Gender-Specific Reproductive Tissue in Ratites and Tyrannosaurus rex Author Mary H. Schweitzer Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695, USA. North Carolina State Museum of Natural Sciences, Raleigh, NC 27601, USA. Museum of the Rockies, Montana State University, Bozeman, MT 59717, USA. schweitzer@ncsu.edu Author Jennifer L. Wittmeyer Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695, USA. Author John R. Horner Museum of the Rockies, Montana State University, Bozeman, MT 59717, USA. text Science 2005 308 1456 1460 journal article 10.1126/science.1112158 3a6b4cb7-b97f-443e-9aec-f5be22e0d320 3742896 A relatively small (femur length, 107 cm) Tyrannosaurus rex [ Museum of the Rockies (MOR) specimen number 1125 ] was discovered at the base of the Hell Creek Formation (dated at ~70 million years ago) as an association of disarticulated elements with excellent preservation (1). At death, MOR 1125 was estimated to be 18 T 2 years (2), on the basis of lines of arrested growth (LAG). 1Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695, USA. 2North Carolina State Museum of Natural Sciences, Raleigh, NC 27601, USA. 3Museum of the Rockies, Montana State University, Bozeman, MT 59717, USA. Interior femur fragments from MOR 1125 were reserved without preservatives for chem­ ical and molecular analyses. Gross examination revealed a thin layer of bone tissue lining the inner (medullary) surfaces of the bone fragments that was structurally distinct from other described bone types ( Fig. 1D ) but possessed characteristics in common with avian medullary bone (MB). MB is an ephemeral tissue, deposited on the endosteal surface of avian long bones (3-10). Its formation in female birds is triggered by increasing levels of gonadal hor­ mones produced upon ovulation ( 4 , 10 , 11 ), but it can also be artificially induced in male birds by the administration of estrogen ( 3 , 4 , 12 ). Because MB is densely mineralized and extremely well vascularized, it provides an easily mobilized source of calcium necessary for the production of calcareous eggshells ( 13 ). We compare MB from emu and ostrich ( 14 ) at different stages of the laying cycle with newly identified dinosaur tissues, because these basal birds share more primitive features with nonavian dinosaurs than do extant neognaths ( 15-18 ). Our investigations show that ratite MB differs from that seen in better-studied neognaths. We observed substantial variation between emu and ostrich MB tissues and between both ratites and reported neognath tissues ( Fig. 1 and fig. S1). MB ( Fig. 1 ) may be thick (chicken and ostrich) or quite thin (emu) at midshaft; and it may be separated by a distinct layer of endosteal laminar bone (ELB) as described by Chinsamy et al. (19) (chicken and emu), or not (ostrich). The innermost layer of MB in the ostrich Eadjacent to the endosteal surface of cortical bone (CB)] appears to arise from dense sheets to form tubular structures that parallel the long axis of the bone ( Fig. 1I ). Thin hairlike spicules ( Fig. 1C ) of mineralized bone protrude from the tubes and may be intimately involved in their formation from the basal layer. Mineralized spicules were also noted arising from emu MB (fig. S2), but the tubelike structures were not so apparent or distinct. The MB tissues are morphologically distinct from overlying CB and are similarly distributed in both dinosaur ( Fig. 1D ) and ratite ( Fig. 1 , E and F) samples. Higher magnifications of T. rex ( Fig. 1G ) and ratite ( Fig. 1 , H and I) tissues show the open, crystalline, and fibrous structure of these highly vascular tissues, in contrast to the denser CB. *To whom correspondence should be addressed. E-mail: schweitzer@ncsu.edu Fig. 1. Extant avian MB and homologous dinosaurian bone tissues. (A ) Domestic laying hen, midshaft femur cross section showing the extension of spongy MB deep into the mar­ row cavity and surrounding preexisting trabeculae (T). ( B) Laying emu, midshaft cross section, with a thin layer of MB on the endosteal bone surface, separated from overlying CB by ELB. (C ) Ostrich MB arising from CB. Convoluted bony projections surround large cavities and form by continued deposition on hairlike spicules of calcified bone (S). ( D ) MB on endosteal surface of MOR 1125 femur fragment delineated from overlying CB by large vascular sinuses and change in color, texture, and density. ( E ) Emu and ( F ) ostrich bone taken at same aspect as (D), showing mor­ phological distinction between bone types. (G ) Higher magnification of dinosaur femur fragment in oblique view shows dense CB lined with newly described bone tissue, also seen in oblique view of emu ( H ) and ostrich ( I ) tibia. Ostrich MB is apparently unique in forming longitudinal tubules. In a fresh fracture, dense and relatively homogenous dinosaur CB is distinct from the loosely organized and highly vascular MB internal to it ( Fig. 2A ). A distinct layer corresponding to ELB (19) separates the two bone types. A large erosion room is visible at this boundary, lined with laminar tissue. An emu bone fragment ( Fig. 2B ) in similar orientation shows MB tissues with a distinctive, less organized and Bcrumbly[texture relative to overlying CB. It is interspersed with or laid down between large erosion rooms within the deep cortex and ELB of the tibial shaft. The dense cortex and laminar structure of the ELB are easily distinguished from surrounding MB. The ostrich MB ( Fig. 2C ) differs in both texture and orientation, with open cavities that are bordered by tubelike bone spicules. The lacy vascularity of the T. rex tissues ( Fig. 2D ) is consistent with the larger vascular canals and whorled pattern of the emu ( Fig. 2E ) and especially with the ostrich ( Fig. 2F ), where wide blood-filled sinuses separate forming bone spicules. In ground section ( Fig. 2G ), MOR 1125 femur cortical bone is characterized by well-developed multigeneration Haversian systems with obvious and defined cement lines, supporting a mature status for this dinosaur (2). A region of decreased vascularity and laminar structure marks the ELB. In contrast, the medullary tissues arising from the ELB are densely vascularized but show no evidence of Haversian remodeling or cement lines, indicating that this tissue is newly deposited, or younger bone. A fresh cut section of emu bone ( Fig. 2H ) in comparable orientation shows dense CB and vascular, crystalline, and loosely organized MB, separated by a thin, dense, and less vascular ELB. MB in the ostrich ( Fig. 2I ) is more extensive than in the emu samples, most likely because shelling had not yet begun (14). No distinct ELB is visible. MB appears laminar rather than spiculated in Fig. 2I , because the tubules formed by bony spicules are oriented longitudinally rather than in cross section as in Fig. 2F , but it is the same tissue. Additional pattern similarities are seen in demineralized (14) ratite ( Fig. 3 , B and C) and T. rex ( Fig. 3A ) medullary tissues. In all cases, the matrix is fibrous and randomly organized. The reddish color in extant tissues is due to blood retained in sinuses that separate the bone spicules. The T. rex tissues are similarly pigmented, due either to diagenetic alteration or to close association of bony tissues with blood-producing marrow during the life of the dinosaur. In all MB tissues shown, large vascular sinuses are easily discerned ( Fig. 3 , D to F), but in the T. rex, vascular openings are surrounded by circumferentially oriented matrix fibers ( Fig. 3D ) that are less apparent in extant bone. The ostrich medullary tissues are denser than either the emu ( Fig. 3E ) or T. rex samples, particularly closer to the cortex, but the variation in the size and density of vascular sinuses ( Fig. 3F ) is similar to that seen in the T. rex tissues. In planar view, MOR 1125 undemineralized tissues show a random orientation of fibers, and vascular openings penetrate deep into the tissues ( Fig. 3G , inset) and exhibit an unusual doublet or triplet pattern, where multiple vessels penetrate an osteonlike core (arrows), also seen in the emu ( Fig. 3 H , arrows). The ostrich medullary tissues ( Fig. 3I ) are more variable, denser, and less random in appearance than those of the emu, but the morphology changes as the tissues extend into the medullary cavity. Close to the cortex ( Fig. 3I , inset, and fig. S3), the bone is sheetlike, relatively dense, and punctured by vascular sinuses exhibiting the doublet pattern (arrows) noted above. As tissues extend into the medullary cavity ( Fig. 3I ), this pattern becomes obscured. Inset bone has been stained (14) for better contrast. Fig. 2. Dinosaur and ratite comparative views. ( A ) Freshly broken fragment of MOR 1125 shows laminar ELB separating CB and MB. Bone tissues decrease in density internal to the ELB, because of increased vascularity. ( B ) Emu tibia, midshaft section. Erosion rooms extending into ELB are secondarily filled by MB. (C ) Ostrich bone, mid­ shaft. MB is distinct from CB, but no obvious ELB is visible and several large vascular sinuses are seen. ( D) Higher magnification of MB region of MOR 1125 , showing increased porosity and more random orientation of MB than CB or ELB. ( E ) Emu, stained (14) to distinguish bone from infiltrating marrow fat. MB is more vascular than overlying CB and exhibits a random, whorled pattern. ( F) Ostrich MB, showing relationship of bony spicules to invading blood sinuses, colored red from remnant blood. ( G ) Ground section of MOR 1125 . Dense cortical Haversian bone shows second- and third-generation remodeling. ELB separates Haversian bone from more vascular MB. ( H ) Similar orientation of emu femur shows dense CB, distinct ELB, and a thin layer of MB. ( I ) Ostrich MB appears more laminar than in (C) or (F) because of the longitudinal orientation of tubelike medullary spicules. In regions of MOR 1125 bone where most of the medullary tissues have eroded ( Fig. 3J ), patches of denser CB can be seen, emphasizing the random mazelike pattern and large vascular sinuses of medullary tissues, a pattern also seen in the emu bone ( Fig. 3K ). The ostrich MB shows a similar pattern of bony spicules surrounding large and small blood sinuses ( Fig. 3L ). Fig. 3 . Dinosaur and ratite MB. (A) MOR 1125 , ( B ) emu, and ( C ) ostrich demineralized (14) MB. The coloration in (B) and (C) results from infiltration of tissues with blood sinuses. ( D ) MOR 1125 , partially demineralized, showing enlarged, randomly arranged vascular openings surrounded by circumferential matrix fibers. Partially demineralized ( E ) emu and ( F ) ostrich medullary tissues show extensive vascular penetration, with randomly spaced and varyingly sized vessel openings. ( G ) Plane view of undemineralized dinosaur tissues shows fibrous matrix and an unusual pattern of vascular doublets or triplets within osteonlike structures (arrows). The inset shows variation in depth and diameter of vascular sinuses. (H) Undemineralized emu MB shows similar doublet pattern (arrows) and fibrous matrix. The greater depth of field makes focusing difficult. ( I ) Ostrich MB is denser closer to the cortex (inset), where the doublet/triplet pattern of vessels (arrows) is evident, but this becomes obscured by the increasing development of bony tubes and spicules as bone extends further into the medullary cavity. (J) Plane view of MOR 1125 shows the partially eroded, fibrous MB distributed across the cortex in a mazelike fashion. (K) Emu MB shows white (chloroform-altered) and cream-colored MB in the same mazelike pattern. ( L ) Thicker, randomly oriented tubular spicules of ostrich MB, showing deep penetration and intimate association of blood-containing sinuses. Scanning electron micrographs (14) reveal the distinctive grainy texture and disorganized morphology of demineralized T. rex and avian MB ( Fig. 4 ). This contrasts with the smooth and fibrous texture of demineralized CB from the same specimens ( Fig. 4, E to H ). Higher magnifications of demineralized CB ( Fig. 4, I to K ) emphasize the smooth, fibrous, and more ordered nature of all specimens, although in MOR 1125 ( Fig. 4I ), degradation is apparent. MB occurs naturally only in extant female birds, although it varies in amount and distribution among taxa and with ovulatory phase (5, 20). It is chemically, functionally, and structurally distinct from both overlying CB and internal trabecular bone (21, 22). Although "medullary" and "trabecular" bone are terms often used interchangeably in the literature, MB has a larger surface area and is more vascular than other bone types, allowing rapid calcium mobilization (5). It is more highly mineralized, with a greater apatite-tocollagen ratio (5, 7, 20-22), and incorporates acidic mucopolysaccharides and glycosaminoglycans that are not present in CB ( 5 , 11 ). Additionally, the matrix of MB is higher in noncollagenous proteins and lower in collagen, and has a higher collagen III-to-collagen I ratio ( 22 ) relative to other bone types. If preservation allows, these characteristics will be used as part of ongoing research to chemically distinguish the two bone types in this dinosaur. The existence of avian-type MB in dinosaurs has been hypothesized ( 9 , 23 ) but not identified. In part, this could be because of taphonomic bias, because the death and fossilization of an ovulating dinosaur would be comparatively rare. Additionally, MB in extant birds is fragile, the spicules separating easily from the originating layer (fig. S1). Dinosaur MB may separate and be lost from overlying CB in a similar manner during diagenesis. The location, origin, morphology, and microstructure of the new T. rex tissues support homology with ratite MB. The T. rex tissues line the medullary cavities of both femora of MOR 1125 , suggesting an organismal response. The tissues are similar in distribution to those of extant ratites, being more extensive in proximal regions of the bone. They are clearly endosteal in origin, and the microstructure with large vascular sinuses is consistent with the function of MB as a rapidly deposited and easily mobilized calcium source. The random, woven character indicates rapidly deposited, younger bone. Finally, the robustly supported relationship between theropods and extant birds ( 15-18 , 24 , 25 ) permits the application of phylogenetic inference to support the identification of these tissues ( 26 , 27 ). The morphology of these dinosaur tissues is not identical to that of extant neognaths (fig. S1), but is more similar to that seen in ratites. T. rex medullary tissues are less extensive than those reported for neognaths, which may be explained by many factors. First, there is a wide range of MB morphologies in extant taxa (20), varying with both reproductive phase and the position of the egg within the reproductive tract (6). Medullary tissues become thinner as shelling progresses and disappear completely with deposition of the last egg. If the same was true of dinosaurs, MOR 1125 may have died toward the end of the laying cycle. Second, MB in extant birds is hypothesized to provide a buffer against excessive and debilitating bone resorption during shelling (4, 28). It is most extensive in smaller taxa with high reproductive rates, because of the demand for rapid mobilization of skeletal calcium for shelling. Extinct theropods produced hardshelled eggs, as did other dinosaurs and all extant birds (29, 30). Although egg size is not known, eggs were most likely smaller relative to overall body size than in extant birds, resulting in less demand for bone calcium reserves and reducing the need to offset resorption. These factors may also contribute to the smaller ratio of medullary to cortical thickness in theropods than in extant birds. Finally, T. rex , although phylogenetically close to extant birds (15-18, 24, 25), was distinct in size, biomechanical constraints, and, to some degree, physiology ( 31 ); therefore, slight variations in bone and tissue types would be expected. Fig . 4. Scanning electron microscope images of demineralized MB [(A) to (D)] and CB [(E) to (K)]. Demineralized, aldehyde-fixed (14) MB tissues from (A) MOR 1125 , ( B) extant laying hen, ( C) emu, and ( D ) ostrich show random, crumbly texture. Organized collagen fiber bundles are not distinct in any sample because of rapid deposition and woven character. Scale bars for (A) and (B), 40 um; for (C) and (D), 20 um. Demineralized fragments of cortical bone from ( E ) MOR 1125 , ( F ) chicken, ( G ) emu, and (H) ostrich are shown. A fibrous character dominates all samples. Scale bars for (E), (F), and (H), 30 m m; for (G), 10 m m. Higher magnification of demineralized CB from ( I ) MOR 1125 , (J) emu, and (K) ostrich CB demonstrates the structural similarity between samples, although the MOR 1125 matrix is highly degraded. Scale bars for (I) and (K), 6 um ; for (J), 5 um . MB most likely first evolved within the lineage in early, small theropods with high productivity. A relatively thicker tyrannosaur bone cortex would reduce the need for MB, and its presence in MOR 1125 may reflect the retention of a primitive trait. This hypothesis may be tested by examination of the limb bones of the recently reported oviraptor containing eggs in the reproductive tract ( 32 ). The existence of MB in crocodiles has been referred to anecdotally ( 3 , 21 ), but although they do resorb CB during shelling, experimental evidence suggests that they do not form MB ( 6 , 9 , 11 , 12 ), even after stimulation with estrogen ( 33 ). The identification of medullary tissues in dinosaurs supports a closer relationship to birds than to other extant archosaurs, sheds light on reproductive strategies of nonavian theropods, and provides an objective means of gender determination in extinct dinosaurs.