Cranial and appendicular ontogeny of Bactrosaurus johnsoni, a hadrosauroid dinosaur from the Late Cretaceous of northern China

Authors


Abstract

Abstract:  The juvenile anatomy of various cranial and appendicular elements of the hadrosauroid dinosaur Bactrosaurus johnsoni is described in detail. Growth changes are documented from juvenile to adult stages for each skeletal element available. In the studied skull, ontogenetic trends consist of: development of features on the ventral surface of the frontal; reduction in the slope of the posteromedial process of the premaxilla; a posterior shift of the dorsal process of the maxilla; development of concavities on the medial surface of the prefrontal; increased robustness and development of the ventral flange of the jugal; decreased curvature of the long axis of the quadrate; increased ventral deflection of the dentary; and changes in the length/width proportions and depth of the anterior surface of the predentary. In the appendicular skeleton, the majority of ontogenetic variation from juvenile to adult occurs in the limb bones, including increased robustness of the deltopectoral crest of the humerus; relative shortening of the ulna; increased development of the fourth trochanter and mediolateral widening of the distal end of the femur; increased expansion of the cnemial crest of the tibia; and the increased prominence of articular protuberances and flanges of the metatarsals. A survey of the phylogenetically informative characters present in B. johnsoni indicates that several characters concerning the frontal, maxilla, jugal, quadrate, predentary, dentary, scapula, humerus and ilium are affected by ontogeny. Nevertheless, the majority of phylogenetic characters are not ontogenetically variable, suggesting that a substantial amount of the information provided by juvenile and subadult specimens for phylogenetic inference is reliable in basal hadrosauroids.

H adrosauroids are the most morphologically derived ornithopod dinosaurs. The clade consists of all iguanodontians closer to Hadrosaurus foulkii than to Iguanodon bernissartensis (Prieto-Márquez 2010a). These animals have been recorded from the late Early to Late Cretaceous (Barremian–Aptian to late Maastrichtian) of all continents except Africa and Oceania (Horner et al. 2004; Norman 2004). Their fossil record is among the richest and best preserved within the Dinosauria, including dozens of articulated skeletons, multi-individual assemblages, eggs and embryonic materials, soft-tissue impressions and footprints (Lull and Wright 1942; Horner 2000).

Knowledge on how the skeletal morphology of these animals changes through ontogeny is instrumental in understanding their palaeobiology and evolutionary history. On one hand, characters that are ontogenetically variable need to be identified and distinguished from those containing information for phylogenetic inference and taxonomic diagnoses. In addition, palaeobiological studies focusing on areas such as life history or heterochrony rest upon the identification of ontogenetic stages for the taxa of interest, which in turn rely upon proper identification of ontogenetically dependent morphological attributes. In comparison with the vast literature on hadrosauroid anatomy, systematics and palaeobiology, works entirely or mostly devoted to the documentation of skeletal changes during ontogeny are uncommon (Rozhdestvensky 1965; Waldman 1969; Maryanska and Osmólska 1981; Horner and Currie 1994; Weishampel and Horner 1994; Dilkes 2001; Grigorescu and Csiki 2006; Brett-Surman and Wagner 2007; Evans et al. 2005; Guenther 2009).

In the case of most speciose and morphologically derived hadrosauroid subclade, the Hadrosauridae, substantially complete ontogenetic series of skeletal elements are available for many taxa (Horner et al. 2000, 2004; Guenther 2009; Evans 2007, 2010). By contrast, the fossil material available for most non-hadrosaurid hadrosauroids is insufficient for revealing morphological changes through ontogeny. Notably, at least half of the currently recognized species of basal hadrosauroids are known from a single specimen, namely: Claosaurus agilis (Carpenter et al. 1995); Equijubus normani (You et al. 2003a); Glishades ericksoni (Prieto-Márquez 2010b); Jinzhousaurus yangi (Wang and Xu 2001); Nanyangosaurus zhugeii (Xu et al. 2000); Penelopognthus weishampeli (Godefroit et al. 2005); Cedrorestes crichtoni (Gilpin et al. 2007); Shuangmiaosaurus gilmorei (You et al. 2003b); Jintasaurus meniscus (You and Li 2009); and Jeyawati rugoculus (McDonald et al. 2010). For other species, the available material is also meagre: Lophorhothon atopus is known from two subadult specimens of similar developmental stage, one represented by cranial and the other by postcranial material (Langston 1960; Lamb 1998); Tanius sinensis is represented by a partial skull and a few postcranial bones, each assigned to a different specimen (Wiman 1929); Probactrosaurus mazongshanensis is only represented by a partial skull and various postcranial bones (Lü 1997; Norman 2002); Altirhinus kurzanovi is known from adult specimens and a few appendicular and axial elements representing two fragmentary subadult specimens (Norman 1998); and Protohadros byrdi is known from a partial skull specimen and two isolated teeth, each assigned to a different exemplar (Head 1998). In the case of the recently described Tethyshadros insularis, various well-preserved and relatively complete specimens exist but no juvenile individuals are present (Dalla Vecchia 2009). Although the hadrosauroid Gilmoreosaurus mongoliensis is also known from a number of exemplars, the vast majority of these (with the exception of three juvenile tibiae and a few pedal phalanges) correspond to individuals that represent approximately the same ontogenetic stage (Gilmore 1933; Prieto-Márquez 2010c). This leaves only a small fraction of known hadrosauroids with enough material to document ontogenetic changes: Bactrosaurus johnsoni (Gilmore 1933; Godefroit et al. 1998), Telmatosaurus transsylvanicus (Weishampel et al. 1993; Grigorescu and Csiki 2006), P. gobiensis (Rozhdestvensky 1966; Norman 2002), Eolambia caroljonesa (Kirkland 1998; Head 2001; Garrison et al. 2007) and Levnesovia transoxiana (Sues and Averianov 2009). Of these, only T. transsylvanicus has been the focus of a detailed study on growth changes (Grigorescu and Csiki 2006).

Bactrosaurus johnsoni is represented by one of the best-preserved and well-represented ontogenetic series available for basal hadrosauroids. This species was named and described by Gilmore (1933) based on an assemblage of disarticulated elements corresponding to various individuals from different ontogenetic stages. They came from a bonebed (locality number 141) in strata corresponding to the Iren Dabasu Formation, near the Chinese town of Erenhot, along the eastern segment of the border between China and Mongolia (Gilmore 1933; Godefroit et al. 1998). Recent studies disagree on the age of the Iren Dabasu Formation: according to Van Itterbeck et al. (2005), the formation is late Campanian to early Maastrichtian in age, whereas in the view of Averianov (2002) and Sues and Averianov (2009), it is as old as late Turonian to early Coniacian. Gilmore (1933) provided very little information on growth changes in B. johnsoni, consisting of a few brief remarks on the number of teeth, the shape of the scapula blade and the shape of the pedal ungual phalanges. More recently, Godefroit et al. (1998) described new materials referable to B. johnsoni that were collected near Erenhot by members of the Sino-Belgian Dinosaur Expedition, from a quarry (locality SBDE 95E5) located within one kilometre of Gilmore’s locality 141.

This study fills a gap in our understanding of the ontogenetic morphological changes that took place in the cranial and appendicular skeleton of basal hadrosauroids. This is accomplished through detailed documentation of the juvenile anatomy of Bactrosaurus johnsoni, with emphasis on those morphological attributes that are variable through ontogeny. Furthermore, this analysis evaluates the impact of ontogenetic variation on phylogenetic characters that have been proposed in the literature for inferring the evolutionary interrelationships of hadrosauroids.

Material and methods

Based on the largest known cranial (e.g. skull roof AMNH 6365, 18 cm in width across postorbitals; dentary and maxilla AMNH 6553, 23 and 22 cm in length, respectively) and postcranial (e.g. femur AMNH 6553, 81 cm in length) elements, the maximum recorded skull and body length for Bactrosaurus johnsoni is estimated at 50–55 cm and 6–6.5 m, respectively. These estimates are based on comparisons with articulated complete skulls and postcrania from other hadrosauroids (e.g. the basal hadrosauroid Equijubus normani, IVPP V12534, has a maxilla that is approximately 28 cm in length and a 63-cm-long skull; according to Dalla Vecchia (2009), an approximately 3.5-m-long articulated skeleton of the basal hadrosauroid Tethyshadros insularis has a femur that is 42 cm in length). For the purpose of this study and to establish a size reference, these estimated maximum lengths for B. johnsoni are regarded as indicative of adult size. According to this, and based on Horner et al. (2000) and Evans (2007), the term ‘juvenile’ is used for specimens that have attained up to 50 per cent the estimated maximum skull or body length. Likewise, the term ‘subadult’ is used for exemplars corresponding to over 50 per cent but less than 85 per cent of the estimated maximum skull or body length.

Skeletal sample of exemplars

This study is based on examination of the disarticulated cranial and appendicular bones of Bactrosaurus johnsoni from the original locality 141 at Iren Dabasu (Gilmore 1933), as well as casts of some of the elements collected at the SBDE 95/E5 locality (Godefroit et al. 1998). Examined cranial and postcranial material corresponding to adult or large subadult individuals included: AMNH 6365 (articulated skull roof and braincase, including frontals, postorbitals, squamosals and parietal), AMNH 6366 (articulated skull roof and braincase, including frontals, postorbitals, parietal and partial squamosals), AMNH 6373 (left jugal), AMNH 6385 (partial right quadrate), AMNH 6396 (partial right quadrate), AMNH 6398 (rostral process of right jugal), AMNH 6553 (comprising more than one individual, including the holotype left maxilla, and left and right dentaries, as well as postcranial elements such as left scapula, left humerus, partial left and right pubes, partial left and right ischia, partial right and left tibiae, left fibula, articulated left metatarsals II–III–IV, articulated right metatarsals II–III, articulated left metatarsals III–IV and numerous pedal phalanges), AMNH 6577 (right frontal) and AMNH 6579 (right astragalus and calcaneum). Examined casts of SBDE specimens include SBDE 95E5/8 (right quadrate), SBDE 12 (left dentary), SBDE 13 (right surangular), 3 SBDE 1 (maxilla), SBDE 22 (right coracoid), SBDE 24 (left humerus) and SBDE 25 (right ilium). Comparisons with other SBDE specimens were made using descriptions and figures in Godefroit et al. (1998).

Examined juvenile material included the following cranial bones: AMNH 6370 (fused opisthotic-exoccipital), AMNH 6372 (predentary), AMNH 6379 (right jugal), AMNH 6380 (left dentary), AMNH 6384 (left quadrate), AMNH 6388 (right maxilla), AMNH 6389 (right maxilla), AMNH 6390 (left maxilla), AMNH 6391(left maxilla), AMNH 6392 (left maxilla), AMNH 6393 (fragment of left maxilla), AMNH 6394 (left surangular), AMNH 6395 (subadult right frontal), AMNH 6396 (left jugal), AMNH 6397 (left jugal), AMNH 6501 (partial right premaxilla), AMNH 6514 (left maxilla), AMNH 6574 (distal fragment of subadult left quadrate), AMNH 6575 (partial left and right premaxillae), AMNH 6580 (right dentary), AMNH 6581 (right dentary), AMNH 6582 (right dentary), AMNH 6583 (right maxilla), AMNH 6586 (partial left prefrontal) and AMNH 6587 (partial left prefrontal). Postcranial juvenile material included: AMNH 6378 (left humerus), AMNH 30239 (left humerus), AMNH 6553 (subadult right tibia), AMNH 6579 (including several subadult and juvenile specimens: two right astragali and one left, and two left calcanea), AMNH 6584 (right prefrontal), AMNH 6585 (left prefrontal) and AMNH 6577 (including various specimens represented by three right coracoids, two right scapulae and one left, one left and two right humeri, right ulna, left and right radii, left metacarpal II, right metarcarpal III, partial left ilium, partial right ischium, two right femora, left and right tibiae, one right and two left metatarsals II, one right and two left metacarpals III, one right metacarpal IV and numerous pedal phalanges).

No significant ontogenetic variation was found in the other elements known for Bactrosaurus johnsoni (i.e. supraoccipital, basioccipital, basisphenoid, parasphenoid, prootic, laterosphenoid, orbitosphenoid, presphenoid, parietal, nasal, lacrimal, squamosal, postorbital, supraorbital, vertebrae, sternal, pubis, fibula, metacarpals) because either this variation was not observed or the element is not preserved in ontogenetic series. For detailed descriptions of all known skeletal elements of this species, the reader is referred to Gilmore (1933), Weishampel and Horner (1986) and Godefroit et al. (1998).

Institutional abbreviations.  AMNH, American Museum of Natural History, New York, USA; IRSNB, Institut Royal des Sciences Naturelles de Belgique, Brussels, Belgium; SBDE Sino-Belgian Dinosaur Expedition (specimens are property of the Inner Mongolian Museum in Hohhot, China; casts and some of the original specimens are temporarily housed at the Institut Royal des Sciences Naturelles de Belgique, Brussels, Belgium); IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing, China.

Juvenile Anatomy and Growth Changes in Bactrosaurus johnsoni

Cranial elements

Opisthotic-exoccipital.  The paroccipital process of the adult (Text-fig. 1D–F) differs from the juvenile in being much thicker anteroposteriorly throughout its length (Text-fig. 1). Moreover, the adult paroccipital process projects further ventrally to the level of the basioccipital, whereas in the juvenile, the distal tip of the process projects only slightly below the level of the suture between the opisthotic-exoccipital and the basioccipital (Text-fig. 1B, E).

Figure TEXT‐FIG. 1..

Bactrosaurus johnsoni, neurocranial bones. A, B, adult braincase, AMNH 6365. A, right lateral and B, posteroventral views. C, adult braincase, AMNH 6366, in dorsal view. D–F, juvenile fused opisthotic-exoccipital, AMNH 6370. D, right lateral, E, posteroventral, and F, dorsal views. G, H, adult specimen, AMNH 6577. G, dorsal and H, ventral views. I, J, juvenile specimen, AMNH 6395. I, dorsal and J, ventral views. ar, annular ridge; bo, basioccipital; boar, articulation surface for basioccipital; cb, cerebral cavity; f, frontal articulation surface; fm, foramen magnum; ns, nasal articulation surface; ob, orbital depression; of, olfactory depression; om, orbital margin; op-ex, fused opisthotic-exoccipital; os, orbitosphenoid articulation surface; pa, parietal; par, parietal articulation surface; pf, prefrontal articulation surface; po, postorbital; poar, postorbital articulation surface; pp(op-ex), paroccipital process of the fused opisthotic-exoccipital; ps, presphenoid articulation surface; sa, supraoccipital; sar, articulation surface for supraoccipital; sq, squamosal; sqar, articular surface for squamosal; sq(pc), precotyloid process of the squamosal; sq(qc), quadrate cotylus of the squamosal. Scale bars represent 20 mm for A–F and 10 mm for G–J.

Frontal.  The frontal (Text-fig. 1G–J) experiences remarkable changes during ontogeny. The cerebral and surrounding concavities of the endocranial surface are further accentuated in adult specimens (e.g. AMNH 6577) when compared to their very gentle relief in the juvenile AMNH 6395. The region for reception of the prefrontal consists of a deep excavation that is bounded posterodorsally by a sharp, crescentic ridge. Notably, this articular region is much reduced in juveniles: the ridge is less prominent, the excavation is reduced to a posterior recess on the anterolateral margin of the frontal, and the entire articular area is both mediolaterally and dorsoventrally narrower (Text-fig. 1G, I). In the adults, the orbital depression becomes much wider mediolaterally and the olfactory depression much more elongated (Text-fig. 1H, J). The cerebral cavity experiences a remarkable increase in depth during ontogeny, as evidenced by the relatively shallower cavity present in the juvenile AMNH 6395. The ridge that separates the cerebral cavity from the orbital depression, bearing the suture for the presphenoid anteriorly and the orbitosphenoid posteriorly, increases substantially in thickness during ontogeny (Text-fig. 1H, J). The large, sharp crenulations, for articulation with the orbitosphenoid and presphenoid, that are present in adult specimens (e.g. AMNH 6577) are narrower, faint and present only anteriorly in juveniles (Text-fig. 1J). Unlike in adults, there is no annular ridge in juveniles; instead, the anterior region of the cerebral cavity is separated from the olfactory depression by a gentle convexity.

Premaxilla.  All examples of this element recovered from the Bactrosaurus type locality consist of the anterior premaxillary regions of two juvenile individuals (Pl. 1, fig. 1–6). In larger subadult specimens (e.g. SBDE 95E5/4-4bis: Godefroit et al. (1998), pl. 3, fig. 4), the premaxilla has a more extensive anteromedial surface than in juveniles, both dorsoventrally and mediolaterally. The juvenile posteromedial process projects posterodorsally forming an angle of 45 degrees with the ventral surface of the premaxilla. Larger specimens (Godefroit et al. 1998, pl. 3, fig. 4C) differ from the AMNH juveniles in having a less steeply oriented posterolateral process of the premaxilla, which forms an angle of approximately 35 degrees with the ventral surface.

Figure EXPLANATION OF PLATE 1.

Figure EXPLANATION OF PLATE 1.


Figs 1–12. Bactrosaurus johnsoni, facial elements. 1–3, juvenile premaxilla, AMNH 6501. 1, anterodorsal, 2, posteroventral, and 3, right lateral. 4–6, juvenile premaxillae, AMNH 6575. 4, anterior, 5, left lateral, and 6, medial views. 7–8, holotype adult specimen, AMNH 6553. 7, left lateral and 8, medial views. 9–10, juvenile specimen, AMNH 6389. 9, right lateral (reversed) and 10, medial (reversed) views. 11–12, juvenile specimen, AMNH 6390. 11, left lateral and 12, medial views. af, row of alveolar foramina; cmn, circumnarial fossa; dm, denticulate oral margin; dp, dorsal process; dtp, dental parapet; er, ectopterygoid ridge; es, ectopterygoid shelf; fo, foramen; mp, mediodorsal process; j, articulation surface for jugal; jp, jugal process; mxar, articulation surface for maxilla; pl, palatine process; pmp, posteromedial process; pmx, articulation surface for premaxilla; pt, pterygoid process; th, tooth battery; vp, anteroventral process; vt, ventral thickening. All scale bars represent 20 mm.

Maxilla.  The jugal articular region of the maxilla (Pl. 1, figs 7–12) is proportionately larger in juveniles than in adults (e.g. AMNH 6553). Specifically, in juveniles, the jugal articular surface extends along 28 per cent of the total length of the maxilla, whereas in adults, it extends along 20–22 per cent of the maxillary length. In juveniles, the dorsal process is centred at maxillary mid-length (e.g. AMNH 6389), whereas it is posteriorly displaced in adults (e.g. AMNH 6553). There is variation in the relative length of the ectopterygoid shelf (e.g. it is 20 per cent of the total length of the maxilla in AMNH 6391 and 25 per cent in AMNH 6389). This variation is also present among the largest specimens (e.g. it is 20 per cent in AMNH 6553 and 25 per cent in SBDE 95E5/31: Godefroit et al. (1998), pl. 4, fig. 1A). The palatine process rises dorsally more prominently in juveniles (e.g. AMNH 6389 and 6391) than in adults (e.g. AMNH 6553 and SBDE 95E5/31: Godefroit et al. (1998), pl. 4, fig. 1A). Only the proximal region of the small pterygoid process is preserved in the juveniles, but this appears proportionately narrower and smaller than in adults.

Jugal.  The ventral margin of the jugal (Text-fig. 2), between the rostral process and the ventral flange, becomes slightly more concave with increasing specimen size. Correlated with this, the posteroventral flange becomes substantially larger and projects further ventrally in larger individuals. Specifically, the ratio between the maximum dorsoventral depth of the jugal across the posteroventral flange and the minimum depth across the posterior constriction is 1.45 in the smallest specimen (AMNH 6397) and 1.75 in the largest nearly complete jugal of the sample (AMNH 6373).

Figure TEXT‐FIG. 2..

Bactrosaurus johnsoni, ontogenetical series of jugals. A, B, subadult jugal, AMNH 6373. A, right lateral (reversed) and B, medial (reversed) views. C, D, juvenile jugal, AMNH 6396. C, left lateral and D, medial views. E, F, juvenile specimen, AMNH 6397. E, left lateral and F, medial views. itm, infratemporal margin; mxar, articulation surface for maxilla; om, orbital margin; pop, postorbital process; qjf, quadratojugal flange; rp, rostral process; vf, posteroventral flange. All scale bars represent 20 mm.

Prefrontal.  The dorsal orbital region of the prefrontal (Pl. 2, figs 4–9) experiences a substantial increase in mediolateral width during ontogeny, as evidenced by the progressively wider exemplars in the ontogenetic series recovered from locality 141 at Iren Dabasu (Pl. 2, figs 2, 5 and 8). The extent of excavation of the medial surface of the prefrontal develops with increasing specimen size (Pl. 2, figs 3, 6 and 9). In the larger prefrontals (AMNH 6584 and 6585), the medial side exhibits two major oval depressions that are separated by an oblique low ridge; the more posterior of these two depressions is subdivided by a gentle convexity (Pl. 2, fig. 3). This configuration of the medial surface of the prefrontal is much less developed in juvenile specimens. For example, in the juvenile AMNH 6586, a very faint convexity separates a posterodorsal depression from the depressed anteroventral region of the specimen (Pl. 2, fig. 6). No such convexity is present in the smallest prefrontal of the sample, AMNH 6587 (Pl. 2, fig. 9).

Figure EXPLANATION OF PLATE 2.

Figure EXPLANATION OF PLATE 2.


Figs 1–15. Bactrosaurus johnsoni, facial elements. 1–3, adult left prefrontal, AMNH 6585. 1, lateral, 2, ventral, and 3, medial views. 4–6, juvenile partial left prefrontal, AMNH 6586. 4, lateral, 5, ventral, and 6, medial views. 7–9, juvenile partial left prefrontal, AMNH 6587. 7, lateral, 8, ventral, and 9, medial views. 10–12, adult or large subadult partial right quadrate, AMNH 6385. 10, lateral, 11, anterior, and 12, ventral view of distal articular surface. 13–15, juvenile left quadrate, AMNH 6384. 13, lateral (reversed), 14, anterior (reversed), and 15, ventral (reversed) view of distal articular surface. ad, anterior depression; alf, anterolateral flange; lcd, lateral condyle; lt, lateral flange; mcd, medial condyle; ob, orbit; obm, orbital lateral margin; or, oblique ridge; pd, posterior depression; pmp, posteromedial process for articulation with frontal; ptf, pterygoid flange; qh, quadrate head; qjn, quadratojugal notch; sqb, squamosal buttress; vqjf, ventral quadratojugal flange. All scale bars represent 25 mm.

Quadrate.  The juvenile quadrate exhibits marked posterodorsal curvature (Pl. 2, fig. 13), which is reduced in the adults (Pl. 2, fig. 10; see also SBDE 95E5/8, Godefroit et al. (1998), pl. 5, fig. 1). In juveniles, the flange that bounds ventrally the quadratojugal notch expands anteriorly, parallel with the larger lateral flange of the quadrate, and its lateral surface is gently convex. By contrast, the ventral flange of the adult specimens (e.g. AMNH 6385: Pl. 2, fig. 11) is anterolaterally oriented, bulging laterally, and mediolaterally thicker along the region adjacent to the quadratojugal notch. Adults show quadrate distal articular surfaces that are anteroposteriorly thicker than in juveniles (e.g. AMNH 6384, AMNH 6574; Pl. 2, figs 12 and 15). In addition, adult specimens exhibit a greater ventral offset of the lateral distal condyle in relation to the position of the medial condyle (Pl. 2, figs 11 and 14).

Predentary.  In juveniles, this element (AMNH 6372: Pl. 3, figs 1–2) exhibits a subrectangular dorsal profile, so that the anterior transverse region is approximately twice as long as each of the lateral processes. These proportions are unlike those of the larger predentary, SBDE 95E5/36, documented by Godefroit et al. (1998, pl. 6, fig. 2). In SBDE 95E5/36, the lateral processes are approximately as long as the anterior transverse bar. The anterolateral corners of this predentary are more smoothly arcuate than in the juvenile (Godefroit et al. (1998), pl. 6, fig. 2). In the juvenile, the anterior surface is very shallow dorsoventrally when compared with the much deeper anterior face of SBDE 95E5/36 (Godefroit et al. (1998), pl. 6, fig. 2D). In AMNH 6372, the depth of the anterior surface is 18 per cent of the mediolateral width of the predentary, whereas in SBDE 95E5/36 reaches 30 per cent of the mediolateral width of the bone. On the oral margin, there are five denticles extending laterally from a central sagittal denticle (Pl. 3, fig. 1), in contrast to the three lateral denticles present in the larger SBDE 95E5/36 (Godefroit et al. (1998), pl.e 6, fig. 2B).

Figure EXPLANATION OF PLATE 3.

Figure EXPLANATION OF PLATE 3.


Figs 1–10. Bactrosaurus johnsoni, mandibular elements. 1–2, juvenile predentary, AMNH 6372. 1, dorsal and 2, ventral views. 3–4, adult left dentary, AMNH 6553. 3, medial and 4, lateral views. 5–6, juvenile right dentary, AMNH 6380. 5, medial (reversed) and 6, lateral (reversed) views. 7–8, juvenile left dentary, AMNH 6581. 7, medial and 8, left lateral views. 9–10, juvenile left surangular, AMNH 6394. 9, lateral and 10, dorsomedial views. an, articulation surface for angular; ap, anterior ascending process; cp, coronoid process; dm, denticulate oral margin; dp, dental parapet; ed, edentulous margin (‘diastema’); lp, lateral process; lap, lateral lap; Mg, Mekelian groove; Mk, Mekelian fossa; oc, occlusal plane; pdt, articular margin for predentary; qar, articulation surface for quadrate; rdg, dorsal ridge; sf, lateral shelf; s, symphysis; sp, symphyseal process; spl, articulation surface for splenial; th, tooth battery; vp, ventral process. All scale bars represent 20 mm.

Dentary.  The ventral deflection of the symphyseal process is greater in the adult AMNH 6553 than in the juveniles (Pl. 3, figs 3–8; see also SBDE 95E5/12, Godefroit et al. (1998), pl. 6, fig. 1). For example, the juveniles AMNH 6581 and 6380 show deflection angles of 12 degrees compared to the 24 degree angle of AMNH 6553. In juveniles, the ratio between the labiolingual extension of the symphyseal process and the labiolingual width of the dentary varies between 1.5 (AMNH 6380) and 2.0 (AMNH 6581). In some juvenile specimens (e.g. AMNH 6380 and 6582), the dental battery is straight, whereas in other examples (e.g. AMNH 6580 and 6581), it is bowed lingually; the dental battery shows a slight lingual bowing in the adult AMNH 6553.

Surangular.  Only one juvenile surangular (AMNH 6394) is known for Bactrosaurus johnsoni (Pl. 3, figs 9–10). The medial margin of the surangular of the juvenile specimen is anteroposteriorly straight, whereas in the adult SBDE 95E5/13 (Godefroit et al. (1998), fig. 19 and pl. 7, fig. 1), this margin is concave in dorsal profile.

Dentition.  The juvenile dentaries recovered at locality 141 in Iren Dabasu (Gilmore 1933) contain 15 (AMNH 6380, 6580 and 6582) or 16 (AMNH 6581) tooth positions. Dentaries AMNH 6581 and 6582 show one functional tooth crown forming part of the occlusal plane, whereas AMNH 6380 and 6580 have two functional teeth in two spots at mid-length of their occlusal planes. There are as many as two teeth per alveolus arranged dorsoventrally at mid-length of the dental battery, in contrast to the three teeth present in adults (e.g. AMNH 6553). Notably, tooth crowns have a height/width ratio of 3.1–3.2, as in AMNH 6553. Maxillary tooth counts for the Bactrosaurus juveniles range from 16 (AMNH 6389) to 17 (AMNH 6514 and 6391). The position of the primary ridge is variable among the teeth of a dental battery, ranging from posterior to a point the crown mid-point (Text-fig. 3D).

Figure TEXT‐FIG. 3..

Bactrosaurus johnsoni, tooth crowns. A, B, adult dentition, AMNH 6553. A, lingual view of dentary tooth crowns and B, labial view of maxillary tooth crowns. C, D, juvenile dentition. C, lingual view of the dentary crowns of AMNH 6580 and D, labial view of the maxillary crowns of AMNH 6390. mr, main ridge; sr, secondary ridge. All scale bars represent 10 mm.

Postcranial elements

Pectoral girdle.  The juvenile coracoid (Pl. 4, fig. 3) is morphologically indistinct from the adult form (e.g. SBDE 95E5/22; Pl. 4, fig. 1). Notably, the lateral margins of the glenoid and the scapular facet meet to form an angle that varies between 105 and 125 degrees.

Figure EXPLANATION OF PLATE 4.

Figure EXPLANATION OF PLATE 4.


Figs 1–13. Bactrosaurus johnsoni, pectoral and forelimb elements. 1, adult right coracoid, cast of SBDE 95E5/22, in lateral view (reversed). 2, adult left scapula, AMNH 6553, in lateral view. 3, juvenile right coracoid, AMNH 6577, in lateral view (reversed). 4, juvenile right scapula, AMNH 6577, in lateral view (reversed). 5, adult right humerus, cast of SBDE 95E5/24, in posterolateral view. 6, juvenile left humerus, AMNH 6577, in posterolateral view (reversed). 7, juvenile right humerus, AMNH 6577, in posterolateral view. 8, juvenile left humerus, AMNH 6378, in posterolateral view (reversed). 9, juvenile right ulna, AMNH 6577, in lateral view. 10, juvenile right metacarpal II, AMNH 6577, in lateral view. 11, juvenile left radius, AMNH 6577, in medial view. 12, juvenile metacarpal III, AMNH 6577, in medial view. 13, adult or large subadult metacarpal III, cast of SBDE 95E5, in medial view. ac, pseudoacromion process; arh, articular head; bct, bicipital tubercle; cor, coracoid facet; db, distal blade; dpc, deltopectoral crest; dtr, deltoid ridge; f, coracoid foramen; glf, glenoid fossa; gln, glenoid; lfg, lateral flange; ol, olecranon process; pcn, proximal constriction; rcd, radial condyle; scp, scapular facet; ucd, ulnar condyle; vp, ventral process. Scale bars represent 20 mm for 1, 3, 10, 12, 13 and 50 mm for 2, 4, 5–9, 11.

In the scapula, the orientation of the pseudoacromion process of the AMNH 6577 juveniles varies from horizontal (e.g. Pl. 4, fig. 4) to slightly oriented anterodorsally. The combined mediolateral width of the dorsal margins of the pseudoacromion process and the coracoid facet is proportionately greater in the adult (AMNH 6553) than in the juveniles (AMNH 6577). For all juveniles, the proximal constriction is relatively narrow, being nearly half as wide as the maximum depth of the proximal region of the scapula at the level of the ventral apex of the glenoid. The proximal constriction is similarly narrow in adult scapulae (e.g. AMNH 6553; Pl. 4, fig. 2). The deltoid ridge becomes more prominent and more clearly defined in AMNH 6553 than in the AMNH 6577 juveniles (Pl. 4, figs 2 and 4).

Humerus.  The juvenile humerus (e.g. AMNH 6577) is elongate and slender in comparison with those of adults (e.g. AMNH 6553 and SBDE 95E5/24; Pl. 4, fig. 5). Specifically, the ratio between the total length of the humerus and the maximum width of the lateral surface of its proximal end ranges from 4.95 (e.g. AMNH 30239) to 5.6 (e.g. AMNH 6577) in the juveniles, whereas it is only 4.5 in the adult SBDE 95E5/24. The juvenile deltopectoral crest is less expanded anteromedially than that of the adult specimens; the ratio between its maximum breadth and the minimum mid-shaft diameter of the humerus ranges from 1.60 to 1.65 in juveniles (e.g. 1.63 in AMNH 30239), compared to 1.74 in SBDE 95E5/24. In juveniles, the scar for the latissimus dorsi muscle located on the posterior surface of the proximal half of the humerus (Dilkes 2000) does not form the prominent protuberance seen in large subadult or adult specimens, but appears restricted to a flat rugose area. The ulnar and radial condyles are further developed in the larger humeri, being more robust and mediolaterally thicker.

Ulna and radius

The juvenile ulna (Pl. 4, fig. 9) has a more elongate shaft than in adults (e.g. SBDE 95E5/42, Godefroit et al. (1998), fig. 29 and pl. 11, fig. 1). Specifically, the total length (excluding the olecranon process)/minimum width at mid-shaft ratio for the juvenile AMNH 6577 is nearly 13, whereas it is only 9.0 in SBDE 95E5/42. In contrast, both the juvenile and adult radii (e.g. SBDE 95E5/43, Godefroit et al. (1998), fig. 29 and pl. 11, fig. 2) show the same length/width proportions (approximately 12–13 times longer than their mid-shaft diameters). The olecranon process of the ulna is longer in the larger SBDE 95E5/42 than in the juvenile AMNH 6577: whereas in the adult ulna, it reaches a length that is 75 per cent of the dorsoventral depth of the proximal region across the lateral flange, in the juvenile, it reaches only 45 per cent of the depth of the proximal ulna. In addition, the dorsal surface of the juvenile olecranon process is anterodorsally oriented to become continuous with the posterior margin of the proximal surface of the ulna, whereas in the adult ulna, the dorsal surface of the olecranon process is anteriorly oriented and horizontal.

Metacarpal III.  The juvenile metacarpal III, exemplified by AMNH 6577 (Pl. 4, figs 12–13), shows a proximodistal length that is seven times greater than the mediolateral width of its mid-shaft. A larger specimen from the SBDE 95E5 collection (Godefroit et al. 1998) is only slightly thicker than the juvenile, with a length/width ratio of six (Pl. 4, fig. 13).

Ilium.  The juvenile ilium is represented by a single partial element missing the postacetabular process and the anterior segment of the preacetabular process (AMNH 6577; Text-fig. 4F). The dorsoventral depth of the proximalmost region of the preacetabular process is half of the depth of the central iliac plate (the latter being measured from the ventral margin of the pubic peduncle to its dorsal margin). This is unlike the condition in the adult ilia, where the proximalmost region of the preacetabular process is deeper, at least 60 per cent of the depth of the iliac central plate (Text-fig. 4E). The preacetabular process of the adults is mediolaterally thicker than that of juveniles.

Figure TEXT‐FIG. 4..

Bactrosaurus johnsoni, pelvic elements. A, B, proximal fragment of adult right ischium, AMNH 6553. A, lateral and B, anterodorsal views. C, D, juvenile proximal fragment of right ischium, AMNH 6577. C, lateral and D, anterodorsal views. E, adult left ilium, cast of SBDE 95E5/25, in lateral view. E, partial juvenile ilium, AMNH 6577, in left lateral view. ac, acetabular margin; ilp, iliac peduncle; isp; ischiac peduncle; ist, ischiac tuberosity; pbp, pubic peduncle; popr, postacetabular process; ppr, preacetabular process; saa, supraacetabular process. Scale bars represent 25 mm for A–D and 50 mm for E, F.

Ischium.  The juvenile ischium is known from a fragment of the proximal region (Text-fig. 4C, D). The posterior curvature of the posterodorsal corner of the iliac peduncle is slightly more pronounced in adult specimens, as in AMNH 6553. The adult ischium (e.g. AMNH 6553; Text-fig. 4A, B) differs from that the juvenile in having a mediolaterally thicker and more distally expanded iliac peduncle. While in the juvenile ischium, the articular surface of the iliac peduncle is subellipsoidal and dorsoventrally elongated (more than twice as long as it is wide), in AMNH 6553, the articular surface is triangular and only 25 per cent longer than it is wide. In the juvenile specimen, the acetabular margin is reduced to a sharp edge uniting the pubic and iliac peduncle, whereas in the large AMNH 6553 ischia, the acetabular margin forms a concave and dorsoventrally elongated facet bounded by parallel lateral and medial borders. The juvenile pubic peduncle is more mediolaterally compressed than the iliac peduncle and subrectangular in lateral profile, being slightly dorsoventrally longer than anteroposteriorly wide. A shallow convexity extends from the proximal region of the pubic peduncle to the proximoventral region of the ischiac shaft. This convexity becomes a thick ridge in the adult exemplar AMNH 6553.

Femur.  In the juvenile femur (Text-fig. 5A, B), the shaft is relatively narrow mediolaterally along the middle third of the element and gradually expands towards the proximal and distal ends. In contrast, in the adult femur (Text-fig. 5C, D), the mediolateral width of the shaft remains approximately constant throughout most of its length. In juveniles, the fourth trochanter is mediolaterally thinner and slightly less extensive proximodistally than in the adult. Notably, the adult fourth trochanter shows a more expanded distal region than in the juvenile. Whereas in the juvenile, it abruptly merges with the shaft distally, in the adult, the distal region of the trochanter extends further distally and merges with the shaft more gradually (Text-fig. 5A, C). The distal end of the femur experiences a substantial increase in mediolateral breadth during ontogeny (Text-fig. 5B, D).

Figure TEXT‐FIG. 5..

Bactrosaurus johnsoni, hindlimb elements. A, B, juvenile right femur, AMNH 6577. A, medial and B, posterior views. C, D, adult left femur, AMNH 6553. C, medial and D, posterior views (reversed). E, proximal half of adult right tibia, AMNH 6553, in lateral view. F, juvenile proximal half of right tibia, AMNH 6553, in lateral view. G, distal half of adult left tibia, AMNH 6553, in anterior view. H, juvenile distal half of right tibia, AMNH 6553, in anterior view (reversed). aig, anterior intercondylar groove; cnc, cnemial crest; fmh, femoral head; gtr, greater trochanter; lc, lateral condyle; lm, lateral malleolus; ltr, lesser trochanter; mc, medial condyle; mm, medial malleolus; 4tr, fourth trochanter. Scale bars represent 50 mm for A–D and 100 mm for E–H.

Tibia.  The tibial shaft (Text-fig. 5F, H) is relatively slender in the juvenile, but thicker in the adult as exemplified by AMNH 6553. Specifically, the ratio between the mid-shaft diameter and the maximum mediolateral width of the distal end is 0.28 in the juvenile and 0.45 in the adult. The proximal posterolateral condyles are proportionately narrower than in the adult tibia, particularly the inner condyle, which experiences a substantial increase in mediolateral thickness during ontogeny (Text-fig. 5E, F). The adult cnemial crest is proportionately thicker both proximally and distally. At the distal end of the tibia, the ventral offset of the lateral malleolus is greater in the adult than in the juvenile tibia (Text-fig. 5G, H). The anterior surface of the lateral malleolus of the adult tibia shows coarser longitudinal striations than in the juvenile.

Astragalus and calcaneum.  The calcaneum is represented by three juvenile and three adult specimens (Text-fig. 6A–C). The adult calcanea differ from those of juveniles in having more deeply concave dorsal and posteromedial surfaces, as well as coarser indentations along the medial articular margin for the astragalus. The only observed ontogenetic change in the astragalus (Text-fig. 6D–I) regards the anteroposterior constriction of the ventral surface, at the level of the ascending process: this constriction is slightly more developed in the adult astragalus.

Figure TEXT‐FIG. 6..

Bactrosaurus johnsoni, proximal tarsal and metatarsal elements. A, adult left calcaneum, AMNH 6553, in medial view. B, juvenile left calcaneum, AMNH 6579, in medial view. C, juvenile right calcaneum, AMNH 6579, in medial view (reversed). D–F, adult right astragalus, AMNH 6579. D, anterior, E, medial, and F, ventral views. G–I, juvenile right astragalus, AMNH 6577. D, anterior, E, medial, and F, ventral views. J, adult right metatarsal II, AMNH 6553, in lateral view. K, adult left metatarsal III, AMNH 6553, in medial view. L, adult left metatarsal IV, AMNH 6553, in medial view. M, juvenile left metatarsal II, AMNH 6577, in lateral view (reversed). N, juvenile right metatarsal III, AMNH 6577, in medial view (reversed). O, juvenile right metatarsal IV, AMNH 6577, in medial view (reversed). aap, anterior ascending process; asar, astragalar articular region; ccar, calcaneal articular margin; ff, fibular articular facet; mfg, mediodorsal flange; mtb, medial tuberosity; pap, posterior ascending process; tf, tibial articular facet; vfg, medioventral flange. Scale bars represent 20 mm for A–I and 50 mm for J–O.

Metatarsus.  Juvenile metatarsals II, III, and to a lesser extent IV are proportionately longer than adult ones (Text-fig. 6J–N). In metatarsal II, the ratio between its total length and the maximum depth of the distal end is 4.5–5.0 in juveniles, whereas it is only 3.0 in adults. In the case of metatarsal III, this ratio is 4.0 in juveniles and 3.2 in adults. The medial flange of metatarsal II is gently developed in juveniles (e.g. AMNH 6577; Text-fig. 6M), whereas it bulges much more prominently dorsally in adults (e.g. AMNH 6553; Text-fig. 6J). Similarly, in juvenile metatarsal III, the medioventral margin of the proximal half of the bone lacks the prominent and mediolaterally thick ridge that is present in adult specimens (e.g. AMNH 6553). In metatarsal IV, a large tuberosity with a rugose texture protrudes medioventrally from the medial surface of the metatarsal. The juvenile tuberosity has a less rugose surface and a less prominent dorsal margin than that of adults (Text-fig. 6L, O).

Pedal phalanges.  Larger specimens show an increase in overall robustness and more coarsely textured muscle scars. As noted by Gilmore (1933), juvenile pedal unguals are slightly more elongated proximodistally than those of adults.

Comparisons with Previously Documented Trends of Ontogenetic Change in Hadrosauroids

Cranium

In Bactrosaurus johnsoni, the length/width proportion of the frontal is maintained from juveniles to adults, in contrast to the decrease in length during ontogeny that occurs in at least one hadrosaurid, Gryposaurus cf. notabilis (Waldman 1969). Maryanska and Osmólska (1981) observed a rounded convexity on the medial side of the frontal in juveniles of the hadrosaurid Saurolophus angustirostris that was absent in adult specimens. No such convexity was observed in either juvenile or adult elements of B. johnsoni.

In the maxilla, Horner and Currie (1994) noted that the premaxillary shelf of neonate Hypacrosaurus stebingeri was steeper than that of adults. In Bactrosaurus johnsoni, the anterodorsal margin of the maxilla shows no substantial decrease in steepness from juveniles to the adult specimen. This indicates that, if the anterodorsal maxillary margin was much steeper in early ontogeny, most of the reduction in steepness occurred before the juvenile stage preserved in the B. johnsoni sample. The position of the dorsal process of the maxilla has been reported to remain constant in juveniles and adults in H. stebingeri (Horner and Currie 1994), unlike the ontogenetic posterior shift observed in B. johnsoni.

One of the most commonly observed trends in hadrosauroid ontogeny is the decrease in the relative size of the orbit, documented in Saurolophus angustirostris (Maryanska and Osmólska 1981), Brachylophosaurus canadensis (Prieto-Márquez 2005), Hypacrosaurus stebingeri (Horner and Currie 1994) and other lambeosaurine hadrosaurids (Evans et al. 2005), as well as in crocodilians (Dodson 1975) and other dinosaurs (Varricchio 1997). In Bactrosaurus johnsoni, the width of the ventral orbital margin, as evidenced by the jugal, appears practically constant throughout the range of sizes present in the sample, suggesting that decrease in orbit size occurred prior to the smallest juvenile state represented for this taxon.

The rostral jugal process becomes anteroposteriorly shorter during ontogeny in at least hadrosaurids Gryposaurus cf. notabilis (Waldman 1969) and Hypacrosaurus stebingeri (Horner and Currie 1994). Waldman (1969) observed that in a juvenile G. cf. notabilis the dorsal border of the rostral jugal process is lower than in adults. Those two changes could not be assessed in Bactrosaurus johnsoni because of the incomplete preservation of the anterior jugal process. Gates and Sampson (2007) noted that the postorbital process of the jugal of G. monumentensis becomes more vertical during ontogeny. This trend also appears to be present in B. johnsoni, although the change is very subtle (Text-fig. 2). As in H. stebingeri (Horner and Currie 1994), the jugal of B. johnsoni becomes more robust throughout ontogeny.

No foreshortening of the frontal occurred in Bactrosaurus johnsoni during ontogeny, in contrast to that reported in the lambeosaurine hadrosaurid Hypacrosaurus stebingeri (Horner and Currie 1994).

In the dentary, different trends have been reported in non-hadrosaurid hadrosauroids and hadrosaurids. Kirkland (1998) noted no substantial increase in the ventral deflection of the dentary of Eolambia caroljonesa between different sized specimens. In contrast, Godefroit et al. (2004) documented an increase in the ventral deflection, as well as a posterior shift of the origin of this deflection along the ventral margin, in the larger dentaries of the lambeosaurine Amurosaurus riabinini. Bactrosaurus johnsoni follows the latter trend. Maryanska and Osmólska (1981) observed that in juveniles of Saurolophus angustirostris, the coronoid process is nearly vertically oriented, whereas in adult specimens of this species, the process is anteriorly inclined. In contrast, the vertical orientation of the coronoid process remains ontogenetically invariable B. johnsoni.

Appendicular skeleton

Horner and Currie (1994) emphasized the very narrow proximal constriction present in the scapulae of Hypacrosaurus stebingeri neonates, a condition further developed in the juveniles than in the adults of Bactrosaurus johnsoni. The width of the distal scapula blade in B. johnsoni juveniles is comparable to that observed in adults (contraGilmore (1933), who regarding the same specimens pointed out that the scapula does not possess the widened blade present in the adult). Additional ontogenetic changes in hadrosauroid scapulae were reported by Brett-Surman and Wagner (2007), of which only the increased robustness and posteroventral elongation of the deltoid ridge was observed in B. johnsoni.

Several authors have documented an increase in the prominence of the deltopectoral crest throughout hadrosauroid ontogeny (Grigorescu and Csiki 2006; Brett-Surman 1989; Godefroit et al. 2004; Brett-Surman and Wagner 2007), a trend that is also present in Bactrosaurus johnsoni. However, the laterodistal displacement of the crest reported by Dilkes (2001) in ontogenetic series of the hadrosaurid Maiasaura peeblesorum was not observed in B. johnsoni. Likewise, Godefroit et al. (2004) observed that the deltopectoral crest of the hadrosaurid Amurosaurus riabinini changes from being slightly convex in juveniles to straighter in adults. This ontogenetic change does not occur in B. johnsoni.Norman (2002) noted a more pronounced curvature of the humeral shaft in smaller individuals of the basal hadrosauroid Probactrosaurus gobiensis, a trend not found in B. johnsoni. The increased robustness of the humerus during ontogeny in B. johnsoni is consistent with that noted by Dilkes (2001) in Maiasaura peeblesorum, but contrary to the ontogenetic reduction in robustness observed by Horner and Currie (1994) in Hypacrosaurus stebingeri. Brett-Surman and Wagner (2007) pointed out a number of ontogenetic changes in hadrosauroid humeri, of which only the increased development of the tuberosities and of both distal condyles was noted in B. johnsoni.

Regarding the ulna of the basal hadrosauroid Telmatosaurus transsylvanicus, Grigorescu and Csiki (2006) observed an increase in robustness and a decrease in the proximal width/minimum width ratio during ontogeny. In Bactrosaurus johnsoni, the ulna also increases in robustness in adult specimens; however, the decrease in the proximal width/minimum width ratio was not apparent. The increasing development of the olecranon process in B. johnsoni has also been observed in other hadrosauroids (Dilkes 2001; Brett-Surman and Wagner 2007).

Brett-Surman and Wagner (2007) found that in general the shape and proportions of the ilium of hadrosauroids remain uniform through ontogeny, as shown here for Bactrosaurus johnsoni. These authors also noted that later in ontogeny the preacetabular process typically experiences substantial mediolateral thickening, a trend also present in B. johnsoni.

Grigorescu and Csiki (2006) showed that in the basal hadrosauroid Telmatosaurus transsylvanicus, the femur experienced an allometric increase in robustness and enlargement of the distal end, a trend also present in Bactrosaurus johnsoni. Likewise, the increased development of the fourth trochanter of B. johnsoni reported here has also been described for other hadrosauroids (Dilkes 2001; Brett-Surman and Wagner 2007). However, the proximal shift of the fourth trochanter reported by Grigorescu and Csiki (2006) in T. transsylvanicus is absent in the ontogenetic series of B. johnsoni.

In agreement with previous observations on hadrosauroids (Dilkes 1996; Grigorescu and Csiki 2006; Brett-Surman and Wagner 2007), no noticeable allometric changes appear to be present in the tibia of Bactrosaurus johnsoni during ontogeny, with the exception of the cnemial crest. The latter becomes more anterolaterally expanded in adults of B. johnsoni, as occurs in the hadrosaurid Maiasaura peeblesorum (Dilkes 1996) and the basal hadrosauroid Telmatosaurus transsylvanicus (Grigorescu and Csiki 2006). However, the crest does not exhibit an increase along its proximodistal extent in B. johnsoni, contrary to the situation reported for hadrosauroids by Brett-Surman and Wagner (2007). Notably, the decrease in robustness of the tibia and femur observed throughout ontogeny in the lambeosaurine Hypacrosaurus stebingeri (Horner and Currie 1994) was not observed between juvenile and adult specimens of B. johnsoni.

Implications of Ontogenetic Variation for Phylogenetically Informative Characters in Basal Hadrosauroids

The following cranial characters, previously used in the inference of hadrosauroid interrelationships, are variable ontogenetically: development of the annular ridge (typically short and sharp in lambeosaurines but low and long in other hadrosauroids; Evans 2007; Evans and Reisz (2007), character 43); position of the maxillary dorsal process (Prieto-Márquez (2010a), character 90); slope of the dorsal margin of the maxillary rostroventral process (Prieto-Márquez (2010a), character 87); extent of curvature of the ventral margin of the jugal (Norman 2002); ventral expansion of the posteroventral flange of the jugal (Wagner 2001); overall jugal robustness (Weishampel et al. (1993), character 13: note, however, that such ontogenetic change in jugal robustness does not occur in all hadrosauroids. For example, it appears to be absent in Lambeosaurus spp. but present in Edmontosaurus spp.; see Prieto-Márquez (2010a), character 114); curvature of the caudal margin of the quadrate (Wagner 2001); length/width proportions of the predentary (Horner et al. (2004), character 13); dorsoventral depth of the anterior predentary surface (Horner et al. (2004), character 14); number of denticles on the oral predentary margin (Prieto-Márquez (2010a), character 27: according to the latter work, and contrary to the situation in B. johnsoni, the number of predentary denticles increases through ontogeny in some hadrosauroids like Edmontosaurus spp., whereas no variation is observed in other taxa such as Prosaurolophus maximus); deflection angle of the ventral margin of the dentary (Prieto-Márquez (2010a), character 36); presence or absence curvature of medial margin of the surangular (Prieto-Márquez (2010a), character 56); and ontogenetic increase in the number of dentary and maxillary teeth and number of teeth per alveolus (Horner et al. (2004), characters 1 and 2; see also Gilmore 1933; Waldman 1969; Horner and Currie 1994).

In the postcrania, the following phylogenetically informative postcranial characters exhibit ontogenetic variation: development of the deltoid ridge of the scapula (Prieto-Márquez (2010a), character 218); lateroventral expansion of the humeral deltopectoral crest (Horner et al. 2004); overall length/width proportion of the humerus (Weishampel et al. (1993), character 36); length/dorsoventral thickness ratio of the ulna (Prieto-Márquez (2010a), character 223); and depth of the proximal region of the preacetabular process of the ilium (Prieto-Márquez (2010a), character 233).

Conclusion

Detailed examination of ontogenetic variation in the basal hadrosauroid Bactrosaurus johnsoni indicates that most morphological changes taking place from juvenile to adult are concentrated in the premaxilla, prefrontal, maxilla, jugal and predentary bones of the skull; in the appendicular skeleton, limb bones appear to experience more variation than pectoral and pelvic girdle elements. Furthermore, this study suggests that, at least for Bactrosaurus johnsoni (and perhaps in basal hadrosauroids in general), most of the phylogenetic characters used for inferring hadrosauroid relationships are not affected by ontogeny. Thus, juvenile and subadult specimens may provide a substantial amount of reliable information for phylogenetic inference in these animals.

Acknowledgements.  I thank M. Norell for the opportunity to examine the holotype and juvenile specimens of Bactrosaurus johnsoni housed at the AMNH and P. Godefroit for allowing access to casts of the SBDE 95E5 specimens housed at the IRSNB. I am also grateful to P. Barrett, T. Gates and A. McDonald for their valuable comments that improved the quality of the manuscript. This study was supported by a Kalbfleisch postdoctoral fellowship provided by the American Museum of Natural History, a Charlotte and Walter Kohler Charitable Trust grant, and the National Science Foundation (EAR 0959029 grant presented to G. Erickson and M. Norell).

Editor. Paul Barrett

Ancillary