Introduction

The Prince Creek Formation (PCF) of northern Alaska preserves one of the most diverse and prolific assemblages of polar dinosaurs known anywhere in the world. To date, evidence for at least 13 different dinosaurian taxa are known from early Maastrichtian horizons of the unit, including five ornithischians, seven non-avian theropods, and an avialan theropod (Gangloff 1994; Druckenmiller et al. 2013). However, only three of these taxa have been identified taxonomically to the species level (Sullivan 2006; Fiorillo and Tykoski 2012, 2014). The remainder have only been tentatively identified at generic or suprageneric levels (Fiorillo et al. 2009; Brown and Druckenmiller 2011; Druckenmiller et al. 2013; Watanabe et al. 2013). Because all three named dinosaurian species are endemic to the Laramidian Arctic, Erickson and Druckenmiller (2011) hypothesized that the Prince Creek Formation supported a distinctive early Maastrichtian dinosaur polar fauna known as the Paaŋaqtat Province. Understanding the overall faunal composition of the PCF is critical in order to test this hypothesis and assess larger questions relating to Laramidian biogeography and dinosaur paleobiology.

Fig. 1.

Study area in northern Alaska, USA (A) and location of the Liscomb bonebed (B). C. Paleogeographic reconstruction of North America at 70 Ma (Blakey .2009); the box indicates the approximate position of Alaska at that time.

Since the 1990s, an abundance of hadrosaurid remains have been recovered from the Prince Creek Formation (Fig. 1). The material is primarily derived from a single well-known horizon known as the Liscomb bonebed (LBB), from which thousands of disarticulated cranial and postcranial dinosaur remains have been excavated (Brouwers et al. 1987; Davies 1987; Nelms 1989; Clemens and Nelms 1993; Gangloff 1994, 1998; Fiorillo and Gangloff 2001; Fiorillo et al. 2007, 2010; Fiorillo 2008a; Flaig 2010; Gangloff and Fiorillo 2010; Erickson and Druckenmiller 2011; Watanabe et al. 2013). The Liscomb remains were first identified as a lambeosaurine hadrosaurid (Brouwers et al. 1987). However, subsequent researchers have reassigned the remains to Saurolophinae (Hadrosauria) and tentatively referred them to the Campanian to Maastrichtian genus Edmontosaurus Lambe, 1917, or more specifically to E. regalis (Xing et al. 2014). However, a comprehensive anatomical survey of this material has not been conducted and the taxonomic referral formally demonstrated. In part, this stems from the fact the majority of the LBB remains represent individuals approximately one-third adult length and it is difficult to identify the taxonomic status of hadrosaurids from juvenile remains because most are differentiated by adult features (Prieto-Márquez 2008, 2010a; Campione and Evans 2011; Campione et al. 2012).

Here, we provide a description of the hadrosaurid material from the Prince Creek Formation. Based on comparisons with other saurolophines, particularly Edmontosaurus, we recognize a new species that can be diagnosed on ontogenetically invariable characters. The recognition of a new taxon from the LBB contributes to a broader understanding of saurolophine diversity, the taxonomic composition of the PCF and provides important new evidence for testing hypotheses of dinosaur provinciality in Laramidia.

Institutional abbreviations.—AENM, Amur Natural History Museum, Blagoveschensk, Russia; AMNH, American Museum of Natural History, New York City, USA; BHI, Black Hills Institute of Geological Research, Hill City, USA; BMNH, The Natural History Museum, London, UK; CM, Carnegie Museum of Natural History, Pittsburgh, USA; CMN, Canadian Museum of Nature, Ottawa, Canada (formerly NMC, National Museums of Canada); DMNH, Denver Museum of Nature and Science, Denver, USA; FMNH, The Field Museum, Chicago, USA; GMV, National Geological Museum of China, Beijing, China; MACN, Museo Argentino de Ciencias Naturales Bernardino Rivadavia, Buenos Aires, Argentina; RAM, Raymond M. Alf Museum of Paleontology, Claremont, USA; ROM, Royal Ontario Museum, Toronto, Canada; SM, Senckenberg Museum, Frankfurt, Germany; TMNH, Toyohashi Museum of Natural History, Toyohashi, Aichi, Japan; UMMP, University of Michigan Museum of Paleontology, Ann Arbor, USA; UAMES, University of Alaska Museum, Fairbanks, USA; USNM, Smithsonian National Museum of Natural History, Washington DC, USA; YPM, Yale Peabody Museum of Natural History, New Haven, USA.

Other abbreviations.—LBB, Liscomb bonebed; PCF, Prince Creek Formation.

Nomenclatural acts.—The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix “ http://zoobank.org/”. The LSID for this publication is: urn:lsid:zoobank.org:pub:84BD5626-4F09-4246-9D86-E831E9465D69.

The electronic edition of this work was published in a journal with an eISSN 1732-2421, and has been archived and is available from the following digital repository:  http://www.app.pan.pl/article/item/app001522015.html

Geological setting and taxonomic Composition

The Prince Creek Formation (formerly referred to as the Kogosukuruk Tongue of the PCF; Gryc et al. 1951) is composed of non-marine sandstone, conglomerate, coal, and mudstone layers representing interbedded fluvial (meandering channels and floodplains) and marginal marine sediments that were deposited on a low gradient coastal plain (Mull et al. 2003; Flaig et al. 2011). Palynological (Frederiksen et al. 1988, 2002; Frederiksen 1991) and biostratigraphic data (Brouwers and Deckker 1993) suggest the entire PCF ranges from the Upper Cretaceous to Eocene in age. The numerical age of the dinosaur-bearing section of the formation, where it is exposed along the lower Colville River and including the LBB, has been dated at 71–68 Ma using ⁴⁰Ar/³⁹Ar methods (McKee et al. 1989; Besse and Courtillot 1991). The age of LBB is further constrained by an ⁴⁰Ar/³⁹Ar age of 69.2±0.5 Ma from a stratigraphically underlying tuff at a locality known as Sling Point (approximately 1 km from the LBB) and from palynological analyses (Flores et al. 2007) consistent with an early Maastrichtian age (Flaig 2010). This age estimate is within the known stratigraphic range of Edmontosaurus, notably falling between the last appearance date of E. regalis Lambe, 1917 (late Campanian; Campione and Evans 2011; Eberth et al. 2013) and the first appearance of E. annectens Marsh, 1892 (late Maastrichtian; Campione and Evans 2011).

The Late Cretaceous paleolatitude of northern Alaska is estimated to range between 67–82° N (Witte et al. 1987; Besse and Courtillot 1991; Lawver et al. 2002), thus the LBB was well within the paleo-Arctic (above approximately 66° N). Although considerably milder than today, paleobotanical evidence indicates a mean annual temperature for the Maastrichtian of northern Alaska at around 5–6°C, with a cold month mean warmer than 2.0±3.9°C (Parrish and Spicer 1988; Spicer and Parrish 1990; Spicer et al. 1992; Spicer and Herman 2010). From pedogenic and paleobotanical evidence Flaig et al. (2013) concluded that the Arctic coastal plain had polar woodlands with an angiosperm understory, and that it experienced both strong dry and wet seasons. In addition to hadrosaurid remains, the PCF preserves a modestly diverse assemblage of ornithischian and saurischian dinosaurs and mammals (Brouwers et al. 1987; Nelms 1989; Clemens and Nelms 1993; Gangloff 1994; Gangloff et al. 2005; Fiorillo 2008a; Fiorillo et al. 2009; Gangloff and Fiorillo 2010; Brown and Druckenmiller 2011; Fiorillo and Tykoski 2012, 2014; Watanabe et al. 2013). However, to date, no unequivocally ectothermic terrestrial vertebrates have been recovered from the formation. This faunal distribution led Clemens and Nelms (1993) to suggest that the climate was too cold for most terrestrial and amphibious ectotherms, including crocodilians, champsosaurs, choristodires, squamates, and turtles that were more common contemporaneously at lower latitudes.

Taphonomically the dinosaur remains from the LBB are dominated by juvenile specimens of hadrosaurids that have been previously assigned to Edmontosaurus (Gangloff and Fiorillo 2010). Other rare elements represented in the bonebed included isolated hadrosaurid material and shed thescelosaurid, tyrannosaurid, and troodontid teeth. The hadrosaurid remains are almost entirely disarticulated, show little evidence of weathering, predation, or trampling, and are typically uncrushed and unpermineralized (Fiorillo et al. 2010; Gangloff and Fiorillo 2010). The LBB occurs in a trunk channel on a distributary channel splay complex and flood plain (Fiorillo et al. 2010; Flaig 2010). The bonebed is posited to reflect a mass mortality event associated with overbank flood deposits (Gangloff and Fiorillo 2010), which could have resulted from rapid snowmelt from the then-rising Brooks Range to the south (Fiorillo et al. 2010).

Material and methods

Material.—More than 6000 hadrosaurid bones from the LBB that form the basis for this study have been collected during expeditions led by the University of Alaska Museum and the University of California Museum of Paleontology (this material is now housed at UAMES). The majority of this material consists of individuals from an immature growth stage (size class 1; humeral length = ∼22 cm; SOM 1, Supplementary Online Material available at  http://app.pan.pl/SOM/app61-Mori_etal_SOM.pdf), approximately onethird the adult humerus length of Edmontosaurus (approximately 60–70 cm). The remaining material represents size class 2 juvenile individuals (SOM 1) (∼29 cm humeral length), and size class 3 subadult individuals (SOM 1) (∼43 cm humeral length). The three size classes may represent a mass mortality event of a herd or herds containing yearly cohorts (Gangloff and Fiorillo 2010).

393 specimens bearing phylogenetically informative characters, including cranial and postcranial elements were examined in detail during the course of this study (SOM 1). Because of the abundance of material, most elements are known from multiple specimens making it possible to account for variation in both size and morphology.

The Alaskan material is compared in detail to Edmontosaurus, which has two recognized species, E. regalis (late Campanian) and E. annectens (late Maastrichtian; Campione and Evans 2011) represented by abundant material from the United States of America and Canada (SOM 2). Edmontosaurus was chosen as the primary comparator, particularly for the regression analyses (see below), for four reasons. First, the Alaskan taxon at a coarse anatomical level appears to be Edmontosaurus and has been referred to as such (Clemens and Nelms 1993; Gangloff 1998; Fiorillo and Gangloff 2001; Prieto-Márquez 2008; Gangloff and Fiorillo 2010; Campione and Evans 2011; Xing et al. 2014). Specifically, the Alaskan material possesses two of the diagnostic characters of the genus, proposed by Xing et al. (2014); (i) the premaxillary margin is strongly folded dorsoventrally, and (ii) the dorsolateral process of the laterosphenoid is truncated. Also, the PCF material has two premaxillary foramina in the prenarial region of the circumnarial fossa as in Edmontosaurus, although we disagree with Xing et al. (2014) in regarding this character as unique to Edmontosaurus, as such a structure is also seen in Maiasaura (Horner 1983). Further, the PCF taxon bears numerous other general anatomical similarities to these taxa such as a nasal lacking ornamentation, as seen in some genera of Saurolophinae (Brown 1913; Lambe 1914; Sternberg 1953; Maryańska and Osmólska 1981; Horner 1983, 1992; Gates and Sampson 2007; Prieto-Márquez 2010b, 2012; Prieto-Márquez et al. 2014), the tooth form bears closest similarity to Edmontosaurus (Prieto-Márquez 2008; Erickson and Druckenmiller 2011), and the frontal is widely exposed laterally. Second, the Alaskan material occurs within the temporal range of the genus Edmontosaurus, although it lies stratigraphically intermediate to both species (Fig. 2). Third, it occurs within the same landmass as Edmontosaurus (Laramidia), although north of the known geographic ranges for both species (Colorado, USA to Alberta, Canada). Finally, the Prince Creek Formation taxon and Edmontosaurus are also clearly distinct from three closely related Asian taxa (Godefroit et al. 2012; Prieto-Márquez 2013, 2014; Xing et al. 2014). Specifically, the PCF taxon and Edmontosaurus differ from Shantungosaurus Hu, 1972 in that the dorsal surface of the postorbital is nearly straight, in possessing a smaller anterior portion of the scapula, a less developed suprailiac crest of the ilium and the absence of well developed boss in the proximal region of the ischiadic peduncle. Both differ from Kerberosaurus Bolotsky and Godefroit, 2004 in lacking a pocket on the basisphenoid process of the prootic, a wide groove for ophthalmic nerve on the laterosphenoid, a prominent palatine process on the maxilla, and a markedly depressed dorsal surface of the postorbital. Finally, the PCF taxon and Edmontosaurus differ from “Kundurosaurus” Godefroit, Bolotsky, and Lauters, 2012 in possessing a ventrally-projecting posterior buttress of the scapula, straight anterior process of the nasal, and a ventrally - -curved preacetabular process of the ilium, although Xing et al. (2014) consider Kundurosaurus to be a junior synonym of Kerberosaurus, with which we concur. For these reasons, we assume the PCF taxon is either referable to, or closely related to Edmontosaurus.

Fig. 2.

Temporal distribution of Edmontosaurus species and Ugrunaaluk kuukpikensis gen. et sp. nov. from the Prince Creek Formation in the Late Cretaceous.

Fig. 3.

Histogram of hadrosaurid bones from the Liscomb bonebed (Prince Creek Formation) for which at least 10 specimens are known. Size is standardized by the mean of size classes 1 and 2 for each bone. The relative size is adjusted so that the mean size of the size class 1 specimens becomes 1. The hypothesis of normal distribution for all samples, and samples whose relative sizes range 0.88–1.52, are both rejected (p « 0.001). The hypothesis of normal distribution for specimens whose relative size range 0.88–1.12 was not rejected (p = 0.25). Relative sizes of juvenile E. annectens specimens are also shown at the bottom. Note that due to insufficient numbers of corresponding bone types, size class 3 individuals of E. annectens are not shown in the histogram. For raw data, see SOM 3.

Size classes in the Liscomb bonebed.—We characterized the size distribution of the individuals in the LBB and a histogram showing size versus specimen counts was prepared depicting size versus the number of specimens (Fig. 3). First, elements for which at least 10 specimens are known were selected (dentary, frontal, humerus, ulna, radius, tibia, metatarsal II–IV; SOM 3) for size standardization. The length of each bone was divided by the mean length of each element in order to compare the lengths of different bone types. When calculating the mean length, specimens that are more than twice as long as the smallest specimens were removed. Distribution of the sizes were tested for normality using the Shapiro-Wilk test in PAST 3.0 (Harper and Ryan 2001).

The PCF elements can be classified into three ontogenetic stages, all of which are interpreted to represent three different developmental stages (Fig. 3). Approximately 85% of the specimens examined are categorized into size class 1. A three-dimensional composite skull (Fig. 4) was reconstructed from casts of size class 1 individuals and found to measure approximately 33 cm long (measured from the anterior end of the premaxilla to the mid-point of the quadrate). This is 36% of the length of the adult paratype specimen of E. annectens, YPM 2182 (91 cm). Size class 2 accounts for about 10% of the examined specimens and the isolated cranial elements are approximately 30% longer than size class 1 specimens, which is equivalent to ∼40% of the skull length of YPM 2182. Size class 3 represents less than 5% of the total specimens, and the cranial elements are approximately 80–100% longer than the size class 1 specimens. These correspond to individuals ∼60% of the size of YPM 2182.

Fortuitously, Edmontosaurus is represented by material spanning a broad ontogenetic range, including specimens of E. annectens that overlap in size with the Prince Creek Formation that permit some direct size-standardized anatomical comparisons. Comparative material of E. annectens (SOM 3) was also categorized by size class after standardization to mean lengths of the PCF material (Fig. 3). While overlapping material of size class 1 for E. annectens and the Alaskan material is not known, there is overlap in size classes 2 and 3 (size class 3 is not included in Fig. 3 because fewer than 10 specimens of a corresponding bone type of the PCF material were available). Among the ontogenetic spectrum of E. annectens available for comparison are individuals approximately equivalent to size class 2 (LACM 23504, skull length = estimated 47 cm), size class 3 (e.g., ROM 53530, disarticulated), to subadult (CMN 8509, skull length = 75 cm), and adult (ROM 57100, skull length = 101 cm; MOR 003, skull length = 118 cm) specimens. It should be noted, however, that there is a limited number of PCF specimens that overlap in absolute size with E. annectens, thus direct comparisons are not possible for some bones. E. regalis specimens are all larger than size class 3 specimens of the PCF taxon and range from subadult (CMN 8399, skull length = 78 cm; BMNH 8937, skull length = 75 cm) to adult (ROM 801, skull length = 107 cm; USNM 12711, skull length = 105 cm).

Regression analysis.—Because the PCF material is generally much smaller than material for E. annectens and E. regalis, quantitative methods were used to compare many elements. Quantitative data on specimens not personally measured were taken from the literature (Lull and Wright 1942; Brett-Surman 1989) or from photographs provided by colleagues. A regression analysis was employed to test if a given feature observed in the PCF material is potentially a juvenile condition of a feature seen in adult E. annectens or E. regalis. Dodson (1975a–c) demonstrated that among the same species of reptiles, log-transformed lengths of various parts show high correlations against body size. However, because the correlation coefficient is not reliable when the data include outliers (specimens which are much smaller than the others), in this case the PCF specimens themselves (Schuyler 2004), testing whether the relatively small specimens represent juveniles requires a different approach.

Fig. 4.

Cranial reconstruction of Ugrunaaluk kuukpikensis gen. et sp. nov. from the early Maastrichtian Prince Creek Formation, Alaska, in left lateral view. Photograph (A) and bone interpretation (B).

In order to address this problem, a new approach was employed in this study. The method is based on the assumption that the growth trend line of Edmontosaurus, from juvenile to adult, can be expressed by the following equation:

A and B are constants. This assumption is based on the observations by Dodson (1975c), who reported linear relationships between the skull length and various parts of the skull in species of lambeosaurines. By logarithmic transformation, the equation above can be transformed as:

If a log-transformed length of any body part correlates with the log-transformed body length, any sets of log-transformed body lengths would also correlate to each other. Using this assumption, we null-hypothesized that the character condition of the PCF taxon is similar to an extrapolated juvenile of E. regalis or E. annectens. From this null-hypothesis, growth trend lines of other Edmontosaurus species + PCF materials combined were prepared. If these hypothetical growth lines are significantly different (p < 0.05) from the growth trend lines of other Edmontosaurus specimens alone, or PCF materials alone, we concluded the PCF materials represent a different ontogenetic trajectory for that feature. For the postorbital and quadratojugal, the length of the dentary was used as a proxy for body size. Because nearly all specimens excavated from the LBB are disarticulated, only size class 1 material of the PCF taxon were compared against the mean size of the dentary of size class 1. Because size class 1 specimens represent the majority of the PCF material, other size class 1 material also likely belongs to individuals of the same size. When statistically significant results were discovered, either maximum or minimum limits of the confidence interval of the dentary length (mean length +/- two standard deviations) is adopted as the proxy of the body size, and the regression analysis was conducted again. When a given element length is too short (or long) relative to dentary length, we used the minimum (or maximum) length of the dentary. This alleviates the problem of using disarticulated bones in the regression analysis, and is more conservative. A reduced major axis (RMA) regression line was chosen to calculate the growth trend line, because it minimizes the errors in the both variables (Pearson 1901; Warton et al. 2006; Smith 2009). SMATR ver. 2.0 (Falster et al. 2006) was used to prepare the RMA regression lines, R2 and P values, and to assess whether two regression lines are statistically different by Wald statistics.

Systematic paleontology

Fig. 5.

Premaxilla comparison of a size class 1 Ugrunaaluk kuukpikensis gen. et sp. nov. from the early Maastrichtian Prince Creek Formation, Alaska (A, UAMES 12995, anterior portion of right premaxillae in dorsal view, lacking the vestibular promontory and lateral cicumnarial cavity); adult Edmontosaurus regalis Lambe, 1917 from the late Campanian Edmonton Formation, Alberta, Canada (B, CMN 2289); and size class 3 Edmontosaurus annectens Marsh, 1892 from the late Maastrichtian Lance Formation, Wyoming, USA (C, ROM 53526). Photographs (A₁–C₁) and explanatory drawings (A₂–C₂).

Fig. 6.

Maxilla comparison of a size class 2 Edmontosaurus annectens Marsh, 1892 from the late Maastrichtian Hell Creek Formation, Montana, USA (A, LACM 23502, right maxilla in lateral view) and a size class 1 Ugrunaaluk kuukpikensis gen. et sp. nov. from the early Maastrichtian Prince Creek Formation, Alaska (B, UAMES 4327, right maxilla in lateral view). U. kuukpikensis has a relatively low maxilla compared to its length. The height (H) to length (L) ratio is 0.32 whereas that of size class 2 E. annectens is 0.41.

Fig. 7.

Postorbital comparison of a size class 3 Ugrunaaluk kuukpikensis gen. et sp. nov. (A, UAMES 33308) from the early Maastrichtian Prince Creek Formation, Alaska, right postorbital in medial (A₁) and anterior (A₂) views; size class 2 Edmontosaurus annectens Marsh, 1982 (B, ROM 53513, reversed) from the late Maastrichtian Lance Formation, Wyoming, USA, left postorbital in medial (B1) and anterior (B2) views; and adult Kundurosaurus nagornyi (C, AENM 2/921-6) from the Maastrichtian Udurchukan Formation, Kundur, Russia, left postorbital in medial view (from Godefroit et al. 2012: fig. 6). Note that size class 3 of U. kuukpikensis does not possess a deep posterior orbital pocket seen in the size class 2 specimens of E. annectens, although the articular surface with the frontal is nearly identical. D. Left postorbital of E. annectens (CMN 8509) from the late Maastrichtian Lance Formation, Saskatchewan, Canada, showing how the jugal process width is defined. E. Postorbital regression analyses. Biplots of the width of the jugal process measured at one-quarter distance from its dorsal end versus dentary length. U. kuukpikensis has a narrower jugal process than Edmontosaurus regalis. For E. regalis and U. kuukpikensis the numbers in the parentheses are those calculated when the minimum dentary length, not average, is adopted as the body size proxy. When the regression lines have statistically indistinguishable slope values but show statistically significant differences in elevation, P values are marked with an asterisk. The 95% confidence interval of U. kuukpikensis dentary length is also shown by the bi-directional arrow.

Fig. 8.

Jugal comparison of Ugrunaaluk kuukpikensis gen. et sp. nov. (A, UAMES 14174, right, size class 3; D, UAMES 4922, left, size class 1) from the Prince Creek Formation, Maastrichtian, Alaska; Edmontosaurus annectens Marsh, 1892 (B, ROM 53518, left, size class 2) from the late Maastrichtian Lance Formation, Wyoming, USA, in medial (B₁) and lateral (B₂) views; and adult Edmontosaurus regalis Lambe, 1917 (C, CMN 2289, left, reversed) from the late Campanian Edmonton Formation, Alberta, Canda. A, B₁, C. Anterior processes in medial views, the dashed lines delineate the posterior border of the anterior process. In U. kuukpikensis this border is angled more strongly than in Edmontosaurus. D, B2. Left jugal in lateral view. Note that U. kuukpikensis has a shallower posterior constriction compared to E. annectens. B₂ shows how the dorsal length and the posterior constriction are defined. E. Jugal regression analyses. Dorsal length versus posterior constriction of jugal. This suggests the shallower posterior constriction of U. kuukpikensis is not attributable to ontogeny.

Fig. 9.

A. Quadratojugal regression analyses. Biplot of the lateral exposure width of the quadratojugal versus dentary length and quadratojugal shown in lateral view of Ugrunaaluk kuukpikensis gen. et sp. nov. Ugrunaaluk kuukpikensis is similar to Edmontosaurus annectens in the lateral exposure width of the quadratojugal, but different from Edmontosaurus regalis. The 95% confidence interval of U. kuukpikensis dentary length is also shown by the bi-directional arrow. When the regressions show statistically indistinguishable slope values but statistically significant differences in elevation, P values are marked with an asterisk. B. Right quadratojugal of U. kuukpikensis (UAMES 4298) from the early Maastrichtian Prince Creek Formation, Alaska, showing how the lateral exposure width is defined.

Cladistic analysis

To test the systematic position of Ugrunaaluk kuukpikensis gen. et sp. nov., we conducted a cladistic analysis using the matrices of the two most recent phylogenetic studies of hadrosaurids (Xing et al. 2014; Prieto-Márquez 2013). First we scored the Alaskan taxon into a modified version of the Xing et al. (2014) character matrix, consisting of 60 taxa and 346 equally weighted and unordered characters. Two versions of the data matrix were prepared. In the first, Ugrunaaluk kuukpikensis gen. et sp. nov. was scored as-is without consideration of possible ontogenetic changes (17% missing data; SOM 6: Matrix 1). In the second, ontogenetically variable characters were scored as missing (30% missing data; SOM 7: Matrix 2) based on a more comprehensive analysis of the Alaskan material and Edmontosaurus (summarized in SOM 5; Mori 2014). We further tested the sister group relationships of Ugrunaaluk by scoring it into the matrix of Prieto-Márquez (2013), consisting of 35 taxa and 265 characters. As for the Xing et al. (2014) matrix, U. kuukpikensis gen. et sp. nov. was scored both as-is (12% missing data; SOM 8: Matrix 3) and with ontogenetically variable characters missing (17% missing data; SOM 9: Matrix 4). Character data was input using Mesquite 3.0.2 (Maddison and Maddison 2011). For both the Xing et al. (2014) and Prieto-Márquez (2013) matrices, we revised the character concerning the lateral premaxillary cavity (characters 81 and 53, respectively) so that it more accurately describes the morphology of the circumnarial ridge and vestibular promontory (see SOM 5).

Fig. 10.

Dentary comparison of size class 2 Ugrunaaluk kuukpikensis gen. et sp. nov. (A, UAMES 4946, left; C, UAMES 12941, right) from the late Maastrichtian Prince Creek Formation, Alaska, dentary in lateral view; and size class 2 Edmontosaurus annectens Marsh, 1892 (B, BHI-6218, cast, reversed) from the late Maastrichtian Hell Creek Formation, South Dakota, USA, right dentary in lateral view. Note that U. kuukpikensis has a shorter edentulous process than E. annectens. C. Right dentary of U. kuukpikensis showing how the proximal edentulous process length and dental battery length are defined. D. Dentary regression analyses. Biplot of posterior symphyseal process length versus dental battery length. U. kuukpikensis has a shorter symphyseal process than in E. annectens. When the regressions show statistically indistinguishable slope values but statistically significant differences in elevation, P values are marked with an asterisk. Because the lines for E. regalis and U. kuukpikensis are statistically indistinguishable, only one is shown.

The cladistic analysis was conducted using TNT 1.1 (Goloboff et al. 2008). Following Xing et al. (2014) and Prieto-Márquez (2013), Ouranosaurus nigeriensis and Equijubus normani were designated as outgroups, respectively. The most parsimonious trees were sought using the “New Technology search” option with “Sectional search” and “Tree fusing options”. The minimal length was searched for 100 times, with 100 random seeds and 5 initially added sequences. The maximum number of trees stored in memory is set to 10000. Bremer support values were calculated using a Bremer Support Script made by Goloboff (2008) with the default setting (1000 replicates). Bootstrap values were calculated using the resampling function of TNT, with the standard (sample with replacement) and traditional search options and 100 replicates, and the results were output as absolute frequencies.

The analyses based on Xing et al. (2014) resulted in 288 MPTs (1059 steps) from Matrix 1 in SOM 6 (Ugrunaaluk kuukpikensis gen. et sp. nov. scored as-is) and 160 MPTs (1036 steps) from Matrix 2 in SOM 7 (ontogenetic characters of Ugrunaaluk kuukpikensis gen. et sp. nov. scored as missing). The strict consensus trees of both analyses (cladograms 1 and 2) are summarized in Fig. 11. In both analyses, Ugrunaaluk kuukpikensis gen. et sp. nov. was recovered as the sister group of Edmontosaurus annectens + E. regalis, although bootstrap and Bremer support values were relatively low in SOM 6: Matrix 1 compared to SOM 7: Matrix 2 (39/1 and 79/2, respectively). Overall tree topology of both analyses was also largely congruent; however, in cladogram 1 (Ugrunaaluk kuukpikensis gen. et sp. nov. scored as-is), Shantungosaurus, Kerberosaurus, and Edmontosaurus + Ugrunaaluk kuukpikensis gen. et sp. nov. collapsed to form a polytomy with the clade Lophorhothon (Prosaurolophus + Saurolophus), and the clade including Secernosaurus, Kritosaurus, and both species of Gryposaurus was also unresolved.

In cladogram 1, five synapomorphies unite Ugrunaaluk kuukpikensis gen. et sp. nov. + Edmontosaurus, including: 27.1, angle between the anteroventral margin of the anterior end and the dorsal margin of the lateral process of the predentary in lateral view is 56–74°; 51.1, presence of a sharp projection on the posterodorsal surface of the coronoid process; 77.1, presence of a premaxillary accessory foramen which shares a common chamber with the premaxillary foramen; 204.1, finger-shaped, anteroposteriorly long and mediolaterally narrow median anterior process of the parietal; 221.1, greatly reduced length of the postorbital process of the laterosphenoid; 234.1, the angle of the ventral deflection of the occipital condyle relative to the horizontal line is less than 155° (Xing et al. 2014). In the better resolved cladogram 2, Ugrunaaluk + Edmontosaurus is supported by two synapomorphies, 51.1 and 221.1 (explained above). Edmontosaurus regalis + E. annectens are united by three synapomorphies: 12.1, the marginal denticles of dentary teeth are composed of a single rounded knob; 81.1, presence of additional premaxillary accessory fossae located lateral to the premaxillary accessory narial fossa and anterior to the circumnarial fossa, separated from the premaxillary anterior fossa by a well developed vestibular promontory (modified character); 182.1, presence of bulging jugal process with a deep pocket of the postorbital.

Fig. 11.

Results of the cladistic analyses of Matrix 1 (SOM 6) and Matrix 2 (SOM 7) based from Xing et al. (2014), showing the phylogenetic position of Ugrunaaluk kuukpikensis gen. et sp. nov. Matrix 1 resulted in 288 most parsimonious trees (MPTs), and Matrix 2 resulted in 160 MPTs. The strict consensus tree based on Matrix 1 (characters scored as-is, cladogram 1) was identical to the strict consensus tree based on Matrix 2 (ontogenetic characters removed, cladogram 3), except that some clades, marked by “X” were collapsed in cladogram 1. Bootstrap, Bremer support values, Consistency and Retention indices of the strict consensus trees are also shown. For the clades collapsed in the cladogram 1, these values are substituted by a dash. Those indices and steps of the parsimonious trees are also shown in the parentheses below the tree.

Fig. 12.

Results of the cladistic analyses based of Matrix 3 (SOM 8) and Matrix 4 (SOM 9) based from Prieto-Márquez (2013), showing the phylogenetic position of Ugrunaaluk kuukpikensis gen. et sp. nov.; Matrix 3 resulted in 2 most parsimonious trees, and Matrix 4 resulted in one fully resolved cladogram. The strict consensus tree based on Matrix 3 (characters scored as-is, cladogram 3) was identical to the most parsimonious tree based on Matrix 4 (ontogenetic character removed, cladogram 4), except that some clades, marked by ”X”, were collapsed in cladogram 3. Bootstrap, Bremer support values, Consistency and Retention indices of the phylogenetic trees are also shown. For the clades collapsed in the cladogram 3, these values are substituted by a dash. Those indices and steps of the parsimonious trees based from Matrix 3 are also shown in the parentheses below the tree.

The analyses based on Prieto-Márquez (2013) resulted in 2 MPTs (637 steps) from Matrix 3 in SOM 8 (characters scored as-is) and 1 MPT (630 steps) from Matrix 4 in SOM 9 (ontogenetic characters of Ugrunaaluk scored as missing). The strict consensus trees of both analyses (cladogram 3 and 4) are summarized in Fig. 12. Again, Ugrunaaluk kuukpikensis gen. et sp. nov. was recovered as the sister group of Edmontosaurus annectens + E. regalis in both analyses, although in cladogram 3 (characters scored as-is), the position of Kerberosaurus was not resolved. Clade support for Ugrunaaluk + Edmontosaurus in cladograms 3 and 4 was strong for both bootstrap and Bremer values (84/4 and 85/5, respectively). There was also less resolution in other clades of Saurolophinae (Fig. 12).

In cladograms 3 and 4, five synapomorphies unite Ugrunaaluk + Edmontosaurus, including: 34.1, presence of a sharp projection on the posterodorsal surface of the coronoid process; 179.1, relative widths of the skull across the postorbitals and squamosals is more than 25%; 196.1, expansion of the scapular blade; 219.1, moderately developed suprailiac crest of the ilium; 242.0, lack of the protuberance on the proximal region of the pubis ischial peduncle. These characters correspond 51.1, 232.1, 258.0, 298.1, and 316.0 respectively in the matrix by Xing et al. (2014). In cladogram 4, Edmontosaurus regalis + E. annectens are united by two synapomorphies (53.1, presence of premaxillary additional accessory fossa (modified), and 119.2, presence of the postorbital pocket, both of which were also identified as synapomorphies in cladograms 1 and 2.

Discussion

Cladistic analysis.—The results of all the cladistic analyses based on the data matrices of both Xing et al. (2014) and Prieto-Márquez (2013) recover Ugrunaaluk kuukpikensis gen. et sp. nov. as the sister taxon to Edmontosaurus. In none of the trees, including those in which characters considered ontogenetically variable are scored as missing, is U. kuukpikensis gen. et sp. nov. recovered as the sister taxon to either E. annectens or E. regalis. The major differences between the analyses of the Xing et al. (2014) and Prieto-Márquez (2013) matrices are found in non-hadrosaurid hadrosauroids, which is beyond the focus of this study.

The analyses of the Xing et al. (2014) and Prieto-Márquez (2013) matrices identify 3 and 2 synapomorphies, respectively, that unite Edmontosaurus regalis and E. annectens, to the exclusion of Ugrunaaluk kuukpikensis gen. et sp. nov. To what extent might ontogenetic status influence the validity of these synapomorhies, and in turn the taxonomic status of Ugrunaaluk kuukpikensis gen. et sp. nov.? The degree to which ontogeny may influence the number and/or morphology of marginal denticles (Xing et al. 2014: character 12.1) is poorly studied and difficult to assess. However, the analyses of both the Xing et al. (2014) and Prieto-Márquez (2013) matrices identify two other identical synapomorphies relating to morphology of the premaxilla (81.1 and 53.1, respectively) and the postorbital (182.1 and 119.2, respectively) that are not biased by the ontogenetic status of the Alaskan material. With respect to the premaxilla, the circumnarial ridge of U. kuukpikensis gen. et sp. nov. is narrow and projects posterolaterally. In marked contrast, the circumnarial ridge in Edmontosaurus annectens and E. regalis has a vestibular promontory, which is also present in juveniles (size classes 2 and 3) of E. annectens. Therefore, this character is not ontogenetically variable in E. annectens, and we reasonably assume it is not so in E. regalis. Likewise, a postorbital pocket is absent in all size classes of Ugrunaaluk kuukpikensis gen. et sp. nov., including a large size class 3 specimen that represents an animal approximately 60% the adult body length of Edmontosaurus. In contrast, a deep postorbital pocket is present even in size classes 2 and 3 of E. annectens (ROM 53513, ROM 53514). A postorbital pocket is also well developed in adult E. regalis, but status of this feature is unknown in juveniles; however, we assume that if the pocket was well developed in all juvenile size classes of E. annectens, then it is likely also present in juvenile E. regalis, although ultimately the discovery of juvenile material of this taxon will be required to test this assumption. Therefore, given the results of both cladistic analyses and available evidence, we consider that both the premaxillary and postorbital characters can be used to distinguish Ugrunaaluk kuukpikensis gen. et sp. nov. from Edmontosaurus.

Table 1.

Comparison of Ugrunaaluk kuukpikensis and Edmontosaurus spp.

Taxonomic status of Ugrunaaluk kuukpikensis .—Although most previous work on the LBB material referred Ugrunaaluk kuukpikensis gen. et sp. nov. to Edmontosaurus, the results of our comparative, quantitative and cladistic analyses indicate important morphological differences between the Alaskan taxon and both species of Edmontosaurus (Table 1). Ugrunaaluk kuukpikensis gen. et sp. nov. differs from Edmontosaurus in that the former has a posterolaterally projected circumnarial ridge of the premaxilla, an angled posterior margin of the anterior process of the jugal, and lacks a deep postorbital pocket. The deep pocket on the jugal process of the postorbital has been considered a diagnostic character of Edmontosaurus (Campione et al. 2011). Recently, Xing et al. (2014) noted that Kerberosaurus and Shantungosaurus possess a shallow postorbital fossa, and that this is a common character seen in Edmontosaurini. The presence of the shallow fossa in Ugrunaaluk kuukpikensis gen. et sp. nov. supports its inclusion in Edmontosaurini, but suggests it is plesiomorphic with respect to the deep pocket that is diagnostic for Edmontosaurus. While Ugrunaaluk kuukpikensis gen. et sp. nov. is similar to Kerberosaurus and Shantungosaurus in possessing a shallow postorbital fossa, it also differs from both taxa in that the dorsal border of the jugal process is nearly straight, more like the condition seen in many specimens of Edmontosaurus (although not all). Additionally, the articular surface on the medial side of the postorbital in Ugrunaaluk kuukpikensis gen. et sp. nov. differs from that of Kerberosaurus, but is nearly identical to that in E. annectens (Fig. 7). In these regards, the postorbital morphology of Ugrunaaluk kuukpikensis gen. et sp. nov. appears to be “intermediate” between that of Kerberosaurus/ Shantungosaurus and Edmontosaurus.

At the species level, Ugrunaaluk kuukpikensis gen. et sp. nov. clearly differs from Edmontosaurus annectens in that the former possesses a dorsoventrally short and elongate maxilla (Fig. 6), a relatively gracile jugal (Fig. 8), and a short symphyseal process of the dentary (Fig. 10). Likewise, U. kuukpikensis gen. et sp. nov. can be differentiated from E. regalis in having a narrower jugal process of the postorbital (Fig. 7E) and wider lateral exposure of the quadratojugal (Fig. 9), although the results of our regression analyses (Fig 7E) are not fully conclusive and overlapping-sized juvenile material with Ugrunaaluk kuukpikensis gen. et sp. nov. are not available. However, with respect to premaxillary morphology, the presence of a vestibular promontory is well developed in juvenile E. annectens and therefore may also have been present in juvenile E. regalis. Similarly, given that adult E. regalis has a much wider jugal process than E. annectens (Fig. 9), it seems likely that juvenile E. regalis had a wider jugal process as well.

Recently, Xing et al. (2014) interpreted the Alaskan material to represent the juvenile form of E. regalis because of its anteroposteriorly short premaxilla and oblique and enlarged postorbital articular surface of the jugal. However, premaxillary length displays positive allometry in Edmontosaurus (Campione and Evans 2011), thus the short premaxillary length seen in juvenile Ugrunaaluk kuukpikensis gen. et sp. nov. specimens likely represent a generalized juvenile state for all Edmontosaurini. Also, based on extensive examination of many specimens of Ugrunaaluk kuukpikensis gen. et sp. nov., we do not consider the postorbital articular surface of the jugal to be enlarged in a manner similar to E. regalis. In light of other significant differences between the two species listed above, we do not consider the Alaskan material to be a juvenile form of E. regalis (Xing et al. 2014).

The results of the cladistic analyses provide additional support that the Alaskan material represents a distinct new hadrosaurid taxon. Ugrunaaluk kuukpikensis gen. et sp. nov. was recovered as the sister taxon to Edmontosaurus regalis + E. annectens in two separate phylogenetic analyses using different matrices (Prieto-Márquez 2013; Xing et al. 2014), even when ontogenetically variable characters were scored as missing (cladograms 2 and 4). None of the analyses recovered Ugrunaaluk kuukpikensis gen. et sp. nov. nested within Edmontosaurus. Because species-specific characters tend not to appear until later stages of development (Baer 1828; Nelson 1978; Kitching et al. 1998), the phylogenetic position of juvenile specimens may be recovered in a more basal position than is actually true (Tsuihiji et al. 2011; Campione et al. 2012). However, Ugrunaaluk kuukpikensis gen. et sp. nov. was consistently recovered as the sister taxon to Edmontosaurus even in analyses where ontogenetically variable characters were removed, and in a relatively derived position compared to Shantungosaurus, which is based on adult material. Furthermore, Prieto-Márquez (2014) recently described and conducted a cladistic analysis of a size class 2 juvenile of E. annectens (LACM 23504), in which its ontogenetically variable characters were scored as missing. It is notable that LACM 23504 was recovered as the sister taxon to E. annectens but not as a sister taxon to E. annectens + E. regalis. This suggests that scoring ontogenetically variable characters as missing can be successfully employed when assessing relationships among juvenile saurolophines, including U. kuukpikensis gen. et sp. nov. Therefore, we interpret that our analysis correctly represents the phylogenetic position of the Alaskan taxon. Our results are also largely congruent with those of Xing et al. (2014) who recover Kerberosaurus and Shantungosaurus as successive sister taxa to Ugrunaaluk kuukpikensis gen. et sp. nov. + Edmontosaurus within Edmontosaurini.

In summary, our results indicate Ugrunaaluk kuukpikensis gen. et sp. nov. is more closely related to Edmontosaurus than any other saurolophine; however, even when ontogenetic effects are taken into consideration, U. kuukpikensis gen. et sp. nov. is not referable to either E. regalis or E. annectens. Ugrunaaluk kuukpikensis gen. et sp. nov. can be diagnosed on a unique combination of characters and becomes the first formally recognized hadrosaurid taxon from the Prince Creek Formation.

Paaŋaqtat fauna.—The recognition of a new saurolophine hadrosaurid, Ugrunaaluk kuukpikensis gen. et sp. nov. provides further evidence in support of a distinct, early Maastrichtian polar fauna, provisionally termed the Paaŋaqtat Province (Erickson and Druckenmiller 2011; Druckenmiller et al. 2013). The Paaŋaqtat fauna is characterized by polar dinosaurian species that are endemic to the Prince Creek Formation, including the pachycephalosaurid Alaskacephale gangloffi (Gangloff et al. 2005; Sullivan 2006), the centrosaurine ceratopsid Pachyrhinosaurus perotorum (Fiorillo and Tykoski 2012), and the tyrannosaurid Nanuqsaurus hoglundi (Fiorillo and Tykoski 2014). Additional evidence suggests a potentially unique orodromine and thescelosaurine thescelosaurid (Brown and Druckenmiller 2011; Druckenmiller et al. 2013) and a distinct large-toothed species of Troodon (Fiorillo 2008b; Druckenmiller et al. 2013) that cannot be referred to any known species from more southerly latitudes. More broadly, U. kuukpikensis gen. et sp. nov. also can add new support for the presence of latitudinally arrayed centers of dinosaurian endemism in Laramidia (Gates et al. 2010; Sampson et al. 2010, 2013).

Conclusions

In this study, we; (i) identify and name a new taxon of saurolophine hadrosaurid, Ugrunaaluk kuukpikensis gen. et sp. nov. from the Prince Creek Formation of Alaska; (ii) demonstrate that skeletally immature specimens can be reliably used in addressing taxonomic problems within Hadrosauridae based on a detailed understanding of the growth patterns of closely related taxa; and (iii) show that Ugrunaaluk kuukpikensis gen. et sp. nov. does not represent a juvenile form of either recognized species of Edmontosaurus. Further morphological description of the new Alaskan taxon and greater clarification of its differences from Edmontosaurus will require discovery of adult material from the Prince Creek Formation and/or description of other juvenile specimens from E. regalis and E. annectens.

The establishment of a new species of hadrosaurid, Ugrunaaluk kuukpikensis gen. et sp. nov. further clarifies the faunal composition of the Prince Creek Formation and contributes to a growing body of evidence that the paleo-Arctic hosted a distinct and endemic polar, early Maastrichtian dinosaurian fauna. Ongoing field work in the formation and taxonomic clarifications of existing material will help to further establish the faunal composition of the unit and add critical new data to test hypotheses of dinosaur provinciality in Laramidia during the latest Cretaceous.