Morphological Variation in the Hadrosauroid Dentary

(1)

Examensarbete vid Institutionen för geovetenskaper

Degree Project at the Department of Earth Sciences

ISSN 1650-6553 Nr 398

Morphological Variation in

the Hadrosauroid Dentary

Morfologisk variation i det

hadrosauroida dentärbenet

D. Fredrik K. Söderblom

INSTITUTIONEN FÖR GEOVETENSKAPER

(2)

(3)

Examensarbete vid Institutionen för geovetenskaper

Degree Project at the Department of Earth Sciences

ISSN 1650-6553 Nr 398

Morphological Variation in

the Hadrosauroid Dentary

Morfologisk variation i det

hadrosauroida dentärbenet

(4)

ISSN 1650-6553

Copyright © D. Fredrik K. Söderblom

(5)

Abstract

Morphological Variation in the Hadrosauroid Dentary

D. Fredrik K. Söderblom

The near global success reached by hadrosaurid dinosaurs during the Cretaceous has been attributed to their ability to masticate (chew). This behavior is more commonly recognized as a mammalian adaptation and, as a result, its occurrence in a non-mammalian lineage should be accompanied with several evolutionary modifications associated with food collection and processing. The current study investigates morphological variation in a specific cranial complex, the dentary, a major element of the hadrosauroid lower jaw. 89 dentaries were subjected to morphometric and statistical analyses to investigate the clade’s taxonomic-, ontogenetic-, and individual variation in dentary morphology.

Results indicate that food collection and processing became more efficient in saurolophid hadrosaurids through a complex pattern of evolutionary and growth-related changes. The diastema (space separating the beak from the dental battery) grew longer relative to dentary length, specializing food collection anteriorly and food processing posteriorly. The diastema became ventrally directed, hinting at adaptations to low-level grazing, especially in younger individuals. The coronoid process became anteriorly directed, and was relatively more elongate, resulting in increased moment arm length, with muscles being re-directed to pull the jaw more posteriorly, and mechanical advantage increasing. Although all hadrosauroid groups went through relative dental battery elongation during growth, by incorporating more teeth into each row, the dental battery became deeper in saurolophids. Previous research supports the interpretation that this is the result of more tooth rows being stacked vertically, allowing the dental battery to work as a shock-absorber during mastication, and allowed teeth to be replaced without interruption to food consumption. The increased anterior inclination of the coronoid process, and relative elongation of the diastema in saurolophids are herein suggested to have evolved through hypermorphosis, a version of peramorphosis where the growth trajectory in the descendant taxon extends beyond the ancestral state, whereas the relative elongation of the coronoid process, the relative deepening of the dental battery, and the increased ventral deflection of the diastema are the result of a novel juvenile condition.

Keywords:

Hadrosauroidea, dentary, morphometric, heterochrony, peramorphosis, mastication

Degree Project E1 in Earth Science, 1GV025, 30 credits

Supervisor: Nicolás E. Campione

Department of Earth Sciences, Uppsala University, Villavägen 16, SE-752 36 Uppsala

(www.geo.uu.se)

(6)

Populärvetenskaplig sammanfattning

Morfologisk variation i det hadrosauroida dentärbenet

D. Fredrik K. Söderblom

Hadrosauroider, även kallade ank-näbbs dinosaurier, var vanligt förekommande och nästan globalt distribuerade växtätare som levde under krita-perioden. Forskning har länge tytt på att deras framgång berott på deras förmåga att processera föda genom att tugga, en förmåga som länge ansågs exklusiv till däggdjur. Denna förmåga förbättrades sedan genom morfologiska förändringar som dök upp i de saurolophida hadrosauroidera. Dessa förändringar ledde till både effektivare födoinsamling och bearbetning. En serie analyser utfördes på 89 dentärben (ett av benen i underkäken) för att testa hur dessas form varierade inom Hadrosauroidea, beroende på gruppers släktskap, ålderstadie, samt individuell variation.

Resultaten tyder på att diastemat, området som i hadrosauroider separerar predentärbenet från dental batteriet (liknande hur fram- och kindtänderna är separerade hos hästar) blev längre i saurolophider. Detta tillät ökad specialisering av predentärbenet till insamling av mat och i dental batteriet (rader av tätt sittande tänder placerade vertikalt på varandra) till att mala ned mat. Diastemats ventrala böjning avtog i alla grupper allt eftersom individen åldrades, men var i helhet mer böjt i saurolophider. Detta tolkas som möjlig anpassning till betande på lågt växande vegetation, som idag ses i vissa hovdjur (Ungulata). Koronoid processen, en vertikalt utstickande del av dentärbenet som är placerad långt bak, lutades framåt och förlängdes i saurolophider. Denna förändring riktade muskler till att dra käken mer bakåt, samt ökade hävstångsverkan, vilket ökade saurolophiders bitkraft. Alla hadrosauroider gick igenom förlängning av dental batteriet genom tillägg av tänder, vilket ökade tuggytans area. Saurolophiders dental batterier var djupare än sina föregångares, vilket har dokumenterats i litteratur som ännu fler vertikalt staplade rader. Fler tandrader betydde att stötar från starkare bitkraft kunde absorberas över en större yta. Dessutom så kunde tandrader därför bytas utan dröjsmål när den översta raden nötts ned, vilket tillät oavbruten födokonsumtion. Anledningen till den ökade lutningen i koronoid processen, och förlängningen av diastemat tros vara hypermorfos, vilket är när tillväxten av något som slutade växa i förfäderna istället fortsätter växa i ättlingarna. Den ökade längden på koronoid processen, det djupare dental batteriet, och det mer ventralt böjda diastemat i saurolophider uppstod p.g.a. att ett nytt juvenilt tillstånd utvecklades.

Nyckelord: Hadrosauroidea, dentärben, morfometrisk, heterokroni, peramorfos, tugga

Examensarbete E1 i geovetenskap, 1GV025, 30 hp

Handledare: Nicolás E. Campione

Institutionen för geovetenskaper, Uppsala universitet, Villavägen 16, 752 36 Uppsala

(www.geo.uu.se)

(7)

Table of Contents (continued)

Appendix ...49

(9)

1 1 Introduction

Hadrosauroid dinosaurs (Hadrosauroidea von Huene 1954) were Cretaceous herbivores capable of processing their food through mastication (chewing) (Weishampel 1983; Norman & Weishampel 1985; Weishampel & Norman 1989; Horner, Weishampel & Forster 2004). Mastication was long thought to be unique to mammals, but it is now indicated to have been present in several in animal groups (e.g. mammals, edaphosaurids, dicynodonts, procolophonians, bolosaurids, ceratopsians, ankylosaurs, extant herbivorous turtles, some lizards and the eusuchian Iharkutosuchus) where it is used to mechanically break down food into smaller particles, thereby increasing its rate of chemical breakdown by increasing the area available for breakdown (Norman & Weishampel 1985; Reilly, McBrayer & White 2001; Ősi and Weishampel 2009; Tanoue et al. 2009; Ősi et al. 2014; Varriale 2016). This in turn increases the organism’s ability to meet metabolic demands. As vertebrates lack the enzymes to break down plant cell walls, they make use of micro-organisms to ferment their food. If food has been masticated before reaching the gut, fermentation becomes more efficient, and the energy-cost of herbivory as a foraging strategy is decreased (Reilly, McBrayer & White 2001).

The first hadrosauroids appeared in the Early Cretaceous of Asia (You et al. 2003; Norman 2014; Shibata et al. 2015) and before they went extinct at the end of the Late Cretaceous hadrosauroids inhabited Asia, Europe, North- and South America, Antarctica (Horner, Weishampel & Forster 2004) and possibly Africa (Fanti et al. 2016). Their ability to successfully inhabit many continents and thrive has often been attributed to a highly efficient masticatory apparatus that evolved through several modifications of the skull (e.g. Weishampel 1984; Norman & Weishampel 1985; Erickson et al. 2012). Evolutionary differences in these morphological modifications are used in phylogenetic systematics, but the morphology of some evolutionary modifications can both arise due to changes in the way ontogeny (growth) progresses (i.e. heterochrony = an evolutionary pattern due to changes in rate, duration or timing of a trait’s development [Schoch 2014]) (e.g. the extreme cranial crest of

Parasaurolophus [Farke et al. 2013]). The morphology of the evolutionary modifications change also

during ontogeny (Maryańska & Osmólska 1981; Bell 2011; Campione & Evans 2011; Prieto-Márquez 2011, 2014; Farke et al. 2013; McGarrity, Campione & Evans 2013; Prieto-Márquez & Gutarra 2016), thereby placing juvenile specimens in a position further back in phylogenies than adult specimens of the same species (Tsuihiji et al. 2011; Campione et al. 2013; Prieto-Márquez 2014).

(10)

2

offspring that create minor differences among individuals (Campione & Evans 2011). On top of taxonomic, ontogenetic and individual variation, it is possible for pathology (illness or injury) (Godefroit, Bolotsky & Bolotsky 2012; Freedman Fowler & Horner 2015), and taphonomy (post-mortem deformation, e.g. crushing [Campione & Evans 2011]) to alter the morphology of hard tissues.

2 Aims

(11)

3

Figure 1. Hadrosauroid skull anatomy exemplified by E. annectens (formerly E. saskatchewanensis [Campione

(12)

4 3 Background

3.1 Cretaceous climate and environment

Exact global temperatures during the Cretaceous are as of the present date not agreed upon, but studies mostly point to higher temperatures than present-day (an extended review in Hay 2017). It is currently debated if the North and South Poles were frozen and to what degree (Hay 2017), but fossil evidence of Antarctic pteridophytes (ferns), bennettitales, ginkgoes, bryophytes (mosses), hepatophytes (liverworts), lycophytes (e.g. clubmosses), conifers and angiosperms (flowering plants) in the late Early Cretaceous (Cantrill & Poole 2002; Nagalingum & Cantrill 2006; Vera 2015), and pteridophytes, bryophytes, bennettitales, cycads, angiosperms and conifers in the Late Cretaceous (Césari, Marenssi & Santillana 2001; Iglesias 2016; Kvaček & Vodrážka 2016), suggest a warmer global climate than today. Due to elevated temperatures and less frozen water, Cretaceous global sea-levels were higher than present and although the exact sea-levels are still debated (Hay 2017), it is hypothesized that several continents, especially Africa, Europe and North America, were partially covered by shallow seas (Markwick & Valdes 2004; references in Müller et al. 2008).

3.2 Taxonomic conventions and the dentary’s use in phylogenetics

It is generally agreed that Hadrosauridae is located within Hadrosauroidea (figure 2), however what taxa are included in Hadrosauridae and how they are distributed within the clade slightly differs depending on what character matrix is used. Two major matrices that are currently used to generate phylogenies were originally created by Prieto-Márquez (2010a), and Xing et al. (2014). Subsequently, the matrices were modified and used to generate phylogenetic trees that are more up-to-date (i.e. Márquez, Erickson and Ebersole 2016a; Xing, Mallon & Currie 2017). The matrix by Prieto-Márquez, Erickson and Ebersole (2016a) has 273 characters, of which 24 (approximately 8.8%) characters are related to the dentary (14 regarding the dentary itself, 10 regarding the dentary teeth). The matrix by Xing, Mallon and Currie (2017) has 346 characters, of which 39 (approximately 11.3%) are related to the dentary (22 regarding the dentary itself, 17 regarding the dentary teeth). The dentary therefore has quite a big influence on phylogenetic analyses, considering that approximately 1/10 of the characters in both matrices are related to the dentary or its teeth.

Although trees generated by the two matrices agree that Hadrosauridae Cope 1870 nests within Hadrosauroidea, they differ in the placement of Hadrosaurus foulkii Leidy 1858, and as such, the names and definitions of other clades within Hadrosauridae also change. Prieto-Márquez, Erickson and Ebersole (2016a) define Hadrosauridae as the clade stemming from the most recent common ancestor of Parasaurolophus walkeri Parks 1922 and Hadrosaurus foulkii (sensu Prieto-Márquez 2010a). In this tree Hadrosauridae includes Hadrosaurus foulkii, Eotrachodon orientalis and Saurolophidae. Saurolophidae (sensu Prieto-Márquez 2010a) is defined as the last common ancestor of

(13)

5

Saurolophidae is divided into Saurolophinae and Lambeosaurinae. Saurolophinae (sensu Prieto-Márquez 2010a) is defined as Saurolophus osborni and all taxa more closely related to it than

Lambeosaurus lambei or Hadrosaurus foulkii. Lambeosaurinae (sensu Prieto-Márquez 2010a) is

defined as Lambeosaurus lambei and all taxa more closely related to it than or Edmontosaurus regalis Lambe 1917, Hadrosaurus foulkii or Saurolophus osborni. Xing, Mallon and Currie (2017) define Hadrosauridae as the least inclusive taxon containing Parasaurolophus and Saurolophus (sensu Sereno 1998). In their tree, Hadrosaurus places within what Prieto-Márquez, Erickson and Ebersole (2016a) call Saurolophinae, which changes the name to Hadrosaurinae Lambe 1918 (sensu Sereno 1998), defined by Xing, Mallon and Currie (2017) as all hadrosaurids closer to Saurolophus than to

Parasaurolophus. Lambeosaurinae is not explicitly defined by Xing, Mallon and Currie (2017) but it

is herein assumed that they follow the definition sensu Sereno (1998), in which Lambeosaurinae is defined as all hadrosaurids closer to Parasaurolophus than to Saurolophus. In their phylogeny, E.

orientalis places in a polytomy just outside Hadrosauridae. As Hadrosauridae then only includes

(14)

6

Figure 2. Taxonomic scheme used herein, based largely on that of Prieto-Márquez, Erickson and Ebersole (2016a) and additional sources (see text). Outgroup taxon is

Iguanodon bernissartensis. Non-hadrosauroid iguanodontian taxa are marked with black-and-white lines. Non-hadrosaurid hadrosauroid taxa are marked with green lines.

(15)

7 3.3 Evolution of the feeding apparatus and jaw mechanics in hadrosauroids

The oldest fossils of hadrosauroids are from the late Early Cretaceous (Aptian) and belong to

Sirindhorna khoratensis (Shibata et al. 2015). Unlike their ancestors, hadrosauroids had (1) more

sophisticated teeth arranged into large grinding surfaces (dental batteries [dentary anatomy in figure 4 in the methods section]) to maximize the efficiency of mastication, (2) larger jaws to maximize the space for mastication that were separated into food gathering in the anterior (premaxillary- and predentary beaks) and food processing in the posterior (dental batteries), which allowed these elements to specialize in their respective tasks, (3) dentaries with more elongate coronoid processes that could generate more leverage and thereby increased bite force, and (4) explored large body sizes, with a posture to compensate for the increased weight (Norman 2014). Further evolution along the lineage, including hadrosaurid hadrosauroids which appeared during the Santonian (Late Cretaceous) (Prieto-Márquez 2010b; Prieto-(Prieto-Márquez, Erickson & Ebersole 2016a), would yield taxa with (5) more closely interlocking teeth in increased numbers of rows (Norman 2014), (6) dental batteries that elongated in a posterior direction medially past the coronoid processes to enable more surface for mastication, and (7) taller coronoid processes for increased leverage, whose apex was expanded in order to recruit more muscles for mastication (Norman 2014).

(16)

8

forms in the Late Cretaceous such as the saurolophine Shantungosaurus giganteus with an estimated mass of approximately 17,000 kg (Benson et al. 2014).

(17)

9

Figure 3. Two hadrosauroid jaw mechanic mastication models. A, pleurokinesis in cross sectional view,

including maxillae (above, middle grey) and dentaries (below, middle grey), with dental batteries (dark grey). B, alternative mastication model in left lateral view (top image), and cross sectional view (bottom image), including maxillae (above, middle grey) and dentaries (below, middle grey), with dental batteries (dark grey). All images based on those of Lambe (1920), and Nabavizadeh (2014).

3.4 Examples of previous research on morphological variation in

hadrosauroid dentaries

3.4.1 Taxonomic variation

Compared to non-hadrosauroid neoiguanodontians, non-hadrosaurid hadrosauroids had more elongate jaws (including dentary), and more extensive elongation of the jaws would not develop until saurolophids appeared (Norman 2014). The dental battery was situated anterior to the coronoid process in early hadrosauroids, however in later forms the dental battery’s posterior edge had migrated medially relative to the coronoid process (Norman 2014).

(18)

10

saurolophids. Kubota and Kobayashi (2009) stated that diastemata in Eolambia caroljonesa and

Shuangmiaosaurus gilmorei were short or non-existent when compared to dentary length, whereas the

diastema in Protohadros byrdi was quite elongate in comparison. The saurolophines Prosaurolophus

maximus and Edmontosaurus were reported to have long diastemata (McGarrity, Campione & Evans

2013) which is in agreement with Kubota and Kobayashi (2009)’s general statement on the presence of longer diastemata in saurolophids; however Prieto-Márquez (2010c) found the saurolophine

Gryposaurus to have a comparatively short diastema.

Norman (2014) suggested that earlier non-hadrosaurid hadrosauroids (e.g. Probactrosaurus,

Protohadros and Altirhinus) started to develop a downturned anterior region of the dentary, which

became even more so in saurolophids. A statement made by Prieto-Márquez (2010c) is in line with Norman (2014)’s general statement in that Gryposaurus exhibits a short but strongly ventrally curved symphyseal process. A strong ventral curvature can also be seen in the dentary of lambeosaurines according to Godefroit, Bolotsky and van Itterbeeck (2004); however Evans (2010) mentioned that the lambeosaurine Parasaurolophus is an exception.

Norman (2014) stated that non-hadrosaurid hadrosauroids possessed tall coronoid processes situated perpendicular to the dentary rami that in hadrosaurids are further elongated. Both Evans (2010) and Blanco, Prieto-Márquez and De Esteban-Trivigno (2015) stated that the coronoid processes in hadrosaurids are anteriorly inclined whereas Norman (2014) mentioned that anteriorly inclined coronoid processes are found in saurolophids (called Euhadrosauria in Norman 2014) and that the apex of the coronoid processes are more expanded. Examples of anteriorly inclined coronoid processes are evident in the lambeosaurines Arenysaurus and Blasisaurus (Blanco, Prieto-Márquez and De Esteban-Trivigno 2015). An odd combination of features pertaining to the coronoid processes are present in the non-saurolophid hadrosaurid Eotrachodon orientalis that possessed a vertically oriented coronoid process with a moderate expansion of the apexes (Prieto-Márquez, Erickson and Ebersole 2016b), although it should be noted that the individual died as a subadult and therefore might not closely reflect adult morphology.

3.4.2 Dentary growth patterns

(19)

11

dental batteries of the saurolophines Edmontosaurus and Saurolophus elongate through ontogeny, which is also true for the non-hadrosaurid hadrosauroid E. caroljonesa according to Kirkland (1998). McGarrity, Campione and Evans (2013) were able to make a similar claim for P. maximus, which exhibited positive allometry in the length of the dentary, indicating that as P. maximus grew, the length of the dentary increased at a relatively faster rate. The same study could not reject isometry for the rate of growth in dentary height. Maryańska and Osmólska (1981) stated that the relative length of the dental battery in Saurolophus angustirostris occupied a larger portion of the mandible in adults than in juveniles.

Maryańska and Osmólska (1981) have also mentioned that the coronoid process of S. angustirostris went from being directed dorsally in juveniles to more anteriorly in adults. Prieto-Márquez and Gutarra (2016) saw similar development of the coronoid process’ angle in Gryposaurus, and also included that the apex expanded in an anteroposterior direction through ontogeny. Unlike S.

angustirostris and Gryposaurus, the coronoid process of Bactrosaurus johnsoni, a non-hadrosaurid

hadrosauroid, remains vertical throughout ontogeny (Prieto-Márquez 2011).

The symphyseal process in the dentary of B. johnsoni has been described as increasing in ventral curvature throughout ontogeny (Prieto-Márquez 2011). However, the same region in another non-hadrosaurid hadrosauroid, E. caroljonesa, does not show increase in symphyseal process curvature with increased growth (Kirkland 1998). In Edmontosaurus, the curvature of the symphyseal process slightly decreases through ontogeny (Prieto-Márquez 2014) whereas in Amurosaurus riabinini, a lambeosaurine, the ventral deflection increases with growth and moves posteriorly (Godefroit, Bolotsky & van Itterbeeck 2004).

Kubota and Kobayashi (2009) measured the length of the diastema compared to the length of the dentary in several specimens and noticed that saurolophids had more elongate diastemata than non-hadrosaurid iguanodontians. The same study also found that saurolophines had an elongated diastema in the subadult stage, whereas lambeosaurines developed an elongated diastema transitioning from subadult to adult. Prieto-Márquez (2014), and Godefroit, Bolotsky and van Itterbeeck (2004) have expressed comparable opinions regarding the diastemata of E. annectens and A. riabinini, respectively, stating that they elongate relative to dentary length during ontogeny. McGarrity, Campione & Evans (2013) stated the same region in P. maximus shows isometric growth, which is contrary to what is interpreted in Kubota and Kobayashi (2009)’s figure 2, however unbeknownst to Kubota and Kobayashi (2009) as specimens belonging to Prosaurolophus blackfeetensis were found to be of a younger ontogenetic stage of P. maximus (McGarrity, Campione & Evans 2013).

3.4.3 Individual variation, taphonomic- and pathologic alteration

(20)

12

(21)

13 4 Methods

4.1 Data collection

Photographs of hadrosauroid dentary specimens in medial view along with information on image source, taxonomic affinity, and ontogenetic stage (juvenile, subadult, or adult) were obtained from the literature, and the personal digital libraries of N. Campione and A. Prieto-Márquez (table 9 in the appendix). The information was then compiled into a dataset consisting of 89 specimens representing 37 species (73 specimens), three genera indeterminate at the species level (ten specimens), two indeterminate hadrosauroids, two indeterminate hadrosaurids, one indeterminate saurolophid, and one indeterminate lambeosaurine. Medial view was chosen so the dental battery could be digitized, as it is not fully visible in lateral view. The choice of specimens included in the dataset was based on specimen completeness. Only specimens deemed to have a well-preserved coronoid process, outline of the dental battery, and portion of the dentary anterior to the dental battery were included. Specimens with minor damage were included if it was believed that missing data could be confidently approximated by knowledge of the anatomy of other specimens. If ontogenetic stage was not stated in literature, it was inferred based on the relative development of the crest or by comparing the size of the dentary following Evans (2010), who suggested an arbitrary division of ontogenetic stages where an individual is juvenile, subadult or adult if skull length is < 50%, 50-85%, or > 85%, respectively, of the maximum skull length in observed that taxon. Taxa known from a single dentary specimen that were not assigned to a particular ontogenetic stage were assumed to be adults due to the absence of comparative specimens, unless a closely related taxon of similar size could be compared with.

4.2 Digitization

(22)

14

converted to homologous landmarks in Notepad. tpsDIG2 was used to derive a scale factor from the amount of pixels contained along scale bars of predetermined length in digitized images. The scale factors for each image would be used in the calculation of centroid size during the generalized procrustes analysis.

Figure 4. Schematic drawing of the landmark configuration, with anatomy of the hadrosauroid dentary included.

Dark blue lines represent equi-distant semilandmark curves. Dark blue numbers are the amount of semilandmarks present in each curve. Yellow points mark the ends of the curves that are homologous and hence do not slide. Yellow points are named using black numbers, and their positions are explained in table 1.

Table 1. Locations of the homologous landmarks and the ends of semilandmark curves (figure 4).

No# Explanation

1 Posteriormost point of the dental battery.

2 Anteriormost point of the dental battery.

3 Anteriormost point of the symphyseal process.

4 Posteriormost point of the angular facet.

5 Intersection of the Meckelian fossa and the base of the anteroposteriorly expanded region of the coronoid process.

6 Anterior base of the anteroposteriorly expanded region of the coronoid process.

4.3 Generalized procrustes analysis and semilandmark sliding

(23)

15

Excel. By sliding semilandmarks, the variation of individual semilandmarks was removed and curve shapes were averaged, making curves comparable between specimens (Bookstein et al. 2002). Centroid size was derived in the GPA using the distance of landmark coordinates from a mean value together with the scale factors. Centroid size was later used to generate bivariate allometry plots.

4.4 Principal component analysis, morphospace diagram and density plots

Axes of major variation, also called principal components (PCs), were isolated in R using the output of the plotTangentSpace() function in the package geomorph. A morphospace diagram was plotted using PCs 1 and 2. The function summary() was used to obtain the proportions of variation (i.e. percentage of variation) explained by each PC axis. Proportions of variation were plotted onto corresponding PCs in the morphospace diagram. The function plotRefToTarget() from the package geomorph was used to generate thin-plate spline (TPS) grids of the shape extremes at the PCs minima and maxima, which added to the morphospace to visualize how certain aspects of morphology changed along the main axes of variance. Density plots separated by taxonomy and ontogenetic stage were generated using PCs with the function density() and were placed to match the principal component axes in morphospace to provide further clarification of groups’ distributions along the PCs.

4.5 Normality tests and principal component group separation

To statistically assess the differences of major groups in morphospace, suitable tests had to be chosen. Normality tests were used to determine the suitability of conducting parametric tests (i.e. analysis of variance and Tukey’s honest significant difference test) on the groupings of the sample herein through use of the function parametric.tests() (function obtained from N. Campione). All normality tests are unable to reject normality, with the exception of non-hadrosaurid hadrosauroids along PC2 that were only weakly non-normal (W=0.8774280, p=0.04344031). As groups generally approach a normal distribution, standard parametric analyses of variance (ANOVA) and Tukey’s honest significant difference (TukeyHSD) tests were carried out on PC1 and PC2 using the functions aov() and TukeyHSD(), which compared mean values of groups, divided both by taxonomy and ontogenetic stage, and tested for significant differences. The aov() and TukeyHSD() functions were also run with the Edmontosaurus annectens specimen AMNH 5730 excluded to investigate its effects on the sample, as its position along PC1 in the morphospace diagram supported previous interpretations that it might have been dorso-ventrally crushed (Campione & Evans 2011). The same tests were also run with

Eotrachodon orientalis as a non-hadrosaurid hadrosauroid, based on the phylogeny by Xing, Mallon

(24)

16

W=0.9365358, p=0.30872298, respectively), and weak normal distributions in PC2 for both groups (W=0.9483623, p=0.0786448, and W=0.8937011, p=0.06376394, respectively).

4.6 Bivariate allometry plots and confidence intervals

(25)

17 5 Results

5.1 Morphospace diagram and density plots

5.1.1 Principal component 1

PC1 is responsible for 31.63% of the total variation. PC1 depicts the variation in inclination of the coronoid process and the relative length of the diastema (figure 5). Specimens closer to the negative end of PC1 have an anteriorly inclined coronoid process, and an elongated diastema relative to dentary length. Specimens closer to the positive end of PC1 have a dorsally inclined coronoid process, and a shortened diastema relative to dentary length.

Figure 5. Morphospace diagram of principal components 1 and 2 with density plots separated by clade for easier

visualization. Major groups in morphospace are divided according to taxonomic affinity, which is visualized by overall color separation (green, red, blue, and yellow shades). Boxes show minimum and maximum extents of groups along respective principal components. Non-hadrosaurid hadrosauroids are shown in green colors and include Altirhinus kurzanovi (A), Bactrosaurus johnsoni (B), Eolambia caroljonesa (E), Jeyawati rugoculus (J),

Plesiohadros djadokhtaensis (P), Probactrosaurus gobiensis (p), Protohadros byrdi (b), Shuangmiaosaurus gilmorei (S), Sirindhorna khoratensis (s) and Telmatosaurus transsylvanicus (T). Indeterminate juvenile

hadrosauroids are marked by black question marks. Non-saurolophid hadrosaurids are shown in red color only include Eotrachodon orientalis (E). Indeterminate juvenile hadrosaurids are shown as grey question marks. Indeterminate hadrosaurids of unknown ontogenetic stage are shown as brown question marks. Juvenile saurolophids are marked by purple question marks. Lambeosaurines are shown in blue colors and include

Arenysaurus ardevoli (A), Amurosaurus riabinini (a), Blasisaurus canudoi (B), Charonosaurus jiayinensis (j), Corythosaurus casuarius (c), Corythosaurus intermedius (i), Corythosaurus sp. (C), Hypacrosaurus stebingeri

(H), indeterminate lambeosaurine (?), Lambeosaurus lambei (L), Olorotitan arharensis (O), Parasaurolophus

(26)

18

Velafrons coahuilensis (V). Saurolophines are shown in yellow colors and include Acristavus gagslarsoni (A), Brachylophosaurus canadensis (B), Edmontosaurus annectens (E), Edmontosaurus regalis (e), Gryposaurus alsatei (a), Gryposaurus latidens (l), Gryposaurus notabilis (n), Gryposaurus sp. (G), Kritosaurus navajovius

(K), Maiasaura peeblesorum (M), Prosaurolophus maximus (P), Saurolophus angustirostris (S), Ugrunaaluk

kuukpikensis (U) and Willinakaqe salitralensis (W). Ontogenetic stages are divided by shade of color (dark,

medium and light) and line type (dotted, dashed and solid). Juveniles are denoted by dark shades and dotted lines, subadults by medium shades and dashed lines, and adults by light shades and solid lines.

Non-hadrosaurid hadrosauroids generally have dentaries with quite dorsally directed coronoid processes and relatively short diastemata compared to the saurolophids (saurolophines and lambeosaurines), with the exception of Plesiohadros djadokhtaensis (MPC-D100/745). Lambeosaurines display dentary morphologies with more anteriorly inclined coronoid processes as well as more elongate diastemata than non-hadrosaurid hadrosauroids. Saurolophines show the most extreme range of coronoid process angles and relative diastema lengths, even if the aberrant

Edmontosaurus annectens specimen at the far negative side of PC1, AMNH 5730, is ignored.

Saurolophinae also displays the most anteriorly inclined coronoid processes and elongate diastemata. When comparing ontogenetic stages, the juveniles generally display vertical to subvertical coronoid processes and relatively short diastemata (figure 5), the adults generally display anteriorly inclined coronoid processes and relatively long diastemata, and the subadults display intermediate coronoid process inclinations and relative diastema lengths, although all ontogenetic stages overlap partially. Lambeosaurine juveniles are an exception, having possessing coronoid processes that are more inclined and diastemata that are relatively longer than some lambeosaurine subadults. However, the restricted extent of the lambeosaurine juveniles along PC1 is perhaps due to low sample size (N= 4).

The non-hadrosaurid hadrosauroid juveniles and lambeosaurine juveniles only show minor similarities to each other, but both are very similar to the saurolophine juveniles. The saurolophine juveniles display a morphological variation in the angle of the coronoid process and relative length of the diastemata equivalent to the entire range of non-hadrosaurid hadrosauroid grade (including juveniles, subadults, and adults). The lambeosaurine subadults display wider morphological variation than the saurolophine subadults in these aspects. The subadults of both clades have more anteriorly inclined coronoid process and relatively longer diastemata than the non-hadrosaurid hadrosauroid subadults, but partially overlap morphologically. The same pattern is also visible when the adult growth stages of the taxonomic groups are compared. The taxa with the most anteriorly inclined coronoid processes and relatively longest diastemata of the adult stages are Plesiohadros

djadokhtaensis in non-hadrosaurid hadrosauroids, Tsintaosaurus spinorhinus and Olorotitan arharensis in the lambeosaurines, and Edmontosaurus in the saurolophines. The juveniles of all

(27)

19

adult non-hadrosaurid hadrosauroids seem to have experienced further increase in coronoid process inclination and diastema elongation, this increase is minor compared to the increase in coronoid process inclination and relative diastema elongation experienced in saurolophids when transitioning from subadult to adult stage.

Individual specimens of the non-hadrosaurid hadrosauroids Eolambia caroljonesa and

Bactrosaurus johnsoni are confined to small stretches along PC1, which could be due to an actual

pattern of limited individual variation, or due to small sample size. One indeterminate juvenile hadrosauroid is located further along the positive extent of PC1 than E. caroljonesa and B. johnsoni and therefore possesses the most vertical coronoid process and relatively shortest diastema in the entire sample. The only non-saurolophid hadrosaurid specimen in the sample belongs to a subadult

Eotrachodon orientalis possessing a vertical coronoid process and a relatively short diastema,

displaying a similar morphology to non-hadrosaurid hadrosauroids, and juvenile saurolophines.

The lambeosaurine Corythosaurus shows a somewhat wide distribution of individuals, especially subadults. Lambeosaurus lambei is represented herein by three adult specimens that show a fairly wide individual variation. The two subadult specimens of Sahaliyania elunchunorum show similar coronoid process inclination and relative diastema length. Tsintaosaurus spinorhinus is represented by two adult specimens that display fairly different inclinations of the coronoid process and elongation of the diastema.

The saurolophine juveniles of the taxon Willinakaqe salitralensis show a wide spread that spans approximately 2/3 of the growth stage’s group along PC1. Ugrunaaluk kuukpikensis displays approximately as much individual variation as W. salitralensis. Subadult Edmontosaurus display relatively little individual variation (but N=2), whereas subadult Brachylophosaurus specimens show moderate individual spread. Adult Edmontosaurus and Brachylophosaurus specimens display moderate individual variation (with AMNH 5730 excluded). Prosaurolophus maximus shows moderate individual variation in the subadult stage and little variation in the adult stage, although it is only represented by two subadult and two adult specimens. Subadult Gryposaurus specimens (two of three specimens belong to G. notabilis) show moderate individual variation, whereas three adult

Gryposaurus specimens belonging to G. alsatei, G. latidens and G. notabilis are positioned within the

individual variation of the subadults. As the adult Gryposaurus sample’s variation is composed of three different species, there is an uncertainty of whether the variation is individual or taxonomic in nature.

5.1.2 Principal component 2

(28)

20

ventrally oriented diastema. Specimens closer to the positive end of PC2 have a short coronoid process, a dorso-ventrally shallow dental battery, and a diastema that is roughly in line with the main axis of the dentary. In general, non-hadrosaurid hadrosauroids have relatively short coronoid processes, shallow dental batteries, and anteriorly angled diastemata; whereas saurolophines and lambeosaurines have relatively tall coronoid processes, deeper dental batteries, and ventrally angled diastemata. Saurolophines and lambeosaurines are similar in these aspects, although saurolophines display slightly more extreme morphologies.

The juveniles generally possess dentaries that are anteroposteriorly short and dorsoventrally elongate, as a result of relatively tall coronoid processes, deep dental batteries and ventrally angled diastemata, whereas adults generally possess more anteroposteriorly elongate and dorsoventrally short dentaries, as a result of relatively shorter coronoid processes, shallower dental batteries, and anteriorly angled diastemata. Subadults display intermediate dentary morphologies. Non-hadrosaurid hadrosauroid juveniles only display morphologies regarding these aspects that are present in non-hadrosaurid hadrosauroid subadults (but not the reverse), likely due to low sample size (N=5) and few taxa being included in the group (Bactrosaurus johnsoni and Eolambia caroljonesa). Lambeosaurine and saurolophine juveniles have similar relative coronoid process lengths, dental battery depths, and diastema angles, whereas the non-hadrosaurid hadrosauroid juveniles are dissimilar to the saurolophid juveniles. The subadults of all taxonomic groups display somewhat similar morphologies. However, the lambeosaurine subadults display morphologies more similar to the saurolophine subadults than to the non-hadrosaurid hadrosauroid subadults. The adults of all taxonomic groups are somewhat similar to each other, and the adults are also morphologically similar to the subadults of the respective taxonomic groups. The only taxonomic group where the adults display very different morphologies from the juveniles are the lambeosaurines, likely due to the juveniles’ low sample size (N=4). The only non-saurolophid hadrosaurid in the sample, Eotrachodon orientalis exhibits a somewhat, albeit not extremely elongate dentary, with a relatively short coronoid process, shallow dental battery, and anteriorly angled diastema, similar to that in subadult and adult non-hadrosaurid hadrosauroids, subadult and adult lambeosaurines, as well as adult saurolophines.

Individual variation of the relative coronoid process length, dental battery depth and diastema angle in E. caroljonesa and B. johnsoni along PC2 is quite limited (figure 5), although the sample size is low (N=4 & 3). The lambeosaurine Corythosaurus shows slightly broader individual variation within separate growth stages than E. caroljonesa and B. johnsoni; however another lambeosaurine,

Lambeosaurus lambei, shows slightly less variation. The individual variation of Sahaliyania elunchunorum and Tsintaosaurus spinorhinus is limited, but the two species are only represented by

(29)

21

taxonomic variation. Subadult Brachylophosaurus is represented merely by two specimens that show minor to moderate variation. Both Brachylophosaurus and Edmontosaurus show somewhat wide individual variation in the adult stages. The Prosaurolophus maximus is limited in individual variation as well, both in subadult and adult stages, but is represented only by two subadults and two adults. Both subadult and adult Gryposaurus specimens show moderate individual variation.

5.2 Normality tests

All normality tests proved insignificant, except for PC2 in the group containing all non-hadrosaurid hadrosauroids which proved weakly significantly different from normal (table 2). As there was only one slightly non-normal distribution, standard parametric ANOVA and TukeyHSD tests were conducted.

Table 2. Significance (p-value) obtained from normality tests on groupings by taxonomy, as well as taxonomy

and ontogenetic stage combined for principal components (PC) 1 and 2. Significant p-values are in bold.

Group PC1 p-value PC2 p-value

(30)

22 5.3 Tukey’s honest significant difference test

5.3.1 Principal component 1

Non-hadrosaurid hadrosauroids show significantly different dentary morphologies concerning coronoid process inclination, and relative diastema length to both saurolophines and lambeosaurines (table 3). Saurolophines and lambeosaurines, however, cannot be statistically differentiated, displaying overall similar morphologies.

Table 3. Significance (p-value) obtained from Tukey’s honest significant difference tests grouped by taxonomy

for principal component (PC) 1. Significant p-values are in bold.

PC1 All non-hadrosaurid

hadrosauroids

All lambeosaurines All saurolophines

All non-hadrosaurid hadrosauroids

- 0.0417973 0.0000225

All lambeosaurines 0.0417973 - 0.2136951

All saurolophines 0.0000225 0.2136951 -

Most of the significant differences along PC1 (table 4) relate to the average dentary morphology of saurolophine adults compared to that of all non-hadrosaurid hadrosauroids, lambeosaurine subadults, and saurolophine juveniles. Saurolophine adults are indistinguishable from that of lambeosaurine adults and saurolophine subadults. Additionally, saurolophine adults are significantly different to that of lambeosaurine juveniles, displaying only a weakly insignificant p-value (approximately 0.066). Non-hadrosaurid hadrosauroid juveniles display significant differences to saurolophine and lambeosaurine adults. When non-hadrosaurid hadrosauroid juveniles are compared to saurolophine subadults, only a weakly insignificant difference (approximately 0.091) is present. All other separations of groups along PC1 are insignificant.

(31)

23

Table 4. Significance (p-value) obtained from Tukey’s honest significant difference tests grouped by ontogenetic stage within taxonomic groups for principal component (PC)

1. Non-had = Non-hadrosaurid hadrosauroid, Lam = Lambeosaurine, Sau = Saurolophine. Significant p-values are in bold. Asterisks (*) indicate group-comparisons that become significant when the specimen AMNH 5730 is removed.

(32)

24 5.3.2 Principal component 2

hadrosaurid hadrosauroids are significantly different from saurolophines along PC2. Non-hadrosaurid hadrosauroids are barely insignificant from the lambeosaurines (table 5). Lambeosaurines and saurolophines cannot be distinguished statistically.

Table 5. Significance (p-value) obtained from Tukey’s honest significant difference tests grouped by taxonomy

for principal component (PC) 2. Significant p-values are in bold.

PC2 All non-hadrosaurid

hadrosauroids

All lambeosaurines All saurolophines

All non-hadrosaurid hadrosauroids

- 0.0503266 0.0000267

All lambeosaurines 0.0503266 - 0.1981771

All saurolophines 0.0000267 0.1981771 -

Non-hadrosaurid hadrosauroid juveniles can be differentiated from non-hadrosaurid hadrosauroid adults and saurolophine juveniles, but not other groups, including non-hadrosaurid hadrosauroid subadults along PC2 (table 6). Non-hadrosaurid hadrosauroid subadults are significantly different from lambeosaurine juveniles, as well as saurolophine juveniles and subadults, but are indistinguishable from all other groups. Non-hadrosaurid hadrosauroid adults are significantly different from all other groups except for non-hadrosaurid hadrosauroid subadults.

The lambeosaurine juveniles are along PC2 significantly different from non-hadrosaurid hadrosauroid subadults and adults, as well as lambeosaurine adults, but are indistinguishable from other groups, including lambeosaurine subadults. The lambeosaurine subadults are significantly different from non-hadrosaurid hadrosauroid adults and saurolophine juveniles, but are indistinguishable from all other groups, including lambeosaurine juveniles and adults. The lambeosaurine adults are significantly different from non-hadrosaurid hadrosauroid adults, lambeosaurine juveniles and saurolophine juveniles, but are indistinguishable from all other groups, including lambeosaurine subadults.

(33)

25

Table 6. Significance (p-value) obtained from Tukey’s honest significant difference tests grouped by ontogenetic stage within taxonomic groups for principal component (PC)

2. Non-had = Non-hadrosaurid hadrosauroid, Lam = Lambeosaurine, Sau = Saurolophine. Significant p-values are in bold.

(34)

26 5.4 Bivariate allometry plots and confidence intervals

Figure 6 consists of bivariate plots and regressions between axes of variation and size, in order to quantify growth patterns. All genera in the PC1 bivariate plots show negative slope coefficients except for the non-hadrosaurid hadrosauroid Eolambia (N=4) that shows a positive slope coefficient (figure 6A). All genera in the PC2 bivariate plots show positive slope coefficients except for the saurolophine

Prosaurolophus (N=4) that shows a negative slope coefficient (figure 6N). Except for the two

(35)

27

Figure 6. Bivariate allometry plots of logged centroid size against principal components (PCs). Plots are

generated by genera to increase sample size but divided to the species level when possible. Regression lines and corresponding slope coefficients were added for increased clarity. Taxonomic division is by color as in figure 5. Juveniles are denoted by dark shades, subadults by medium shades, and adults by light shades. Non-hadrosaurid hadrosauroids are shown in green and include Eolambia caroljonesa (E) seen in A (PC1) and B (PC2). Lambeosaurines are shown in blue and include Corythosaurus casuarius (c), C. intermedius (i) and

Corythosaurus sp. (C) seen in C (PC1) and D (PC2). Saurolophines are shown in yellow and include Brachylophosaurus canadensis (B) seen in E (PC1) and F (PC2); Edmontosaurus annectens (E) and E. regalis

(36)

28

(PC1) and J (PC2); Gryposaurus alsatei (a), G. latidens (l), G. notabilis (n) and Gryposaurus sp. (G) seen in K (PC1) and L (PC2); as well as Prosaurolophus maximus (P) seen in M (PC1) and N (PC2).

Table 7. Significance (p-value) of regression lines in allometry plots (figure 6), divided by genus and principal

component (PC). Significant p-values are in bold.

Genus PC1 p-value PC2 p-value

(37)

29 6 Discussion

6.1 Evolution and growth of the hadrosauroid dentary

6.1.1 Taxonomic variation

The results of the morphospace analysis and statistical tests (figure 5; tables 3 and 5) add quantitative support to the previously published patterns stating that throughout the evolution of hadrosauroids the coronoid process increased in relative length and anterior inclination, the diastema became relatively more elongate and ventrally rotated, and the dental battery deepened (Evans 2010; Blanco, Prieto-Márquez & De Esteban-Trivigno 2015; Norman 2014). The relatively increased separation of food collection by the predentary and premaxillary beaks in the anterior, and food processing by the teeth of the dental battery in the posterior in saurolophids as is supported herein (figure 5; table 3), would have allowed both of these areas to become more specialized and effective in their respective tasks (Norman 2014). Functional specialization of the dentary’s anterior in saurolophids evidenced herein by increased curvature of the diastema (figure 5; table 5). This morphological change is suggestive of functional modifications regarding food gathering abilities (Norman 2014). Similarly to the suggestion that the ventrally extending rhamphoteca (keratinous beak) of the premaxilla allowed saurolophids to crop plants at ground level more efficiently without needing to bend the neck as extensively (Farke et al. 2013), the increased curvature of the diastema could have served a similar purpose by moving the anterior area of the jaws lower. This idea is further supported by several studies on ungulates indicating that a ventrally curved snout is indicative of low-level grazing (Spencer 1995; references in Mallon et al. 2013). If this was the case in hadrosauroids, then saurolophids display a diastema curvature indicative of a more grazing diet than non-hadrosaurid hadrosauroids (with lambeosaurines being very weakly insignificant to non-hadrosaurid hadrosauroids), and further, saurolophines were a little more adapted to low-level graze than lambeosaurines (figure 5), although the latter cannot be statistically recognized herein (table 5). Additionally, it has been discussed that saurolophines possessed adaptations for living in more open habitats, whereas lambeosaurines were morphologically more adapted to closed habitats (Carrano, Janis & Sepkoski 1999), which would lend credit to the idea that saurolophines were more adapted to low-level graze than lambeosaurines. The anteroposterior motion of the lower jaw (Cuthbertson et al. 2012) that worked to move food posteriorly and at the same time grind it was augmented in saurolophids, through relatively taller and more anteriorly inclined coronoid processes that resulted increased moment arm length. This re-directed muscles connecting the coronoid process to the skull’s posterior and posterodorsal areas to pull in a more posterior direction, and lead to increased mechanical advantage, which made mastication more effective (Nabavizadeh 2014, 2016; Norman 2014) (figure 5, tables 3 and 5).

(38)

30

hadrosauroid taxa Altirhinus (Norman 1998), Eolambia (Kirkland 1998), Equijubus (You et al. 2003),

Probactrosaurus (Norman 2002) and Protohadros (Head 1998) whereas up to seven rows were

present in some saurolophid hadrosaurids (Xing, Mallon & Currie 2017). Previous research has shown that the pulp in the teeth of the upper-most (functional) tooth row in at least saurolophid hadrosaurids was replaced with dentine to allow the entire tooth to be ground down through mastication and afterward be replaced by a new row of teeth (LeBlanc et al. 2016). It could be the case that the multiple adaptations mentioned herein that lead to more efficient mastication likely caused to increased tooth wear due to the increased force generated. Increased tooth wear meant that additional rows of teeth enabled saurolophids to eat throughout their lives and as such increased the individual’s lifespan and reproduction potential. This idea is supported by research stating that development of the dental battery in saurolophids started early; already being similar in embryonic and neonatal individuals to that found in adults (LeBlanc et al. 2016), and would additionally decrease the need for dental battery development needed after hatching. Additionally, saurolophid teeth were likely connected by periodontal (soft, non-mineralized) tissue and ligaments rather than (rigid and mineralized) cementum, which gave the teeth the ability to work as shock absorbers during dental occlusion (contact of the maxillary and dentary teeth during jaw closure) (LeBlanc et al. 2016). It has not been investigated how the teeth in non-hadrosaurid hadrosauroid were connected; however major differences are herein thought to be unlikely. A deeper dental battery would mean more space for tooth rows that worked to dissipate the force of chewing, as well as made the grinding of plant matter more efficient and allowed continuous food intake throughout life.

The current results cannot resolve exactly when during hadrosauroid evolution that these morphological changes took place. E. orientalis’ phylogenetic placement would suggest that it should look more like a hadrosaurid, as it is just outside of Saurolophidae (Prieto-Márquez, Erickson & Ebersole 2016a). The fact that it doesn’t means that either (1) major shifts in the evolution of the lower jaw occur after Eotrachodon, within saurolophids, (2) major shifts occur before Eotrachodon which autapomorphically reverts to a more primitive state, or (3) the Eotrachodon specimen’s subadult nature makes it look like it has a more primitive condition. Alternatively, Xing, Mallon and Currie (2017)’s placement of E. orientalis as a non-hadrosaurid hadrosauroid could be considered affirmed by the placement of E. orientalis within the non-hadrosaurid hadrosauroids in morphospace, however when E. orientalis was added to the non-hadrosaurid hadrosauroid grade in the TukeyHSD tests, no significant change took place. If an adult E. orientalis were to be found and incorporated into the analysis performed herein it might help to shed light on the species’ development and hadrosauroid evolution. Morphometric analyses of other skeletal elements could also be performed in order to assess

(39)

31 6.1.2 Growth patterns and their taxonomic differences

Non-hadrosaurid hadrosauroids experienced little-to-no change in coronoid process inclination and relative diastema length throughout growth (PC1). Even though their occupation in morphospace seems to expand with increasing growth stage (figure 5), statistical tests could not detect significant differences coronoid process inclination and relative diastema length between stages (table 4). Most of the non-hadrosaurid hadrosauroid taxa herein are represented only by one growth stage, and only two taxa are represented by both juvenile and adult stages (Bactrosaurus johnsoni and Eolambia

caroljonesa), suggesting that there might be a taxonomical component to this pattern. Nevertheless,

the constrained position of these two taxa reflect what was also seen in the statistical tests for the stages of the entire grade (table 4), and even though isometry cannot be rejected, nor proven, for non-hadrosaurid hadrosauroid individual taxa (table 7), the low slope coefficient of E. caroljonesa (slope=0.02, figure 6A) vaguely hints at very low amounts of change during growth for the taxon (but N=4), and possibly for the grade.

Even though lambeosaurines seem to increase in coronoid process inclination and relative diastema length (figure 5), statistical tests cannot distinguish the stages’ shifts in morphospace from random (table 4). However morphospace occupation is quite narrow and sample size is quite low in the juvenile lambeosaurines (N=4), meaning that more extreme morphologies might have been excluded, which therefore alters the outcome of the statistical tests. In the saurolophines the increase in coronoid process inclination and relative diastema length (figure 5), is also reflected in the statistical tests (table 4), where the juveniles are significantly different from the adults, but the subadults cannot be distinguished from the juveniles, nor from the adults. This suggests a gradual increase in coronoid process inclination and relative diastema length in saurolophines (and possibly lambeosaurines which show a similar trend in morphospace occupation).

(40)

32

suggested for lambeosaurines by Mallon and Anderson (2015), due to juveniles not being able to generate as much bite force as adults. Their results however, were statistically insignificant, possibly due to sample size.

When coronoid process inclination and relative diastema length of the growth stages in a taxonomic group are compared to the same growth stages in the other taxonomic groups (figure 5; table 3), only the non-hadrosaurid hadrosauroid adults are significantly different from the lambeosaurine and saurolophine adults (when the taphonomically altered Edmontosaurus annectens specimen AMNH 5730 is excluded). As the lambeosaurine adults are insignificantly different from the saurolophine adults, it is interpreted that saurolophids went through additional increases in coronoid process inclination and relative diastema length in the transition from subadult to adult stage, compared to non-hadrosaurid hadrosauroids. The results therefore suggest possible long-term peramorphosis of these morphological aspects along the lineage leading from non-hadrosaurid hadrosauroids to saurolophids. Peramorphosis is defined as when the ontogenetic development of a trait in the descendant species extends beyond that of the ancestor species (Schoch 2014). The statement is meant to be very broad, as deviations from the overall trend are likely to have taken place through geological time. The statement is also made with caution, as the full sample size (89 specimens with AMNH 5730) is somewhat low. Even with the aforementioned shortcoming in mind, the results still fit with already published knowledge about taxonomic changes in coronoid process inclination and diastema elongation along the non-hadrosaurid hadrosauroid to saurolophid lineage (Norman 2014), suggesting that at least a general evolutionary pattern was still captured by the analyses herein, despite the sample size. Furthermore, there are three processes that lead to peramorphosis, all of which allow the descendant species to grow beyond that of the ancestral species. These include: (1) acceleration, whereby development is sped up, (2) hypermorphosis, whereby extension of the developmental trajectory takes place, and (3) pre-displacement, whereby developmental events take place earlier in growth (Alberch et al. 1979; Schoch 2014). The angle of the coronoid process and the relative length of the diastema in saurolophid juveniles and subadults are insignificantly different from that of non-hadrosaurid hadrosauroid juveniles and subadults (table 4), and thereby do not develop a difference in earlier stages, rejecting both acceleration and pre-displacement of development. That there is a significant difference between non-hadrosaurid hadrosauroid adults, and saurolophid adults (figure 5; table 4), suggests that the evolutionary increase coronoid process inclination and relative diastema length development in saurolophids relative to non-hadrosaurid hadrosauroids took place in the adult stage. This would be considered an extension of the developmental trajectory’s end, supporting hypermorphosis in coronoid process inclination and relative diastema length.

(41)

33

new tooth row could erupt below it without any interruption to mastication (LeBlanc et al. 2016). Lambeosaurines overall are thought to have gone through pre-displacement of the cranial crest, and

Parasaurolophus is thought to have gone through both pre-displacement and hypermorphosis of its

cranial crest compared to other lambeosaurines (Farke et al. 2013). The lambeosaurine Hypacrosaurus and the saurolophines Brachylophosaurus and Maiasaura all show pre-displacement in the development of the supraacetabular process of the ilium (Guenther 2009). Peramorphosis is not unique to hadrosauroids however, as several instances have been found dinosaurs, e.g. hypermorphosis in the limbs and feet of sauropods compared to other saurischians (Lockley & Jackson 2008), pre-displacement in the bone micro-structure of the titanosaur Lirainosaurus astibiae compared to other sauropods (Company 2011), and possible acceleration within the skull of carcharodontosaurid theropods (Canale et al. 2014).

(42)

34

and neonatal saurolophids (LeBlanc et al. 2016), which created more tooth rows in saurolophids (Xing, Mallon & Currie 2017) than in non-hadrosaurid hadrosauroids (Head 1998; Kirkland 1998; Norman 1998, 2002; You et al. 2003), and therefore a deeper dental battery relative to its length. This is why in all three taxonomic groups, as growth progresses, the coronoid process looks shorter, and the dental battery looks shallower, but these are relative to the elongation of the dental battery. This means that relative dental battery elongation and diastema angle straightening throughout growth progressed in the same direction and extent in all three taxonomic groups, although the morphologies exhibited at the same growth stages were different in saurolophids and non-hadrosaurid hadrosauroid, and that is why saurolophids do not reach the juvenile non-hadrosaurid hadrosauroid morphology until subadult/adulthood.

The combined growth- and heterochrony changes in hadrosauroids all suggest food selection- and processing-related adaptations. The increased anterior inclination of the coronoid process and the relatively longer diastema in saurolophids was achieved not merely by evolution alone but due to heterochrony, which increased mechanical advantage, directed muscles to pull posteriorly, and created functional specialization of the predentary and dental batteries (Nabavizadeh 2014, 2016; Norman 2014), in older growth stages compared to non-hadrosaurid hadrosauroids. All of the three taxonomic groups experienced relative elongation of the dental battery as well, indicative of increasing amounts of teeth per row, leading to larger surfaces for the grinding of plant matter which were appropriate considering the increase in size with advancing age. When this is considered with the increased relative diastema length in saurolophids, the resulting growth-related changes created adult herbivores with relatively very anteroposteriorly elongate dentaries. Furthermore, a growth-related straightening of the diastema hints at possible dietary niche partitioning between younger and older growth stages, similar to what was suggested by Mallon et al. (2013) on feeding height stratification. With a lower muzzle, young saurolophids would have reached low vegetation without needing to excessively bend their necks, whereas older (and therefore larger) saurolophids would have benefitted from the increased reach enabled by a straighter diastema when feeding from more elevated vegetation. This is further supported by the presence of multiple saurolophid pes prints, but absence of manus prints, around conifer and palm tree roots, and fallen logs in the Campanian-aged Blackhawk Formation of Utah, suggesting that saurolophids reared up against the trees to consume high-growing foliage (Mallon et al. 2013 and references therein).

(43)

35

annectens, and therefore chose to assign them to a separate genus. Recently, Xing, Mallon and Currie

(2017) declared Ugrunaaluk kuukpikensis a nomen dubium, as the Prince Creek Formation material in Mori, Druckenmiller and Erickson (2016)’s paper was solely composed of juveniles that were compared to adult E. annectens and E. regalis, meaning that Xing, Mallon and Currie (2017) considered the characters used by Mori, Druckenmiller and Erickson (2016) to differentiate the taxa as possibly being ontogenetically variable. As previous research has shown that using ontogenetically variable characters can cause non-adult specimens to place further back in phylogenies than adults of the same taxa (Tsuihiji et al. 2011; Campione et al. 2013; Prieto-Márquez 2014), it is herein considered that the decision made by Xing, Mallon and Currie (2017) to re-classify the Prince Creek Formation specimens as Edmontosaurus sp. was sensible, especially as no adult Prince Creek Formation specimens have been found to compare with adult Edmontosaurus specimens.

To investigate similarities and differences in the dentaries of the Prince Creek Formation specimens to Edmontosaurus, the Prince Creek Formation specimens were added to bivariate allometry plots with Edmontosaurus. When the combined Ugrunaaluk and Edmontosaurus plots (figures 6I and 6J) are compared to the Edmontosaurus plots, they show that (1) inclusion of

Ugrunaaluk specimens does not greatly change the slopes obtained from just Edmontosaurus

specimens (only Edmontosaurus -0.19 and 0.10, vs. Ugrunaaluk/Edmontosaurus -0.12 and 0.09, for PC1 and PC2, respectively), and (2) by increasing the size range, isometry could be rejected in this group (p=0.001127 for PC1, and 0.00006904 for PC2). As Ugrunaaluk and Edmontosaurus specimens both plot very closely to the regression line, the growth trajectories with Ugrunaaluk included or excluded are similar and are impossible to distinguish from each other as confidence intervals overlap, it could be taken as an indication that the Ugrunaaluk specimens belong to Edmontosaurus, in which case the dentary Edmontosaurus would show an allometric growth pattern of increase in the inclination of the coronoid process, relative dental battery length, as well as relative diastema length and straightness. Xing, Mallon and Currie (2017) suggested that the Prince Creek Formation specimens might belong to a new northernmost extent of E. regalis as they display some similarities, but mentioned that there are no E. regalis specimens of similar size to the Prince Creek Formation specimens to compare with. If the Prince Creek Formation specimens belong to E. regalis, the taxon’s geological range would extend from the present Late Campanian into the Early Maastrichtian (Mori, Druckenmiller & Erickson 2016). Alternatively, the inclination of the coronoid process, relative elongation and straightening of the diastema, and the relative elongation of the dental battery progressed very similarly during growth in the clade encompassing both Edmontosaurus and

Ugrunaaluk. Even though the results herein suggest close similarities in the dentaries of these taxa,

(44)

36

premaxilla, maxilla, jugal and quadratojugal to distinguish Ugrunaaluk from Edmontosaurus, they are worthy of further morphometric study.

6.2 Implications for phylogenetic characters

The results presented here reveal much about the nature of variation in the hadrosauroid dentary and, as a result, have important implications for understanding the use of this structure in phylogenetic systematics. Characters associated with the dentary have been included in all hadrosauroid analyses, representing anywhere from 8.8% to 16.2% of the total number of characters (table 8). Given that these analyses assume variation to be solely taxonomic, it is important that the nature of variation be understood. In the following section I focus on the data sets of two recent phylogenetic analyses: Prieto-Márquez, Erickson and Ebersole (2016a), and Xing, Mallon and Currie (2017).

Table 8. Amount of dentary-related characters in literature. Percentage is of total amount of characters in that

publication.

Source Dentary teeth Dentary bone Dentary teeth +

bone Total amount of characters Weishampel, Norman & Grigorescu 1993 4 (10.8%) 2 (5.4%) 6 (16.2%) 37 Godefroit et al. 2008 3 (5.4%) 2 (3.6%) 5 (8.9%) 56

Evans & Reisz 2007 (and Evans 2010) 5 (5.3%) 4 (4.3%) 9 (9.6%) 94 Prieto-Márquez 2010a 14 (3.8%) 21 (5.7%) 35 (9.5%) 370 Xing et al. 2014 15 (4.3%) 21 (6.1%) 36 (10.4%) 345 Prieto-Márquez, Erickson and Ebersole 2016a 10 (3.7%) 14 (5.1%) 24 (8.8%) 273 Xing, Mallon and Currie 2017 17 (4.9%) 22 (6.4%) 39 (11.3%) 346