Norntates PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, NY 10024 Number 3265, 36 Pp., 15 Figures May 4, 1999

Home , Bird, Hesperornis, Ichthyornis, Oviraptor, Oviraptoridae, Sinornithoides

AMERICANt MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3265, 36 pp., 15 figures May 4, 1999

An Oviraptorid Skeleton from the Late Cretaceous of Ukhaa Tolgod, Mongolia, Preserved in an Avianlike Brooding Position Over an Oviraptorid Nest

JAMES M. CLARK,I MARK A. NORELL,2 AND LUIS M. CHIAPPE3

ABSTRACT The articulated postcranial skeleton of an ovi- presence of a single, ossified ventral segment in raptorid dinosaur (Theropoda, Coelurosauria) each rib as well as ossified uncinate processes from the Late Cretaceous Djadokhta Formation associated with the thoracic ribs. Remnants of of Ukhaa Tolgod, Mongolia, is preserved over- keratinous sheaths are preserved with four of the lying a nest. The eggs are similar in size, shape, manal claws, and the bony and keratinous claws and ornamentation to another egg from this lo- were as strongly curved as the manal claws of cality in which an oviraptorid embryo is pre- Archaeopteryx and the pedal claws of modern served, suggesting that the nest is of the same climbing birds. The skeleton is positioned over species as the adult skeleton overlying it and was the center of the nest, with its limbs arranged parented by the adult. The lack of a skull pre- symmetrically on either side and its arms spread cludes specific identification, but in several fea- out around the nest perimeter. This is one of four tures the specimen is more similar to Oviraptor known oviraptorid skeletons preserved on nests than to other oviraptorids. The ventral part of the of this type of egg, comprising 23.5% of the 17 thorax is exceptionally well preserved and pro- oviraptorid skeletons collected from the Dja- vides evidence for other avian features that were dokhta Formation before 1996. The lack of dis- previously unreported in oviraptorids, including turbance to the nest and skeleton indicate that the the articulation of the first three thoracic ribs specimen is preserved in the position in which with the costal margin of the sternum and the the adult died. Its posture is the same as that

' Research Associate, Department of Vertebrate Paleontology, American Museum of Natural History; Assistant Professor, Department of Biological Sciences, George Washington University, Washington, D.C. 20052. 2 Chairman, Department of Vertebrate Paleontology, American Museum of Natural History. 3Chapman Fellow and Research Associate, Department of Ornithology, American Museum of Natural History.

Copyright ( American Museum of Natural History 1999 ISSN 0003-0082 / Price $3.90 2 AMERICAN MUSEUM NOVITATES NO. 3265 commonly taken only by birds among tetrapods indicating that the behavior of sitting on open that brood their nest, and its close proximity to nests in this posture evolved before the most re- the eggs indicates that the nest was not covered, cent common ancestor of modern birds.

INTRODUCTION Among the most surprising and revealing but only circumstantial evidence for reiden- specimens collected from Upper Cretaceous tifying the eggs was forwarded. The discov- Djadokhta-like beds in Mongolia by the joint ery of Ukhaa Tolgod in 1993 provided the American Museum of Natural History-Mon- evidence crucial to deducing a definitive an- golian Academy of Sciences expeditions are swer-an oviraptorid embryo within the those of the peculiar dinosaurs of the Ovi- same type of egg as those beneath the Ovi- raptoridae (Norell and Clark, 1997). Unlike raptor philoceratops holotype (Norell et al., the skulls of most other nonavian theropods, 1994). With this reidentification of the eggs the unusually short, often highly pneuma- the association of the skeleton with eggs of tized skull of oviraptorids lacks teeth and its own kind became explicable as evidence may bear a crest, resembling superficially the of parental care, rather than predation. skull of the living cassowary. The abundance The specimen described here, IGM 100/ of oviraptorid specimens at the extraordi- 979 (mistakenly labeled 100/972 in fig. 1 of narily rich locality of Ukhaa Tolgod (Dash- Norell et al., 1995), provides further evi- zeveg et al., 1995) is most unexpected in dence that this association is indeed a result light of their rarity in other deposits (Bars- of parental behavior. The specimen (fig. 1) bold et al., 1990). This wealth of new ma- was discovered in 1993, and after its prepa- terial already has answered a 75 year old ration a preliminary note was published (No- enigma, and the specimen described here rell et al., 1995). A second specimen on a played a crucial role in this story (Norell et nest was collected from Ukhaa Tolgod in al., 1994). 1995 (see Webster, 1996), and another from The bizarre and poorly preserved holotype correlative beds at Bayan Mandahu in Inner specimen of Oviraptor philoceratops was Mongolia, China, was reported by Dong and discovered at Bayn Dzak (the "Flaming Currie (1996). These three specimens and the Cliffs") by the American Museum's Central 0. philoceratops holotype provide compel- Asiatic Expedition in 1923, fortuitously en- ling evidence that the close association be- countered while excavating the first dinosaur tween adults and nests in Oviraptoridae has nest found at this famous locality (Andrews, a biological explanation. 1932). Because eggs similar to those in this The specimen is remarkably intact, and in nest are abundant at Bayn Dzak, and because addition to its importance in preserving this of the preponderance of skeletal material of individual's relationship to the nest it pro- Protoceratops andrewsi at this locality, the vides important new data on the thoracic eggs were identified as belonging to Proto- skeleton in oviraptorids. Many of these fea- ceratops (Osborn, 1924). The skeleton over- tures were revealed by preparation following lying the nest therefore presented an enigma. publication of the preliminary note, and thus To explain this association, Henry Fairfield were not reported or illustrated there. Ovi- Osborn (Osborn, 1924) speculated that the raptorids are among the closest relatives of adult animal had been preserved while rob- Avialae, the group comprising Archaeopter- bing the nest (although there was no evi- yx, extant birds, and related taxa (Aves of dence to indicate predation), hence he en- some other authors: see Gauthier, 1986), and dowed it with a name meaning "egg seizer these new features help determine the precise fond of ceratopsians." relationships of oviraptorids among Thero- For many years the identity of the eggs as poda and the homology of these features. those of Protoceratops was accepted with little Institutional abbreviations: AMNH- doubt. In recent years this identification was American Museum of Natural History, New questioned (Sabath, 1991; Mikhailov, 1991), York; IGM-Mongolian Institute of Geolo- 19991999~CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA33 gy, Ulaan Baatar; IVPP-Institute of Verte- Ukhaa Tolgod, traces of burrowing invertebrate Paleontology and Paleoanthropology, brates are preserved with the skeleton. They Beijing; MIAE-field numbers of the collec- are recognizable mainly because they are tions made by the Mongolian Academy of more strongly cemented than the surrounding Sciences-American Museum of Natural His- matrix and are lighter in color. Five burrows tory expeditions to the Gobi Desert. were exposed during preparation, including a long burrow on the dorsal surface of the ster- GEOLOGIC SETTING OF THE FIND nal plates (fig. 2). These burrows are filled with white particles which may be pieces of The specimen is from a thick-bedded digested bone, and two in the shoulder region sandstone in the lower part of the local sec- are nearly white. Four of the five are slightly tion at the Ankylosaur Flats sublocality, curved, but the smallest one is straight. One Ukhaa Tolgod. These sediments and the fau- is preserved above the right coracoid, one to na they entomb closely resemble sediments the left of the fourth digit of the left pes, and and assemblages at localities in the Djadokh- another between the right ischium and right ta Formation, such as Bayn Dzak and Bayan pes, and one was removed from above a right Mandahu, but some similarities with the fau- ventral rib just behind the right sternal plate. na of the Barun Goyot Formation have also Most are similar in shape, ranging from 3-4 been noted (Dashzeveg et al., 1995; Gao and cm in length and 4-8 mm in diameter, but Norell, 1996). Currently the stratigraphic re- the one above the sternum is over 6 cm long lationships of this locality relative to other (fig. 2). Although other evidence of contem- localities, especially the type section of the poraneous fossil arthropods at Ukhaa Tolgod Barun Goyot Formation, are unclear and un- (Loope et al., 1998) and other Djadokhta lo- der study (L. Dingus et al., in prep.). calities (see Johnston et al., 1996) may be At Ukhaa Tolgod, the specimen was col- from the taxon responsible for the burrows, lected from a unit with a lithology that has it is also possible that burrows associated been described elsewhere as a "structureless with IGM 100/979 were formed significantly sandstone" and interpreted as eolian in origin after deposition of the bed in which it is bur- (e.g., Eberth, 1993). In the Djadokhta For- ied. mation at Ukhaa Tolgod this facies initially The age of the Djadokhta Formation at was interpreted as eolian in origin (Dashzev- Ukhaa Tolgod and elsewhere in Mongolia eg et al., 1995). Recently, Loope et al. (1998) and China currently is not constrained by ra- reinterpreted these deposits as representing diometric dates, marine invertebrates, or pa- low-energy debris flows from eolian deposits leomagnetic polarity patterns. Comparisons that formed thick alluvial fans. The source of of the vertebrates suggest a correlation with these deposits was an extensive field of sta- the late Campanian to early Maastrichtian bilized sand dunes that destabilized when in- marine invertebrate stages (Lillegraven and undated with water during large rainstorms. McKenna, 1986; Jerzykiewicz and Russell, They interpreted the ancient Ukhaa Tolgod 1991). However, the Darabasa Formation of environment as dominated by large, stabi- Kazakhstan has a mammalian fauna similar lized sand dunes separated by interdunal en- to that of the Djadokhta Formation and is vironments such as small, ephemeral ponds. interbedded with Early Campanian marine Loope et al. (1998) suggested that sand invertebrates (Averianov, 1997). flowed at relatively low energy from the dunes into the interdunal zone following DESCRIPTION large rainstorms. The low-energy flows cov- THE SKELETON ered dead, or dying, animals and may have entombed smaller animals in burrows. Low- The skeleton (fig. 1) is incompletely ex- energy depositional conditions may therefore posed, as further preparation would compro- be responsible for the excellent preservation mise the integrity of the specimen or endan- of the Ukhaa Tolgod fossils, including IGM ger bones by removing their support. Miss- 100/979. ing are nearly the entire vertebral column, As with many of the specimens from the skull, and the ilia and left femur. The 4 AMERICAN MUSEUM NOVITATES NO. 3265

Fig. 1. IGM 100/979, in dorsal view. See Appendix for abbreviations. exposed portions comprise a complete right gastralia and the distal ends of seven pairs of forelimb; a partial left forelimb including the ribs. Most of the bones are uncrushed except proximal end of the humerus, the distal ends the right radius, ulna, femur, and tibia. Rem- of the radius and ulna, and all the bones of nants of the keratinous claws of the manus the manus; the nearly complete furcula; the are preserved on the ends of the second and ventral portion of both scapulae fused with third unguals on the right side and the first the coracoids; the left and right sternal plates; and third ungual on the left side. what appears to be the ventral part of a ver- PECTORAL GIRDLE AND FORELIMB: The fur- tebral centrum in the shoulder region; the cula (fig. 3) is missing the distal half of its distal ends of the ischia and pubes; the distal left side and a small part of the distal tip of end of the right femur and the proximal half its right side, which is otherwise preserved of the tibia and fibula in articulation; the in articulation with the right scapula. It is right pes distal to the middle of the metatar- completely fused, with no indication of a su- sals (the second and third digits are not ex- ture between the two clavicles. It is a robust posed); the left fibula, tarsus, and pes in ar- bone that curves posterodorsally at its distal ticulation, and what may be a fragment of end (as preserved on the right side). The bro- the proximal end of the tibia; and a series of ken edge of the left side of the bone reveals 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 5

Imd

Imd 1I rmd I

Imd III

rmd 11

Fig. l.-Continued. no indications of a pneumatic space within shelf is nearly horizontal and dorsally con- it, as is found in the furcula of many birds. cave, with a well-developed semicircular rim The distal end flattens where it articulates anteriorly. with a distinct shelf on the anterior edge of The right coracoid is partly exposed and the scapula. An unusually long, dorsoven- the left is almost completely exposed in me- trally flat hypocleidium extends posteroven- dial view. The exposed portion is similar in trally from the midline of the furcula, direct- size and shape to that of other oviraptorosau- ly toward the sternum (fig. 3). It is nearly rians (Barsbold et al., 1990), which is more half as long as is each half of the furcula. It elongate ventrally than in basal theropods, tapers distally from a broad base, and is forming a quadrangular shape in medial and about twice as long as the base is wide. lateral views. The articulation with the ster- Remnants of the scapulae are poorly pre- num is exposed only on the left side, where served. A well-developed shelf is preserved the coracoid has been displaced slightly an- on the anterior edge of both elements dorsal terodorsally. The coracoid contacts the ster- to the glenoid fossa, and the furcula articu- num along the anterior edge near the midline. lates with this shelf on the right side. The The humerus (fig. 4) has a well developed AMERICAN MUSEUM NOVITATES NO. 3265

could also be accentuated by crushing. The ulna is relatively straight with no evidence of bowing, and there is no evidence of quill knobs. The olecranon process is only weakly developed, and the proximal end is not as robust as in an undescribed oviraptorid skeleton from Ukhaa Tolgod, IGM 100/1002. The distal end is expanded into a well- formed, convexly rounded end. The radius is slightly more slender than the ulna, and shows no evidence of injury. The ends of the radius are not well exposed, but the distal third has a slight dorsal bend to the shaft. The carpals are poorly exposed (fig. 4). The large distal ("semilunate") carpal is separated slightly from the proximal ends of the metacarpals, and could not have been fused to them. The left element appears to have rotated somewhat out of position. A second, smaller carpal is preserved laterally, opposite the proximal end of the third metacarpal. It is triangular in shape, with a proximal apex. There is no evidence of proximal carpals, but this area is not completely exposed on either side. The manus comprises only three digits (figs. 5, 6), homologous with 1-111 in comparison with other Theropoda. The first metacarpal is relatively robust and is approximately one-third the length of the second and third. The second and third parallel one another, the third being more slender than the second. The lateral edge of the first metacarpal is in contact with the medial surface of X,. the second, but as preserved on both sides the third articulates on the ventrolateral, rather than lateral, edge of the proximal end of Fig. 2. IGM 100/979, invertebrate trace on the dorsal surface of the sternal plates. Anterior the second. The proximal end of the third end of skeleton toward top. metacarpal is well developed rather than being reduced as in some oviraptorids (such as Ingenia). deltoid crest, extending nearly the proximal The digital formula of the manus is 2-3-4- third of the bone. Opposite the deltoid crest, X-X. The first digit is about two-thirds the near the medial edge of the humerus, is a low length of digits II and III (including meta- prominence, the ventral (internal) tuberosity. carpals), which are equal in length. The prox- Only the anterior surface of the distal end is imal phalanx of the first digit is long and exposed. robust, similar in length to, but more robust The shaft of the only complete ulna, the than, metacarpals II and III. The proximal right, is apparently broken and healed, as in- and medial phalanges of the second digit are dicated by an expanded area two-thirds of the also elongate, each being approximately two- way down the shaft with a rugose surface. A thirds the length of the proximal phalanx of deep longitudinal groove medial to the ru- digit I. The proximal two phalanges of the gose area may also be due to this injury, but third digit are smaller, each being approxi- 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 7

Fig. 3. IGM 100/979, furcula in anterodorsal view. Note the elongate, flat hypocleidium (hc). mately one third the length of the proximal placed on both sides. The claw of digit I is phalanx of digit I. The penultimate phalanx larger than that of digit II (89 mm in greatest of digit III is approximately 25% longer than length versus 77 mm), which is larger than the proximal phalanges. that of digit III (71 mm). The articulations between the proximal Remnants of the unossified, presumably phalanges and the metacarpals indicate that keratinous, part of the claw are preserved on little rotation was possible. The surfaces are the second and third unguals on the right side not well rounded, and they do not extend far and the first and third ungual on the left side. beyond the minimum area over which the They indicate that the keratinous claw ex- two bones contact. The articulations between tended significantly beyond the end of the the phalanges, however, are all strongly con- bone. In all four cases the keratin is pre- vexo-concave and extend beyond the mini- served as a thin ribbon of fibrous tissue dor- mum area of contact, indicating a much sal to the end of the ungual, suggesting that greater degree of anteroposterior rotation. this region of the claw differed from others The manal claws are strongly curved, in some way (perhaps in density or thickness deep, and laterally compressed with well de- of the keratin). (A small clump of white ma- veloped flexor tubercles. The proximal end terial beneath the ungual of left digit III near of unguals II and III has a well-developed lip its tip may also be from the keratinous claw.) dorsal to the articular surface. The groove for The longest example is on right digit II, al- the claw sheath on each side is situated near though the distal part of the keratin is sepa- the ventral edge proximally and rises to the rated by a large gap from the proximal part. dorsal surface distally, being symmetrically The total length of the tissue, including the 8 AMERICAN MUSEUM NOVITATES NO. 3265

rdr rmc 3

rmt 2 rmt 3 rmt 4 Fig. 4. IGM 100/979, right fore and hindlimbs on nest in dorsal view. Note pairing of eggs. Pho- tographed prior to preparation of sternum. See Appendix for abbreviations. gap, is 32 mm, and it extends 23 mm beyond somewhat anteroventrally. The pubic boot is the end of the bone in a direction continuous not exposed and must occupy the region in with the curvature of the dorsal edge of the the center of the nest. claw. The distal end of the ischium (fig. 7) is PELVIS AND HINDLIMB: Only the distal part expanded and the two bones form a horizon- of the pubes are preserved, and their distal tal symphysis with a deeply concave dorsal ends are not exposed (fig. 1). In cross section surface. The ischia are in firm contact over the lateral part of each bone is subcircular, the length of the symphysis (116 mm), and and a thin lamina extends medially towards the suture involves a complex interdigitation the midline. The two bones are strongly su- of large processes and concavities in a si- tured into a midline symphysis, forming a nusoidal pattern. An ischial symphysis has relatively flat anterior surface. Proximally the not been reported in any other oviraptorid medial edge of each bone becomes more pos- specimen (Barsbold et al., 1990). The bone teriorly oriented, creating a concavity on the thins dramatically distally, in the posterior anterior surface of the symphysis, and pre- third of the symphysis, and ventrally along sumably a pubic apron posteriorly. The ori- the midline. The preserved portion of the is- entation of the bones is similar to the posi- chia lies distal to the region an obturator pro- tion of these bones in articulated pelves of cess would be expected. oviraptorids, nearly vertical but projecting The distal end of the right femur is too 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 9

.1.~

Fig. 4.-Continued. 10 AMERICAN MUSEUM NOVITATES NO. 3265

p 1

kc 2 cm

Fig. 5. IGM 100/979, right manus in lateral view. Note remnant of keratinous claw on digit II and III unguals. See Appendix for abbreviations. poorly preserved to offer reliable evidence of ascending process of the astragalus is a broad its original condition. It is crushed dorsoven- sheet of bone that would have covered the trally (fig. 8). entire anterior surface of the distal end of the The proximal end of the right tibia is part- tibia, but it is unclear how far along the tibia ly covered by the femur, but a strong lateral it extended proximally. The distal tarsals are deflection proximally indicates the presence not exposed, presumably lying beneath the of a lateral cnemial crest. The shaft is proximal tarsals. crushed and exhibits few features of interest. As in other Theropoda, the proximal end Both fibulae exhibit a distinct bend ap- of the first metatarsal is greatly reduced. proximately one-third of the way down the Metatarsals II-IV are similar to each other in shaft, some (but not all) of which may be due size, and only the proximalmost part of meta- to post-depositional distortion. The portion tarsal III narrows between II and IV, although proximal to the bend expands proximally and much less so than in, for example, Troodon- is mediolaterally flattened, articulating with tidae. None of the metatarsals are fused to the lateral cnemial crest of the tibia. Distal one another. Metatarsal III is slightly longer to the bend, the fibula becomes extremely than IV, which is slightly longer than II. The slender, and it extends to the tarsus to contact articulating surfaces on the distal ends of the calcaneum (fig. 9). these metatarsals are smooth and do not The poorly preserved left proximal tarsals show a ginglymoid condition. They extend indicate that the calcaneum is fused to the anterodorsally onto the shaft, indicating the astragalus distally but not proximally. The capacity for at least limited digitigrade pos- 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA ture. Metatarsal I articulates on the postero- plate and very close to this edge on the right, medial edge of metatarsal II three-fourths of and they undoubtedly articulated to these ir- the way down its shaft, and is preserved in regularities. The fourth ventral rib terminates what appears to be its natural position on far from this surface (11.5 cm from it on the both feet. Its distal end is twisted anteriorly, right side), and its distal end tapers to a point so that the toe is not reversed as it is in many rather than being expanded like the first three birds. The splintlike metatarsal V is exposed ribs, strongly suggesting it did not articulate on the left side, and is approximately one- with the costal margin of the sternum. third the length of metatarsal III. It articulates On the right side, which is better exposed, with metatarsal IV on its posterolateral edge, the more posterior of the two lateral pro- and extends proximally to the same level as cesses is square, terminating in an antero- metatarsals II-IV. posteriorly broad, gently rounded edge, and The digital formula of the pes is 2-3-4-5- it extends further laterally than the anterior 0. The relative lengths of the complete digits process. Its anterior edge is short and slightly are III>IV>II>>I. The phalanges are short concave anteriorly, and the posterior edge is and robust with strongly curved articulation straight and continuous medially with the surfaces. The unguals are broad and robust, posterior edge of the plate. The posterior with very deep grooves for the keratinous edge of the plate is somewhat irregular but claw. The unguals are stout and only weakly is much thinner than the costal margin, and curved ventrally except that of the first toe. there is no evidence for the articulation of AXIAL SKELETON: The only vertebra pre- ribs. Near the midline the sternal plate has a served, located between the scapulae, is too short, broad posterior process. poorly preserved to provide significant in- Only the ventral portions of the posteri- formation. ormost cervical rib and the first six thoracic The two sternal plates (fig. 10) are very ribs are preserved, but they are nearly undis- well preserved, exposed in dorsal view, and turbed and preserved in exceptional detail. are generally similar to those of other ovi- These seven ribs are folded backward ante- raptorids with paired elements (Barsbold, rior to the pubes on the right side, which is 1981). They abut along the midline, without the basis for this description because the left overlapping. The right plate is slightly longer ribs are incompletely exposed (fig. 10). The anteriorly than the left, but otherwise the ex- second through fourth preserved ribs articu- posed portions of both sides are symmetrical late with the sternum, and because the tho- across the midline. The anterior edge of each racic series is typically defined to begin with plate is concave anterolaterally, and the cor- the first rib that articulates with the sternum acoid articulates with only a small area near (e.g., Romer, 1956: 227) they belong to this the midline. The medial edge of each plate series. The posteriormost cervical is thinner is straight without indentations, although an- than the first thoracic, but is comparable in teriorly it is obscured by the trace fossil cov- thickness to the second thoracic. ering most of this region. The lateral edge is A separate ventral (sternal) segment is pre- complex but is dominated by two lateral pro- sent in the first three dorsal ribs and articu- cesses: an anteriorly placed one with a pos- lates ventrally with the sternum (fig. 10). The terolateral costal margin, and a blunt, poste- fourth thoracic rib has a ventral segment that riorly placed process comparable to the lat- does not articulate with the sternum, and it eral xiphoid process (or trabecula) on the tapers to a point ventrally. The fifth and sixth sternum of many avians. ribs lack ventral components, and there is no The anterior process has a concave ante- indication of an articulation surface distally rior edge and a straight posterolateral edge, for a cartilaginous segment. All of the ventral forming a triangle in dorsal view. The ante- rib segments are well ossified, as is evident rior edge is smooth, while the posterolateral from cancelous bone exposed within the ven- edge is thick, rugose, and with several ex- tral end of several segments and the presence pansions. The first three ventral ribs (i.e., the of haversian systems (fig. 11). ventral segments of the thoracic ribs) are pre- The articulation surfaces between the ven- served in contact with this edge on the left tral and dorsal rib segments are expanded AMERICAN MUSEUM NOVITATES NO. 3265

I' 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 13 with nearly flat articulation surfaces (fig. 10). They are separated from each other by several millimeters of matrix, suggesting the presence of connective tissue between them. All of the ventral rib segments are flattened in the same directions as the corresponding dorsal component to the rib. They are preserved close to one another, as a bundle, and the ventral end of each is expanded where it meets the sternum. The ventral segment of the first dorsal rib is the stoutest and shortest, and is straight. The left ribs are preserved in articulation with the sternum (those on the right are displaced a few centimeters poster- omedially), and the robust ventral segment of the first thoracic rib is the most lateral of the three articulating on the costal margin of the sternal plate (as inferred in Sinraptor; Currie and Zhao, 1993). The ventral segments of the second and third thoracic rib are progressively longer and thinner and have a distinct medial bend near the articulation with the dorsal segment (exposed only on the right side). The ventral segment of the fourth thoracic rib is much shorter and tapers to a point, and it is preserved far from the sternum (see above), indicating it did not articulate with it. The fourth, fifth, and sixth thoracic ribs are progressively shorter than those articulated with the sternum, but because they are incomplete proximally their actual lengths are Fig. 7. IGM 100/979, ischia in dorsal view, not known. Nevertheless, the dramatic short- anterior toward the top of the figure. Note ischial ening in these ribs and the lack of distur- symphysis (is) and apposition of ischia and egg. bance to the skeleton strongly suggest that the rib cage was anteroposteriorly short in with the expanded base abutting the rib an- comparison with, for example, crocodylians. terior to them. Associated with the dorsal segments of the A few segments of gastralia are partly ex- first four preserved ribs are four mediolater- posed near the ventral midline beneath the ally flattened uncinate processes that expand ventral ribs and sternal plates and anterior to ventrally (fig. 12). The uncinate processes the left pes. They continue anteriorly beneath are all incomplete dorsally but their propor- the sternal plates, and are easily distin- tions suggest they were at least as long as in guished from the ventral ribs by their smaller Velociraptor (see Norell and Makovicky, in diameter, complex articulations, and lack of press). The posterior edge is straight and the articulation with the edge of the sternum. anterior edge is concave anteriorly. They are Four segments exposed behind the posterior preserved extending posterodorsally over the edge of the right sternal plate converge an- lateral surface of the rib posterior to them, teromedially and are much larger than seg-

Fig. 6. IGM 100/979, left manus in lateral view. Note remnant of keratinous claw (kc) on unguals of digits I and III and strong curvature of claw. AMERICAN MUSEUM NOVITATES NO. 3265

eggs are horizontal or subhorizontal, aligned radially within the nest, and, as exposed beneath the right arm, arranged in at least two layers. The remaining thickness of the block beneath the lower layer suggests that no more than two layers are present beneath the right arm, but the bottom of the nest is not 5 cm exposed. Eight eggs are arranged in pairs Fig. 8. IGM 100979, right hindlimb in dor- (figs. 1, 4), and all may be paired. Where solateral view. Note close proximity of fourth dig- eggs are exposed in two layers, beneath the it and egg. See Appendix for abbreviations. specimen's right forelimb, matrix is absent between some areas of the juxtaposed eggs, but an egg in the upper layer beneath the left ments exposed more posteriorly. Otherwise manus is suspended at least partly by matrix they are similar to those of Dromaeosauridae and not by the lower layer of eggs. Further- (Norell and Makovicky, 1997). more, at least one pair of eggs in the upper layer is situated farther away from the center THE NEST of the nest than the eggs upon which they Fifteen eggs are partially exposed beneath rest (fig. 4), unlike the arrangement of at least the skeleton, but a similar number are esti- two other nests of this type (Dong and Cur- mated to remain unexposed in the matrix. rie, 1996). Most of the center of the nest is The nest appears to be intact, and with a few not exposed, but there is no evidence for exceptions breakage is consistent with com- eggs in the center at the same level as the pression of the sediment rather than preburial uppermost eggs exposed further out from the disturbance because there is little displace- center. ment of the fragments. All of the exposed The eggs are elongate, parallel sided, with 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA

Ica 3cm Fig. 9. IGM 100/979, left fibula and tarsus in posterior view, left pes in dorsal view. Note egg fragments beneath unguals. See Appendix for abbreviations. rounded ends of similar shape, a shape bones visible in the few in which the interior termed elongatoolithid (Mikhailov et al., is exposed. Two eggshell fragments, near the 1994). Although previously we implied that left ischium, suggest the presence of a bro- they are less rounded at one end than the ken egg, but this area is insufficiently ex- other (Norell et al., 1995), in those eggs posed to interpret definitively. A histologic where both ends are exposed they are sym- study of the egg shell will be published else- metrical. The outer surface has a series of where (Bray et al., in press). longitudinally oriented fine ridges that grade None of the eggs are exposed in their en- into dimples or smooth areas at the poles. tirety, so all of their dimensions cannot be None of the well-exposed eggs show indi- measured accurately. A pair of eggs beneath cations of hatching or predation, nor are the right manus differ in their width, one be- AMERICAN MUSEUM NOVITATES NO. 3265 ing only 6.5 cm wide, the other 7.2, and the difference is likely due to the effects of crushing. An egg beneath the left carpus is nearly completely exposed, and as preserved is at least 18 cm in length, but less than 19 cm. Eggs are not preserved within the pelvic region, much of which is missing. If the animal had been in the act of laying an egg or preparing to do so, it is expected that one or two would be preserved here. The incompleteness of the pelvis makes this assessment uncertain, but to have been lost to erosion an egg must have been "floating" in matrix well above the ischiadic symphysis. There is no evidence from the outlying sediments as to the shape or size of the nest, although little sediment was removed with the block. There is no obvious structure to the sediments, nor changes in macroscopic details away from the eggs (however, they have not been studied microscopically). There is also no evidence for plant matter in the nest, but as the sediments are highly ox- rf idized and poorly consolidated with large pore spaces it is unclear whether it could 3cm have survived diagenesis and the percolation of water Fig. 10. IGM 100/979, sternal plates in dor- ground (D. Loope, pers. comm.), sal view, anterior toward top of figure. Note the and plant remains are not known from the articulations of the three ventral rib segments with Djadokhta Formation at Ukhaa Tolgod. the sternum, dislocated slightly posteriorly on the right side but intact on the left. Also note the ap- POSTURE OF THE SKELETON ON THE NEST position of the gastralia and ventral ribs with a partially exposed egg posterior to the right sternal The most remarkable aspect of the skele- plate. See Appendix for abbreviations. ton is the birdlike posture in which it is preserved (figs. 1, 13). The body would have completely covered the nest, with the abdom- egg is exposed beneath the ribs and gastralia inal region over the center, and both limbs immediately posterior to the right sternal el- are placed in positions nearly symmetrical ement (fig. 10), and the bones rest directly across the midline. The positions of the tibia, on the egg surface. The more posterior ribs fibulae, femur, pelvis, and ribs suggest that are above the highest level of eggs, suggest- the body was shifted slightly to the left of ing that the center of the nest may have been center, lying on the left side of its chest, rath- filled. er than being exactly symmetrical. Further- The forelimbs extend laterally and then more, the right forelimb is shifted medially, posteriorly to overlie the perimeter of the and the proximal end of the humerus has nest (fig. 1). The humeri extend laterally and come to lie near the midline. This shifting of slightly posteriorly from the glenoid fossa. the body and right forelimb to the left sug- The right antebrachium is extended, but the gest that the sediment deposited upon the distal ends of the left indicate that the elbow skeleton may have come toward the animal on that side was flexed about 900. The right from its right hand side. carpus is strongly flexed, but the left is only The gastralia, ventral ribs, and apparently slightly flexed. The hands are strongly pro- the sternal elements rest directly on eggs. An nated and lay on their medial surface, and 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA

_ It ^ i.' ...4 1K ->

.11. 1, AL

I..i. t-l".. W a5 e- I

-K I,a}0'.A4

. I-,.

1K T 1.: is

Fig. 10.-Continued. AMERICAN MUSEUM NOVITATES NO. 3265

in several places (at the base of phalanges II- 1, 111-2, and IV-5), while the fourth toe of the right side is completely extended (the second and third are not exposed). The bones of the left pes are intermixed with pieces of egg shell, and the fourth toe of the right pes is less than 1 cm from an egg beneath it. The skeleton shows little indication of disturbance after deposition. All of the exposed elements are in their natural articulations, and the few cases in which they are slightly separated (e.g., the ungual of left pedal digit III) are to the limited extent consistent with postdepositional compaction. The region showing the greatest disturbance is the shoulder girdle, where the coracoids are separated from the sternal elements and the furcula is moved anteriorly, apparently due to a medial movement of the right, and to a lesser extent the left, forelimbs.

DISCUSSION PHYLOGENETIC AFFINITIES OF THE SKELETON The precise identification of IGM 100/979 Fig. 11. IGM 100/797, thin sections of a dor- is made difficult by the lack of a skull and sal segment of a thoracic rib (top) and a ventral vertebrae, but several features allow identi- segment of the left first thoracic -rib (bottom). fication both to Oviraptoridae and to a group Note the presence in both of osteons with a cen- within it. Skeletons of Oviraptoridae are tral lacuna, identified by arrows in the ventral rib section, indicative of ossification rather than cal- common at Ukhaa Tolgod (Dashzeveg et al., cification. 1995), and at least two taxa are present (Clark and Norell, in prep.). Oviraptoridae are members of the theropod group Coelu- the ventral displacement of the first and third rosauria, but do not belong to the largest metacarpals relative to the second is most group of coelurosaurs, the birds (i.e., Avi- likely due to the vertical compression that alae). came to bear after deposition. The right car- The coelurosaurian affinities of the skele- pus rests on top of, or very close to, an egg, ton are evident from the presence of elongate but the left forearm is several centimeters forelimbs, the absence of manus digit IV and above the level of the eggs. the presence of the semilunate carpal (Gau- The ischia lie directly on top of an egg, thier 1986). Among nonavialan Coelurosau- and the pubes extend into the matrix anterior ria, only Dromaeosauridae, Therizinosauri- to this egg, toward the center of the nest. The dae, Troodontidae and Oviraptorosauria (in- pubes have not been completely exposed, but cluding Oviraptoridae) are known to have a the preserved portions are from the pubic subquadrangular coracoid (Gauthier, 1986; symphysis and therefore near the ventral end Russell and Dong, 1993; Sues, 1997). Dro- of the bone. Although the expanded "boot" maeosaurids and Oviraptorosauria both have at the end of the bone is not evident, the two broad sternal plates that are sometimes pubes probably do not extend much further fused (Barsbold, 1983; Norell and Makov- ventrally than their exposed portions. icky, 1997), but dromaeosaurids have an op- Both legs are oriented parasagittally with isthopubic pelvis, a thinner fibula, metatar- the knee and ankle joints strongly flexed. The sals II-IV with ginglymoid distal ends, a re- toes of digits II-IV on the left pes are flexed tractable second pedal digit with a greatly en- 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 19

| l - _ .g .r .dB -| I 11f >.S _| *- | E.l .^aXw 111|1 | | | _ llg i | Eppl!j.p X 1111 | l l ''YZ * Z | ... ' 'r .s .e w_ - S | | :g.n 1 . 'iEll I_ g | Fig. 12. IGM dorsal 1l_115 100/979, segments of right thoracic ribs and_ associated_., uncinate|.|_.processesF.K.;;_- in right dorsolateral view, anterior to the right. _Bil|_>air larged claw, and shallower manal claws with (Currie, 1990).__However,lll|troodontids||in*;have an> asymmetric grooves for the keratinous extremely reduced third metatarsal,*11l11;.,,.,an*sj;e/>,un-@ sheath. The manus also differs in having a usually robust___11!!fourth1111111metatarsal,XBIS|.KN and..,,,e,ajFre- shorter third digit relative to the second, and tractable second pedal digit (Osm6lska and a more slender, slightly bowed third meta- Barsbold, 1990), and are reported to lack a carpal. The deep, strongly curved manal furcula, the clavicles instead being slender claws of this specimen, with a dorsal lip over and paired (Russell and Dong, 1993). the articulating surface, are similar only to Oviraptoridae and the poorly known Caen- those of oviraptorosaurians and troodontids agnathidae (including Chirostenotes, Caen-

Fig. 13. Schematic reconstruction of the oviraptorid as it might have appeared on the nest in lateral view, based on the posture of the skeletal remains of IGM 100/979. The preserved portion of the skeleton is shaded darker than the reconstructed portion. The arrangement of eggs in the nest is in part hypothetical, because most of the nest is not exposed, and the degree to which the outer parts of the nest were buried is hypothetical. 20 AMERICAN MUSEUM NOVITATES NO. 3265 agnathasia, and Elmisaurus; Sues, 1997) and Coelurosauria (Barsbold, 1981; Barsbold et Microvenator (Makovicky and Sues, 1998) al., 1990), features that are absent in IGM together comprise Oviraptorosauria (Osmol- 100/979. The sternum of Ingenia also has a ska and Barsbold, 1990). Many postcranial less developed lateral trabecula than does elements are unknown for Caenagnathidae, that of Oviraptor (Barsbold, 1983, fig. 15). and the relationships of Microvenator are un- The postcranial skeleton of Conchoraptor certain, limiting the unambiguous diagnosis also lacks the specializations of Ingenia of Oviraptoridae. Nevertheless, the metatar- (Barsbold et al., 1990), but the numerous sals of Caenagnathidae differ from those of specimens of this taxon preserved in the Oviraptoridae and IGM 100/979 in that Mongolian Institute of Geology are all much metatarsal III is much narrower dorsally, and smaller than those of IGM 100/979, whereas in Elmisaurus rarus, at least, the proximal several specimens of Oviraptor are of a sim- metatarsals are fused together with the distal ilar large size, much larger than any other tarsals (Osmolska, 1981). Furthermore, ma- oviraptorosaurian. Pending revision of this nus digits II and III are subequal in length group based upon the new material from and robustness in Oviraptoridae (Barsbold et Ukhaa Tolgod, the present evidence therefore al., 1990), as they are in this specimen, but suggests that IGM 100/979 is an oviraptorid in Caenagnathidae and most other Coeluro- most closely related to Oviraptor. sauria (with the exception of ornithomimo- The specimen also presents some features saurs) digit III is shorter and more gracile differing from conditions in other Ovirapto- than digit II (Gauthier, 1986). The only ridae. The posterolateral flattening and bend- known skeleton of Microvenator celer shares ing of the furcula and the horizontal orien- few elements in common with IGM 100/979, tation of the shelf on the anterior edge of the but the deltopectoral crest of the humerus of scapula with which it articulates are similar Microvenator is shorter and less pronounced to, but more developed than, these regions in (Makovicky and Sues, 1998). Oviraptor philoceratops (IGM 100/42). The Oviraptoridae are unique among nonavi- strongly sutured ischial symphysis is also alan theropods in the shape of their furcula, unique among oviraptorid specimens, but the but this element is not known in Caenag- bone along the symphysis is very thin and nathidae or Microvenator. A furcula is pre- therefore easily damaged, and none of the sent in several nonavialan theropods (Currie known oviraptorid skeletons with complete and Zhao, 1993; Norell et al., 1997; Chure ischia are as large as this one. and Madsen, 1996; Makovicky and Currie, 1998), but in Oviraptoridae it is particularly PHYLOGENETIC AFFINITIES OF THE EGGS robust and often has a distinct hypocleidium (Barsbold, 1983), as on this specimen. Many types of eggs are known from the Among theropods the most similar furcula is Djadokhta Formation of Mongolia (Mikhai- that of Archaeopteryx (de Beer, 1954) and lov et al., 1994), but the identification with Confuciusornis (Peters, 1996; Chiappe et al., particular groups of vertebrates is tentative submitted), although these (unlike that of for all except the single type for which di- many birds) lack a hypocleidium. The hy- agnostic embryonic remains are known. The pocleidium of IGM 100/979 is longer than in single exception (IGM 100/971) is an ovi- any described oviraptorid, but this delicate raptorid embryo preserved within an incom- process may have been lost or damaged in plete egg from Ukhaa Tolgod (Norell et al., other specimens. 1994). The eggs of IGM 100/979 are similar Of the three named genera of Oviraptori- in all respects to the egg with the oviraptorid dae-Oviraptor, Ingenia, and Conchorap- embryo. tor-the skeleton of Ingenia is the most spe- A parataxonomy of eggs and eggshell has cialized. The manus of Ingenia is unusual in been developed by Zhao (e.g., 1975), and that digit I is nearly as long as digit II, and Mikhailov (1991) identified histologic mor- metacarpal I is half as long as the metacarpal photypes correlated with this taxonomy. The II, rather than being one third the length of eggs of IGM 100/979 are of the type classi- metacarpal II as in Oviraptor and most other fied as elongatoolithid in Zhao's system as 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 21 modified by Mikhailov, based upon their Oviraptoridae and unknown or rare in other large size, elongate shape with parallel or close relatives of birds: a single ossified ven- sub-parallel sides, and sculpture pattern tral rib segment in the first four thoracic ribs, ("discretituberculate") of nodes and short the articulation of three ribs with the costal longitudinal ridges on the long edges and margin of the sternum, and ossified uncinate nodes on the ends. The eggs from the nest processes. The specimen also confirms fea- (AMNH 6508) beneath the holotype of Ovi- tures noted previously in other oviraptorids raptor philoceratops at Bayn Dzak (Norell et (Barsbold, 1981), such as the articulation of al., 1994) and those from beneath the ovirap- the coracoid with the anterior, rather than lat- torid skeleton at Bayan Mandahu (Dong and eral, edge of the sternum and the presence of Currie, 1996) are also of this type. a robust furcula. The eggs of IGM 100/979 are unusually The pattern of rib segmentation in IGM long, at least 18 cm in length. Elongatoolith- 100/979, with a single, ossified ventral seg- id eggs are generally 15-17 cm long (Mik- ment, is remarkably similar to the pattern in hailov et al., 1994); those from Bayan Man- the ribs of birds and unlike that in crocodyli- dahu are only 15 cm in length, and the eggs an ribs. In the thoracic rib cage of nonavian from beneath the 0. philoceratops holotype reptiles, such as crocodylians, one or two (AMNH 6508) are 14 cm long (although cartilaginous segments are present between crushing may have shortened them by up to the ossified dorsal rib and the sternum, but 2 cm). The egg from Ukhaa Tolgod with the they do not ossify (Parker, 1868; Hoffstetter oviraptorid embryo (IGM 100/97 1) was cited and Gasc, 1969). In crocodylians, the first 9 as being only 12 cm long (Norell et al., or (in Gavialis) 10 thoracic ribs have two 1994), but it is too incomplete to estimate its cartilaginous segments ventral to the dorsal length accurately. rib (i.e., three segments in all), whereas the Because of the paucity of embryonic re- following two ribs lack the intermediate el- mains in fossil eggs the precise identity of ement. The two ventral segments calcify in most is unclear, and the taxonomic level at some individuals, but they are not reported which elongatoolithid eggs are diagnostic to ossify (e.g., they do not have haversion may extend beyond Oviraptoridae. Further- systems). The ribs of birds, however, have more, the only known embryo is incomplete, only one ventral segment, and it is well os- and cannot be identified below the family sified (Bellairs and Jenkin, 1960). An appar- level. The skull and mandible of the embryo ently ossified, single ventral rib segment is possess the short, vertical rostrum, edentu- present in fossils of some basal avialans (e.g., lous margins, and tall mandible that distin- Iberomesornis, Confuciusornis, and Hespe- guish oviraptorids from all other theropods. rornis) although they are not discernible on The mandible is much shorter and taller than any known specimen of Archaeopteryx. that of caenagnathids, placing it within the Ossified ventral ribs were identified ten- Oviraptoridae. Unfortunately, the lack of tatively in the dromaeosaurid Deinonychus dorsal roofing bones and evidence of the ma- on the basis of isolated elements (Ostrom, nus (which may be covered by the rest of the 1969: 84-86), in the dromaeosaurid Veloci- skeleton) precludes more precise identifica- raptor (Barsbold, 1983: 34), and in the basal tion. troodontid Sinornithoides on the basis of CT scans of an articulated skeleton (Russell and POTENTIAL SKELETAL HOMOLOGIES Dong, 1993). Comparing Deinonychus with WITH AVIALAE crocodylians, Ostrom tentatively inferred the presence of two ventral segments in Deinon- This specimen is significant in providing ychus, but the isolated elements he studied new information concerning the ventral part do not provide evidence sufficient to reach of the thorax and the keratinous claws in ovi- this conclusion, especially considering the raptorids, and in suggesting that oviraptorids evidence of only a single ventral segment in behaved in ways similar to their living rela- their relatives, birds and oviraptorids. An os- tives, birds. The ribs and sternum provide ev- sified ventral segment is definitively present idence for features hitherto poorly known in in Velociraptor (Norell and Makovicky, in 22 AMERICAN MUSEUM NOVITATES NO. 3265

0. . S°,.v0'Psd eP5 t -AOOJV \\1 0 90 s Oss. ventral ribs o 0 0 7 ? 1 1 0? 1 1 1 Sternal rib # '3-4 0-6 2 2 2 3 3 ? 5 5 2-9 Ossified stemum 0 0 0 1 1 1 1 1 1 1 1 Oss. unc. proc. 1 0 0 7 ? 1 1 0? 1 1 2(1)

3 rib articulations \ fl to the sternum; ossified ventral ribs*; ossified uncinate processes*

Fig. 14. Cladogram of relationships among theropod taxa with well-preserved sternal regions and distribution of osteological characters discussed in the text, and the level at which selected features are synapomorphous (under DELTRAN optimization). Relationships are supported by the analyses of Gau- thier (1986), Holtz (1994), and Sues (1997), and the interpretation of Currie and Zhao (1993). Ossified ventral ribs: rib ossification confined to dorsal segment (0) or both dorsal and ventral rib segments ossified (1); Sternal rib number: number of rib articulations to the body of the sternum (the praesternum); Ossified sternum: sternum unossified (0) or ossified (1); Ossified uncinate processes: ossified uncinate processes absent (0), present and separate from ribs (1), present and fused to ribs (2). Paren- theses indicate the occurrence in a group of a condition that is most simply interpreted as secondarily derived, asterisk indicates characters that apply to more inclusive groups under different optimization procedures. Sources: Parker, 1868; Marsh, 1880; Furbringer, 1888; Howes and Swinnerton, 1901; Lam- be, 1917; Hoffstetter and Gasc, 1969; Currie and Zhao, 1993; Wellnhofer, 1993; Norell and Makovicky, in press; Chiappe et al., submitted. press), but is not otherwise reported in non- parison with the typically larger number in avialan Coelurosauria. birds and the smaller number in more basal The ventral segments on IGM 100/979 de- theropods. In birds a variable number of ribs finitively confirm the presence of these struc- articulate with the sternum, but with a single tures in nonavialan theropods, and they cor- exception it is greater than two. The number roborate the identification of these elements of rib attachments to the avian sternum is in Deinonychus, at least one example of reported to range from 2 in some species of which is similar in shape and has expanded Dinornithidae (moas) to 9 in, for example, ends (Ostrom, 1969: fig. 52b). The absence Cygnus (Furbringer, 1888: table 21). The of these elements in other taxa may be an sternum of some moas is reported to have artifact of ontogenetic sampling, preserva- facets for only two ribs, but those of other tion, or the manner in which they were col- species in this family are reported to have lected. In any case, the presence of ossified three (Owen, 1879; Oliver, 1949). The rela- ventral rib segments is a synapomorphy of tionships of moas, within Paleognathae (Lee birds with a group including at least ovirap- et al., 1997), imply that the condition of hav- torids and dromaeosaurids among dinosaurs ing only two rib articulations represents an (fig. 14). evolutionary reversal from a larger number The articulation of three ribs with the ster- of articulations, given that tinamous, neo- num in IGM 100/979 is significant in com- gnaths, and basal avialans, where known 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 23

(e.g., Hesperornis and Ichthyornis, Marsh, The precise phylogenetic relationships 1880; Confuciusornis, Chiappe et al., sub- among theropods is disputed, but all recent mitted), have a larger number. In Archaeop- studies (Currie and Zhao, 1993; Holtz, 1994; teryx a sternum is preserved on only one Sues, 1997) indicate that oviraptorosaurs and specimen (Wellnhofer, 1993), and it is too dromaeosaurids are more closely related to poorly preserved to determine the number of extant birds than are tyrannosaurids, and tyr- rib articulations. annosaurids are more closely related to a In the few nonavialan theropods for which group comprising these taxa than is Sinrap- this feature is known (Albertosaurus and Sin- tor. The optimization of the number of rib raptor), distinct facets for only two rib artic- articulations to the sternum on a cladogram ulations are evident on the sternum (Lambe, of these phylogenetic relationships (fig. 14) 1917; Currie and Zhao, 1993). The sternum indicates that (1) two rib articulations with is known in representatives of all of the other the ossified portion of the sternum is ple- major groups of dinosaurs (Sauropodomor- siomorphic for Theropoda, (2) three rib ar- pha, Thyreophora, Ornithopoda, and Margin- ticulations with the sternum is a synapomor- ocephalia), but it is unclear how many ribs phy of dromaeosaurids, oviraptorids, and attach (Weishampel et al., 1990). A specimen Avialae (and perhaps some other coelurosau- of Velociraptor mongoliensis from Tugru- rians more closely related to them than to geen Shireh (IGM 100/976) has three closely tyrannosaurs and more basal dinosaurs), and appressed ventral ribs preserved adjacent to 3) the higher number of sternal articulations the right side of the right sternal plate (Norell in many birds is an evolutionary increase. et al., 1997; Norell and Makovicky, in press), The identification of the costal margin of providing evidence that three ribs articulated the sternum allows for a more precise un- with the sternum in this taxon as well. The derstanding of the morphology of this struc- articulation of more than two ribs with the ture in oviraptorids (see Barsbold, 1981, for sternum in the oviraptorid specimen de- descriptions of other oviraptorid sternae). scribed here and in Velociraptor therefore is The costal margin is in a position comparable a synapomorphy of these two taxa with birds, to that of most birds, on the anterior half of but the lack of information about the distri- the lateral surface, rather than posteriorly as bution of this feature among other extinct in crocodylians and lepidosaurs. Although Coelurosauria greatly limits its implications there is a great deal of variation among birds at this time. in the morphology of the lateral edge of the In crocodylians (Parker, 1868; Hoffstetter sternum there is often a lateral or posterolat- and Gasc, 1969), the first two thoracic ribs eral process immediately posterior to the cos- articulate (via cartilaginous segments) with tal margin, termed the external xiphoid pro- the broad body of the cartilaginous sternum cess of the lateral xiphisternum by Parker (the praesternum of Parker, 1868), and the (1868) and Bellairs and Jenkin (1960) but following 6 or 7 ribs articulate with a much identified as the lateral trabecula by Furbrin- narrower posterior extension (the mesoster- ger (1888) and Baumel and Witmer (1993). num of Parker, 1868). In extant birds, ribs The posterior of the two lateral processes of articulate only with the anterior part of the the sternum described above in IGM 100/979 lateral edge of the broad, ossified sternal is in this same position, and its similarity plate, and a structure similar to the meso- with the structure in extant birds and the sternum of crocodylians is absent (Parker, presence of a similar process in some basal 1868; Furbringer, 1888). The ossified ster- avialans (e.g., Concornis; Sanz et al., 1995) num of theropods (including birds) is there- is sufficient to hypothesize that it may be ho- fore most comparable to the body (praester- mologous among these taxa. The absence of num) of the crocodylian sternum, and the such a process in alvarezsaurids (Perle et al., presence of two ribs in articulation with the 1994) suggests that this feature may have sternum in basal theropods and with the body been lost and regained in the course of avi- of the sternum in crocodylians in conse- alan evolution, however. quence establishes this as the primitive con- The articulation of the coracoid to the an- dition for Archosauria. terior edge of the sternum is a further feature 24 AMERICAN MUSEUM NOVITATES NO. 3265 that oviraptorids and some other nonavialan 1969; Bellairs and Jenkin, 1960). Uncinate theropods share with birds. In crocodylians processes are not reported in nontheropodan and lepidosaurs the coracoid articulates along diapsids nor in Archaeopteryx, but it is dif- the lateral surface of the sternum, whereas in ficult to ascertain whether the apparent ab- birds it articulates along the anterior edge sence of these elements is actually due to the (Parker, 1868). The sternum is poorly known vagaries of preservation because it requires in dinosaurs, but in Ankylosauria (Coombs, exceptional preservation to differentiate 1990) and Iguanodon (Norman, 1980) the these elements from the ribs proper. In any coracoid contacts the anterolateral edge of case, the elements in IGM 100/979 are sim- the sternum (Coombs, 1990) whereas in sau- ilar to the elongate elements in basal birds ropods it is unclear where the coracoid artic- such as Hesperornis rather than to the short, ulates (McIntosh, 1990). In Sinraptor (Currie unossified elements of crocodylians, and the and Zhao, 1993), Albertosaurus (Lambe, distribution of elongate, ossified uncinate 1917), and the dromaeosaurid Velociraptor processes indicates that it is a synapomorphy (Norell et al., 1997; Norell and Makovicky, of Avialae and a more inclusive group of di- in press), as well as in other oviraptorids nosaurs including oviraptorids and dromaeo- (Barsbold, 1981), the articulation is with the saurids (fig. 14). anterior edge of the sternum. In extant birds, movements of the rib cage The presence of ossified uncinate process- and sternum are critical to the passage of air es in IGM 100/979 broadens the taxonomic through the specialized respiratory system distribution of this character within Thero- (McLelland, 1989), and it is has been sug- poda (fig. 14). Free uncinate processes as- gested that this respiratory system was absent sociated with the thoracic ribs are present in in nonavialan theropods (Ruben et al., 1997). Velociraptor (Paul, 1988; Norell and Makov- The presence in oviraptorids of ossified un- icky, in press) and in some basal avialans, cinate processes, a furcula, ossified ventral being especially well developed in Hespe- ribs in articulation with a broad, ossified ster- rornis regalis (Marsh, 1880), Confuciusornis num, and an anteriorly placed coracoid-ster- (Chiappe et al., submitted) and Chaoyangia num articulation, on the other hand, suggests (Hou and Zhang, 1993). In extant birds these that these animals were capable of the move- elements develop independent of the rib but ments required by the avian respiratory sys- usually fuse to it, except in Apteryx and pen- tem. Cataloging the osteological components guins where they remain free (Parker, 1868; of the avian respiratory mechanism does not Furbringer, 1888; Bellairs and Jenkin, 1960). constitute a critical test of the biomechanical The incomplete elements preserved on this capabilities of these structures, however, and specimen are similar to those of Velociraptor demonstrating the capability of a behavior is but are more expanded ventrally. Those of not equivalent to demonstrating that the be- Hesperornis are as broad ventrally but are havior was present. There is at present no shorter. Although the evidence presented by strong evidence for the detailed morphology IGM 100/979 for identifying these as unci- of the respiratory system in nonavian dino- nate processes is not definitive, owing to its saurs worthy of detailed consideration. In incompleteness dorsally, at least one other any case, the assertion that ossified ventral oviraptorid specimen from Ukhaa Tolgod, rib segments and a broad sternum are absent IGM 199/1002, a nearly complete articulated in nonavialan theropods (Ruben et al., 1997) skeleton, preserves similar elements inter- is false, and the inference that a respiratory laced with the ribs, corroborating this iden- system like that of avians was absent in these tification. taxa is not supported by any reliable evi- The evolution of uncinate processes in dence (the interpretation of incompletely pre- diapsids is somewhat ambiguous, as they are pared soft tissue residue in Sinosauropteryx present and ossified in birds and the basal notwithstanding). lepidosaur Sphenodon, present but unossified The remains of the keratinous claws of the in crocodylians, and altogether absent in ex- manus are the first reported for an ovirapto- tant squamates (Parker, 1868; Howes and rid, and indicate a highly curved claw Swinnerton, 1901; Hoffstetter and Gasc, (Chiappe, 1997). The curvature of the kera- 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 25 tinous pedal claws of living birds and its re- The position of IGM 100/979 strongly lationship to lifestyle was studied by several evokes the posture taken by birds while sit- authors, most recently Feduccia (1993), in an ting on their nests, and the occurrence of four attempt to interpret the lifestyle of Archae- specimens directly on nests among the first opteryx and other theropods. Feduccia noted thirty excavated oviraptorid skeletons indi- that in Archaeopteryx the curvature of the cates that this association is the result of a pedal claws is similar to that of perching consistent behavior of these animals, so a birds, and the curvature of the manal claws careful consideration of its implications is is similar to that of climbing birds. Using necessary. We review other specimens here, Feduccia's method of measuring curvature, because a discussion of other specimens of the claw of digit II on the right side of IGM oviraptorids preserved on nests is critical to 100/979 spans a minimum of 1600, similar to corroborating the evidence provided by our the manal claws of Archaeopteryx and the specimen. pedal claws of climbing birds (Chiappe, (1) AMNH 6517, the holotype of Ovirap- 1997). Furthermore, the bony manal claws of tor philoceratops Osborn, 1924. The speci- the dromaeosaurids Deinonychus (Ostrom, men is from the Djadokhta Formation at 1969) and Velociraptor (Barsbold, 1983; No- Bayn Dzak. The skeleton is missing the hind rell and Makovicky, in press) are as strongly limbs, but the anterior part of the skeleton is curved as that of IGM 100/979. This sug- preserved lying on its side rather than in a gests either that Deinonychus and ovirapto- posture similar to that of IGM 100/979. Os- rids were climbers, or that there was no cor- born (1924) noted that the skull was pre- relation between claw curvature and climb- served four inches above the nest, but the ing behavior in these taxa. Correlations be- precise position of the skeleton over the nest tween form and function are a dubious means is unclear because they were separated dur- of inferring behavior, because form does not ing preparation. The nest from beneath the always indicate behavior where this relation- skeleton (AMNH 6508) is incomplete, com- ship can be examined in extant taxa (Lauder, prising only 15 eggs when collected (two of 1995), and we are unaware of any evidence which were not intact). The preserved por- that tests critically the climbing abilities of tion of the nest indicates that the eggs were oviraptorids and dromaeosaurids. In any arranged in pairs and in at least three tiers, case, the lack of curvature in the manal claws and the preserved semicircular pattern sug- of nonavialan theropods reported by Feduc- gests the nest was originally circular. cia (1993) is contradicted by this specimen. (2) IVPP specimen V9608, described by Dong and Currie (1996). Most of the skele- IMPLICATIONS OF THE SPECIMEN ton is absent, but preserved are several ver- FOR OVIRAPTORID BEHAVIOR tebral fragments and portions of the right fore and hind limbs overlying six elonga- The specimen described here presents rare toolithid eggs. There is no indication among evidence of behavior in an extinct animal, the preserved bones that the skeleton or nest but the interpretation of behavior in fossils is was incomplete before erosion, but the eggs more speculative than in living organisms. occur in only one layer. The limb elements Because behavior-self-initiated movement indicate the skeleton was in a posture similar of living organisms-cannot be observed di- to that of IGM 100/979, with the arm around rectly in fossils it must be inferred, and these the perimeter of the nest and the pes in the inferences involve untestable assumptions center of the nest. beyond those involved in observing behavior (3) An unprepared skeleton from Ukhaa in living organisms. However, the success of Tolgod, field number MAE 95-97. As ex- studies of insect nests, spider webs, and writ- posed in the field, the specimen consists of ten languages demonstrates that the products much of a skeleton (except the skull) over- of behavior are interpretable, and in some lying a nest that may be complete. Although ways more easily interpreted than is the be- incompletely exposed, the eggs have discre- havior producing them because they can be tituberculate sculpturing on the poles similar observed repeatedly without being affected. to that of elongatoolithid eggs. A photograph 26 AMERICAN MUSEUM NOVITATES NO. 3265 of the specimen was published by Webster which this specimen and the three other ovi- (1996: 80). raptorids directly overlying nests were col- In the most recent review of Oviraptori- lected suggest sudden burial, and represent dae, Barsbold et al. (1990) listed the total an environment in which the animal could number of specimens known at that time plausibly have been living (rather than one from all formations as 13, including the first to which it was carried after death). All are one ever collected, AMNH 6517. Dong and from similar facies of the Djadokhta For- Currie (1996: 632) reported that three spec- mation, a facies termed a structureless sand- imens of oviraptorids were collected from stone (Eberth, 1993). Although such sand- the Djadokhta Formation at Bayan Mandahu stones often are continuous laterally with by the Sino-Canadian Dinosaur Project, in- crossbedded sandstones, indicative of sub- cluding IVPP V9608. The American Muse- aerial, eolian deposition, there is no evidence um of Natural History-Mongolian Academy of crossbedding within them. The lack of in- of Sciences expeditions in 1991 and 1992 ternal structure to these sandstones, which collected three oviraptorid specimens from may reach 15 m in thickness, has been in- the Barun Goyot Formation at Ikh Khongil terpreted as the result of sudden accumula- (also known as Nemegt) and Khermeen Tsav, tions of wind-borne sand (Eberth, 1993). A and in 1993 discovered the wealth of fossils recent model suggests instead that these at Ukhaa Tolgod (Dashzeveg et al., 1995). A structureless sandstones were deposited as complete census of oviraptorids from Ukhaa mass movements from standing dunes during Tolgod must await preparation of all of the rain, rather than wind, storms (Loope et al., theropod material collected thus far, but the 1998). two specimens overlying eggs collected in Regardless of which of these two deposi- 1993 and 1995 were among the first 12 spec- tional models is correct, the horizontal ori- imens collected that were identifiable as ovi- entation of the nest and skeleton, the lack of raptorid (including one embryo). Thus, the disturbance to them, and their occurrence four oviraptorid skeletons overlying nests within a single bed of structureless sandstone were among the first 30 adult specimens of indicate that they were in place before burial this family collected, a ratio of 13.3%. If and then covered by a flow. The common only the remains from the structureless sand- occurrence of apparently undisturbed nests of stones of the Djadokhta Formation are con- dinosaur eggs in the sandstones of the Dja- sidered, then the four specimens on nests are dokhta Formation (Mikhailov et al., 1994), among only 17 adult specimens collected be- and to a lesser extent that of undisturbed, ar- fore 1996, a ratio of 23.5%. ticulated skeletons (which, unlike the nest, The state of preservation of IGM 100/979 may be transported as a unit when the car- indicates that it was little disturbed after the cass has not decomposed), indicate that the death of the animal, and its posture is likely sandstones formed the substrate of the envi- that taken by the animal as it died (fig. 1). ronment in which the adult oviraptorids The skeleton is remarkably complete and ex- lived. hibits no evidence of scavenging or other dis- The direct apposition of the skeleton on turbance after death that would have altered the nest in IGM 100/979 provides strong ev- the posture of the skeleton (other than the idence that the nest was not completely cov- worm burrows, which are unlikely to have ered. The adult skeleton is separated from the altered it significantly). This is unusual for eggs in most places by little sediment, and in vertebrate fossils, which are often disarticu- some areas (e.g., the right carpus, left pedal lated and show other evidence of transpor- digits, right fourth pedal digit, ischial sym- tation, such as hydrodynamic wear. Both the physis, and sternal elements) they are in con- skeleton and the nest are so little disturbed tact or nearly in contact. The position of the as to leave little doubt that they were not posterior ribs, pubes, and the medial part of transported after death and, thus, that the po- the right foot above sediments in the middle sition in which they are preserved reflects the of the nest indicate that the center of the nest position of the skeleton when it died. was not open. The sedimentology of the deposits from Unfortunately, it is not possible to deter- 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 27 mine precisely how much of the nest was (Wenzel, 1992; Greene, 1994), because both buried, because the sediments forming the are based upon the simplest interpretation of nest are not distinguishable from those that similarities shared among groups of taxa later buried them with the skeleton. Further- (Rieppel, 1994). Because of its ephemeral more, although most parts of the forelimbs nature, however, behavior is more difficult to are separated from the top of the nest by sev- document (e.g., Drummond, 1981; Miller, eral centimeters of sediment it is unclear that 1988). Sampling is a critical issue because this necessarily implies these areas of the the condition in all relevant taxa-including nest were covered by this sediment, because those supposedly lacking the behavior-must it could have been deposited as the animal be ascertained. These problems are exacer- was being buried. During burial it is much bated in the fossil record, because in addition less likely that sediment was removed from to the more speculative documentation of be- between the skeleton and nest to bring them havior it offers, sampling is biased by un- closer together, rather than added to separate known factors related to the genesis of the them, so the separation of the skeleton and geologic deposits in which they are pre- nest probably indicate the minimal amount served, and the taxonomic identity of the or- of separation (except for the relatively minor ganism is often poorly resolved or, in the compaction of the sediments). case of trace fossils, conjectural. Neverthe- The precise relationship between the skel- less, these considerations should not neces- eton and nest is known at present for only sarily exclude the use of fossil evidence of one other nesting oviraptorid, IVPP V9608. behavior in comparative studies if they offer Because several skeletal elements are at or reliable evidence of particular behaviors. below the level of the eggs, Dong and Currie Among extant oviparous reptiles parental (1996) infer that the nest was not buried. care of eggs after deposition is common and They also infer from the position of the feet, taxonomically widespread only in archo- in the center of the nest but at the same level saurs. Parents remain close to their eggs after as the top of the eggs, that the center of the they are laid, in some cases in direct contact nest was filled in. Although the skeleton is with them, in birds (Kendeigh, 1952), croc- incomplete, the preserved portions are in po- odylians (Magnusson et al., 1989; Thorbjar- sitions similar to those of IGM 100/979, in- narson, 1996), and some squamates (Tinkle dicating a symmetrical posture. and Gibbons, 1977; Shine, 1988) but not in The symmetrical posture exhibited by turtles (Shine, 1988) or Sphenodon (Moffat, IGM 100/979 and by IVPP V9608 is similar 1985). Because parental care is pervasive in to the position birds take when they sit on mammals (although only one group is ovip- nests (Skutch, 1976; Campbell and Lack, arous) and is common and widespread 1985). Other oviraptorid skeletons from the among amphibians (Duellman and Trueb, structureless sandstones of Ukhaa Tolgod 1994), it may be primitive for the Tetrapoda. that are not on nests are not in this position, Optimizing these conditions to a clado- lying instead on their sides with the limbs gram of tetrapod relationships (fig. 15), the extended in various positions (e.g., IGM 199/ condition in Squamata is critical to determin- 1002). Indeed, we are unaware of any non- ing whether parental care of eggs was lost in avialan dinosaur specimens other than IGM Reptilia and then re-evolved in some Squa- 100/979 and IVPP V9608 preserved in pre- mata and in Archosauria or whether instead cisely this posture. it was lost independently in turtles, rhyncho- The similarity between the posture of cephalians, and within Squamata. If the dis- these two specimens with the posture of birds tribution of this feature among squamates is clear, but the important question is whether implies that it evolved in the common an- they are homologous-is this condition "the cestor of the entire group, then the two al- same"l as a result of their close evolutionary ternatives just mentioned are equally parsi- relationship? The inference of homology be- monious. If, however, it is simpler to infer tween behaviors shared among living organ- that parental care was not present in the com- isms is no more problematic than is the in- mon ancestor of squamates then parental care terpretation of morphological homologues is most parsimoniously considered to have 28 AMERICAN MUSEUM NOVITATES NO. 3265

NXA. 0. 'lip,IV 6\11". -<..ePsl..31 0

Parental + _ _ _ care "

Brooding + _

7 parents brood eggs

parents do not brood eggs Fig. 15. Cladogram of relationships among extant Reptilia illustrating character distributions and inferences of homology discussed in the text, and the level at which they are synapomorphous (under DELTRAN optimization). Relationships are supported by the analysis of Eernisse and Kluge (1993). Above: Presence or absence of parental care of eggs after deposition. Note that this assumes the condition in Squamata to be variable; it is one of two equally parsimonious optimizations if brooding is inferred to be plesiomorphic for Squamata, the other being separate losses of brooding behavior in Testudines, Rhynchocephalia, and some Squamata. Below: Presence or absence of direct contact between parent and eggs after deposition (i.e., brooding). Parentheses indicate the occurrence in a group of a condition that is most simply interpreted as secondarily derived. Conditions refer only to oviparous taxa (i.e., monotremes, all testudines, both species of Rhynchocephalia, most squamates, and all archosaurs). Sources: Shine, 1988; Moffat, 1985; Collias and Collias, 1984.

been lost in the common ancestor of Reptilia 1988). Unfortunately, because of the cryptic and re-evolved in the Archosauria (fig. 14). behavior of most squamates it is unclear Among Squamata, parental care of eggs is whether the absence of reports of parental known to occur in Iguaninae, the scincid ge- care in the other 98% of squamate species nus Eumeces, Anguidae, the teiid genus Tup- (Shine, 1988) is due to the absence of this inambis, Varanidae, Boidae, Colubridae, and behavior or to insufficient opportunities for Elapidae (Tinkle and Gibbons, 1977; Shine, observation of these species. Taken at face 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 29 value (i.e., assuming brooding is absent in eggs with the body is not a consistent part of those groups for which it has not been re- the behavioral repertoire. ported), the distribution of brooding behavior Among birds, with a few interesting ex- in squamates implies that it evolved separate- ceptions (e.g., megapodes), one or both par- ly in each of these groups. However, the ac- ents habitually sit directly on their eggs, and, tual distribution of this behavior among squa- again with only a few interesting exceptions mates is undoubtedly broader. It is therefore (e.g., emperor penguins), take a characteristic uncertain whether parental care is plesiom- posture over the nest (Campbell and Lack, orphic for Squamata, and this is how we have 1985). The body is centered over the nest treated this group in optimizing this character with the hind legs folded beneath, the ab- to the relationships of amniotes (fig. 14). domen contacts the eggs broadly, and often If parental care is a synapomorphy of ar- the forelimb is folded back along the sides chosaurs, the crocodylian condition might of the body. In extant birds this behavior usu- mistakenly be considered to represent the ally involves a brooding patch, a highly vas- primitive condition for the group. But the cularized area on the abdomen over which primitive condition for Archosauria compris- the feathers are shed and body heat is trans- es only those features of parental care com- ferred to the eggs more efficiently than else- mon to both crocodylians and birds that are where on the body. Brooding behavior is pre- absent in outgroups. Thus, crocodylian pa- sent in tinamous and ratites (although the lat- rental care may include specializations of ter lack a brooding patch) as well as in the Crocodylia, just as parental care of birds in- putatively most basal neognath taxa (Camp- volves specializations of Aves identifiable by bell and Lack, 1985), and is therefore prim- their absence in crocodylians and other avian itive for modem birds. outgroups. Although it is possible that oviraptorid A specialized form of parental care is the nests were buried and occasionally uncov- brooding of eggs, in which the parent brings ered (a behavior exhibited by the Egyptian its body into direct contact with the eggs for plover; Howell, 1979), there is no compel- prolonged periods of time. In the two archo- ling evidence that any oviraptorid nest was saur groups, only bird parents are known to ever completely buried. It is unlikely that ev- habitually brood their eggs, typically by rest- idence is forthcoming for specimens without ing directly on top of the nest (Skutch, 1976; nesting adults, given the difficulty of distin- Campbell and Lack, 1985; Gill, 1995). Croc- guishing the sediments in which the nest was odylian eggs are instead buried en masse ei- deposited from those that later came to bury ther within a convex mound or a hole (Greer, them. In any case, because open nests are 1970; Thorbjarnarson, 1996) covered by sed- present in basal ratites and neognaths (Col- iment that in some cases is mixed with plant lias and Collias, 1984, appendix 1) and the debris. Although crocodylian parents (most oviraptorid nest provides evidence that in commonly the female) often stay near the this group the nest was open at least some of nest and, in some species at least, often lie the time during the brooding period, it cor- on the side of the nest (Cott, 1961, 1971), roborates the hypothesis that an open nest is they are not reported to lie directly on eggs the plesiomorphic condition for extant Aves. (which are buried) or with the body centered The presence of two oviraptorid adults di- on the nest (Magnusson et al., 1989). There rectly on nests suggests that adults of this is, thus, no evidence for brooding, or even taxon habitually sat on nests, but it does not for a consistent position or orientation of the necessarily imply that they were endothermic adult's body relative to the nest. Crocodylian and provided heat to the eggs on a regular behavior is difficult to study in the wild, and basis, as birds do through their brooding the possibility that this particular behavior patch. There is no evidence available to test occurs occasionally cannot be dismissed en- this hypothesis, and the assumption that the tirely, but sufficient field studies and numer- behavior associated with brooding and the ous observations of captive animals indicate act of brooding itself was correlated in ovi- that in crocodylians, unlike in birds, sitting raptorids is just that-an untestable assump- on the center of the nest and covering the tion. 30 AMERICAN MUSEUM NOVITATES NO. 3265

It has been suggested that the posture in position between pairs (Dong and Currie, this specimen is similar to that occasionally 1996). Although this pattern suggests conti- taken by crocodylians (Geist and Jones, nuity in the process of egg deposition, and is 1996), but few similarities are apparent on plausible, it is also possible that the mother close comparison. Indeed, the differences be- (or mothers) simply aimed well when they tween the many different postures taken by returned to the nest over time. It is again un- crocodylians and that of IGM 100/979 are so clear that any evidence is available that could obvious-the latter lies over the center of the refute either of these hypotheses. nest in direct contact with eggs and with its The behavior of egg turning, in which legs folded beneath it and its arms spread birds manipulate the eggs after they have around the perimeter of the nest-that this been laid, is common to nearly all living suggestion is easily dismissed. birds and, as far as is known, to all that di- The decidedly nonrandom pattern in rectly brood their eggs (Campbell and Lack, which the eggs within oviraptorid nests are 1985; Deeming, 1991). It has been suggested arranged also provides evidence of the be- that the apparent partial burial of the eggs in havior involved in their deposition. The pair- oviraptorid nests "preclude[s] the possibility ing of eggs within nests of this type (i.e., of egg rotation as in birds" (Varicchio et al., elongatoolithid) strongly suggests that each 1997: 249), and indicates the absence of a pair was laid simultaneously using both ovi- structure-chalazea, fibers that support the ducts (Varicchio et al., 1997). The arrange- embryo within the egg-correlated with this ment of the pairs of eggs is also highly struc- behavior. As discussed above, it is uncertain tured, much more so than the positions of to what extent the eggs were buried while the eggs in crocodylian nests. The pairs of eggs adult was alive, but in any case it is unclear are spaced around the perimeter of a circle why burial of the eggs necessarily precludes in two layers, with a central area devoid of periodic unburying and rotation. The repeat- eggs. We considered the circular arrangement ed burial of eggs by the Egyptian plover of the eggs to be evidence of manipulation (Howell, 1979) demonstrates the possibility by an adult after deposition (Norell et al., of this behavior, and although the large num- 1995), but alternatively it could be the result ber of eggs in oviraptorid nests would require of highly precise positioning by the mother, a great deal of activity to rotate all of them, standing in the middle of the nest, during the there seems no reason to consider egg turn- egg laying process (Dong and Currie, 1996). ing to have been impossible or improbable It is unclear if there is any evidence that in the absence of information about ovirap- could falsify either of these hypotheses. torid metabolism and behavior. Indeed, It has been suggested that the pairing of which of the more unusual behaviors exhib- eggs in oviraptorid nests is evidence either ited by living animals would be deemed un- that they were laid at intervals of one day or likely if only their effects were known? more ("monoautochronic ovulation") or in a Stronger evidence against egg rotation is much shorter period of time, but there is no the pairing of the eggs, because rotation of strong evidence against either alternative. the eggs would presumably disrupt this pat- Deposition of the eggs on more than one day, tern, assuming it to have been created when as in birds, was advocated by Varicchio et al. the eggs were lain. It is of course possible (1997), who cited in support the pairing of that the eggs were laid so soon before their eggs in lizards and an abnormal duck that terminal burial that rotation had not yet taken laid pairs of eggs over long time intervals. It place, or that the pairing was due to precise is unclear why the pairing of eggs necessarily parental manipulation after egg turning, so at precludes more rapid deposition, however, best the hypotheses of paired oviducts and given our lack of information about the re- egg turning are only weakly tested by the productive biology of oviraptorids. The al- observation of egg pairs. ternative scenario, in which the eggs were There remain many obvious desiderata for laid at one time, is supported mainly by a understanding these fossils. In addition to the nest in which the pairs of eggs are arranged unanswered questions just surveyed, we do in a spiral pattern, with gradual changes in not know the sex of the adults on the nests, 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 31 their precise genealogical relationship to the versity systematics discussion group for eggs (e.g., whether more than one set of par- fruitful discussion. We thank Dong Zhi- ents contribute to the nest), how often and Ming for access to the Bayan Mandahu ovi- for how long adults attended the nest, what raptorid, and C. Brochu and D. Weishampel was the gestation period of the eggs, or how for their thoughtful reviews. The histologi- long hatchlings stayed in the nest. It is tempt- cal preparations of the ribs were prepared ing to speculate, but a more fruitful approach by Marco van Gemeran, Mineral Optics is to continue searching the sands of Ukhaa Laboratory, and we thank the AMNH De- Tolgod and elsewhere for specimens that will partment of Mineral and Planetary Sciences provide evidence that might address these for the use of their microscope. The fossil questions. was skillfully prepared by Amy Davidson, and the illustrations are by Michael Ellison. ACKNOWLEDGMENTS Funding for this work was provided by NSF grant DEB 9407999, the Frick Laboratory We thank Hans-Dieter Sues, Chris Mc- endowment, IREX, and Richard, Lynnette, Gowan, John Wenzel, Lowell Dingus, David and Byron Jaffe. Finally, we thank the other Loope, Michael Novacek, Darrel Frost, and members of the 1993 field crew of the Mon- Anusuya Chinsamay for advice on this golian Academy of Sciences-American Mu- work, and M. Kearney, E. Strong, and other seum of Natural History Expedition for their participants in the George Washington Uni- help in collecting the fossil.

REFERENCES Andrews, R. C. Bray, E. S., D. K. Zelenitsky, K. F Hirsch, and 1932. The new conquest of Central Asia. M. A. Norell New York: Am. Mus. Nat. Hist. In press. Correlation of eggshell ratite mor- Averianov, A. 0. photype with dinosaurs based upon the 1997. New late Cretaceous mammals of find of a Late Cretaceous embryonic southern Kazakhstan, Acta Paleontol. oviraptorid. Am. Mus. Novitates. Pol. 42(2): 243-256. Campbell, B., and E. Lack (eds.) Barsbold, R. 1985. A dictionary of birds. Vermillion, SD: 1981. Toothless dinosaurs of Mongolia. Joint Buteo Books. Soviet-Mongolian Paleontol. Exped. Chiappe, L. M. Trans. 15: 28-39 [in Russian]. 1995. The first 85 million years of avian evo- 1983. Carnivorous dinosaurs from the Creta- lution. Nature 378: 349-355. ceous of Mongolia. Ibid. 19: 1-120 [in 1997. Climbing Archaeopteryx? A response Russian]. to Yalden. Archaeopteryx 15: 109-112. Barsbold, R., T. Maryanska, and H. Osmolska Chiappe, L. M., Q. Ji, S.-A. Ji, and M. A. Norell In prep. Anatomy and systematics of the 1990. Oviraptorosauria. In D. B. Weishampel, Confuciusornithidae (Aves) from the P. Dodson, and H. Osm6lska (eds.), The Mesozoic of northeastern China. Am. Dinosauria: 249-258. Berkeley: Univ. Mus. Novitates California Press. Chure, D. J., and J. H. Madsen, Jr. Baumel, J. J., and L. M. Witmer 1996. On the presence of furculae in some 1993. Osteologia. In J. J. Baumel, A. S. King, non-maniraptoran theropods. J. Vertebr. J. E. Breazile, H. E. Evans, and J. C. Paleontol. 16: 573-577. Vanden Berge (eds.), Handbook of avi- Clark, J. M., A. Perle, and M. Norell an anatomy: Nomina anatomica avium: 1994. The skull of Erlicosaurus(sic) andrew- 45-132. 2nd ed. Publ. Nuttall Ornithol. si, a Late Cretaceous "segnosaur" Club 23. (Theropoda: Therizinosauridae) from Bellairs, A. d'A., and C. R. Jenkin Mongolia. Am. Mus. Novitates 3113: 1960. The skeleton of birds. In A. J. Marshall 39 pp. (ed.), Biology and comparative physi- Collias, N. E., and E. C. Collias ology of birds 1: 241-300. New York: 1984. Nest building and bird behavior. Academic Press. Princeton, NJ: Princeton Univ. Press. 32 AMERICAN MUSEUM NOVITATES NO. 3265

Coombs, W. P., Jr., and T Maryanska Dong, Z.-M., and P. J. Currie 1990. Ankylosauria. In D. B. Weishampel, P 1996. On the discovery of an oviraptorid Dodson, and H. Osmolska (eds.), The skeleton on a nest of eggs at Bayan Dinosauria: 456-483. Berkeley: Univ. Mandahu, Inner Mongolia, People's California Press. Republic of China. Can. J. Earth Sci. Cott, H. 33(4): 631-636. 1961. Scientific results of an inquiry into the Drummond, H. ecology and economic status of the 1981. The nature and description of behavior Nile Crocodile (Crocodilus niloticus) in patterns. In P. G. Bateson and P. H. Uganda and northern Rhodesia. Trans. Klopfer (eds.), Perspectives in ethology Zool. Soc. London 29: 211-357. 4: 1-33. New York: Plenum Press. 1971. Parental care in Crocodilia, with special Duellman, W. E., and L. R. Trueb reference to Crocodylus niloticus. Pro- 1994. Biology of amphibians. Baltimore: ceedings of the First Working Meeting Johns Hopkins Univ. Press. of Crocodile Specialists. IUCN Publ. Eberth, D. A. N. ser. 32: 166-180. 1993. Depositional environments and facies Cracraft, J. transitions of dinosaur-bearing Upper 1976. The species of moas (Aves, Dinorni- Cretaceous redbeds at Bayan Mandahu thidae). Smithsonian Contrib. Paleo- (Inner Mongolia, People's Republic of biol. 27: 189-205. China). Can. J. Earth Sci. 30(10-11): Currie, P. J. 2196-2213. 1990. Elmisauridae. In D. B. Weishampel, P. Eernisse, D., and A. G. Kluge Dodson, and H. Osmolska (eds.), The Dinosauria: 245-248. Berkeley: Univ. 1993. Taxonomic congruence versus total ev- California Press. idence, and amniote phylogeny inferred 1995. New information on the anatomy and from fossils, molecules, and morphol- relationships of Dromaeosaurus alber- ogy. Mol. Biol. Evol. 10: 1170-1195. tensis (Dinosauria: Theropoda). J. Ver- Feduccia, A. tebr. Paleontol. 15: 576-591. 1993. Evidence from claw geometry indicat- Currie, P. J., S. J. Godfrey, and L. Nessov ing arboreal habits of Archaeopteryx. 1993. New caenagnathid (Dinosauria: Thero- Science 259: 790-793. poda) specimens from the Upper Cre- Furbringer, M. taceous of North America and Asia. 1888. Untersuchungen zur Morphologie und Can. J. Earth Sci. 30: 2255-2272. Systematik der Vogel. Specieller Theil. Currie, P. J., and X.-J. Zhao Amsterdam: T.J. Van Holkema. 1993. A new carnosaur (Dinosauria, Thero- Galton, P. M. poda) from the Jurassic of Xinjiang, 1990. Basal Sauropodomorpha-prosauro- People's Republic of China. Can. J. pods. In D. B. Weishampel, P. Dodson, Earth Sci. 30: 2037-2081. and H. Osmolska (eds.), The Dinosau- Dashzeveg, D., M. J. Novacek, M. A. Norell, J. ria: 320-344. Berkeley: Univ. Califor- M. Clark, L. M. Chiappe, A. R. David- nia Press. son, M. C. McKenna, L. Dingus, C. C. Gao, K., and M. A. Norell Swisher III, and A. Perle. 1995. A rich lizard assemblage from Ukhaa 1995. Unusual preservation in a new verte- Tolgod, Gobi Desert, Mongolia. J. Ver- brate assemblage from the Late Creta- tebr. Paleontol. 16, suppl. to 3: 36A. ceous of Mongolia. Nature 374: 446- Gauthier, J. A. 449. 1986. Saurischian monophyly and the origin Deeming, D. C. of birds. In K. Padian (ed.), The origin 1991. Reasons for the dichotomy in egg turn- of birds and the evolution of flight: 1- ing in birds and reptiles. In D. C. 56. San Francisco: California Acad. Deeming and M.W.J. Ferguson (eds.), Sci. Egg incubation: 307-324. New York: Geist, N. R., and T. D. Jones Cambridge Univ. Press. 1996. Juvenile skeletal structure and the re- de Beer, G. R. productive habits of dinosaurs. Science 1954. Description of the British Museum 272: 712-714. specimen of Archaeopteryx and com- Gill, E B. parison with the Berlin specimen. Lon- 1995. Ornithology. New York: W.H. Free- don: Trustees of the Br. Mus. man. 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 33

Greene, H. W. Kendeigh, S. C. 1994. Homology and behavioral repertoires. 1952. Parental care and its evolution in birds. In B. K. Hall (ed.), Homology. The hi- Illinois Biol. Monogr. 22: 1-356. erarchical basis of comparative anato- Lambe, L. M. my: 369-391. New York: Academic 1917. The Cretaceous theropodous dinosaur Press. Gorgosaurus. Geol. Surv. Can., Mem. Greer, A. 100: 1-84. 1970. Evolutionary and systematic signifi- Lauder, G. V., Jr. cance of crocodilian nesting habits. Na- 1995. On the inference of function from ture 227: 523-524. structure. In J. J. Thomason (ed.), Hoffstetter, R., and J.-P. Gasc Functional morphology in vertebrate 1969. Vertebrae and ribs of modem reptiles. paleontology: 1-18. New York: Cam- In C. Gans, A. d'A. Bellairs, and T. S. bridge Univ. Press. Parsons (eds.), Biology of the Reptilia Lee, K., J. Feinstein, and J. Cracraft 1: 201-310. New York: Academic 1997. The phylogeny of ratite birds: resolving Press. conflicts between molecular and mor- Holtz, T. R., Jr. phological data sets. In D. P. Mindell 1994. The phylogenetic position of the Tyr- (ed.), Avian molecular evolution and annosauridae: implications for theropod systematics: 173-211. New York: Ac- systematics. J. Paleontol. 68: 1100- ademic Press. 1117. Lillegraven, J. A., and M. C. McKenna 1996. Phylogenetic taxonomy of the Coelu- 1986. Fossil mammals from the "Mesaver- rosauria (Dinosauria: Theropoda). Ibid. de" Formation (Late Cretaceous, Judi- 70: 536-538. thian) of the Bighom and Wind River Hou, L., and J. Zhang basins, Wyoming, with definitions of 1993. A new fossil bird from the Lower Cre- Late Cretaceous North American land- taceous of China. Vertebr. PalAsiat. mammal "ages." Am. Mus. Novitates 31(3):217-224. [in Chinese, English 2840: 68 pp. summary] Loope, D. B., L. Dingus, C. C. Swisher III, and Howell, T. R. C. Minjin 1979. Breeding biology of the Egyptian Plo- 1998. Life and death in a Late Cretaceous ver, Pluvianus aegyptius. Univ. Califor- dunefield, Nemegt Basin, Mongolia. nia Pub. Zool. 113: 1-76. Geology 26: 27-30. Howes, G. B., and H. H. Swinnerton MacIntosh, J. S. 1901. On the development of the skeleton of 1990. Sauropoda. In D. B. Weishampel, P. the tuatara, Sphenodon punctatus, with Dodson, and H. Osmolska (eds.), The remarks on the egg, on the hatching, Dinosauria: 345-401. Berkeley: Univ. and on the hatched young. Trans. Zool. California Press. Soc. London 16(1):1-87. Magnusson, W. E., K. A. Vliet, A. C. Pooley, and Jerzykiewicz, T, P. J. Currie, D. A. Eberth, P. A. R. Whitaker Johnston, E. H. Koster, and J.-J. Zheng 1989. Reproduction. In C.A. Ross (ed.), 1993. Djadokhta Formation correlative strata Crocodiles and alligators: 118-135. in Chinese Inner Mongolia: an over- New York: Facts on File. view of the stratigraphy, sedimentary Makovicky, P. J., and P. J. Currie geology, and paleontology and compar- 1998. The presence of a furcula in tyranno- isons with the type locality in pre-Altai saurid theropods, and its phylogenetic Gobi. Can. J. Earth Sci. 30(10-11): and functional implications. J. Vertebr. 2180-2195. Paleontol. 18: 143-149. Jerzykiewicz, T, and D. A. Russell Makovicky, P. J., and H.-D. Sues 1991. Late Mesozoic stratigraphy and verte- 1998. Anatomy and phylogenetic relation- brates of the Gobi Basin. Cretaceous ships of the theropod dinosaur Micro- Res. 12: 345-377. venator celer from the Lower Creta- Johnston, P. A., D. A. Eberth, and P. K. Anderson ceous of Montana. Am. Mus. Novitates 1996. Alleged vertebrate eggs from Upper 3240: 27 pp. Cretaceous redbeds, Gobi Desert, are Marsh, 0. C. fossil insect (Coleoptera) pupal cham- 1880. Odontornithes: a monograph of the ex- bers: Fictovichnus new Ichnogenus. tinct toothed birds of North America. Can. J. Earth Sci. 33(4):511-525. Washington, D.C.: U.S. Gov. 34 AMERICAN MUSEUM NOVITATES NO. 3265

McLelland, J. Oliver, W. R. B. 1989. Anatomy of the lungs and air sacs. In 1949. The moas of New Zealand and Austra- A. S. King and J. McLelland (eds.), lia. Dom. Mus. Bull. 15: 1-206. Form and function in birds, Vo. 4: 221- Osbom, H. F 279. New York: Academic Press. 1924. Three new Theropoda, Protoceratops Mikhailov, K. E. zone, central Mongolia. Am. Mus. 1991. Classification of fossil eggshells of am- Novitates 144: 12 pp. niotic vertebrates. Acta Palaeontol. Pol. Osmolska, H. 36: 193-238. 1981. Coossified tarsometatarsi in theropod Mikhailov, K. E., K. Sabath, and S. Kurzanov dinosaurs and their bearing on the 1994. Eggs and nests from the Cretaceous of problem of bird origins. Palaeontol. Mongolia. In K. Carpenter, K. Hirsch, Pol. 42: 79-95. and J. Homer (eds.), Dinosaur eggs and Osmolska, H., and R. Barsbold babies, 88-115. New York: Cambridge 1990. Troodontidae. In D. B. Weishampel, P. Univ. Press. Dodson, and H. Osmolska (eds.), The Miller, E. H. Dinosauria: 259-268. Berkeley: Univ. 1988. Description of bird behavior for com- California Press. parative purposes. Current Ornithol. 5: Ostrom, J. H. 347-394. 1969. Osteology of Deinonychus antirrhopus, Moffat, L. A. an unusual theropod from the Lower 1985. Embryonic development and aspects of Cretaceous of Montana. Peabody Mus. reproductive biology in the tuatara, Nat. Hist. Yale Univ. Bull. 30: 1-165. Sphenodon punctatus. In C. Gans, E 1976. Archaeopteryx and the origin of birds. Billett, and P. E Maderson (eds.), Bi- Biol. J. Linn. Soc. 8: 91-182. ology of the Reptilia, Vol. 14, Devel- Owen, R. opment A, 493-522. New York: Wiley. 1879. Memoirs on the extinct wingless birds Norell, M. A., and J. M. Clark of New Zealand, with an appendix on 1997. Birds are dinosaurs. Science Spectra 8: those of England, Australia, New- 28-34. foundland, Mauritius and Rodriguez. 2 Norell, M. A., and P. Makovicky vols. London: John van Voorst. 1997. Important features of the dromaeosaur Parker, W. K. skeleton: information from a new spec- 1868. A monograph on the structure and de- imen. Am. Mus. Novitates 3215: 28 pp. velopment of the shoulder-girdle and In press. Important features of the dromaeo- sternum. London: R. Hardwicke. saur skeleton H: information from new Paul, G. specimens of Velociraptor mongolien- 1988. Predatory dinosaurs of the world. New sis. Am. Mus. Novitates. Norell, M. A., J. M. Clark, D. Dashzeveg, R. York: N.Y. Acad. Sci. Barsbold, L. M. Chiappe, A. R. David- Perle A., L. M. Chiappe, R. Barsbold, J. M. Clark, son, M. C. McKenna, and M. J. No- and M. A. Norell vacek 1994. Skeletal morphology of Mononykus 1994. A theropod dinosaur embryo, and the olecranus (Theropoda: Avialae) from affinities of the Flaming Cliffs dinosaur the Late Cretaceous of Mongolia. Am. eggs. Science 266: 779-782. Mus. Novitates 3105: 29 pp. Norell, M. A., J. M. Clark, L. M. Chiappe, and Peters, S. D. D. Dashzeveg 1996. Ein nahezu vollstandiges Skelett eines 1995. A nesting dinosaur. Nature 378: 774- urtumlichen Vogels aus China. Nat. 776. Mus. 126(9): 298-302. Norell, M. A., P. Makovicky, and J. M. Clark Rieppel, 0. 1997. A Velociraptor wishbone. Nature 389: 1994. Homology, topology, and typology: 447. The history of modem debates. In B.K. Norman, D. B. Hall (ed.), Homology. The hierarchical 1980. On the ornithischian dinosaur Iguan- basis of comparative anatomy: 64-101. odon bernissartensis from the Lower New York: Academic Press. Cretaceous of Bernissart (Belgium). Romer, A. S. Inst. R. Sci. Nat. Belg. Mem. 178: 1- 1956. Osteology of the reptiles. Chicago: 103. Univ. Chicago Press. 1999 CLARK ET AL.: NESTING OVIRAPTORID FROM MONGOLIA 35

Ruben, J. A., T D. Jones, N. R. Geist, and W. J. Thorbjarnarson, J. Hillenius 1996. Reproductive characteristics of the Or- 1997. Lung structure and ventilation in the- der Crocodylia. Herpetologica 52: 8- ropod dinosaurs and early birds. Sci- 24. ence 278: 1267-1270. Tinkle, D. W., and J. W. Gibbons Russell, D. A., and Z.-M Dong 1977. The distribution and evolution of vivi- 1993. A nearly complete skeleton of a new parity in reptiles. Misc. Pub. Mus. troodontid dinosaur from the People's Zool., Univ. Mich. 154: 1-54. Republic of China. Can. J. Earth Sci. Varicchio, D. J., F Jackson, J. Borkowski, and J. 30(10-11): 2163-2173. R. Homer Sabath, K. 1997. Nest and egg clutches of the dinosaur 1991. Upper Cretaceous amniotic eggs from Troodon formosus and the evolution of the Gobi Desert. Acta Palaeontol. Pol. avian reproductive traits. Nature 385: 36: 151-192. 247-250. Sanz, J. L., L. M. Chiappe, and A. D. Buscalioni Webster, D. 1995. The osteology of Concornis lacustrix 1996. Dinosaurs of the Gobi. Nat. Geog. (Aves: Enantiornithes) from the Lower Mag. 190(1): 70-89 Cretaceous of Spain and a reexamina- Weishampel, D. B., P. Dodson, and H. Osmolska tion of its phylogenetic relationships. (eds.) Am. Mus. Novitates 3133: 23 pp. Univ. Cali- Shine, R. 1990. The Dinosauria. Berkeley: 1988. Parental care in reptiles. In C. Gans and fornia Press. R. B. Huey (eds.), Biology of the Rep- Wellnhoffer, P. tilia, Vol. 16, Ecology B: 275-329. 1993. Das siebte Exemplar von Archaeopter- New York: Wiley. yx aus den Solnhofner Schichten. Ar- Skutch, A. F chaeopteryx 11: 1-47. 1976. Parent birds and their young. Austin: Wenzel, J. Univ. Texas. 1992. Behavioral homology and phylogeny. Sues, H.-D. Annu. Rev. Ecol. Syst. 23: 361-381. 1997. On Chirostenotes, a Late Cretaceous Zhao, Z. K. oviraptorosaur (Dinosauria: Theropo- 1975. The microstructure of the dinosaurian da) from western North America. J. eggshells of Nanxiong Basin, Guang- Vertebr. Paleontol. 17(4): 698-716. dong. Vertebr. PalAsiat. 7: 249-260. 36 AMERICAN MUSEUM NOVITATES NO. 3265

APPENDIX: ABBREVIATIONS USED IN lt lateral trabecula of sternum ILLUSTRATIONS lul left ulna lvr left ventral rib cm costal margin of sternum pl proximal phalange, digit I of right e egg manus fu furcula Ps pubic symphysis g gastralia rdr dorsal segment of right thoracic rib hc hypocleidium rf right femur i invertebrate burrow rfi right fibula is ischial symphysis rh right humerus it invertebrate trace rmcl-3 metacarpals of right manal digits I-III kc keratinous claw remnant rmdl-III right manus digits 1-111 la left astragalus rmtl-4 metatarsals of right pedal digits I-IV lca left calcaneum lfi left fibula rpdl-IV right pedal digits I-IV lh left humerus rr right radius lmdI-III left manus digits I-III rs right sternal plate lmtl-4 metatarsals of left pedal digits I-IV rsc right scapula lpd I-IV left pedal digits I-IV rt right tibia lr left radius ru right uncinate process ls left sternal plate rul right ulna lsc left scapula rvr ventral segment of right thoracic rib lslc left semilunate carpal v vertebra

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last five years are available at World Wide Web site http://nimidi.amnh.org. Or address mail orders to: American Museum of Natural History Library, Central Park West at 79th St., New York, NY 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

G) This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).