Involved in writing sections of the paper: PCS, RNM, JAW, DJV, OAA, HCEL. Participants in the fieldwork: PCS, RNM, JAW, DJV, OAA, HCEL.
The authors have declared that no competing interests exist.
Living birds possess a unique heterogeneous pulmonary system composed of a rigid, dorsally-anchored lung and several compliant air sacs that operate as bellows, driving inspired air through the lung. Evidence from the fossil record for the origin and evolution of this system is extremely limited, because lungs do not fossilize and because the bellow-like air sacs in living birds only rarely penetrate (pneumatize) skeletal bone and thus leave a record of their presence.
We describe a new predatory dinosaur from Upper Cretaceous rocks in Argentina,
We present a four-phase model for the evolution of avian air sacs and costosternal-driven lung ventilation based on the known fossil record of theropod dinosaurs and osteological correlates in extant birds:
(1) Phase I—Elaboration of paraxial cervical air sacs in basal theropods no later than the earliest Late Triassic.
(2) Phase II—Differentiation of avian ventilatory air sacs, including both cranial (clavicular air sac) and caudal (abdominal air sac) divisions, in basal tetanurans during the Jurassic. A heterogeneous respiratory tract with compliant air sacs, in turn, suggests the presence of rigid, dorsally attached lungs with flow-through ventilation.
(3) Phase III—Evolution of a primitive costosternal pump in maniraptoriform theropods before the close of the Jurassic.
(4) Phase IV—Evolution of an advanced costosternal pump in maniraptoran theropods before the close of the Jurassic.
In addition, we conclude:
(5) The advent of avian unidirectional lung ventilation is not possible to pinpoint, as osteological correlates have yet to be identified for uni- or bidirectional lung ventilation.
(6) The origin and evolution of avian air sacs may have been driven by one or more of the following three factors: flow-through lung ventilation, locomotory balance, and/or thermal regulation.
The respiratory tract of birds has an elaborate series of pneumatic (air-filled) outgrowths that include
All pneumatic spaces are paired except the clavicular air sac, and the lungs are shaded.
In this paper, we describe a new large-bodied theropod from the Late Cretaceous of Argentina,
The new theropod is particularly interesting for another reason; it represents a previously unrecorded lineage of large-bodied predator in the early Late Cretaceous (Santonian, ca. 84 Ma) of South America. Most large-bodied Cretaceous theropods on southern continents (South America, Africa, Madagascar, India) pertain to one of three distinctive and contemporary clades: abelisaurids
The purpose of the present paper is not to determine the precise phylogenetic position of
Avian air sacs arise directly from the lungs (
The remaining air sacs are involved in lung ventilation and include the median
The cervical air sacs lie alongside the vertebral column in living birds and sometimes invade the vertebrae and ribs via pneumatopores of variable size. Invaginated pneumatic spaces that enter the lateral aspect of the centra, called
For more than a century, paleontologists have agreed on the existence of cervical air sacs, or some comparable pneumatic structure, in saurischian dinosaurs on the basis of pleurocoels which closely resemble those in some living birds. The longstanding question is the fossil record of noncervical air sacs—sacs that are involved in avian lung ventilation.
Ventilatory air sacs, unfortunately, are the least likely to leave evidence of their presence in skeletal bone. Paleontologists and comparative anatomists, as a result, have focused on axial pneumaticity, and opinion has split as to the meaning of observed patterns. In extant birds, pneumatic invasion by cervical air sacs is usually restricted to the cervical and anterior thoracic vertebrae and their respective ribs
We are inclined to support the latter, more conservative interpretation that pleurocoels in nonavian dinosaurs are a product of paraxial cervical air sacs and provide, at best, ambiguous evidence for intrathoracic ventilatory air sacs. First, pleurocoels are rare in birds, and no living bird has an unbroken cervical-to-caudal series of pleurocoels as occurs in some nonavian dinosaurs, including the one we describe below
Second, cervical air sacs have been observed extending to the posterior end of the vertebral column in birds. Several authors have described cervical air sacs extending posteriorly beyond the abdominal air sacs in the ostrich (
Finally, the posterior portion of the avian axial skeleton is completely transformed by extensive coossification of vertebrae and girdle bone and by posterolateral rotation of the pubes away from the midline, which allows unobstructed distal extension of the viscera and abdominal air sac under the synsacrum and caudal vertebrae. The pelvic space in nonavian saurischians, in contrast, is not nearly as open and continuous with the thoracic cavity, offering comparably less space for these air sacs
More conclusive evidence of intrathoracic ventilatory air sacs in fossils, in sum, will require an osteological signature of pneumatic structures in nonaxial (appendicular) bones, which that cannot be dismissed as an elaboration of the paraxial cervical air sacs. We now turn attention to previous reports of such pneumaticity in the appendicular skeleton of nonavian dinosaurs.
Appendicular pneumaticity in nonavian dinosaurs is poorly documented. An ilium of the Middle Jurassic theropod
A deep depression on the brevis shelf on an ilium of the Cretaceous carcharodontosaurid theropod
Recently, the furcula in the Early Cretaceous dromaeosaurid
“Cancellous” bone was described in the ilia of the titanosaurs
In sum, the evidence currently documenting the presence of nonaxial pneumatic structures in nonavian dinosaurs is poorly established. Several years ago, we reported the pneumatic spaces in the furcula and ilium described below
The holotypic and referred specimens were prepared using pin vice, pneumatic air scribe, and air abrasive. The white-colored bones are embedded in a fine-grained, poorly sorted, hematitic siltstone matrix that in places was very hard. To reduce color distractions in photographic images between the white bone and red-brown matrix, some bones were molded and cast in matt grey-colored casting epoxy.
Computed tomography of the furcula, gastral elements, and ilium was undertaken to observe internal pneumatic spaces using a Philips Brilliance 64-slice scanner at 80 Kv in the University of Chicago Hospitals.
We employed traditional, or “Romerian,” anatomical and directional terms over veterinarian alternatives
FMNH | Field Museum of Natural History, Chicago, Illinois, United States of America. |
LH | Las Hoyas collection, Museo de Cuenca, Cuenca, Spain. |
MCNA | Museo de Ciencias Naturales y Antropológicas (J. C. Moyano) de Mendoza, Mendoza, Argentina. |
MNN | Muséum National du Niger, Niamey, République de Niger. |
Dinosauria Owen, 1842
Theropoda Marsh, 1881
Tetanurae Gauthier, 1986
Allosauroidea Marsh, 1878
Isolated crown from the maxillary or dentary series (MCNA-PV-3137; cast). (A)-Side view of crown. (B)-Enlarged view of crown tip. Scale bars equal 1 cm.
Right postorbital (MCNA-PV-3137; cast) in right lateral (A), medial (B), anterior (C), posterior (D), and dorsal (E) views. Scale bar equals 5 cm.
Left quadrate (MCNA-PV-3137; cast) in left lateral (A), medial (B), anterior (C), posterior (D), dorsal (E), and ventral (F) views. Scale bar equals 5 cm.
Atlas and cervical 3 centrum (MCNA-PV-3137; cast) in left lateral view. (A)-Atlas. (B)-Cervical 3 centrum. Scale bars equal 5 cm.
Cervical vertebrae 4 and 6 (MCNA-PV-3137; cast) in left lateral view. (A)-Cervical 4. (B)-Cervical 6. Scale bar equals 5 cm.
Dorsal vertebrae 1, 4 and 8 (MCNA-PV-3137; cast) in left lateral view. (A)-Dorsal 1. (B)-Dorsal 4. (C)-Dorsal 8. Scale bar equals 10 cm.
Dorsal vertebrae 11 and 14 (MCNA-PV-3137; cast) in left lateral view. (A)-Dorsal 11. (B)-Dorsal 14. Scale bar equals 10 cm.
Anterior and mid caudal centra (MCNA-PV-3137; cast) in left lateral view. (A)-Anterior caudal centrum. (B)-Mid caudal centrum. Scale bar equals 10 cm.
Coossified medial gastral elements from anterior end of cuirass (MCNA-PV-3137). (A)-Coossified gastralia (cast) in ventral view. (B)-Stereopairs of the medial portion of one gastralium showing the pneumatopore and lumen inside the shaft in ventrolateral view. Scale bars equal 10 cm in A and 5 cm in B.
Furcula (MCNA-PV-3137; cast) in anterior (A) and posterior (B) views. Scale bar equals 10 cm.
(A)-Stereopairs of the furcula of
Left ilium (MCNA-PV-3137; cast) in left lateral (A) and medial (B) views. Scale bar equals 20 cm.
Detail views of the left ilium (MCNA-PV-3137). (A)-Pneumatopores on the base of the preacetabular process in lateral view. (B)-Pneumatopores on the central iliac blade in medial view. (C)-Stereopairs of the pubic peduncle in lateral view showing pneumatopore complex. (D)-Pneumatopores on the brevis fossa of the postacetabular process in ventral view; largest pneumatopore (4 cm in transverse diameter) opens posteriorly (to the right) just posterior to five smaller ventrally-facing pneumatopores (marked). A, B, and D are from a cast of MCNA-PV-3137 to reduce color distraction. Scale bars equal 5 cm in A and 10 cm in B, C and D. Arrows point to pneumatopores.
Pubes (MCNA-PV-3137; cast) in left lateral (A) and anterior (B) views. Scale bar equals 20 cm.
(A)-Silhouette reconstruction in left lateral view showing preserved bones of the holotype and referred specimens (MCNA-PV-3137-3139); body length approximately 9-10 m. (B)-Left quadrate in posterior view. (C)-Dorsal 14 in left lateral view with enlarged cross-sections of the neural spine and transverse process. (D)-Furcula in anterior view with sagittal cross-section. (E)-Cross-section of medial gastral element from the anterior end of the cuirass showing pneumatocoel. (F)-Left ilium in lateral view with enlarged cross-section of pubic peduncle. Scale bars equal 5 cm in B, 10 cm (3 cm for cross-sections) in C, 10 cm (same for cross-section) in D, 2 cm in E, and 20 cm (6 cm for cross-section) in F.
MCNA-PV-3137; one maxillary or dentary crown, left prefrontal, right postorbital, left quadrate, posterior portion of the left pterygoid, right prearticular, partial or complete vertebrae including C1, C3, C4, C6, C8, D1, D4–11, D14, S2–5, CA1, one mid and distal caudal centrum, several cervical and dorsal ribs, gastralia, furcula, left scapulocoracoid, left ilium, and right and left pubes. The bones were found disarticulated but in close association over an area of 11 m2. The specimen is an immature individual with many open neurocentral sutures. The specimen, however, may have been approaching maturation, as indicated by fusion of the centrum and neural arch of several vertebrae and fusion of the coracoid and scapula. Body length is approximately 9–10 m, or roughly equivalent to the larger body size range for
Cañadon Amarillo (S 37.5°, W 70.5°), north of Cerro Colorado, 1 km north of the Río Colorado near the southern border of Mendoza Province, Argentina.
Anacleto Formation, Neuquén Group; Santonian (ca. 84 Mya)
These features distinguish
MCNA-PV-3138, left metatarsal 2; MCNA-PV-3139, articulated left tibia, fibula, astragalus and calcaneum lacking only the proximal end of the fibula. This material is tentatively referred to
We briefly comment below on the morphology of
Bone | Measurement | Length |
Crown | Height Base, mesiodistal length Base, maximum labiolingual width | 3.8 1.7 1.0 |
Prefrontal | Maximum length Maximum width | 6.8 2.3 |
Postorbital | Maximum height Maximum anteroposterior length | 11.5 11.4 |
Quadrate | Maximum height Head, anteroposterior length Head, transverse width Quadrate foramen, maximum height Quadrate foramen, maximum width Transverse width across condyles | 16.3 2.6 2.4 2.7 1.7 7.9 |
Vertebra | Centrum Length |
Posterior Centrum Height | Posterior Centrum Width | Pleurocoel Morphology |
C1 | 2.5 |
3.3 |
6.5 |
— |
C3 | 9.6 | 8.4 | 9.2 | Two; dorsal smaller, septum inclined anterodorsally |
C4 | 9.8 | 8.4 | 9.2 | Two; subequal, septum inclined anterodorsally |
C6 | 9.1 | 8.1 | 8.4 | Two; dorsal is rudimentary, septum inclined anterodorsally |
D1 | 8.5 | 8.8 | 9.4 | One; oval, large |
D4 | 7.1 | 8.7 | 8.8 | One; oval, medium-sized |
D8 | 8.8 | 10.8 | 9.8 | One; oval, medium-sized |
D10 | 8.4 | 13.1 | 11.3 | One; oval, medium-sized |
D11 | 8.4 | 12.1 | 11.1 | Two; posterior is rudimentary, septum inclined posterodorsally |
D14 | 10.2 | 15.0 | 13.5 | Two; subequal in size, septum inclined posterodorsally |
CA1 | 9.3 | 11.8 | 12.8 | One; large with recessed septae |
mid CA | 10.0 | 7.7 | 7.7 | Two; posterior is rudimentary |
Measured along ventral edge excluding anterior convexity of centrum when present.
Measurement pertains to intercentrum.
Intercentrum does not appear to be pneumatic; neural arch has single lateral pneumatocoel.
Right side has only one pleurocoel (left side shown in
Bone | Measurement | Length, Angle (cm, degrees) |
Scapulocoracoid | Coracoid, maximum height Posterior process (glenoid to tip of process) Scapular length Scapular blade, minimum width Scapular blade, distal width Height (glenoid to acromion) Acromion width Glenoid, maximum length | 27.6 10.0 57.0 7.6 11.0 21.0 9.5 8.7 |
Furcula | Maximum transverse width Intrafurcular angle | 25.2 120° |
Ilium | Iliac blade, maximum length Iliac blade, height above acetabulum Preacetabular process length Preacetabular process, height at base Preacetabular process, height at distal end Postacetabular process length Postacetabular process, height at mid length Pubic peduncle, anteroposterior length Pubic peduncle, maximum transverse width Acetabulum, maximum width | 76.8 30.0 16.8 24.4 33.5 28.5 25.0 16.9 8.1 10.3 |
Pubis | Length (acetabulum to foot) Iliac peduncle length Ischial peduncle length Pubic foot, maximum length Pubic foot, maximum width | 62.0 18.0 9.0 47.2 19.3 |
Pelvic outlet (between pubes) | Maximum width Minimum width between ischial peduncles Minimum width between iliac peduncles | 14.0 11.2 7.0 |
Of the preserved cranial bones (prefrontal, postorbital, quadrate, posterior pterygoid part, prearticular), the prefrontal, postorbital and quadrate are the most informative. The prefrontal is complete and lacks a long ventral process along the orbital margin, unlike most basal tetanurans such as
The triradiate postorbital has a short posterior process, a subtriangular ventral process, and a blunt medial process (
The quadrate (
All of the vertebrae exhibit camellate (honeycomb) internal structure in both the centrum and neural arch. The expression of a “hyperpneumatic” condition in the vertebrae is matched elsewhere by the marked pneumatic invasion of the quadrate, appendicular elements and gastralia. The heightened state of pneumatic invasion across these skeletal divisions suggests genetic control of the degree or general extent of pneumaticity
The atlas has a single slit-shaped pneumatopore in the center of the neural arch that opens dorsally (
The pneumatic features in dorsal 14 are extreme. The neural spine has a large central lumen (
There appear to be only five sacral vertebrae in
Three caudal vertebrae are preserved including caudal 1, a mid caudal centrum, and a distal caudal centrum (
Cervical and dorsal ribs have pneumatic fossae or pneumatocoels opening near the junction of the capitulum and tuberculum on the anterior side of the rib head. The most extreme condition occurs in the last dorsal vertebra, in which a pneumatic canal extends within the transverse process to the rib (
Gastralia are composed of medial and lateral elements. The suture between the medial and lateral elements is sinuous and long and in many cases is fused. Medial elements, with rare exception, fuse in the midline (
Several pneumatopores open from the external side of the gastralia into the body of the element, passing internally along the gastral shaft without septae. In one case, an oval pneumatopore opens into a space within the gastralium approximately 1 cm in diameter (
The scapulocoracoid has a subrectangular blade comparable in proportions to that in many allosauroids, such as
The furcula (wishbone) is V-shaped with a broad intrafurcular angle of 120° and epicleideal processes that arch slightly to each side (
The ilium has a deep preacetabular process, a tapering postacetabular process, and an arched brevis fossa (
The articulated pubes, in contrast, do not show any external pneumatopores and are composed of dense or cancellous bone (
The pelvic outlet is rather narrow dorsoventrally and transversely as in
The form of the pectoral and pelvic girdles in the holotypic skeleton clearly indicate that
Placing
Inferences about postcranial pneumaticity in
Anterior axial pneumaticity (cervical to anterior dorsal vertebrae plus their associated ribs) has long been attributed to cervical air sacs in saurischian dinosaurs like
In living birds, anterior and posterior axial pneumaticity develops during growth as diverticulae invade the axial skeleton from cervical air sacs, lung diverticulae, and caudal (abdominal) air sacs. These separate sources for axial pneumaticity can leave a pneumatic hiatus in the thoracic vertebrae, where the approaching diverticulae fail to fully anastomose. In
Wedel
Recently O'Connor proposed that axial pneumaticity in the abelisaurid theropod
Review of the posterior dorsal-to-sacral vertebral series in
The mid sacral vertebrae are preserved, and each has two fossae visible on the ventral aspect of the diapophysis
It appears, in sum, that the “reduction” and “enhancement” of pneumatic fossae between posterior dorsal and sacral vertebrae in
The situation in
O'Connor and Claessens
An extraordinary feature of
The external (ventral) position of the pneumatopores suggests that the pneumatic diverticulae lay in superficial tracts outside the gastral cuirass. It seems unlikely that pneumatic diverticulae would penetrated the ventral thoracic wall to access external pneumatopores, when entering the gastralia directly from their internal (dorsal) surface would be much easier. A plausible explanation may be that these ventral pneumatic tracts are part of a subcutaneous system, which is present to varying degrees in birds and is composed of diverticulae from cervical, clavicular, and abdominal air sacs
Pneumaticity in pectoral and pelvic girdles may provide a more straightforward signpost for ventilatory air sacs, because these bones are pneumatized exclusively by ventilatory sacs in living birds. The median pneumatocoel in the furcula (wishbone) of
The pneumatic openings on lateral and ventral aspects of the ilium in
Some of the iliac pneumatic sculpting, nevertheless, may have been associated with posterior extension of the cervical air sac system. In particular, the pneumatopores on the medial aspect of the iliac blade (
Tracking pneumatic patterns in the fossil record is complicated by the one-sided nature of outgroup comparison, which is restricted to birds among extant vertebrates, and the ambiguous meaning of the absence of a soft structure that only sometimes leaves an osteological imprint
Pneumatic sculpting (fossae, pneumatocoels) of the cervical and anterior dorsal vertebrae in saurischians is widely understood as evidence of the presence of paraxial cervical air sacs, given the close correspondence with pneumatic structures observed in extant avians
Soft Anatomy or Functional Capacity | Osteological Correlates |
Pneumatic diverticulum | Pneumatic sculpting, preferably with a pneumatopore characterized by a smooth rim and an invaginated space with smooth walls and interconnected cells that are considerably larger than those in cancellous bone |
Cervical air sac | Pneumatic invasion of cervical and dorsal (thoracic) vertebrae (centrum, neural arch) and ribs |
Clavicular air sac | Pneumatic invasion of the furcula, coracoid, sternal ribs or humerus; furcular invasion preferably median on central body or parasagittal on epicleideal processes |
Abdominal air sac | Pneumatic invasion of the pelvic girdle, preferably in areas removed from its contact with the sacrum (to avoid potential confusion with axial invasion by cervical air sacs) |
Subcutaneous pneumaticity | Pneumatic invasion on an external bone surface at some distance from an air sac that would require superficial transmission |
Costosternal pump | Ossification of sternal ribs and sternum; joints (synovial) between vertebral ribs, sternal ribs and sternum |
Advanced costosternal pump | Concavoconvex joint (synovial) between coracoid and sternum; uncinate processes; dorsal (thoracic) column shortened |
Flow-through ventilation (rigid lung) | Evidence of pneumatic invasion by at least one avian ventilatory air sac (clavicular, anterior or posterior thoracic, abdominal) |
Uni- or bidirectional lung ventilation | None |
In
In
Complete, articulated skeletons of other ornithomimids show no trace of sternal ossifications
In
Two general models have been proposed for lung ventilation in nonavian dinosaurs. The first infers the presence of compliant lungs with crocodile-like diaphragmatic ventilation, based in part on stained areas in two theropod skeletons purported to represent a diaphragm separating thoracic and abdominal cavities
Some of the initial reconstructions of the pulmonary condition in nonavian dinosaurs were characterized as the “stepwise transformation of a crocodilian-like lung into an avian airsac system”
Perry
Based on the osteological correlates we have assembled (
Experimental studies on ventilation in birds have shown slight changes in the orientation of air sacs relative to the connecting air passages can convert bidirectional to unidirectional flow in a poorly valved respiratory tract
Only the abdominal air sac in birds extends into the abdominal cavity, the other air sacs occupying more anterior spaces separated by septae
Research on the gastral cuirass in archosaurs led to the suggestion that it may have functioned as an accessory aspiration pump in nonavian dinosaurs
In
Most or all of the gastral cuirass, in addition, is either fused or solidly articulated in the midline, which would prevent expansion of the cuirass as an accessory aspiration pump as proposed
Avian lung ventilation is driven by muscles that expand and contract thoracic volume by deforming the ribcage and rocking a large bony sternum
The factors driving the origin and evolution of the functional capacity of avian air sacs and lung ventilation remain poorly known and tested. We briefly mention these in the light of the new evidence from
The second is that air sacs may have arisen originally for locomotory control in bipeds as a means to lower the center of mass and reduce rotational inertia
Finally thermoregulatory control may have played a role in the origin of air sacs
In sum, although we may never be able to sort out the most important factors behind the origin and evolution of the unique avian pulmonary system, discoveries such as
We present a four-phase model for the evolution of avian air sacs and costosternal-driven lung ventilation based on the known fossil record of theropod dinosaurs and osteological correlates in extant birds:
Phase I—Elaboration of paraxial cervical air sacs in basal theropods no later than the earliest Late Triassic.
Phase II—Differentiation of avian ventilatory air sacs, including both cranial (clavicular air sac) and caudal (abdominal air sac) divisions, in basal tetanurans during the Jurassic. A heterogeneous respiratory tract with compliant air sacs, in turn, suggests the presence of rigid, dorsally attached lungs with flow-through ventilation.
Phase III—Evolution of a primitive costosternal pump in maniraptoriform theropods before the close of the Jurassic.
Phase IV—Evolution of an advanced costosternal pump in maniraptoran theropods before the close of the Jurassic.
In addition, we conclude:
The advent of avian unidirectional lung ventilation is not possible to pinpoint, as there are no known osteological correlates for uni- or bidirectional lung ventilation.
The origin and evolution of avian air sacs may have been driven by one or more of the following three factors: flow-through lung ventilation, locomotory balance, and/or thermal regulation.
We thank members of the 1996 expedition for discovery of the material, Q. Cao and R. Masek for fossil preparation, E. Fitzgerald and T. Keillor for molding and casting, C. Straus for CT-scanning of fossil bones, and C. Abraczinskas for specimen drawings and executing final drafts of reconstructions. We thank B. Britt for comments on an earlier version of this paper. We are indebted to the Museo de Ciencias Naturales y Antropológicas (J. C. Moyano) de Mendoza (Argentina) for permission to conduct fieldwork and to the Field Museum of Natural History for loan of avian skeletal material.