Histologic Examination of an Assemblage of Psittacosaurus (Dinosauria: Ceratopsia) Juveniles From the Yixian Formation (Liaoning, China)

The basal ceratopsian Psittacosaurus was first discovered in 1922 during the Third Asiatic Expedition to Mongolia and was originally based on two well preserved complete skeletons (Osborn, 1923, 1924). Since this initial discovery, Psittacosaurus has become one of the most abundant dinosaurs known with specimens from across Asia, including China, Mongolia, Russia, and possibly Thailand (Buffetaut and Suteethorn, 1992; Sereno, 2010). The Yixian Formation in Liaoning, China is famous for the excellent preservation of feathered avian and non-avian theropods, other dinosaurs, plants, insects, fishes, non-dinosaurian reptiles, and mammals (see Xu and Norell, 2006 for review). The Yixian Formation is also one of the most abundant localities preserving Psittacosaurus remains in both two and three dimensions (Hedrick and Dodson, 2013). This includes a beautiful specimen of a larger Psittacosaurus individual preserved with at least 24 smaller juveniles that are all approximately the same size (DMNH D2156) (Meng et al., 2004; Hedrick et al., 2014a). Considering that Psittacosaurus is known from such a large amount of complete material, it is an important taxon for answering complex questions about dinosaur biology, especially considering that the majority of named dinosaur species are based on single specimens (Dodson, 1990; Wang and Dodson, 2006). Given the large sample size for Psittacosaurus from different ontogenetic stages, it has been possible to fully analyze a wide range of sizes of Psittacosaurus using bone histology (Erickson and Tumanova, 2000; Erickson et al., 2009; Zhao et al., 2013).

Bone histology and skeletochronology provide insights into animal biology, including growth rates, age of sexual maturity, and age of individuals (Chinsamy, 1990, 1993a; Chinsamy et al., 1995; Sander, 2000; Erickson and Tumanova, 2000; Chinsamy and Elzanowski, 2001; Horner et al., 2001, 2000; Starck and Chinsamy, 2002; Erickson et al., 2009; Sander et al., 2011; Hedrick et al., 2014b). Without bone histology, many of these important aspects of dinosaur life history would remain unknown. Growth curves have been calculated for a large number of dinosaurs including Massospondylus (Chinsamy, 1993a), Tyrannosaurus (Erickson et al., 2004) and many others (see Erickson et al., 2001 and Chinsamy-Turan, 2005 for results of several taxa), including two species of Psittacosaurus, P. mongoliensis and P. lujiatunensis (Erickson and Tumanova, 2000; Erickson et al., 2009). These data were further used to estimate life history and survivorship curves for Psittacosaurus. More recent histologic analyses of Psittacosaurus have identified growth trends potentially related to ontogenetic transitions from bipedality to quadrupedality (Zhao et al., 2013). However, histologic variation across a number of similarly sized Psittacosaurus from the same locality has not yet been assessed nor has serial sectioning of juvenile long bones, which would allow for an understanding of histologic variation across entire bones rather than just at the midshaft region.

DMNH D2156 is an assemblage of at least 24 articulated juvenile psittacosaurs associated with a larger skull and partial postcranium (Meng et al., 2004; Hedrick et al., 2014a). Because the range of variation in juvenile size is small (femur length: 30–36 mm), it is possible to examine a number of Psittacosaurus specimens from the same assemblage that were presumably the same age. Although developmental plasticity has been noted in other dinosaurs (Sander and Klein, 2005; Klein and Sander, 2007; Cerda and Chinsamy, 2012; Chinsamy et al., 2012), it does not appear to occur in Psittacosaurus (Erickson and Tumanova, 2000; Erickson et al., 2009; Zhao et al., 2013). Hedrick et al. (2014a) hypothesized that the small psittacosaurs in the assemblage were not embryonic on the basis that the ends of their long bones were well defined (Horner and Makela, 1979) and the lack of eggshell material in and around the assemblage. However, certain taphonomic factors allow for the preservation of bone without the preservation of eggshells, as was the case for Bagaceratops juveniles from the Iren Dabasu Formation (Dong and Currie, 1993). By analyzing the bone microstructure, it is possible to qualitatively evaluate the stage of development of the bone and check for the presence or absence of cartilage, which may allow inference on whether the animals were embryonic or neonates, and whether they were precocial or altricial (Barreto et al., 1993; Barreto, 1997; Horner et al., 2001; Hübner, 2012).

This study aims to (1) document the bone microstructure of Psittacosaurus within its first year of life; (2) confirm that the juveniles preserved in DMNH D2156 are approximately the same age; (3) understand variation in microstructure among different similarly aged specimens and also within a single specimen; (4) determine whether the specimens were embryonic or neonates; and (5) compare the early ontogeny of Psittacosaurus with other juvenile dinosaurs that have been studied histologically, including other psittacosaurs, to better understand basal ceratopsian growth.

Institutional Abbreviations: DMNH- Dalian Museum of Natural History, Dalian, Liaoning, China. IVPP- Institute of Vertebrate Paleontology and Paleoanthropology, Beijing, China. LPM- Liaoning Paleontological Museum, Shenyang, China. PKUVP- School of Earth and Space Sciences, Peking University, Beijing, China. YPM- Yale Peabody Museum, New Haven, CT, USA.

MATERIALS AND METHODS

To analyze juvenile Psittacosaurus bone microstructure, five femora were taken from the juvenile assemblage of DMNH D2156 for sectioning (Meng et al., 2004; Hedrick et al., 2014a) (Fig. 1). All specimens were preserved together in a single block and were buried during the same depositional event (Hedrick et al., 2014a). Hedrick et al. (2014a) generated a numbering scheme for the juveniles (Supporting Information Fig. S1) and the juvenile Psittacosaurus specimens chosen for sectioning correspond to specimens 4, 9, 12, 16, and 17 using that scheme. These specimens were chosen because they had relatively complete femora that were not overlain by other specimens. The specimens were then redesignated A–E in this study for easier reference (Fig. 1) (A = 9, B = 4, C = 12, D = 16, E = 17). Prior to sectioning, all femora were photographed and cast at the University of Pennsylvania Museum of Archaeology and Anthropology. Casts were returned to the Dalian Museum of Natural History to be reposited with DMNH D2156.

Femur A was missing both the proximal and distal ends and was the most poorly preserved of the five femora. Therefore, the bone was cut longitudinally as there was no way to establish the exact location of the midshaft for transverse sectioning (Fig. 1). Femora B, C, D, and E were cut transversely at the mid-diaphysis just distal to the fourth trochanter (Fig. 1). This is the location where dinosaur femora are usually sectioned and allows for comparison with other studies (Chinsamy, 1990; Chinsamy, 1993a; Sander, 2000; Horner et al., 2001; Erickson and Tumanova, 2000; Erickson et al., 2009; Zhao et al., 2013) as studies have shown that bone microstructure can vary based on the location of the section (Reid, 1990) possibly due to differential growth rates, whether or not the bone is weight bearing, or due to differing functional demands on the bone (Lee, 2004; Hübner, 2012). This can affect microstructural patterns and age calculations (Horner et al., 1999). Femora B, C, and D were cut longitudinally from the proximal metaphyses to the proximal ends and from the distal metaphyses to the distal ends to evaluate the microstructure at the epiphyses (Fig. 1). It should be noted that the ends of the bones are not true epiphyseal ends because these ends are not true articular surfaces (Reid, 1997; Chinsamy-Turan, 2005; Cerda et al., 2014), but we follow previous dinosaurian histology studies in referring to the distal end of the long bones as epiphyses (Padian et al., 2004). Femur E was cut transversely at the proximal and distal ends for comparison with the longitudinal sections made from femora A–D (Fig. 1). Because of the small sizes of the bones, all cuts were made using an Isomet 11-1180 low speed saw. Thirteen individual slices were then mounted on 25 × 50 mm² slides. Methods for thin sectioning follow Chinsamy and Raath (1992), and the terminology follows Francillon-Vieillot et al. (1990). Vascular canals were used as a proxy for degree of vascularization, though they contain other connective tissue in addition to blood vessels (Starck and Chinsamy, 2002). Slides were viewed and imaged using a Nikon Eclipse 6600 Pol petrographic microscope fitted with a Nikon DXM1200F digital camera.

Prior to sectioning, femora were measured using 300-mm Mitutoyo 500-173 digital calipers. Measurements were each repeated three times to verify each value to within one tenth of a millimeter to reduce the possibility of mismeasurement. To determine the age of each individual, femur length was plotted against age using data from Erickson et al. (2009) (n = 26) and the additional femora in this study (n = 5) with a reduced major axis regression (Supporting Information Table S1). This was done using the custom gmregress function in MATLAB (Mathworks, 2012), which was written by Antonio Trujillo-Ortiz (www.mathworks.com/matlabcentral). The percent of vascularity was also calculated and compared to data published by Horner et al. (2001). Chinsamy (1993b) also examined percent of canal space as a proxy for the percent of vascularization. However, this was done for a given field of view and not the entire bone. As dinosaur bone is now known to be widely variable in terms of vascularity across a single section (e.g., Cerda and Chinsamy, 2012), these results are not directly comparable to ours and are not included. The percent of vascularity was calculated using custom code written for MATLAB (Supporting Information File S1).

RESULTS

Microstructural Analysis

The bone microstructure is variably preserved across the five bones and some regions have undergone intense diagenetic alteration obscuring variable degrees of the microstructure. The alteration manifests as a dark brown or black infilling and is readily distinguishable from the bone microstructure. This is not uncommon in dinosaur taxa and has occurred in other small ornithischians (Horner et al., 2009) as well as other Psittacosaurus specimens (Erickson and Tumanova, 2000; Erickson et al., 2009). However, the microstructure is clear enough on each bone to permit inferences, especially femora C, D, and E (Fig. 2). Both femora D and E provided the best preserved microstructure, but we describe all five femora in order to evaluate variation in microstructure when apparent. No growth lines or annuli are present in any section in line with the hypothesis that these individuals are all <1 year of age (Hedrick et al., 2014a). Secondary osteons were also absent from all sections. There is a wide range of vascular orientations present within the same bone (Fig. 3A). Additionally, the levels of vascularity do not differ across the sections such that the inner cortex and outermost cortex have similar levels of vascularization.

Femur A was particularly poorly preserved throughout both internally and externally, lacking both its proximal and distal ends (Fig. 2A). There is partial histologic structure preserved along the midshaft showing primarily longitudinally oriented canals (Fig. 2A), as well as some structure on the distal end, similar to that seen in other femora. The bone is composed of woven tissue. Individual osteocyte lacunae are not visible. However, the majority of the bone has no discernable form or structure due to diagenetic alteration and this bone is not discussed further in favor of the better preserved femora.

Femur D has the best preserved transverse section. It is composed of woven bone with numerous flattened osteocyte lacunae that are randomly distributed throughout the matrix (Fig. 3B). The medullary cavity is filled with black and brown mineralized material that does not have any internal structure similar to femora A, B, and C. The medullary cavity does not have a clear outline such that the black and brown mineralized material grades to bone without a definitive line separating them. It also has a long axis twice the length of the short axis suggesting some postmortem crushing. The cortical bone has well preserved microstructure across the section though the vascular patterns are not uniform throughout similar to femur B and E (Fig. 2). The caudomedial region has clear radial organization and this is the region with the thickest cortex (∼1 mm) (Fig. 3A). The craniomedial and lateral regions have a combination of a simple reticular and longitudinal vascularization (Fig. 3A). The lateral part of the section has the thinnest cortex (∼0.5 mm) and has a highly organized circumferential organization. The development of the fourth trochanter is likely the cause of the radial organization and thicker cortex in the caudomedial region.

The proximal longitudinal section of femur D has well preserved microstructure along its entire length. The medullary cavity is mineralized with the brown diagenetic infilling making it readily identifiable as in the transverse section. Toward the midshaft, there is a combination of radially and longitudinally organized vascular canals. Approximately half way up the proximal section, the orientation of the canals becomes entirely longitudinal with the canals radiating outward proximally similar to femora B and C. These canals exit the bone proximally as in femur B (Fig. 3C). The distal section is not well preserved. It is not possible to identify the medullary cavity and no canals or microstructure can be identified.

The transverse section of femur B is made up of woven tissue lacking any calcified cartilage. It primarily has simple reticular vascular organization with interspersed longitudinal, circumferential, and radial canals. No growth lines are present and the bone is highly vascularized. Based on the shape of the medullary cavity, it is evident that the bone has been diagenetically compacted as in femur D. The long axis of the medullary cavity is twice as long as the short axis in this femur, but in juvenile dinosaurs (and other juvenile Psittacosaurus femora studied here) it is typically circular since the specimens had not yet undergone cortical drift. The overall periosteal outline is still circular, but the caudomedial region has a thicker cortex (0.6 mm) than the rest of the cortex (∼0.4 mm) and has radially organized vascularization. The cranial region shows simple reticular or circumferential vascularization and has a thinner cortex (0.4 mm) suggesting that this region was growing relatively more slowly as in femur D (Fig. 3A), though it is apparent that all regions were growing quickly. Although no fourth trochanter was preserved on this femur, it is clearly developed in some similarly sized specimens (see femur D, Fig. 1). This could be related to either preservation or plasticity in growth. Because this section was taken at midshaft, it is likely that the thicker cortex is related to faster growth due to development of the fourth trochanter as has been demonstrated for other taxa (e.g., Reisz et al., 2013).

Both the proximal and distal longitudinal sections of femur B have well preserved histologic structure. The bone is composed of woven tissue. The canals are longitudinally oriented with a slight radial orientation. The proximal end has clearly preserved structure with longitudinally oriented canals that open to the periosteal surface proximally (Fig. 3C). Horner et al. (2001) identified longitudinally oriented canals in perinatal dinosaurs and posited them to be cartilage canals; these are potentially homologous with the longitudinal canals preserved in femur B. However, femur B does not appear to preserve any cartilaginous structure as would be expected based on comparisons with Maiasaura long bones (Horner et al., 2000) and the canals may simply be longitudinal vascular canals as reported for juvenile birds (Tumarkin-Deratzian et al., 2006) (see Discussion). The distal end of the distal section is diagenetically altered so similar canals cannot be confirmed (Fig. 2B).

The midshaft of femur E is well preserved and composed of woven bone, though ∼50% of the caudomedial outer cortex was diagenetically altered (Fig. 2E). Flattened osteocyte lacunae are present and are randomly distributed throughout the section (Fig. 3B). The osteocyte lacunae are oblong, as demonstrated for Mussaurus (Cerda et al., 2014). The beginnings of primary osteons are also evident with several layers of lamellar bone surrounding them, which fluoresce under cross-polarized light (Fig. 3D). The vascular organization across the section is primarily simple reticular, though it grades to radial in the caudomedial part of the section similar to both femora B and D (Fig. 3E). Also as in femora B and D, the region with radial canals is the region with the thickest cortex. Unlike the other transverse sections, the medullary cavity is filled with calcite and is clearly defined. The medullary cavity is relatively small in comparison with adult psittacosaurs and the medullary margin is resorptive suggesting expansion of the medullary cavity.

Femur E is the only femur where transverse sections were taken along the proximal and distal regions rather than just at the midshaft (Figs. 1 and 2E). The proximal transverse section is diagenetically altered on the medial side and well preserved on the lateral side. The cortex is much thinner (<0.3 mm) in the proximal section than in the midshaft (0.4–0.9 mm) and the majority of the section is composed of trabeculae similar to Maiasaura tibial proximal transverse sections (Horner et al., 2001, p 50). The trabeculae are composed of bony struts and the outer cortex is made up of woven bone (Fig. 3F). There is no evidence of chondrocytes. The outer cortex is along the fringe of the section and is primarily composed of longitudinal canals (Figs. 2E and 3F). A similar organization is seen in the distal section, though there is no diagenetic alteration. The medullary cavity is not clearly differentiated from the outer cortex in either the distal section or proximal section (Fig. 2E).

The transverse section of femur C is made up of woven bone tissue and has simple reticular organization of vascular canals on the lateral side of the bone, but the bone is highly fragmented on the medial side (Fig. 2C). The proximal longitudinal section is also poorly preserved. Close to the midshaft of the proximal longitudinal section, there are several longitudinal canals, but more proximally the preservation precludes comment. The distal longitudinal section is well preserved close to the midshaft of that section. There are clear longitudinal canals that radiate outward as they go further distally similar to the longitudinal canals in the longitudinal sections of femur B (Fig. 2B). The canals appear as black lines making them readily identifiable, especially toward the midshaft (Fig. 2C). However, the canals are lost half way down the distal section and are replaced by the diagenetically altered structure that characterizes the proximal longitudinal section.

Age Determination

Erickson et al. (2009) developed an equation for calculating age based on femur length in P. lujiatunensis with histologic data from 26 specimens. The equation given by Erickson et al. (2009) was y = 0.0615 x – 1.9214. However, as pointed out by Myhrvold (2013, 2015), age should have been used as the independent variable (x-axis) and femur length should have been used as the dependent variable (yiaxis) as has been done in previous histologic studies (Chinsamy, 1993a; 2005). Reduced major axis (RMA) regressions are preferred over ordinary least squares regressions when both axes are subject to error (Sokal and Rohlf, 2011). Age has an associated variance considering that the ages were obtained by retrocalculation (Castanet et al., 1993; Erickson et al., 2009) and the age is inherently uncertain because LAGs only record ages to the nearest year (assuming annual deposition of LAGs) (Castanet, 1994) so the actual age of the animal within that year is unknown. Therefore, we recalculated Erickson et al. (2009)'s regression using an RMA regression with age as the independent variable. This gave the equation y = 17.25 x + 23.31 with upper and lower confidence intervals (slope: 16.35–18.15; y-intercept: 19.82–26.79; R² = 0.981) at α = 0.05 (Fig. 4; Supporting Information Table S1). Using either the equation from Erickson et al. (2009) or the revised equation presented here, the DMNH D2156 specimens in this study were less than 1 year old (RMA ages: 0.39–0.74, Erickson et al. (2009) ages: −0.076–0.29).

DISCUSSION

While the microstructure of Psittacosaurus appendicular bones has been extensively studied, far fewer studies have focused on ornithischian histology as a whole in comparison with either Theropoda or Sauropodomorpha (Fig. 5). Only 17 ornithischian genera have been examined from an appendicular histologic perspective. A greater variety of ornithischian taxa must be examined to determine potential phylogenetic trends in histology for the clade. This study seeks to compare the juvenile microstructure of Psittacosaurus with that of other ornithischians to increase further the number of taxa that have been fully evaluated within Ornithischia.

A number of P. lujiatunesis specimens examined by Erickson et al. (2009) were shown to be 1 year of age or younger (IVPP 14155, PKUVP V1058 (n = 2), LPM R00142 (n = 4), LPM R00118). The Erickson et al. (2009) analysis focused primarily on the establishment of life history and survivorship curves for Psittacosaurus rather than the description of primary histologic structure through development. Zhao et al. (2013) described Psittacosaurus microstructure across a growth series composed primarily of juvenile and subadult specimens. Of their 16 specimens, 11 were considered subadults and three of these specimens (IVPP V16902.1, V16902.2, V16902.3) were slightly smaller than our specimens with femur lengths of 22, 25, and 26 mm respectively. These individuals, like those in our study are likely <1-year old based on the absence of LAGs and overall size. The other specimens were all 1 year of age or older based on LAG counts from various limb bones (Zhao et al., 2013). However, that study focused on ontogenetic postural shifts rather than primary histologic description. In spite of these previous studies (Erickson and Tumanova, 2000; Erickson et al., 2009; Zhao et al., 2013), the microstructure of small Psittacosaurus specimens has yet to be described in detail, and is here presented based on multiple sections of five small juveniles based on not only histology for the midshaft region, but also for the metaphyses and epiphyseal regions. This allows for inferences to be made about (1) histological variation between different similarly sized juvenile Psittacosaurus specimens from the same locality; (2) histological variation within a single bone of juvenile Psittacosaurus; and (3) comparison with previous studies that examined Psittacosaurus histology as well as studies that examined juvenile dinosaur microstructure more generally.

The DMNH D2156 juveniles have little histologic variation across the five specimens in bone microstructure, except in terms of preservation. In general, the longitudinal sections are all composed of woven bone tissue and all preserve well developed longitudinal canals that radiate slightly outwards as they proceed away from the midshaft and continue to the metaphyses (Fig. 3F). The only exception is femur D, which has a combination of longitudinal canals and radiating canals near the midshaft in longitudinal section. However, these canals become longitudinally oriented proximally (Fig. 2D). The better preserved transverse sections demonstrate numerous flattened, randomly distributed osteocyte lacunae. All regions have similarly high levels of vascularity in the inner and outer cortex, though the orientation of the vascular canals varies regionally. No regions are avascular or nearly avascular. Based on the bone tissue, vascularity levels, and comparisons with adults (Erickson et al., 2009), it is clear that these animals were growing rapidly prior to the postnatal slowdown that is evident in dinosaurs and birds (Chinsamy and Elzanowski, 2001).

A number of studies have previously examined ornithischian metaphyseal histology in depth including Maiasaura (Horner et al., 2000), Scutellosaurus (Padian et al., 2004), Orodromeus (Padian et al., 2004), and Gasparinisaura (Cerda and Chinsamy, 2012). A common finding in many metaphyseal sections was calcified cartilage, especially in perinate material (Barreto et al., 1993; Barreto, 1997; Horner et al., 2001). Haines (1969) performed a comparative study of the epiphyses of reptiles and found a wide range of variation in epiphyseal development. Horner et al. (2001) expanded upon this study by including birds and dinosaurs. They found that turtles have large epiphyseal cartilage cones with uncalcified, hypertrophied chondrocytes whereas Alligator has calcified cartilage along the periphery of the cartilage cone. Birds (Struthio, Dromaius, Sturnella) were shown to have large cartilage canals running into the metaphyses, with a cartilage cone that ossified after hatching (Horner et al., 2001, 2000). Horner et al. (2001) also examined Orodromeus, Troodon, Maiasaura, and indeterminate hadrosaurs and found what appear to be cartilage canals at the epiphyses, which had previously been reported in extant reptiles (Haines, 1969). The epiphyses in other dinosaurs such as Mussaurus were composed entirely of hypertrophied calcified cartilage perforated by cartilage canals as well demonstrating that these structures are widespread in the Dinosauria (Cerda et al., 2014). Similar canals are readily visible in the longitudinal sections taken from femora B and D (Fig. 3C) and in the transverse sections taken from femur E (Fig. 3F). The longitudinal canals in femora B and D seem to be much more proximally located than those in Maiasaura or Gasparinisaura based on longitudinal sections and it is evident that these canals did not extend very far from the proximal end in Psittacosaurus and more closely resembled the extent seen in Orodromeus (Horner et al., 2001; Cerda and Chinsamy, 2012). However, no calcified cartilage or chondrocytes were found in the canals in DMNH D2156 precluding definitive attribution of these canals to cartilage canals in these juveniles. Given the poor preservation of the bone, it is possible that the calcified cartilage was not preserved. We consider this the most likely scenario given the small size of the Psittacosaurus specimens here examined and the prevalence of calcified cartilage in juvenile dinosaurs (Barreto et al., 1993; Horner et al., 2001; Padian et al., 2004; Cerda and Chinsamy, 2012; Cerda et al., 2014) and even as islands of calcified cartilage in ontogenetically older dinosaurs (Padian et al., 2004; Cerda and Chinsamy, 2012).

Horner et al. (2001) suggested that the relative size of the cartilage zones in dinosaur perinates might be related to precocial versus altricial behavior. Both Troodon and Orodromeus had relatively well formed epiphyseal surfaces in comparison with the hadrosaurs examined suggesting that Orodromeus and Troodon may have been precocial (Horner et al. 2001). Horner and Makela (1979) used the poor development of long bones to infer altriciality for Maiasaura. The well developed epiphyseal surface, reduced region for the longitudinally oriented canals if identified as cartilage canals, and strong development of structures such as the fourth trochanter and lesser trochanter in the DMNH D2156 juveniles (Meng et al., 2004) is similar to what has been found for Orodromeus and Troodon. This suggests that Psittacosaurus had a similar degree of precociality to Orodromeus and Troodon and the individuals in this study were probably neonates. Evidence that the DMNH D2156 specimens are post-hatchlings comes from the lack of eggshell remains at the site (Hedrick et al., 2014a), the morphological development of the bones (see Hedrick et al., 2014a), and now the microstructure of the bones.

Variation in vascular canal orientation within the DMNH D2156 elements is much more pronounced in the transverse sections than the longitudinal sections, but overall variation is small between elements. All bones preserve the same general pattern. Femora B, D, and E all preserve a combination of radial, circumferential, longitudinal, and simple reticular canals (Figs. 2 and 3A,E). Horner et al. (1999), Padian et al. (2004), and Cerda and Chinsamy (2012) found substantial variation in single sections of ornithopod dinosaurs as well (e.g., Hypacrosaurus, Gasparinisaura). Erickson and Tumanova (2000) reported only longitudinal canals for femora of P. mongoliensis specimens of 1 year of age. However, the smallest animal sectioned by Erickson and Tumanova (2000) was 3 years of age and the microstructure of ages 1 and 2 were inferred from the microstructure near the medullary cavity in older animals, which may have been resorbed. Two-year-old P. mongoliensis specimens were inferred to have a mixture of longitudinal, reticular, and radiating reticular organization (Erickson and Tumanova, 2000), which is much more similar to that of the DMNH D2156 femora. This discrepancy could be related to interspecific variation in Psittacosaurus (P. mongoliensis versus P. lujiatunensis), or could be due to slight miscalculations in the retrocalculated vascularization patterns. After age 2, P. mongoliensis settled on a reticular vascular pattern and retained this pattern between ages 3 and 6 until redeveloping a radial vascular pattern later in life possibly related to varying local apposition rates due to osseous drift (Erickson and Tumanova, 2000). At around age 5, it is suggested that Psittacosaurus underwent a growth spurt resulting in increased vascularization and a shift from reticular to radial vascular orientation (Erickson and Tumanova, 2000; Padian et al., 2004). It is clear that Psittacosaurus grew relatively quickly as a juvenile as well based on the woven bone matrix and radial bone orientation in parts of the sections of DMNH D2156. This demonstrates that juveniles of P. lujiatunensis definitely had a substantial amount of variation in a single section similar to that suggested for 2-year old, and possibly younger, specimens of P. mongoliensis.

Other dinosaurs show vascular patterns different from what we found in DMNH D2156. Padian et al. (2004) first hypothesized that smaller dinosaurs have slower growth and thus less extensive vascularization than larger dinosaurs (but see Redelstorff et al. (2013) for alternate hypotheses in Thyreophora). Padian et al. (2004) commented on the relative growth rate of Psittacosaurus based on its size using data from Erickson and Tumanova (2000). However, juvenile Psittacosaurus growth was unknown at that time. Adult Scutellosaurus has mostly circumferential and longitudinally oriented canals with little fibrolamellar bone and low vascularization (Padian et al., 2004). The slightly larger, small ornithopod adult Orodromeus (femur length in Scutellosaurus is 5 cm in comparison with 17 cm in Orodromeus (Padian et al., 2004) has primarily longitudinal canals throughout its ontogeny and is less vascularized than other taxa (18% vascularity; Table 1), but more so than Scutellosaurus (Horner et al., 2009). In contrast, Maiasaura appears to have grown much faster than these other taxa (26–34% vascularity for juveniles; Table 1) (Horner et al., 2000; Padian et al., 2004). Psittacosaurus (18% vascularity; Table 1) likely grew at rates similar to Scutellosaurus or Orodromeus and somewhat slower than Maiasaura and Dysalotosaurus (Hübner, 2012), which is in line with the hypothesis that larger dinosaurs grew faster than smaller dinosaurs. In spite of the lower vascularity percentage, juvenile Psittacosaurus grew quite fast given that their bone was made up of woven tissue rather than more structured fibrolamellar bone, as is common in adult dinosaurs. It is necessary to consider the nature of the bone tissue when examining vascularization percentages, though the vascularization percentages are also meaningful (Chinsamy, 1993b).

De Margerie et al. (2002) demonstrated that a difference in osteon orientation is not statistically significantly correlated with growth rate in the mallard duck (Anas platyrhynchos) and that the differences in orientation may relate instead to biomechanical factors on the bone. Given that all three well preserved Psittacosaurus femora in this study have a single region of radial vascularization combined with a thickened cortex in the caudomedial region of the bone, it would appear that radial organization in juvenile Psittacosaurus specimens is related to the development of the fourth trochanter with a combination of simple reticular and longitudinal organization dominating in other regions. A similar pattern is found in Orodromeus, Maiasaura, Hypacrosaurus, Kentrosaurus, Mussaurus, and Gasparinisaura (Horner et al., 1999, 2000, 2009; Cerda and Chinsamy, 2012; Redelstorff et al., 2013; Cerda et al., 2014), but is less evident in Troodon (Horner et al., 2001), likely related to the smaller fourth trochanter in theropods compared with ornithischians. Mussaurus is somewhat unique in having large resorptive cavities in the fourth trochanter region (Cerda et al., 2014), but this is not seen in these Psittacosaurus juveniles. De Margerie et al. (2004) did find significantly accelerated growth rates for radial bone in the King Penguin (Aptenodytes patagonicus) so it is possible that early in development the fourth trochanter grows at an accelerated rate in dinosaurs. It is also possible that the radial bone is present in the fourth trochanter as necessitated by biomechanical stresses, though radial bone has poor biomechanical resistance in comparison to other vascular orientations (de Margerie et al., 2004). More work on extant species will be needed to further test these hypotheses, possibly on the development of the fourth trochanter in Alligator or the third trochanter in perissodactyls or primates.

Dinosaur embryos often have extensive vascularization that is open to the periosteal surface (e.g., Sauropodmorpha indet., Theropoda indet., Maiasaura) (Horner et al., 2001, Ricqles et al., 2001; Reisz et al., 2013). None of the Psittacosaurus specimens examined as part of this study demonstrated vascularization open to the surface, except at the epiphyses. Two “nestling” stages were examined for Maiasaura (Horner et al., 2000) allowing for a more detailed comparison with Psittacosaurus. The smallest animal was represented by a single femur (YPM-PU 22432) and had a spongy bone matrix with large vascular canals. The outermost vascular canals were open to the periosteal surface and there was no centripetal organization along the margins of primary osteons (Horner et al., 2000). The larger nestlings have a cortex that is clearly differentiated from the marrow cavity as in the DMNH D2156 psittacosaurs. The bone is densely vascular with woven bone making up the matrix also similar to the DMNH D2156 juveniles. This suggests that the juveniles in the DMNH D2156 assemblage were post-hatchlings as was likely the case for the larger nestling stages of Maiasaura. Sauropodomorph embryos found with eggshell material had asymmetric radial growth of the femoral shaft (Reisz et al., 2013) in very small specimens, but a pattern similar to that seen in DMNH D2156 specimens in larger specimens. The combination of different vascular orientations in addition to radial growth in DMNH D2156 potentially suggests a slower growth rate, possibly related to a later ontogenetic stage as in the larger sauropodomorph specimens. Indeterminate titanosaurs were shown to have extremely high growth rates in comparison with other dinosaur embryos, much higher than what is seen in DMNH D2156 (García and Cerda, 2010). These were suggested to be embryonic material not close to hatching (García and Cerda, 2010). Additionally, no Sharpey's fibers were evident in the DMNH D2156 sections. This may be due to poor preservation, but Horner et al. (2000) did not find Sharpey's fibers in perinatal Maiasaura specimens (femur length = 70 mm) that had fully differentiated fourth trochanters. Maiasaura regarded as larger nestlings (femur length = 120 mm) did have Sharpey's fibers (Horner et al., 2000) and Cerda et al. (2014) found Sharpey's fibers in Mussaurus juveniles along the fourth trochanter. Petermann and Sander (2013) recently demonstrated that muscle insertions were only detectable in 60% of thin sections of extant animals so the lack of Sharpey's fibers is not necessarily related to poor muscle development and may not be a valuable criterion for defining whether a specimen is precocial or altricial, or whether it is a neonate or embryo. These factors together suggest that the DMNH D2156 specimens are ontogenetically older than the earliest Maiasaura and sauropodomorph specimens (Horner et al., 2001; Reisz et al., 2013). A potential issue with this interpretation is that Maiasaura has been interpreted as altricial (Horner and Makela, 1979; Horner and Weishampel, 1988) while Psittacosaurus is here interpreted as more likely precocial, similar to Orodromeus and Troodon (Horner et al., 2001). The evaluation of ontogenetic stages in juvenile dinosaurs is complicated by the fact that the same characteristics that suggest a particular animal is a neonate also suggests that it is precocial. An embryonic precocial animal close to hatching may resemble an older altricial animal that was post-hatching. Therefore it is necessary to treat these conclusions with caution. More work on extant animals will be needed in order to better define the histologic factors that define embryos and neonates as well as altriciality and precociality so that they can be better applied to fossil taxa.

Jiang et al. (2014) hypothesized that the Lujiatun animals were caught in pyroclastic flows resulting in the excellent preservation of fossils from the Lujiatun beds. This hypothesis was supported by the fact that a number of animals from the Lujiatun beds that Jiang et al. (2014) examined histologically had modifications to their periosteal surface as a result of burning including pits, cracking, and elimination of microstructure close to the periosteal surface (see supplemental figure 20 of Jiang et al., 2014). On the basis of the previous evidence, Hedrick et al. (2014a) could not refute the hypothesis that DMNH D2156 was preserved in a pyroclastic flow, but preferred a lahar or fluvial flow due to mineralogical indicators. The lack of scorch marks or modifications to the periosteal surface of any DMNH D2156 specimen examined as a part of this study further supports the lahar or fluvial mode of deposition for the assemblage rather than the pyroclastic flow mode.

CONCLUSIONS

The large number of Psittacosaurus specimens known provides an excellent opportunity to study dinosaur biology, exemplified by specimens such as DMNH D2156. By examining five juvenile Psittacosaurus femora from DMNH D2156, it was possible to evaluate the microstructure of similarly sized juvenile animals all living closely together that were possibly from the same clutch (Meng et al., 2004) and compare it with that of other juvenile dinosaurs for which bone microstructure has been studied. The Psittacosaurus specimens all have similar microstructural patterns suggesting limited developmental plasticity in the taxon at this early age of development and that all of the juveniles in DMNH D2156 were the same age. The microstructural patterns including the lack of calcified cartilage in the epiphyses, the development of the fourth trochanters, and well-defined ends of the long bones suggest that the animals were neonates and possibly precocial (Hedrick et al., 2014a). In general, juvenile Psittacosaurus had microstructural patterns and a percentage of vascularization similar to Orodromeus, but somewhat less so than Maiasaura, hadrosaurs, embryonic titanosaurs, or Mussaurus, which presumably had higher growth rates (Horner et al., 2001; García and Cerda, 2010; Cerda et al., 2014). Finally, we were able to support previous studies (Hedrick et al., 2014a) that suggested DMNH D2156 was buried in either a fluvial or lahar flow rather than a pyroclastic flow based on the lack of modification to the microstructure near the periosteal surface. Ornithischia is one of the two main dinosaurian clades, but is far less studied than Saurischia in terms of bone histology. Given the panoply of information that can be garnered from paleohistology studies with regard to animal biology, we hope that more workers will examine ornithischian paleohistology to unravel more about this important clade.

ACKNOWLEDGEMENTS

The authors are thankful for helpful discussions with P. Martin Sander (University of Bonn, Germany), Sam Cordero (University of Pennsylvania), and Eric Morschhauser (Drexel University). They thank Samantha Cordero and Adam Laing (both University of Pennsylvania) for help collecting data on DMNH D2156 and Janet Monge and Lisa Gemmill (both University of Pennsylvania) for casting the material prior to sectioning at the UPenn Museum of Archaeology and Anthropology. They thank Dennis Terry (Temple University) and Gomaa Omar (University of Pennsylvania) for microscope access and James Ladd (Temple University) for assistance in making thin sections. Finally, they thank Jeffrey Laitman (editor) and an anonymous reviewer for helpful comments greatly improving the quality of the manuscript.