One important fact to highlight from our summary above is that two species that belong to the extant phylogenetic bracket of the Dinosauria, the chick, Gallus gallus, and the American alligator, Alligator mississippiensis, possess CB (Hall, 1971, Hall, 1972, Lengelé et al., 1990 and Vickaryous and Hall, 2008). This strongly suggests that CB could have been involved in the craniofacial development of non-avian dinosaurs (Witmer, 1995). Note that there is always a possibility that CB arose in parallel in crocodilians and birds. Whether avian and crocodilian CB is homologous or analogous, it is important to investigate whether or not non-avian dinosaurs had the ability to form CB and how ancient this tissue was within the Archosauria. Therefore, the skull of some Lambeosaurine and Hadrosaurine dinosaur embryos and juveniles were analyzed by means of paleohistological techniques and microradiography (Table 1). Paleontological crews of the Museum of the Rockies (MOR, Bozeman, Montana) have unearthed hundreds of remains of young hadrosaurs from nesting grounds (Horner, 1982, Horner and Makela, 1979, Horner and Currie, 1994 and Horner et al., 2000), and this abundant material was therefore selected for this investigation. Reporting CB in hadrosaurs would bring two important insights into the growth of their skull: •

first, even though it is commonly accepted that most hadrosaurs grew fast, fast growth has never been reported in the cranium, and most studies concern the post-cranium (e.g., Cooper et al., 2008, Horner et al., 2000 and Padian et al., 2001). Moreover, the presence, absence or relative abundance of this tissue in hadrosaurs would give qualitative insights into the growth rate of their skull;

Indeed, sutures are often mentioned in paleontological studies because they are used for maturity assessment (e.g., Bakker and Williams, 1988, Longrich and Field, 2012 and Sereno et al., 2009). However very little is known about the sutures of dinosaurs (or even extant archosaurs) from a histological perspective and only a few living mammalian species have been sectioned (i.e., some humans, Kokich, 1976, Koskinen et al., 1976, Latham, 1971, Miroue and Rosenberg, 1975, Opperman, 2000, Persson and Thilander, 1977 and Sitsen, 1933; and some rats and rabbits, Moss, 1958, Persson, 1973, Persson and Roy, 1979, Persson et al., 1978 and Pritchard et al., 1956). For future paleontological studies, it is important to document the osteohistology of sutures in order to understand their morphology and their (potential) relationship to ontogeny.

Finally a third point to note is the possible importance of the study of CB to understand the evolution of skeletal tissues within the Archosauria or the evolution of skeletal tissues in general. Indeed, as mentioned earlier, CB has been observed in birds (e.g., Lengelé et al., 1990), reptiles (Vickaryous and Hall, 2008) and mammals (e.g., Goret-Nicaise, 1984) but neither in anuran nor in urodele amphibians (e.g., Beresford, 1981 and Hall, 2003). This study could give more insights into the phylogenetic distribution of this intermediate tissue.

In extant species, CB can be identified histologically with the use of stains on decalcified sections (e.g., with Mallory's trichrome, Vickaryous and Hall, 2008; or methylene blue, Lengelé et al., 1990). Although it is possible to stain decalcified archaeological bone to observe histological structures (e.g., with Toluidine blue, Garland, 1989), stains used on dinosaur bones cannot show histological structures or different types of tissues per se, as they are usually histochemical (e.g., Sudan black can show lipids, Pawlicki, 1977; the Hoescht 33258 dye reveals DNA; Schweitzer et al., 1997). Therefore, we could not use a specific stain that would allow the identification of CB in these hadrosaurs. Instead, CB was identified qualitatively on undecalcified and unstained sections (i.e., standard paleohistological sections). As mentioned earlier, CB cells are round (cartilage-like) and embedded in a bone-like matrix. It is possible to differentiate CB from bone or cartilage under natural light (see section 3 for complete explanation).

For this study, we re-analyzed the paleohistological thin sections that were made for two previous studies on secondary cartilage (Bailleul et al., 2012 and Bailleul et al., 2013) because secondary cartilage and CB are often found in the same areas. About a hundred sections were analyzed, from a total of 21 separate skull bones (see Table 1). All elements were found disarticulated and were collected from nesting horizons or isolated nests in the Two Medicine and Judith River formations (Upper Cretaceous), Montana, USA (see Bailleul et al., 2012 and Bailleul et al., 2013 and Horner and Currie, 1994 for more information about the localities).

Paleohistological methods were employed according to the procedure for small specimens presented in Lamm (2013). The embryonic bones were embedded in epoxy resin and cut with a Norton 5′′ diamond blade on an Isomet precision saw. Thick sections (between 1.0 and 1.3 mm) were mounted on plastic (Plexiglas) slides with cyanoacrylate glue, then ground and polished by hand on a Buehler Ecomet grinder with silicon carbide paper of decreasing grit sizes. Finished thin sections (with a thickness surrounding 100 microns) were studied by light microscopy under normal and polarized light with a Nikon Optiphot-Pol polarizing microscope. Photographs were taken with a Nikon DS-Fi1 digital sight camera and the NIS Elements BR 4.13 software.

In extant species, another efficient method to identify CB is microradiography, an X-ray technique that shows the mineral distribution in calcified tissues at the microscopic scale (e.g., Goret-Nicaise and Dhem, 1982). Under microradiography, CB appears as radioopaque mineralized struts or islets containing irregular patches of confluent cellular lacunae that are radiotransparent (i.e., they appear dark). These struts and islets can be adjacent to woven bone or surrounded by lamellar bone, which have a distinct radiopacity and cell organization (Fig. 1A–D). Microradiography is also used on archaeological bone (e.g., Garland, 1989) and to our knowledge, only one other study performed microradiography on Mesozoic fossils (on reptilian and amphibian teeth from the Triassic, Wyckoff et al., 1963). Since paleohistological and histological methods are not standardized, we encountered problems while performing microradiography analysis, and we were only able to microradiograph five slides successfully (Table 2). Even though this is a small sample size, this study introduces a powerful tool that can be used by paleohistologists in the future.

All microradiography experiments were performed at the ‘Université catholique de Louvain’ (UCL, Brussels, Belgium). The pre-made paleohistological thin sections (70 to 90 microns thick, except for slide DE1-12 that was as thick as 175 microns in some places) were microradiographed in contact with a fine grain emulsion (VRP-M, Slavich-Geola, Lithuania), exposed to long-wavelength X radiations produced by a Machlett tube (Baltograph BF- 50/20, Balteau, Liège, Belgium) at 14 kV and 15 mA. In standardized sections (i.e., on extant material), the exposure usually lasts around 50 min, but because the X-rays did not go through the Plexiglas used at the MOR (i.e., it was too radio-opaque), the exposure lasted 4 hours for a film-focus distance of 106 mm. At UCL, the sections were previously flipped and the Plexiglas (not the bone) was ground down to about 100 microns in order to make it less radio-opaque. The microradiographs were observed and photographed with a Nikon Digital Sight DS-5MC camera (NIS Elements BR 3 software) attached to a microscope Leitz DMRB (Leica).

The microradiographs obtained from these hadrosaurs look exactly like those of humans (Fig. 1) and chicks (see Lengelé et al., 1990, Lengelé et al., 1996a) that possess CB (compare Fig. 1 with Fig. 3, Fig. 4, Fig. 5 and Fig. 6). Both the extant species and the hadrosaurs show islets of mineralized matrix with clusters of confluent cellular lacunae, typical of CB. This suggests that our identification in the fossil bone is correct. When comparing microradiographs with their corresponding natural light pictures, we found two noteworthy results. Firstly, while CB always has the same appearance under microradiography, different corresponding morphologies can be observed under natural light: the ECM of CB can be much brighter than the ECM of the surrounding bone (Fig. 2 and Fig. 4), it can have the appearance of normal woven bone (Fig. 3 and Fig. 6), or of a diagenetically altered bone (Figs. 4F, G, I). Moreover, the cell lacunae can appear round, i.e., almost like hypertrophied cartilage cells (Fig. 2E), or more irregular (Fig. 4G, I). Note that CB cells are sometimes known to present features of hypertrophied chondrocytes (e.g., Tuominen et al., 1996). In all the cases, the cellular density was always higher than that of the surrounding bone, or that of calcified cartilage (Fig. 7). Second, some areas that look like CB under natural light (with clusters of round cells closely packed together) look similar to bone under microradiography (see Fig. 3C-D, blue arrows). These conflicting results did not enable us to find a clear-cut relationship between microradiography and natural light pictures of CB. Nevertheless, we propose four characteristics that CB presents under natural light in paleohistological ground sections. Note, however, that these characteristics are not necessarily accurate for CB in ground- or decalcified sections of extant animals. Also note that those are preliminary results since further examination and discrimination of this tissue need to be made with paleohistological sections. The characteristics of CB are as follows: •

it presents large ‘cartilage-like’ cell lacunae (similar in morphology to those of hypertrophic cells of extant vertebrates), or irregular cell lacunae;

This diversity of morphologies could directly reflect skeletal diversity (i.e., there would be multiple types of CB). Indeed, Beresford (1981) classified two types of CB (type I and type II, see Chapter 1) and noted that it is difficult to draw limits between intermediate tissues, as they can form a “jungle of overlapping categories”. Another example of reported skeletal diversity is that in Benjamin, 1990, where six different types of cranial cartilage where reported in the black molly, Poecilia sphenops (also see Witten et al., 2010). To demonstrate that hadrosaurs possess different types of CB is beyond the scope of this paper, but we would like to note that this is a plausible hypothesis.

It cannot be ruled out that this morphological diversity does not reflect reality but instead is due to diagenesis (e.g., the latter could have altered the overall appearance of the tissues but not the orientation of their crystals). However, as mentioned in the results sections, both the microradiographs and the natural light images of the periosteal borders of the sampled elements showed good preservation (e.g., compare Fig. 4H with Fig. 1A,C). According to our results, diagenetic alterations seem to have been minor. Perhaps a bigger sample size, or a comparison between microradiographs and serial sections of some extant species stained with Masson's trichrome and/or Mallory's trichrome (i.e., what was used in Gillis et al., 2006 and Vickaryous and Hall, 2008) could help answer our questions about this morphological diversity.

In analogy to the five locations presented in this study, CB has been found: •

on the surangular of chick embryos (Lengelé et al., 1996a);

Finding this tissue in these duck-billed dinosaurs has two important implications for their skull growth: •

as mentioned earlier, CB is an adaptation to rapid growth (Gillis et al., 2006, Goret-Nicaise, 1986, Huysseune and Verraes, 1986 and Taylor et al., 1994). The fact that CB was found in large amounts in these dinosaur embryos suggests a rapid embryonic skull growth. Alas for now, we can only make qualitative statements since a comparison between microradiographs of slow and fast-growing animals (with quantitative information) has never been undertaken. This rapid development is not surprising because it is already known that the post-cranium of hadrosaurs grew fast and that they were more endotherms than ectotherms (Cooper et al., 2008, Horner et al., 2000 and Padian et al., 2001);

Our results suggest that CB might also have been involved in the sutural growth of hadrosaurs. Many dinosaur studies use the degree of closure of sutures to assess the maturity of their specimens and in turn, if they are identified as ‘adults’, this sometimes is considered grounds for naming a new species (e.g., Bakker and Williams, 1988 and Sereno et al., 2009). However, such conclusions should be reconsidered since there is a lack of understanding of sutures in non-mammalian species (including the phylogenetic bracket of the Dinosauria) in terms of: •

pattern (i.e., the order in which sutures fuse through ontogeny);

Therefore, it is important to start documenting and understanding the osteohistology of sutures in extant and extinct archosaurs. Our results suggest that sutural growth was very similar in birds, mammals and hadrosaurs (at least during embryonic development), but other preliminary results suggest that this similarity fades through ontogeny and leads to more unique modes of fusion (Bailleul and Horner, 2013). Even though CB was only found in one suture of Hypacrosaurus (i.e., the only suture that was microradiographed), this is the first step in understanding more about sutural growth in non-avian dinosaurs, a process that we need to comprehend to accurately assess their maturity.

CB has not been studied in many clades of vertebrates, therefore hasty and preliminary conclusions should not be made. However, it is not surprising to find CB in non-avian dinosaurs, because members of their extant phylogenetic bracket also form this tissue (Hall, 1971, Hall, 1972, Lengelé et al., 1990 and Vickaryous and Hall, 2008). It is parsimonious to say that the CB in Gallus gallus, Alligator mississippiensis, and hadrosaurs are homologous and that it was present in their most recent common ancestor. If our hypothesis of homology holds true, CB would be present in the other clades of dinosaurs as well, but only more paleohistological and microradiography examinations could verify this.

It is not clear exactly when CB arose within the Archosauria, and if it is homologous to the CB found in less derived vertebrates (e.g., in Gillis et al., 2006 and Ørvig, 1951). Nevertheless, this study sheds some new light on an old problem, reporting the presence of CB in an additional clade within the Vertebrata. This may be of importance for those who study skeletal diversity and the evolution of skeletal tissues.

One last issue that we would like to discuss is the relationship between CB and secondary cartilage, an intermediate tissue that arises secondarily on membrane bones and that has previously been reported in some hadrosaur skulls (Bailleul et al., 2012 and Bailleul et al., 2013).

Secondary cartilage and chondroid bone are often found together throughout the skull of mammals and birds (Goret-Nicaise and Dhem, 1982, Hall, 1971, Hall, 1972 and Lengelé et al., 1996b). In our small sample size, we found CB (present study) where dinosaurian secondary cartilage had previously been reported (Bailleul et al., 2013, in the alveoli of MOR 559). We also discovered a new secondary cartilage location while we were looking for CB, on the coronoid process of a post-hatching Hypacrosaurus (slides DE10 and DE13 from project 1988-13, MOR 548, data not shown).

The possibility of the existence of multiple types of CB, as presented in the first part of the discussion, was partly suggested by the fact that CB cell lacunae present two different morphologies: they can have the appearance of hypertrophied cartilage (Fig. 2E) or can appear irregular (e.g., Fig. 4G, I). Even though it has been reported that CB cells sometimes have the appearance of hypertrophied chondrocytes (Mizoguchi et al., 1997 and Tuominen et al., 1996), since we could not perform microradiography on the element present in Fig. 2 (MOR 1038), the possibility that those cells are actually secondary cartilage cells, and not CB cells, cannot be ruled out. Secondary cartilage is usually organized as nodules on the periosteal surface of membrane bones (Hall, 1967 and Hall, 1968), while here, it is present as a long central strut. The absence of resorption suggests that it is not an external secondary cartilage nodule that has been relocated from the periosteal surface to the interior of the bone. Rather, these CB cells could be transforming into cartilage cells via metaplasia (Beresford, 1981). Lengelé et al. (1996b) showed that CB and secondary cartilage were two autonomous tissues, arising independently from the cephalic mesenchymal cells under different biomechanical inductions. However, if indeed there is not one, but many types of CB, perhaps metaplasia between CB and secondary cartilage cells is possible in some cases. Metaplasia, the permanent transformation of cell identity from one cell type to another (Beresford, 1981 and Hall, 2005), is a common mechanism in vertebrate lineages in skeletal and non-skeletal tissues: for e.g., hypertrophied chondrocytes can transform directly into osteoblasts in turtle and lizard long bones (Haines, 1969); fibroblasts can transform into fibrocartilage (e.g., at mammalian tendon or ligament insertions, Gao et al., 1996, Hurov, 1986 and McLean and Bloom, 1940; during antlerogenesis; and at the anterior tip of the rat penile bone and during the fracture repair of bones in some frogs and lizards; see Beresford, 1981 and Hall, 2005 and Hall, 2014 in press for full review). Metaplasia has also been reported in the extant phylogenetic bracket of non-avian dinosaurs: the osteoderms of Alligator mississippiensis arise by in situ transformation of dense irregular connective tissue (perhaps a population of fibroblasts; M. Vickaryous, personal communication) into bone (Vickaryous and Hall, 2008). Tenocytes of avian tendons can transform into osteoblasts (Adams and Organ, 2005 and Agabalyan et al., 2013) and most importantly, secondary cartilage can transform into CB in paralyzed embryonic chicks (Hall, 1972). Moreover, even though it is not possible to know the exact mode of formation of fossilized tissues, due to the unusual histology of some dinosaur mineralized tissues, metaplastic transformations have been hypothesized to play a role in the formation of dinosaur skulls (Goodwin and Horner, 2004;Hieronymus and Witmer, 2008, Horner and Lamm, 2011 and Horner et al., 2015) and osteoderms (Main et al., 2005, Scheyer and Sander, 2004). It appears also that this unusual histology was much more abundant in the skulls of dinosaur than in those of extant mammals and birds (J. Horner personal observations). Therefore, these numerous examples of metaplasia suggest that it is reasonable to hypothesize the transformation of CB cells into secondary cartilage cells in our sample. Of course, this is beyond the scope of this paper, and we are simply considering additional possibilities. More neontological studies on extant archosaurs are necessary for the current and futures advances in the growing field of dinosaur paleohistology.