Comparative bone microstructure of modern birds has been well researched (Enlow and Brown, 1956), and in recent years, our knowledge of their bone depositional rates and growth has increased substantially (Chinsamy, 1995; Castanet et al., 1996; de Margerie et al., 2002). Likewise, knowledge of the histology of fossilized bones of nonavian dinosaurs (Seitz, 1907; de Ricqlès, 1980; Reid, 1983, 1986; Chinsamy-Turan, 2005) and Mesozoic birds (Houde, 1987; Chinsamy et al., 1994; Chinsamy and Elzanowski, 2001; Chinsamy, 2002; de Ricqlès et al., 2003) has grown considerably. Today, we have a fairly good understanding of the bone tissues present among the Dinosauria, and there is much discussion about the paleobiological significance of the bone tissues preserved (Erickson and Tumanova, 2000; Chinsamy and Hillenius, 2004; Chinsamy-Turan, 2005; Erickson, 2005). In recent years, there have been reports of two unusual bone tissue types in dinosaurs: radial fibrolamellar bone tissue and medullary bone tissue.
Radial fibrolamellar bone tissue is characterized by a large number of radially oriented channels that run perpendicularly to the circumference of the bone wall and is considered to be rapidly formed (Erickson and Tumanova, 2000; Klein, 2004). Radial fibrolamellar bone tissue was first reported in the Cretaceous ornithischian Psittacosaurus mongoliensis (Erickson and Tumanova, 2000: Figs. 1C, 2C). This tissue was observed in late stages of Psittacosaurus ontogeny and appeared to be restricted to the outermost compacta (in two femora and a tibia) of three of the largest individuals (Erickson and Tumanova, 2000). Subsequently, Klein (2004) and Klein and Sander (2007) reported the occurrence of radial fibrolamellar bone tissue in several bones (tibia, femur, vertebra, and ischium) from mature individuals of the Late Triassic prosauropod Plateosaurus engelhardti. Radial fibrolamellar bone tissue was also reported in a limb bone of a late Triassic dinosaur from Norway (Hurum et al., 2006). A common and perplexing feature of radial fibrolamellar bone was the localized formation of this rapidly formed tissue in mature individuals, rather than in young individuals where rapid growth and bone deposition are expected (Erickson and Tumanova, 2000; Klein, 2004).
Unusual endosteal bone tissues have been described from the medullary cavities of three dinosaurs: a femur of Tyrannosaurus rex (Schweitzer et al., 2005), a tibia of Allosaurus, and a femur and tibia of Tenontosaurus (Lee and Werning, 2008). Although recognizing that the unusual endosteally formed bone observed within the medullary cavities of these dinosaurs is morphologically dissimilar to modern avian medullary bone, and also differs among the three taxa, the researchers interpreted these endosteally derived bone tissues as homologous to avian medullary bone, and suggested that all three dinosaurs were reproductively active females (Schweitzer et al., 2005; Lee and Werning, 2008).
Recently, a fortuitous observation was made of associated periosteal and endosteal pathologies in the bone of a modern turkey vulture (Cathartes aura), which showed a striking similarity to features present in a Late Cretaceous Transylvanian dinosaur long bone from the Baron Nopsca collection of the British Museum. Dinosaur bone pathologies are well recognized in the literature (Moodie, 1923; Rothschild and Martin, 1993). Initially, the Nopsca specimen was described by Swinton (1934) and exhibited in the Dinosaur Gallery of the British Museum as an example of a periosteal sarcoma. However, Campbell (1966) re-evaluated this section and suggested that the “lesions” present resembled those formed by osteopetrosis in birds.
Here, we describe the periosteal and endosteal bone tissues in the modern vulture and the Cretaceous nonavian Transylvanian dinosaur, and their implications for the identification and interpretation of radial fibrolamellar bone tissues and avian style medullary bone in nonavian dinosaurs.
The similarity of the radial “sunburst” pattern of pathological bony growths (Lehrer et al., 1970) on the peripheral margins of the turkey vulture and the dinosaur bone is striking. Periosteal outgrowths perpendicular to cortical bone are well established as a pathological characteristic (Lehrer et al., 1970; Lenehan and Fetter, 1985; Lorigan et al., 1989). Curiously, several observations of radially formed fibrolamellar bone have been reported in various subadult/adult dinosaurs (Psittacosaurus, Erickson and Tumanova, 2000; Plateosaurus, Klein, 2004; Klein and Sander, 2007; Allosaurus, Bybee et al., 2006; and a Norwegian Triassic dinosaur, Hurum et al., 2006) and have been variously suggested to indicate rapid growth spurts, high depositional rates, and osseous drift. Given that these dinosaurs are generally already mature individuals (Klein, 2004), and that the tissue is locally restrictive within the individuals and bones studied (i.e., located in the outermost part of the cortex, and when present in long bones, unevenly distributed along shafts), it seems more likely that these tissues reflect compacted periosteal reactive bone tissue produced as a result of disease.
Lenehan and Fetter (1985) mention that in systemic diseases such as hypertrophic osteopathy diffuse periosteal reactive bone results, and that lamellar deposits can later occur to form a more compacted tissue. Lenehan and Fetter (1985) also found that in these cases with periosteal reactive bone caused by hypertrophic osteopathy, no endosteal bone deposits were present in the medullary cavities of the affected bones. These features seem to correspond with those observed for the nonavian dinosaurs in which radial fibrolamellar bone has been observed, but in which no associated endosteal deposits were reported (Erickson and Tumanova, 2000; Klein, 2004; Hurum et al., 2006; Klein and Sander, 2007).
The turkey vulture and Transylvanian dinosaur bones both have endosteal deposits along the margins of the medullary cavity. Such endosteal deposits of bone seem to be a feature that accompanies the radial “sunburst” pattern of bony growths associated with avian osteopetrosis (Bell and Campbell, 1961; Campbell, 1966: Fig. 7). Admittedly, there is some difference in the actual morphology of the bone tissues deposited in the medullary cavity of the turkey vulture and the nonavian dinosaur; notably, the lack of a predominant radial spicule orientation in the latter, but in both cases the tissues are clearly endosteally derived and have accompanying periosteal reactive bone growths.
It is well known that reproductively active female birds deposit “medullary bone” within the medullary cavity, that is, a specialized bone tissue that is well vascularized and better mineralized than cortical bone (Simkiss, 1967). Medullary bone can be metabolized at 10–15 times the rate of cortical bone (Simkiss, 1967) to cope with the high calcium demands of eggshell production (Miller and Bowman, 1981; Wilson and Thorp, 1998). Given that it is well documented that there is substantial variation in the morphology of medullary bone among extant birds (Schweitzer et al., 2005), the presence of the endosteal bone within the medullary cavity in the adult female turkey vulture could have been easily interpreted as medullary bone. Indeed, the tissue meets all three of the criteria used by Lee and Werning (2008, Supplementary Information) to determine medullary bone homology: the spicules form a woven network, it is internal to endosteal lamellae, and is of endosteal origin. However, dissection of DMNH 82686 revealed that the vulture was not reproductively active at the time of death. The endosteal bone tissue in this individual is unrelated to ovulation or egg shelling, and is the result of a pathological condition. The similar association of the endosteal bone in the medullary cavity of the Transylvanian dinosaur with periosteal reactive tissue suggests that in this case, as well, the endosteal deposition is also likely caused by a pathology (e.g., osteopetrosis), rather than being homologous to avian medullary bone.
Given our findings here that endosteally formed pathological bone with similar morphology to medullary bone can also be formed within the medullary cavity, caution is warranted in interpreting all unusual endosteally derived tissue in dinosaurs as homologous to avian medullary bone. Interestingly, one of the three dinosaur bones with purported medullary bone (Allosaurus tibia, UUVP 5300) also shows evidence of a periosteal pathology, similar in morphology to that observed in both the turkey vulture and the nonavian dinosaur described herein (Bybee et al., 2006: Fig. 1F,H; Lee and Werning, 2008, Supplementary Information: Fig. 8B), raising the question that, at least in this specimen of Allosaurus, the unusual endosteal tissues might have a pathological origin. Indeed, the microstructure of the endosteally derived tissue in the Allosaurus tibia (Lee and Werning, 2008, Supplementary Information: Fig. 4B) resembles the long radially oriented bony spicules present in the turkey vulture.
Lee and Werning (2008) also reported medullary bone in a femur and tibia of Tenontosaurus, and although such distribution in long bones is well known for avian medullary bone (Dacke et al., 1993), it should be noted that osteopetrosis is also known to affect multiple long bones in the skeleton of modern birds (Bell and Campbell, 1961). No obvious pathological features were observed periosteally in either the Tenontosaurus specimens (Lee and Werning, 2008) or the T. rex specimen (Schweitzer et al., 2005) in which medullary bone was reported. However, it is worth noting that some birds with osteopetrosis are known to have had negligible periosteal reactions (such that they are described as normal from superficial examination) but upon radiography, cortical and medullary osteopetrosis reactions are observed (Holmes, 1958).
In their study of osteopetrosis in fowls, Bell and Campbell (1961) mention specifically that the periosteal osteogenesis occurs in layers suggesting periodic cycles of bone deposition. This seems to be similar to the periosteal bone deposition of radial fibrolamellar bone reported in Psittacosaurus (Erickson and Tumanova, 2000) and in Plateosaurus (Klein, 2004; Klein and Sander, 2007).
It is also worth noting that of the published examples of purported medullary bone in nonavian dinosaurs, none of the three individuals was directly associated with nests, eggs, or eggshell fragments that can be attributed to the same animal. [As noted by Lee and Werning (2008), a single egg is known from the same locality as the Allosaurus specimen, but the egg possesses unique and pathological shell morphology and cannot be linked to Allosaurus (Hirsch et al., 1989).] Although absence of association with eggs and nests does not falsify the authors' hypotheses of medullary bone in these individuals, it raises another interesting question. There are relatively few examples of dinosaurs directly associated with eggs or egg clutches (Dong and Currie, 1995; Norell et al., 1995; Varricchio et al., 1997; Clark et al., 1999). The most famous of these are the specimens of the Mongolian oviraptorid Citipati that were found directly on top of nests (Norell et al., 1995; Clark et al., 1999). If nonavian dinosaurs formed medullary bone during egg laying, it might be expected that these nesting dinosaurs would have this characteristic tissue. Curiously, however, the long bones of four taxa of dinosaurs directly associated with eggs or nests that have been studied histologically [femora and fibulae of two Citipati osmolskae and a fibula of a troodontid (Erickson et al., 2007; Erickson and Norell, personal communication), a femur and tibia of Troodon formosus (Erickson et al., 2007; Varricchio et al., 2008)], do not show any endosteally derived tissues suggestive of medullary bone. The absence of medullary bone in all examined long bone elements of dinosaurs associated with nests or eggs could be interpreted in several ways: First, as hypothesized by Varricchio et al. (2008), it is possible that the nesting dinosaurs are not females, but rather brooding males, in which case medullary bone would not be expected to be present. Second, if the nesting dinosaurs are females, it is entirely possible that medullary bone had originally been present, but was resorbed soon after the eggs were laid, and therefore was not present at the time of death.
The persistence of medullary bone post egg laying in extant birds has not been comprehensively studied, and isolated reports are available only for a few species. In chickens (Gallus gallus), medullary bone develops ∼1–2 weeks before the first egg is laid, and then it is gradually resorbed over a period of 1–3 weeks after the last egg is laid (Simkiss, 1967; Rick, 1975; Lentacker and van Neer, 1996). In pigeons (Columba livia), medullary bone is totally resorbed about 10–20 days after the last egg is laid (Simkiss, 1967). Complete resorption of medullary bone has also been noted in C. livia during the shelling process of one egg, with repeated formation before the next egg is shelled (Bloom et al., 1958; Dacke et al., 1993). In house sparrows (Passer domesticus), medullary bone is only present when the ovary is near maximum activity or when eggs are in the oviduct (Kirschbaum et al., 1939; Pfeiffer et al., 1940). In Canada geese (Branta canadensis), medullary bone has been reported during the last 6–7 days of rapid ovarian development, with rapid depletion after the clutch is laid, such that medullary bone is absent at the onset of the incubation period (Raveling et al., 1978; Gotfredsen, 2002). Even within the same species, there is individual variation in terms of the amount of medullary bone produced, and amounts of medullary bone deposited may also fluctuate within a single individual during the egg-laying period. There is some suggestion that these latter fluctuations may be more extreme in species with longer gaps between the laying of successive eggs in a clutch (Rick, 1975; Cohen and Serjeantson, 1996). Thus, the lack of medullary bone in the brooding nonavian dinosaurs could simply indicate that complete resorption occurs by the time the eggs are laid, similar to the pattern in extant Branta canadensis.
The third possibility is that perhaps avian style medullary bone was not a feature of reproductively active female dinosaurs; at least not in all taxa. The question then becomes one of how would female dinosaurs that did not form medullary bone cope with the high calcium demands of egg-laying? Interestingly, it appears that under special circumstances such as a calcium-deficient diet, chickens can resorb up to 40%–50% of their cortical bone (Bloom et al., 1958; Simkiss, 1967; Rick, 1975; Lentacker and van Neer, 1996), leaving only enough of the bone wall to be biomechanically sound. The calcium mobilized from the cortical bone is not directly used in the shelling of the eggs, as is the case in crocodilians (Wink and Elsey, 1986; Wink et al., 1987; Schweitzer et al., 2007), but rather is used to deposit medullary bone, which is then resorbed for eggshell calcium (Bloom et al., 1958; Simkiss, 1967; Rick, 1975; Lentacker and van Neer, 1996). These findings suggest that reproductive osteoporosis is likely plesiomorphic for archosaurs. The question remains though, where among the evolutionary history of birds and/or nonavian dinosaurs did avian-style medullary bone originate?
Medullary bone may be an adaptation to compensate for relatively thin bone walls (Miller and Bowman, 1981; Wilson and Thorp, 1998), because extensive erosion of thin-walled compacta would seriously compromise biomechanical strength. Given the close phylogenetic relationship between birds and nonavian dinosaurs, it is reasonable to hypothesize that the origin of avian-style medullary bone occurred in the more derived maniraptoran dinosaurs close to the transition from nonavian dinosaurs to birds. However, given the fact that the dinosaurs so far purported to exhibit medullary bone (two nonmaniraptoran theropods and an ornithischian) (Schweitzer et al., 2005; Lee and Werning, 2008) are large animals with thick compact bone walls, it seems unlikely that they would have needed to deposit a specialized bone tissue to prevent excessive osteoporosis because of reproductive calcium demands.
Ovulating female crocodilians obtain the necessary calcium by direct resorption from their thick-walled bones (Wink and Elsey, 1986; Wink et al., 1987; Schweitzer et al., 2007). No tissue analogous to avian medullary bone has been observed in modern crocodilians (Elsey and Wink, 1986; Schweitzer et al., 2007, and references therein). We suggest that it is quite likely that nonavian dinosaurs with relatively thick bone walls (especially those that were not small derived maniraptoran theropods on the evolutionary line to birds) would have been able to draw calcium directly from their skeletons, as crocodilians do, and may not have needed to form a special store of calcium specifically for the demands of egg laying.
In conclusion, among a variety of modern animals, periosteal reactive bone appears to be the result of a pathologic disease process, such as hypertrophic osteopathy or osteopetrosis. In the latter case, endosteal reactions accompany the periosteal features, whereas in the former reactive growth appears to be restricted to the periosteal surface. On the basis of origin, location, and morphology, it appears that the radial fibrolamellar bone tissues identified in nonavian dinosaurs (Erickson and Tumanova, 2000; Klein, 2004; Hurum et al, 2006; Klein and Sander, 2007) could also have resulted from a disease such as hypertrophic osteopathy. Moreover, our observations of pathological endosteal deposits within the medullary cavity associated with periosteal pathologies in a turkey vulture and a nonavian dinosaur are similar to the periosteal and endosteal bone tissues reported in Allosaurus (UUVP 5300) (Bybee et al., 2006; Lee and Werning, 2008). In addition, the fact that endosteal reactive growth can result from osteopetrosis with virtually no associated periosteal deposits (Holmes, 1958), suggests that caution is needed in the interpretation of all endosteal bone deposits in nonavian dinosaurs as homologous to that of avian medullary bone, particularly when they occur in large-bodied taxa phylogenetically distinct from the maniraptoran clade.