Palaeobiodiversity and Palaeoenvironments

, Volume 93, Issue 1, pp 65–75

Glishades ericksoni’, an indeterminate juvenile hadrosaurid from the Two Medicine Formation of Montana: implications for hadrosauroid diversity in the latest Cretaceous (Campanian-Maastrichtian) of western North America

Authors

    • Department of Ecology and Evolutionary BiologyUniversity of Toronto
    • Department of Natural HistoryRoyal Ontario Museum
  • Kirstin S. Brink
    • Department of Ecology and Evolutionary BiologyUniversity of Toronto
  • Elizabeth A. Freedman
    • Museum of the Rockies and Department of Earth SciencesMontana State University
  • Christopher T. McGarrity
    • Department of Ecology and Evolutionary BiologyUniversity of Toronto
  • David C. Evans
    • Department of Natural HistoryRoyal Ontario Museum
Original Paper

DOI: 10.1007/s12549-012-0097-1

Cite this article as:
Campione, N.E., Brink, K.S., Freedman, E.A. et al. Palaeobio Palaeoenv (2013) 93: 65. doi:10.1007/s12549-012-0097-1

Abstract

Glishades ericksoni was named on the basis of partial paired premaxillae collected from the Late Campanian Two Medicine Formation of Montana, and was described as a non-hadrosaurid hadrosauroid. This interpretation of G. ericksoni has significant implications for hadrosauroid diversity and distribution because it represents the first occurrence of a non-hadrosaurid hadrosauroid in the Late Campanian of North America, and therefore implies either a prolonged period of sympatry between these forms and hadrosaurids or a biotic interchange with Asia. Given its small size, and therefore potential juvenile status, the taxonomic identity of G. ericksoni is re-evaluated here. Comparison with similarly-sized, taxonomically determinate, and coeval hadrosaurid specimens from the Two Medicine Formation (Prosaurolophus, Gryposaurus, and Maiasaura) suggest that the combination of characters used to distinguish G. ericksoni as a non-hadrosaurid hadrosauroid are more widely distributed or individually variable in hadrosaurids, or can be explained as the result of ontogenetic variation. In particular, the unique combination of characters used to diagnose G. ericksoni is also found in juvenile individuals of Prosaurolophus, Gryposaurus, and Maiasaura. Inclusion of juveniles of these taxa, scored on the basis of comparable anatomy, in the original phylogenetic analysis recovers the juvenile hadrosaurid specimens outside Hadrosauridae. Consequently, G. ericksoni cannot be confidently differentiated from a juvenile saurolophine, which are common in the upper and middle sections of the Two Medicine Formation, and is thus considered a nomen dubium. Given their absence in well-sampled Late Campanian and Maastrichtian deposits, non-hadrosaurid hadrosauroids appear to have been completely replaced by hadrosaurids in western North America by the Late Campanian.

Keywords

Hadrosauridae Hadrosauroidea Dinosaur systematics Dinosaur diversity Ontogeny

Introduction

Non-hadrosaurid hadrosauriforms (Sereno 1997) were abundant and diverse during the Early Cretaceous and the early stages of the Late Cretaceous. Their geographically widespread fossil record indicates that they attained a global distribution and represented an integral part of the dinosaur fauna of this time (Godefroit et al. 2005; McDonald et al. 2010; Norman 1980; Norman et al. 2004; Prieto-Márquez 2010b; Ramirez-Velasco et al. 2012; Sues and Averianov 2009; Taquet 1976; Wang and Xu 2001; You and Li 2009; You et al. 2011). In comparison, fossils of non-hadrosaurid hadrosauriforms are rare during the Campanian and Maastrichtian of Laurasia, where hadrosaurids dominate, with only a few but notable exceptions in Eurasia (Dalla Vecchia 2009; Evans et al. 2011a; Horner et al. 2004; Prieto-Márquez 2010b; Ryan and Evans 2005).

In North America, two possible occurrences of non-hadrosaurid hadrosauriforms have been noted in latest Cretaceous deposits, including a reported iguanodontian tooth from the Late Maastrichtian Scollard Formation of Alberta (Russell 1987) and the recently named Glishades ericksoni from the Late Campanian Two Medicine Formation of Montana (Prieto-Márquez 2010a). Recently, Evans et al. (2011a) noted that the Scollard tooth had been misidentified, and that it pertains to a ceratopsid. Glishades ericksoni therefore represents the only latest Cretaceous (Late Campanian–Maastrichtian) occurrence of a non-hadrosaurid hadrosauriform in North America. This occurrence therefore implies a geographic range extension of this grade of hadrosauroid into North America during the Late Campanian, and suggests the sympatry of hadrosaurids and non-hadrosaurid hadrosauriforms in North America at this time.

The holotype and only specimen of Glishades ericksoni (AMNH 27414) consists of a united pair of partial premaxillae, and is notable for its very small size (86 mm total width). The holotype was collected along the Milk River, 48 km (30 miles) west (though not necessarily due west) of Sweetgrass, Montana, Glacier County, in the Two Medicine Formation (Prieto-Márquez 2010a). This description is the approximate location of Landslide Butte, a series of exposures that generally correspond to the upper 60 m of the formation (Horner et al. 2001), likely dated between 74.5 and 74 million years old (Foreman et al. 2008; Rogers et al. 1993). Landslide Butte, along with other nearby localities, preserves abundant remains of the hadrosaurids Prosaurolophus and Hypacrosaurus, including juvenile material (Horner 1994; Horner and Currie 1994; Rogers 1990). It is worth noting, however, that the exact locality data for AMNH 27414 is not known, and therefore a definitive locality and horizon cannot be established with confidence.

Although the potential juvenile nature of AMNH 27414 was discussed by Prieto-Márquez (2010a), it was dismissed on the grounds that its combination of characters is not found in any other hadrosauroid and the individual characters did not change significantly through ontogeny in hadrosaurids. However, recent quantitative allometric analyses indicate important morphological changes in the snout of hadrosaurids through ontogeny that call into question the nature of characters used to diagnose Glishades (Evans 2010; Campione and Evans 2011). Given this new information, we hypothesise that Glishades is a juvenile hadrosaurid and that the incomplete nature of the holotype (partial premaxillae), and the tendency of juveniles to exhibit unique combinations of primitive and derived characters that are not present in adults due to ontogenetic and/or size-related variation, may be responsible for the non-hadrosaurid phylogenetic position of Glishades, as well as its putatively diagnostic combination of characters.

Due to the possible juvenile status of AMNH 27414 and the biogeographical and ecological implications of its occurrence in the Late Campanian of North America, we compare this taxon to premaxillae that pertain to juvenile individuals of three well-known saurolophine taxa, Prosaurolophus, Gryposaurus, and Maiasaura that occur in the Late Campanian of Montana. Using new information on the juvenile morphology of these taxa, we reassess the combination of characters used to diagnose G. ericksoni within a phylogenetic context, and discuss them within the context of ontogeny and variation in hadrosaurids.

Institutional abbreviations

AMNH

American Museum of Natural History, New York, New York, USA

CMN

Canadian Museum of Nature, Ottawa, Ontario, Canada

MOR

Museum of the Rockies, Bozeman, Montana, USA

ROM

Royal Ontario Museum, Toronto, Ontario, Canada

SM

Senckenberg Museum, Frankfurt, Germany

TMP

Royal Tyrrell Museum of Palaeontology, Drumheller, Alberta, Canada

USNM

National Museum of Natural History, Smithsonian Institute, Washington D.C., USA.

Materials

Anatomical comparisons presented in this study are based on the juvenile premaxilla of Gryposaurus notabilis (CMN 8784) and the subadult premaxillae of Prosaurolophus maximus (TMP 1983.064.0003) and Maiasaura peeblesorum (ROM 65035). The growth stages of these specimens are defined as per Brink et al. (2011).

CMN 8784 includes a complete articulated skull (skull length from the tip of the snout to the quadrate: 355 mm) from the Dinosaur Park Formation, Alberta, and was assigned to G. notabilis based on nasal morphology by Waldman (1969). This skull is less than half the length of largest known adult skull of this species (CMN 2278: 862 mm). TMP 1983.064.0003 was collected from the Bearpaw Formation in Alberta (Late Campanian; Eberth and Hamblin 1993) and comprises a nearly complete sub-adult skull that preserves a well-demarcated circumnarial excavation within a low crest on the nasal that allows it to be confidently assigned to Prosaurolophus. Based on measurements of the premaxillae (total paired width: 153 mm), TMP 1983.064.0003 is half the size of a large individual of Prosaurolophus (TMP 1984.001.0001 total paired width: 306.5 mm). Finally, ROM 65035 is the isolated rostral portion of a right premaxilla and was recovered from the Lister bonebed within the Two Medicine Formation, Montana, known to preserve numerous specimens of M. peeblesorum, and can therefore be confidently assigned to the latter (Schmitt et al. 1998). Based on the prenarial length of the premaxilla (87 mm), this specimen is approximately 60 % the size of a typical adult Maiasaura premaxilla (ROM 44770 prenarial length: 141 mm).

These three specimens are all moderately larger than Glishades ericksoni (Table 1), suggesting that AMNH 27414 originally pertained to a skull with a length approximately three-quarters that of CMN 8784, or roughly 260 mm.
Table 1

Select measurements (in mm) of the premaxilla indicating the relative size difference between specimens

Taxon

Specimen number

Prn.L

Prm.W

Prm.T.W

Glishades ericksoni

AMNH 27414

69

61

86

Gryposaurus notabilis

CMN 8784

82

Prosaurolophus maximus

TMP 1983.064.0003

69.1

100.1

153

Maiasaura peeblesorum

ROM 65035

87

128

181a

Prn.L prenarial length: from the rostral-most point of the naris to the tip of the snout; Prm.W premaxilla width: from the midline to the lateral-most region of the oral margin, perpendicular to the midline, along the dorsal surface; Prm.T.W premaxillary total width: straight width between the lateral-most portion of the right and left premaxilla

aOnly the right is preserved; straight width from the midline to the lateral-most portion of the premaxilla, measured along the ventral surface, multiplied by two

Anatomical description and comparisons

Glishades was diagnosed by a unique combination of premaxillary characters including “absence of recurvature (“reflection”) of oral margin; arcuate oral margin with wide and straight, obliquely oriented, and undeflected anterolateral corner; foramina on anteromedial surface above oral edge and adjacent to proximal end of narial bar; and grooved transversal thickening on ventral surface of premaxilla posterior to denticulate oral margin.” (Prieto-Márquez 2010a, p. 3). We therefore focus here on the morphology of these characters in the juvenile hadrosaurid specimens.

The general morphology of the premaxillae is similar in all four specimens (Fig. 1). The premaxillae are mediolaterally broad and arcuate along the oral margin in dorsal view. In rostral view, the oral margin is denticulate ventrally and it is dorsoventrally thickened near the midline and gradually thins laterally and caudally. In lateral view, the rostral margin of the premaxillae is deflected ventrally. This margin is vertical rostrally and gently slopes caudally to form the dorsal margin of the snout above the bony naris.
https://static-content.springer.com/image/art%3A10.1007%2Fs12549-012-0097-1/MediaObjects/12549_2012_97_Fig1_HTML.gif
Fig. 1

Premaxillae of Glishades ericksoni and juvenile/subadult hadrosaurid dinosaurs, in right lateral view (row 1), rostral view (row 2), dorsal view (row 3), and ventral view (row 4). a, e, i, m AMNH 27414, holotype of Glishades ericksoni (left premaxilla reflected). b, f, j, n ROM 65035, Maiasaura peeblesorum. c, g, k, o TMP 1983.064.03, Prosaurolophus maximus. d, h, l, p CMN 8784, Gryposaurus notabilis (left premaxilla reflected). pg peripheral groove, pmf premaxillary formamen, pmf’ accessory premaxillary foramina, pmt premaxillary trough, rpmf rostral premaxillary foramina. Scale bars 2 cm

Small foramina are present in the three hadrosaurid specimens along the rostral surface of the oral margin adjacent to the proximal end of the narial bar (Fig. 1e–h), similar to that described in AMNH 27414 (Prieto-Márquez 2010a), as well as in Bactrosaurus (Prieto-Márquez 2011). The ventral aspect of the oral margin in all four specimens (Fig. 1m–p) is similarly denticulate with a deep peripheral groove that results in the ‘double-layer’ morphology present in all hadrosaurids and some non-hadrosaurid hadrosauroids (Horner et al. 2004; Prieto-Márquez 2010a). The oral margin is not strongly reflected dorsally in any of the specimens. This is in contrast to the morphology seen in larger specimens of Prosaurolophus (e.g. USNM 12712) and Gryposaurus (e.g. CMN 2278), which exhibit a definite dorsal reflection of the premaxillary margin (Fig. 2a, b). A reflected premaxillary margin is not present at any growth stage in Maiasaura, as the morphology of ROM 65035 is similar to that of a much larger individuals (e.g. ROM 44770; Horner 1983; Fig. 2c), nor is it present in Brachylophosaurus (e.g. CMN 8893; Cuthbertson and Holmes 2010), lambeosaurines (e.g. Hypacrosaurus stebingeri, MOR 549), and non-hadrosaurid hadrosauroids (e.g. Head 1998; Prieto-Márquez 2011).
https://static-content.springer.com/image/art%3A10.1007%2Fs12549-012-0097-1/MediaObjects/12549_2012_97_Fig2_HTML.gif
Fig. 2

Left lateral view of the premaxilla in adult saurolophine specimens. a CMN 2870, Prosaurolophus maximus. b CMN 2278, Gryposaurus notabilis. c ROM 44770, Maiasaura peeblesorum (right premaxilla reflected). rpm reflected margin of premaxilla. Scale bars 10 cm

In dorsal view, the lateral oral margins of TMP 1983.064.0003 (Fig. 1k) and CMN 8784 (Fig. 1l) are not sharply deflected laterally and they gradually narrow caudally to form the post-oral constriction. In comparison, AMNH 27414 (Fig. 1i) and ROM 65035 (Fig. 1j) exhibit an abrupt constriction caudal to the oral margin. As described by Prieto-Márquez (2010a), these differences are also clear in adult skulls of these and other hadrosaurid taxa. A gradual constriction is typical of Prosaurolophus (e.g. USNM 12712), Saurolophus (e.g. AMNH 5221; Bell 2010), and Gryposaurus (e.g. CMN 2278), whereas an abrupt constriction caudal to the oral margin is seen in Maiasaura (ROM 44770) and Brachylophosaurus (e.g. CMN 8893; Cuthbertson and Holmes 2010), as well as in lambeosaurines (e.g. Hypacrosaurus stebingeri, MOR 549) and non-hadrosaurid hadrosauroids (e.g. Head 1998; Prieto-Márquez 2011).

In all specimens, the naris is only partially preserved along its rostral and ventral margins, but would have been elongate, with its long axis parallel to the dorsoventral premaxillary process. Rostral to the naris, all specimens possess a premaxillary foramen, typical of hadrosaurids (Fig. 1i–l). Although the premaxillary foramen was described as definitively absent by Prieto-Márquez (2010a), our inspection of the specimen indicates that this region is damaged in AMNH 27414. However, a trough is present at the base of the caudodorsal process (Fig. 1a) and is equivalent to a trough in ROM 65035 (Fig. 1b), which leads to the premaxillary foramen. We therefore interpret this morphology to indicate that the premaxillary foramen is present in AMNH 27414. An accessory premaxillary foramen, rostral to the premaxillary foramen, is present in the subadult Maiasaura (ROM 65035; Fig. 1j), and in the larger individual (ROM 44770). The accessory foramen is also present in juvenile incomplete premaxillae of Maiasaura (MOR 547 W-57-1 and MOR 547 W), suggesting that its presence is not ontogenetically variable. Comparisons with other taxa indicate that this foramen is also present in both species of Edmontosaurus (USNM 12711: E. regalis; SM R4050: E. annectens), as well as in Brachylophosaurus canadensis (MOR 1071-7-23-99-179), and in the newly named Acristavus gagslarsoni (MOR 1155; Gates et al. 2011). Despite its presence in these taxa, the coding of this character appears to be highly variable in recent analyses, with only Edmontosaurus consistently coded as possessing an accessory premaxillary foramen, whereas it is generally coded as absent in Maiasaura, Brachylophosaurus, and Acristavus (Gates et al. 2011; Horner et al. 2004; Prieto-Márquez 2010c). Prieto-Márquez (2010a) codes this foramen as absent in G. ericksoni; however, as described earlier, the holotype is not well preserved in this region, and, as a result, we interpret the presence or absence of an accessory premaxillary foramen in AMNH 27414 as equivocal.

Phylogenetic analysis

Methods

A phylogenetic analysis of Glishades ericksoni, as presented in the strict consensus of Prieto-Márquez (2010a), recovered Glishades ericksoni (AMNH 27414) in a polytomy (with Telmatosaurus and various non-hadrosaurid hadrosauroids) outside Hadrosauridae (as defined by Prieto-Márquez 2010c). However, given its possible juvenile status it is important to consider the potential effects and limitations that ontogenetic variation may introduce to a phylogenetic analysis (Butler and Zhao 2009; Evans et al. 2011b; Fowler et al. 2011; Tsuihiji et al. 2011). We hypothesize that the incomplete nature of the specimen, and the tendency of juveniles to exhibit unique combinations of primitive and derived characters that are not present in adults due to ontogenetic and/or size-related variation, may be responsible for the phylogenetic placement of Glishades outside Hadrosauridae. As the ontogenetic age of AMNH 27414 is difficult to determine without destructive histological sampling, we coded the premaxillae of a known juvenile Gryposaurus (CMN 8784), a subadult Prosaurolophus (TMP 1983.064.03), and a subadult Maiasaura (ROM 65035) as described above into the matrix used in Prieto-Márquez (2010c) and modified by Prieto-Márquez (2010a). We predict that, given their ontogenetic status, these juvenile saurolophines will fall outside Hadrosauridae in a similar phylogenetic position as the holotype of Glishades ericksoni. The analysis was run under a heuristic search (20,000 trees saved) in TNT v.1.1 (Goloboff et al. 2008). Coding for all specimens can be found in Table 2.
Table 2

Character codings for the immature hadrosaurid specimens discussed in this study, as well as the original and modified codings of Glishades ericksoni. Character numbers correspond to those described in Prieto-Márquez (2010c); all other characters were coded as ‘?’ in the matrix

Taxon

Sp. no.

62

63

64

65

66

67

287

Glishades ericksoni a

AMNH 27414

0

1

0

0

0

0

1

Glishades ericksoni b

AMNH 27414

0

1

1

?

0

0

1

Gryposaurus notabilis

CMN 8784

0

1

1

0

0

0

1

Prosaurolophus maximus

TMP 1983.064.0003

0

1

1

0

0

0

1

Maiasaura peeblesorum

ROM 65035

0

1

1

1

0

0

1

aOriginal coding of AMNH 27414 used in Prieto-Márquez (2010a)

bModified coding based on reinterpretation in this study

Results

The analysis resulted in 10,752 most parsimonious trees (Fig. 3), with a tree length of 980, a consistency index of 0.438, a retention index of 0.707, and rescaled consistency index of 0.31. Given the high amount of homoplasy within this dataset, a strict consensus reveals little resolution, especially outside Hadrosauridae (Fig. 3). Nevertheless, the monophyly of Hadrosauridae (as defined by Prieto-Márquez 2010c) is supported, along with the main hadrosaurid sub-clades, excluding Hadrosaurus foulki (as in Prieto-Márquez 2010c). All three juvenile hadrosaurid premaxillae (TMP 1983.064.03, CMN 8784, and ROM 65035) fall well outside Hadrosauridae in the strict consensus tree. A 50 % majority consensus indicates that the juvenile hadrosaurids cluster together, and form a polytomy with Glishades ericksoni. In this consensus, the well-known hadrosauroid Bactrosaurus johnsoni is recovered as the sister-taxon to this group. The clade that includes Bactrosaurus, Glishades, and the juvenile hadrosaurid premaxillae occurs in a polytomy with Nanyangosaurus zhugeii and a large clade that includes Gilmoreosaurus and Hadrosauridae. Most importantly, the analyses clearly posits that TMP 1983.064.0003, CMN 8784, and ROM 65035 are not hadrosaurids, despite the unequivocal taxonomic assignment of these specimens to Prosaurolophus maximus, Gryposaurus notabilis, and Maiasaura peeblesorum, respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs12549-012-0097-1/MediaObjects/12549_2012_97_Fig3_HTML.gif
Fig. 3

Phylogenetic analysis based on the matrix of Prieto-Márquez (2010c) and modified by Prieto-Márquez (2010a). New inclusions to the analysis include TMP 1983.064.03, a subadult Prosaurolophus maximus; CMN 8784, a juvenile Gryposaurus notabilis; and ROM 65035, a subadult Maiasaura peeblesorum (codings shown in Table 2). The analysis resulted in 10,752 most parsimonious trees with a tree length of 980, a consistency index of 0.438, a retention index of 0.707, and rescaled consistency index of 0.31. The strict consensus is shown on the left and the 50 % majority on the right. Colours and shapes indicate specimens from the same taxon; open shapes represent the adult individuals and the solid shapes indicate the juvenile or subadult specimens

Discussion

Glishades ericksoni was named and diagnosed on the basis of a unique combination of primitive and derived characters: (1) the absence of a reflected margin of the premaxilla; (2) the presence of foramina on the rostral surface of the premaxilla; (3) a groove on the ventral surface of the premaxilla caudal to the oral margin, which results in a “double-layer” morphology; and (4) a straight and wide oral margin, which is undeflected along the rostrolateral corner. Based on this combination of characters, it was hypothesized to have a systematic position outside of Hadrosauridae (Prieto-Márquez 2010a).

Although the small size of AMNH 27414 is suggestive of juvenile status, Prieto-Márquez (2010a, 2011) argued that these characters do not vary ontogenetically in hadrosauroids, and thus could be confidently coded in a phylogenetic analysis and used as evidence of the unique taxonomic identity of Glishades. However, as we show above, the vast majority of the characters used to diagnose G. ericksoni are present in a similar combination in immature specimens of Prosaurolophus, Gryposaurus, and Maiasaura. As a result, the differences in morphology between G. ericksoni and the juvenile specimens described here can be explained by ontogenetic allometry, as well as by individual variation.

Rostral premaxillary foramina

Prieto-Márquez (2010a) describes the presence of foramina along the rostral surface of the premaxilla as rarely occurring in hadrosauroids, being present in only a few, mostly subadult, saurolophine specimens, and in G. ericksoni. These foramina likely represent osteological correlates for the medial branch of the ophthalmic nerve (V1) and associated vasculature that innervate and supply the integument associated with the oral margin in archosaurs (Witmer 1995). Evans (2006) and Evans et al. (2009) identified these foramina in lambeosaurines, suggesting that they are likely more prevalent in hadrosauroids than considered by Prieto-Márquez (2010a). Examination of ROM, CMN, TMP and MOR specimens listed by Prieto-Márquez (2010a; Table 1) reveals a number of instances where specimens were miscoded with respect to the presence or absence of foramina in this region. Numerous specimens that preserve these rostral premaxillary foramina were coded as absent, including: ROM 845, Corythosaurus intermedius; ROM 801, Edmontosaurus regalis; ROM 873, Gryposaurus notabilis; MOR 553 S-7-18-91-107 and MOR 553 L-7-23-8-57, Gryposaurus sp.; and MOR 447-7-8-7-9 (labeled MOR 454-7-8-2-9 in Prieto-Márquez 2010a) and ROM 787, Prosaurolophus maximus (Fig. 4). In addition, several specimens listed by Prieto-Márquez (2010a; Table 1) are either poorly preserved or reconstructed in this region, yet were coded as definitively having a character state which, based on our interpretation, cannot be accurately assessed. In total, of the 20 specimens listed by Prieto-Márquez (2010a; Table 1) that were re-examined here, 65 % (13 specimens) are considered by us to be miscoded.
https://static-content.springer.com/image/art%3A10.1007%2Fs12549-012-0097-1/MediaObjects/12549_2012_97_Fig4_HTML.gif
Fig. 4

Reassessment of character 287 of Prieto-Marquez (2010a). a ROM 782, Lambeosaurinae indeterminate, displays well-preserved rostral premaxillary foramina in a lambeosaurine. b–e Examples of specimens erroneously listed as not having foramina present on the rostral surface of the premaxilla next to the midline. Foramina are indicated by arrows. b ROM 845, Corythosaurus casuarius. c ROM 787, Prosaurolophus maximus. d ROM 873, Gryposaurus notabilis. e MOR 553 L-7-23-8-57, Gryposaurus sp. Scale bars 2 cm

The presence or absence of foramina along the rostral surface of the premaxilla near the midline appears to be widespread in hadrosaurids. Their presence may even be individually variable in certain taxa (Prieto-Márquez 2010a), such as Edmontosaurus annectens, in which they are clearly present in CMN 8509 and appear absent in SM R4050, and similarly in Brachylophosaurus canadensis (CMN 8893, present; MOR 1071-7-23-99-179, absent), and Corythosaurus intermedius (ROM 845, present; CMN 8704, absent). As a result, this character appears to be of limited taxonomic and phylogenetic value until a more complete survey of this feature is conducted, and cannot be used to diagnose G. ericksoni.

Premaxillary oral margin

Hadrosaurids have often been diagnosed by a transverse or peripheral groove on the ventral surface of the premaxilla, caudal to the denticulate rostral margin. This morphology is commonly referred to as the “double-layer” structure (Horner et al. 2004; Sues and Averianov 2009). This morphology is also present in Bactrosaurus johnsoni, Telmatosaurus, all hadrosaurids, and may be present in even earlier non-hadrosauroid forms, such as Ouranosaurus nigerensis (Taquet 1976). Given the widespread occurrence of this character, its presence in AMNH 27414 may be plesiomorphic for the clade of hadrosauroids that includes Glishades (as suggested by our phylogenetic results), and it is therefore not taxonomically informative.

Ventral deflection of the oral margin

The oral margin of AMNH 27414 is ventrally deflected along its rostral portion as well as its lateral portion. In comparison, the rostrolateral region of the oral margin is not ventrally deflected. This morphology was considered as distinct from Maiasaura, Brachylophosaurus, Edmontosaurus, and lambeosaurines, which exhibit a ventral deflection along the entire oral margin (Prieto-Márquez 2010a). Examination of the oral margin of a subadult Maiasaura (ROM 65035; Fig. 1b) reveals an intermediate morphology between that shown by AMNH 27414 and a large Maiasaura exemplar (ROM 44770; Fig. 2c). The rostrolateral portion of the oral margin in ROM 65035 is slightly ventrally deflected compared to the rostral and lateral portions. This variation in Maiasaura indicates that the development of the ventral deflection may change through ontogeny or simply be individually variable. As a result, this character is also of limited taxonomic value with reference to the possible juvenile status of AMNH 27414.

Reflected margin of the premaxilla

Finally, Prieto-Márquez (2010a) noted that exclusion of G. ericksoni from Hadrosauridae was unambiguously supported by the lack of a reflected oral margin of the premaxilla in AMNH 27414. In many saurolophine species, the oral margin of the premaxilla is reflected caudodorsally, which results in a ‘lip’-like appearance. This morphology is most notable in Edmontosaurus (Campione and Evans 2011), but is also present in Gryposaurus, Prosaurolophus, and Saurolophus. This feature is not present in the holotype specimen of Glishades ericksoni and was therefore considered distinct from these taxa (Prieto-Márquez 2010a). Our examination of juvenile/sub-adult Gryposaurus and Prosaurolophus (Fig. 1) reveals only a very weakly developed dorsal reflection of the lateral premaxillary margin compared to larger individuals of these taxa (Fig. 2a, b). This indicates that the distinctive reflected margin of the premaxilla is poorly developed in juveniles of Prosaurolophus and Gryposaurus, and it may be absent in individuals of comparable size to the Glishades holotype. Campione and Evans (2011) note similar ontogenetic changes in the reflected margin of Edmontosaurus. Therefore, the small size of AMNH 27414 and its unreflected premaxillary margin can be interpreted as an ontogenetic character, and the ontogenetic stage will affect its coding in the matrix of Prieto-Marquez (2010c). Importantly, some saurolophine hadrosaurids including Maiasaura and Brachylophosaurus from Montana lack reflection of the oral margin of the premaxilla at all ontogenetic stages.

Phylogenetic reassessment of AMNH 27414

In its original analysis, AMNH 27414 was recovered in a polytomy outside Hadrosauridae, and therefore G. ericksoni was interpreted as a non-hadrosaurid hadrosauroid. However, several studies have recognised that some phylogenetically informative characters can also vary ontogenetically and, as a result, can have a significant effect on the phylogenetic affinities of juvenile specimens when they are included in these analyses (Butler and Zhao 2009; Evans et al. 2011b; Fowler et al. 2011). This issue was most recently discussed in a description of a juvenile tyrannosaurid, Tarbosaurus baatar, where it was shown that the juvenile specimen, when coded into a phylogenetic matrix, was recovered near the base of Tyrannosauridae (Tsuihiji et al. 2011). We present a very similar situation here: inclusion of juvenile and subadult premaxillae of common Late Campanian hadrosaurids in the same phylogenetic analysis used to distinguish Glishades revealed that equivalently sized hadrosaurids, with a presumed similar ontogenetic age, were recovered outside Hadrosauridae (Fig. 3). Therefore, we confirm our prediction that incompletely known juvenile hadrosaurid taxa known only from isolated premaxillae will have a tendency to appear primitive with respect to adult exemplars in Saurolophinae. Since G. ericksoni cannot be distinguished from three common saurolophine taxa of equivalent size, we cast serious doubt on the putative non-hadrosaurid phylogenetic position of G. ericksoni.

The polyphyletic relationship between juvenile and subadult saurolophines recovered in our analysis underscores the importance of understanding the ontogenetic status of a particular fossil specimen, as comparisons between exemplars at different life-stages, or ‘semaphoronts’ as defined by Hennig (1966), can have a significant effect on the results of a phylogenetic analysis. For example, this concept is often considered in studies of fossil amphibians (e.g. Steyer 2000; Fröbisch et al. 2010), where ontogenetic series are used to inform on phylogenetic relationships. Given the results presented in this study, as well as in others (Butler and Zhao 2009; Evans et al. 2011b; Fowler et al. 2011; Tsuihiji et al. 2011), it is evident that phylogenetic inferences using semaphoronts preserved at different stages of maturation can significantly affect studies of non-avian dinosaur systematics, especially since numerous taxa have been named on the basis of immature holotype specimens (Bakker et al. 2006; Bonaparte and Vince 1979; Burnham et al. 2000; Dodson 1986; Gates et al. 2007; Godefroit et al. 2004; Suzuki et al. 2004).

Concluding remarks

With the exception of Glishades ericksoni, the Two Medicine Formation preserves a typical Late Cretaceous dinosaur assemblage from western North America, including a diversity of ceratopsians (Sampson 1995), hadrosaurids (Horner 1983; Horner et al. 1992), troodontids (Varricchio et al. 2002), other theropods, pachycephalosaurids, and ankylosaurs (Horner et al. 1992, 2001; Trexler 2001). Most importantly, the Two Medicine Formation is known for the preservation of rare juvenile ontogenetic stages of numerous taxa, notably hadrosaurids (Brink et al. 2011; Dilkes 1993; Horner and Currie 1994; Rogers 1990; Varricchio and Horner 1993). Given the possible, if not probable, juvenile nature of AMNH 27414, and its questionable phylogenetic affinities, we find no compelling evidence to support the hypothesis that AMNH 27414 represents a non-hadrosaurid hadrosauroid. Furthermore, our analysis shows that the holotype of G. ericksoni cannot be confidently distinguished from juveniles of previously known saurolophine hadrosaurids. It is virtually identical to the sub-adult Maiasaura premaxilla (ROM 65035); however, its provenance in the Two Medicine Formation, though uncertain, suggests a younger age than the known Maiasaura localities (Horner et al. 2001). Penecontemporaneous taxa may include Prosaurolophus and Gryposaurus (Varricchio and Horner 1993), the subadults and juveniles of which are also morphologically very similar to AMNH 27414. We therefore suggest that the unusual combination of characters in AMNH 27414 is better explained as an early ontogenetic stage of an indeterminate saurolophine hadrosaurid, rather than invoking the presence of a Late Campanian non-hadrosaurid hadrosauroid in the Two Medicine assemblage.

Given the incomplete nature of AMNH 27414, assignment to any of these hadrosaurids is equivocal at this time; therefore, we refer this specimen to an indeterminate juvenile saurolophine (=hadrosaurine), and we consider Glishades ericksoni a nomen dubium. Based on these conclusions, and those presented by Evans et al. (2011a), there is no unequivocal evidence to indicate that non-hadrosaurid iguanodontians were present during the Late Campanian and Maastrichtian of western North America. Furthermore, their absence in the well-sampled Late Campanian and Maastrichtian deposits indicates that these non-hadrosaurid taxa were likely completely replaced by hadrosaurids in western North America by the Late Campanian.

Acknowledgements

We thank M. Currie and K. Shepherd (CMN) and B. Strilisky (TMP) for access and loan of specimens needed in this study, and K. Seymour and B. Iwama for assistance with ROM specimens. We are grateful to I. Morrison and R. Bavington-Sanchez for cleaning and preparing CMN 8784 and TMP 1983.064.0003, respectively. Finally, we thank C. Brown, K. Chiba, D. Fowler, J. Horner, D. Larson, A. Prieto-Márquez, C. VanBuren, and M. Vavrek for useful discussions. We thank T. Ciotka for the donation of ROM 65035. Special thanks to D. Weishampel and A. McDonald for reviewing the manuscript. Funding for the project was provided by a Queen Elizabeth II Graduate Scholarship in Science and Technology (to N.E.C.), a National Sciences and Engineering Research Council Post-Graduate Scholarship (to K.S.B.), the Museum of the Rockies and its Horner Fund (to E.A.F.), and a National Sciences and Engineering Research Council Discovery Grant (to D.C.E.).

Copyright information

© Senckenberg Gesellschaft für Naturforschung and Springer 2012