have been proposed for retrocalculation of growth marks lost to remodeling and medullary cavity expansion [35,94], but these require that growth marks can be traced along the full circumference of the bone, something that is not possible with the sectioned rib of FMNH PR 3847. Instead, we took a simplified approach to estimating the number of growth markers that have been erased by secondary remodeling by measuring the distance along the transect defined above from the long axis of the bone cross section to the innermost definitive LAG, and dividing by the greatest distance measured between observed LAGs along the same transect defined above (Fig 13). This rough calculation indicates 4.3 missing growth markers. Rounding down to account for the hatching size of the rib, we conservatively estimate that three or four growth markers were lost to secondary remodeling, suggesting an age of eight or nine years at death for this individual.
Histological analysis of the trunk rib shaft indicates that FMNH PR 3847 was still growing at the time of death, an observation that is consistent with other skeletochronological evidence such as the unfused and even fully open neurocentral sutures in the presacral column [95] and lack of fusion between sacral centra and ribs. The examined rib section exhibits decreasing spacing between LAGs toward the periphery, indicating a slow-down in rib growth. Without limb bones for FMNH PR 3847, it is difficult to ascertain whether the slow-down in rib growth mirrors patterns in faster growing limb elements. Horner et al. [33] found a greater possible number of annual lines in the EFS of a rib than in those of hind limb bones in a somatically mature specimen of Hypacrosaurus stebingeri, suggesting that ribs may cease growing before limb bones do in hadrosaurs.
FMNH PR 3847 is still a growing subadult with an estimated age of eight to nine years based on five observed and three to four retrocalculated growth markers. This is slightly older than the mean age for attainment of somatic maturity in Maiasaura [36], but still within the observed range of individual of growth trajectories for that taxon (Figure 2 in [36]), and also comparable to results published for a mature specimen of Hypacrosaurus stebingeri [33], which records five to six LAGs in rib sections, but seven to eight LAGs in the tibia prior to deposition of an EFS in each bone.
Most of these clades are geographically restricted and suggest that brief periods of endemism might have been a recurring phenomenon throughout styracosternan evolution. As in prior versions of this analysis, Hippodraco and Theiophytalia form a clade of basal styracosternans restricted to the Barremian-Albian? of western North America [4,98]. The analysis recovered two small basal hadrosauroid clades restricted to the Aptian (Jinzhousaurus and Bolong [5,63,82]) and Albian (Equijubus and Xuwulong [68,69,96]) of east-central Asia.
In contrast to these small, geographically and temporally restricted clades, a widespread and long-lived novel clade of basal hadrosauroids also was recovered by this analysis. This clade consists of taxa from North Africa (Ouranosaurus [59]), Europe (Morelladon [100] and Proa [41]), and Asia (Altirhinus [67], Batyrosaurus [14], Koshisaurus [45], and Sirindhorna [101]). Most members of this clade are of Early Cretaceous age (Barremian-Albian); however, if the Santonian age of Batyrosaurus proposed by Godefroit et al. [14] is accurate, then this clade persisted into the Late Cretaceous. There are several theropod clades that exhibit similar geographic and temporal distributions, with members found in the Early and Late Cretaceous of North Africa, Europe, and Asia, including Spinosauridae [102], Carcharodontosauridae [103,104,105], and Neovenatoridae [30,106,107].
It should be noted that Verdú et al. [108] recently and independently conducted a phylogenetic analysis using a prior version of the same basic data matrix employed herein. There are some noteworthy similarities between our results and those of Verdú et al. [108], such as a Late Cretaceous "Asian clade" including Shuangmiaosaurus, Zhanghenglong, and Plesiohadros (Fig 15). Verdú et al. [108] also recovered a Late Cretaceous "North American clade" that included Eolambia and Protohadros; however, their analysis placed Jeyawati in this clade, while our results position Jeyawati as a more derived hadrosauromorph (Fig 15). Finally, Verdú et al. [108] recovered a "Eurasian clade" similar to the clade of North African, European, and Asian taxa in our results. However, the clade found by Verdú et al. [108] excluded Ouranosaurus and Sirindhorna, but included Xuwulong and Gongpoquansaurus, which we find to be more derived hadrosauroids (Fig 15). The phylogenetic and biogeographic history of Styracosterna is clearly very complex and will require much additional analysis to refine.