MAURICIO ANTÓN 1 *, MANUEL J. SALESA 1, JUAN FRANCISCO PASTOR 2, ISRAEL M. SÁNCHEZ 1, SUSANA FRAILE 1 and JORGE MORALES 1 INTRODUCTION

Transcription

1 Lin- Blackwell Science, LtdOxford, UKZOJZoological Journal of the Linnean Society The nean Society of London, 2004? ? Original Article M. ANTÓN ET AL. MASTOID ANATOMY IN FELIDS Zoological Journal of the Linnean Society, 2004, 140, With 7 figures Implications of the mastoid anatomy of larger extant felids for the evolution and predatory behaviour of sabretoothed cats (Mammalia, Carnivora, Felidae) MAURICIO ANTÓN 1 *, MANUEL J. SALESA 1, JUAN FRANCISCO PASTOR 2, ISRAEL M. SÁNCHEZ 1, SUSANA FRAILE 1 and JORGE MORALES 1 1 Departamento de Palaeobiología, Museo Nacional de Ciencias Naturales. C. José Gutiérrez Abascal, Madrid, Spain 2 Museo Anatómico, departamento de Anatomía Humana, Universidad de Valladolid, C. Ramón y Cajal 7, Valladolid, Spain Received November 2002; accepted for publication July 2003 Muscle attachments in the mastoid region of the skull of extant felids are studied through dissection of two adult tigers Panthera tigris (Linnaeus, 1758) Pocock, 1930, a lion Panthera leo (Linnaeus, 1758) Pocock, 1930 and a puma Puma concolor (Linnaeus, 1771) Jardine, 1834, providing for the first time an adequate reference for the study of the evolution of that region in sabretoothed felids. Our study supports the inference by W. Akersten that the main muscles inserting in the mastoid process in sabretooths were those originating in the atlas, rather than those from the posterior neck, sternum and forelimb. Those inferences were based on the anatomy of the giant panda, Ailuropoda melanoleuca (David, 1869) Milne-Edwards, 1870, raising uncertainties about homology, which were founded, as revealed by our results. The mastoid muscle insertions in extant felids differ in important details from those described for Ailuropoda, but agree with those described for domestic cats, hyenas and dogs. The large, anteroventrally projected mastoid process of pantherines allows a moderate implication of the m. obliquus capitis anterior in head-flexion. This contradicts the widespread notion that the function of this muscle in carnivores is to extend the atlanto-cranial joint and to flex it laterally, but supports previous inferences about the head-flexing function of atlanto-mastoid muscles in machairodontines. Sabretooth mastoid morphology implies larger and longer-fibred atlanto-mastoid muscles than in pantherines, and that most of their fibres ran inferior to the axis of rotation of the atlanto-occipital joint, emphasizing head-flexing action The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, ADDITIONAL KEYWORDS: basicranial morphology Felinae hunting behaviour Machairodontinae neck muscles skull. INTRODUCTION The predatory behaviour of sabretoothed carnivorous mammals has been a matter of debate for nearly a century, and it remains one of the most intriguing issues in modern palaeontology. With no extant carnivore displaying such adaptations, we are left with the fossil remains, and our knowledge of functional anatomy, to propose credible hypotheses about the way those predators used their elongated upper canines for prey procurement. *Corresponding author. mfanton@terra.es The most recent group to evolve sabretooth adaptations were the machairodontines, members of the subfamily Machairodontinae within the true-cat family, Felidae. Like all derived sabretoothed mammals, the felid sabretooths differed from their non-sabretoothed relatives in showing modified mastoid regions, and these modifications were recognized by early workers as a key feature for understanding the animal s mode of attack. The precise interpretation of the meaning of the sabretooths mastoid morphology has varied considerably. In the early 20th century, Matthew (1910) proposed his stabbing theory, according to which the 207

2 208 M. ANTÓN ET AL. Figure 1. Schematic drawing of the skull and cervical vertebrae of the scimitar-toothed cat Homotherium latidens showing the hypothetical motions of the stabbing bite (top) and the canine shear bite (bottom). In the first case the main rotation is around a point behind the thoraco-cervical joint (white circle) and the posterior cervicals, whereas in the second case, the main rotation occurs at the atlantooccipital joint (white circle). In the stabbing model (top), the pull of the brachiocephalic muscles (single headed arrow) and of the scalenes (two headed arrow) provides the main force for the strike. In the canine shear-bite model (bottom), the pull of the atlanto-mastoid muscles (short two-headed arrow) is the most important force for the penetration of the upper canines. sabretooths used the neck-flexing and head-flexing muscles to power a downward stroke or stab, thus allowing the upper canines to penetrate into the flesh of prey. The mandible would have virtually no role in this kind of attack, and the clear adaptations for gape seen in sabretooths would be meant to keep the lower jaw out of the way of the stabbing motion. In Matthew s scheme, the supposed hypertrophy of several muscles that depress the head and neck was the key to the success of the stabbing motion. As evidence for that adaptation, he mentioned the enlarged transverse processes of cervical vertebrae, which house the insertions for a group of neck-flexing muscles called the scalenes, and the enlarged mastoid process, which in turn houses the insertions for the sterno-cleidomastoid group, a set of muscles that depress the head, pulling from their origins at the sternum and upper arm (Fig. 1). Matthew s theory was refined by Simpson (1941), who proposed that the inertia of the predator s rush toward prey would be incorporated into the stabbing motion, and it was accepted by many specialists for several decades. However, other workers (Bohlin, 1947) argued that, among other problems, the long, flattened upper canines of sabretoothed cats were too fragile to withstand such violent use (Biknevicius & van Valkenburgh, 1996). Most problems of the stabbing theory were satisfactorily overcome with the canine shear-bite hypothesis (Akersten, 1985), according to which the sabretooths killed prey by means of a modified type of bite. In Akersten s scenario, the jaw-closing muscles have a supporting role during the bite, but the main force for penetration of the upper canines is provided by the head-flexing action of the atlanto-mastoid musculature. In the canine shear-bite, the mandible has a clear part to play as support for the opposing action of the upper canines, and the modifications for wide gape are seen as adaptations to allow the lower jaw to play such a role while allowing clearance between the upper and lower canines. In addition, a recent study (Antón & Galobart, 1999) has shown that the cervical morphology of sabretooths reflects adaptations for strong muscular control of motions of the neck including extension and lateral rotation, as would be necessary to provide support for a canine shear-bite, rather than the exclusive emphasis on efficient neck flexion required by the stabbing theory. A central difference between Matthew s hypothesis and Akersten s is that the former considers that the modifications in the mastoid region of sabretooths can be explained as related to changes in the insertion area, and function, of the brachiocephalic and sternomastoid muscles, whereas in Akersten s view, the atlanto-mastoids were the most important muscles involved. In other words, Akersten gave more importance to rotation of the head around the atlantooccipital joint, whereas advocates of the stabbing theory emphasized the importance of flexion of the whole neck and head turning around the thoraco-cervical joint and posterior cervical intervertebral joints (Fig. 1). But when Akersten needed to support his ideas on anatomical grounds, he found a paucity of

3 MASTOID ANATOMY IN FELIDS 209 detailed studies about the muscle insertions in the mastoid region of large cats. In his own words, Descriptions of modern felid and canid anatomy often ignore, or are not consistent in the terminology and interpretation of those muscles (Akersten, 1985: 5). As a result, he based his hypothesis on a published study on the anatomy of the giant panda (Davis, 1964). Akersten s choice was supported by the common-sense assumption of similarity in the disposition of muscle insertions across different families and even suborders within the Carnivora, but it gave rise to uncertainties about homology, creating an unsatisfactory situation, which has lasted to this day. Over 15 years later, we still lack detailed comparative data from the anatomy of the animals that would provide the most adequate comparison, i.e. the living, large members of the family Felidae, to which the extinct machairodontines belong. The most useful study of the muscles of a large felid is Barone s (1967) paper on the myology of the lion, in which the neck musculature is described in detail, but insertion areas in the mastoid region are only mentioned and not described or figured. Similarly, for the domestic cat, anatomical studies abound, but again the insertion areas of muscles have been described (Reinhard & Jennings, 1935; Barone, 1989) but not figured in sufficient detail; regardless, the differences in morphology between small cats and pantherines go beyond mere size and, as commented below, may imply differences in muscle function. A further complication lies in the fact that all anatomy treatises state that the action of the muscles most relevant to Akersten s analysis, the obliquus capitis cranialis, is to extend and to rotate laterally the atlanto-cranial joint, rather than flexing the head as was assumed by Akersten (Evans & Christensen, 1979; Barone, 1989). A more recent, specific study of the suboccipital muscles in the neck of the domestic cat shows that the atlanto-mastoids generate little torque in the saggital plane (Selbie, Thomson & Richmond, 1993). With these data in mind, Akersten s hypothesis would imply a case of function shift in a muscle group for the machairodontines (Bryant, 1996). But it is also feasible that these muscles have a head-flexing action in some modern carnivores, reflected in mastoid morphology. This study intends to provide detailed information about the muscle insertions in the mastoid area of modern pantherines, and then to discuss the likely implications of our findings in terms of the evolution and functional anatomy of sabretoothed cats. MATERIAL AND METHODS An adult lioness Panthera leo, two adult tigers Panthera tigris, one male and one female, an adult female puma Puma concolor and an adult female genet Genetta genetta (Linnaeus, 1758) Cuvier, 1816 were dissected by the authors at the Anatomy Department of the University of Valladolid. All specimens were captive animals coming from the Zoological Park of Matapozuelos, Valladolid province. We also incorporated information from a previous dissection of an adult female striped hyaena Hyaena hyaena (Linnaeus, 1758) Brünnich, 1771 and an adult male binturong Arctictis bingturong (Raffles, 1821) Temminck, 1824 from the Madrid Zoological Park. We also studied osteological material of extant and extinct carnivores to compare the morphology of the mastoid region. The skulls of extant carnivores studied (besides those of the dissected species) include specimens of domestic cat Felis catus Linnaeus, 1758, spotted hyena Crocuta crocuta (Erxleben, 1777) Kaup, 1828 and wolf Canis lupus Linnaeus, Reference is also made of published material of extinct machairodontine felids of the genera Homotherium Fabrini, 1890 and Smilodon Lund, 1842, as well as descriptions of giant panda Ailuropoda melanoleuca (David, 1869) Milne-Edwards, 1870 and other ursids. The dissected specimens were fresh and were not prepared with any preservation products, so the observation of muscle fibres was excellent but dissection had to proceed quickly before the tissues deteriorated. Each specimen was dissected in a single session, except in the case of the tigress and the puma, which were frozen and subjected to a second session. In all specimens, dissection proceeded from the most external muscle layer, except in the lioness, which had been partly defleshed before our dissection, so that muscles external to the splenius were damaged or missing, but the deeper layers relevant for this study were intact. In each specimen, we dissected the neck muscles that insert on the mastoid region and on adjacent areas of the skull and we identified origin and insertion areas of each muscle. We made both digital and conventional photographs of each subsequent muscle layer in each specimen, and relevant details of muscle structure and insertion were sketched in pencil directly from the dissected specimens. Skulls of large pantherines were available for comparison and identification of skull features that emerged during dissection, and the skulls and skeletons of each dissected specimen were subsequently prepared for detailed observation of the relevant osteological details. We followed the muscular nomenclature used by (Barone, 1989) in his treatise on the anatomy of domestic mammals.

4 210 M. ANTÓN ET AL. RESULTS MUSCLE ATTACHMENTS IN THE MASTOID AREA OF EXTANT LARGER FELIDS The most external muscle inserting in the mastoid area is the brachiocephalicus, which originates on the humerus and has an aponeurotic insertion extending from near the tip of the mastoid up the mastoid crest and nuchal crest to the midline (Fig. 2). Immediately internal to this muscle appear the fibres of the rhomboideus and splenius, which originate at the anterior thoracic spinous processes and terminate in a common aponeurosis extending just internal to that of the brachiocephalic, and with broadly the same extension (Fig. 3). At this same level, the fibres of the sternomastoid muscle are seen, extending from their origin at the sternum and terminating in a tendinous insertion on the tip of the mastoid process. Some fibres of the sternomastoid extend upwards, sharing with the brachiocephalic a common aponeurotic insertion along the mastoid crest (Fig. 3) Internal to the m. splenius is the muscle mass of the extensors of the neck, including the biventer cervicis, semispinossus capitis, complexus and longissimus capitis, which have fleshy insertions on the occipital area, above the mastoid region. Inferior to these can be observed the obliquus muscles: the obliquus capitis caudalis extending from the spine of the axis to the dorsal surface of the atlas wings, and the obliquus capitis cranialis, extending from the ventral side of the atlas wings to the mastoid process. The mastoid insertion of the latter muscle is fleshy and extensive, covering much of the area of the process (Fig. 4A, B). In the atlas, the origin area of the obliquus capitis cranialis occupies most of the ventral surface of the wing. The superior fibres of this muscle are short, owing to the small distance between their origin and insertion, but the inferior fibres span a considerable distance between the atlas wing and the lower part of the mastoid process. The insertion area in the skull is broadest at its inferior tip. Superiorly, the insertion area is not restricted to the mastoid process, and continues in a narrowing band up the nuchal crest. In the felid specimens dissected by us, we did not find evidence of an accessory portion of this muscle extending dorsally from the lip of the atlas wing, as described for the dog (Evans & Christensen, 1979; Done et al., 1995). In the puma specimen there were some additional fibres extending dorsal to the atlas wing, in a position similar to the mentioned accessory portion. Unlike the main portion of the muscle, these fibres form a thin layer, superficial to the m. obliquus capitis caudalis and covered by the m. splenius. Cranially, these fibres mingle with the latter muscle before it inserts on the nuchal crest (Fig. 4B). The origin of the m. digastricus on the paroccipital process is found to be variable in extent. In all cases the main fibres of the muscle attach to the paroccipital process tip, but whereas in the male tiger the insertion was apparently restricted to that process, on the tigress some fibres of the muscle went to the tip of the mastoid, and in the lioness the insertion extended to the area around the base of the process, even into the adjacent part of the tympanic bulla (Fig. 5A) In the mandibular ramus, the digastric muscle insertion extends from the level of the first molar anteriorly, reaching the level of the symphisis. The muscle rectus capitis lateralis originates on a small area of the ventral side of the atlas wing, medial to the attachment of the m. obliquus capitis cranialis, and inserts on the medial side of the paroccipital process, medial to the insertion of the m. obliquus capitis cranialis. In the puma, the insertion reaches the tip of the paroccipital process, adjacent to the insertion of the m. digastricus (Fig. 5B). In the genet, some fibres attach to the tympanic bulla. MUSCLE ATTACHMENTS IN MACHAIRODONTINES AND IN OTHER CARNIVORES The mastoid area in machairodont felids differs from that of extant larger felids in several ways (Fig. 6). The mastoid process itself is more developed, and it projects in an antero-ventral direction to varying extent, almost touching the postglenoid process in the more derived machairodont genera such as Smilodon and Homotherium. The paroccipital process is variably reduced, especially in some of the more derived machairodont genera, and as a result the tip of the mastoid process has a greater ventral projection than the tip of the paroccipital, whereas the reverse is true for the extant felids. The texture of the mastoid process is often more rugose than in extant felids, and more so in the more derived taxa. The orientation of the tip of the process is variable, facing anteriorly in some taxa, just as in extant felids, whereas in other taxa it faces inferiorly. In most cases the process tip is proportionally broader and more distinctly sculpted than in extant felids. The mastoid crest is usually more developed with a greater lateral expansion, and as a result the surface of the mastoid process itself tends to face more inferiorly and posteriorly, and less laterally, than in extant felids (Fig. 7). Compared with machairodontine and extant felids, the mastoid area of canids shows a relatively small mastoid process and a relatively large paroccipital process. The mastoid process tip in canids is not placed inferior but actually level with, or dorsal to, the axis of rotation of the atlanto-occipital joint, whereas the paroccipital process tip projects considerably more ventrally. In bears and in the giant panda, the mastoid

5 MASTOID ANATOMY IN FELIDS 211 Figure 2. Photograph and schematic representation of superficial layer of head and neck muscles of male puma. 1, m. brachiocephalicus.

6 212 M. ANTÓN ET AL. Figure 3. Photograph and schematic representation of intermediate plane of neck muscles of male puma. 2, m. trapezius; 3, m. splenius; 4, m. sternomastoideus. process is much larger and projects far more ventrally than in the wolf, and indeed more than in extant felids. In hyaenids, the paroccipital is ventrally projected and anteriorly placed and thus very close to the mastoid process, and the insertion area for the m. obliquus capitis cranialis is contained in the vertically orientated band between the paroccipital and the mastoid crest.

7 MASTOID ANATOMY IN FELIDS 213 In his description of the giant panda, Davis (1964) figures a bundle of muscle fibres similar in position to the accessory portion of the m. obliquus capitis cranialis described in the dog (Evans & Christensen, 1979; Done et al., 1995). As noted above, we did not find evidence of such a portion in the studied pantherines, although a thin band of muscle fibres was observed in the puma. In a binturong and a striped hyaena previously dissected by us there was no accessory portion, but in the genet, as in the puma, we observed a thin band of muscle stemming from a small aponeurosis that envelopes the lateral border of the atlas wing. These fibres appear to be continuous with the m. obliquus capitis cranialis but they join cranially with the m. splenius and they lack a separate insertion area in the skull. DISCUSSION Our observations of the muscle insertions in the mastoid region of modern felids support most of Akersten s interpretations about the myology of extinct sabretooths, but also allow us to update and revise his discussion of the head-flexing musculature. As he correctly extrapolated from Davis s (1964) study of the giant panda, the muscle fibres having a most extensive attachment area in the mastoid process are those originating on the ventral side of the atlas wings. Given the distant relationship between felids and the giant panda, Akersten cautioned that the muscles extending between the ventral surfaces of the atlas wings to the postero-lateral margins of the mastoid process in Smilodon may or may not be homologous with those of Ailuropoda (Akersten, 1985: 5). Our results show that those concerns about homology were founded, because there are important differences in detail between the positions of the muscle insertions in extant felids and those described by Davis (1964) for Ailuropoda. In figure 19 of Davis (1964) the insertion of the m. obliquus capitis cranialis is seen to occupy much of the surface of the mastoid and paroccipital processes, extending to an area medial to the latter, whereas the m. rectus capitis lateralis is seen to insert on a portion of the mastoid process lateral to the insertion of the previous muscle. Thus there appear to be two atlanto-mastoid muscles in the giant panda. By contrast, in extant felids dissected by us, the m. obliquus capitis cranialis insertion occupies most of the mastoid process, whereas the m. rectus capitis lateralis inserts outside the mastoid process, in and near the medial surface of the paroccipital process and medial to the insertion of the previous muscle. The latter condition agrees with that described for the domestic cat (Reinhard & Jenning, 1935; Barone, 1989), lion (Barone, 1967), and with that described and figured for the domestic dog (Barone, 1989), striped hyena (Spoor & Badoux, 1986) and genet (this paper). Consequently, it can be said that in all these animals the only atlanto-mastoid muscle is the m. obliquus capitis cranialis. These differences suggest an apomorphic condition for the panda or for the whole Ursidae, or alternatively Davis (1964) may have been wrong about the identity of the muscles in the specimen available to him. In any event, dissection of additional specimens of Ailuropoda would be desirable in order to clarify this subject. Some of the observed differences among the reviewed carnivores in the disposition of some relevant muscles may have a functional significance: the presence of an accessory dorsal portion of the m. obliquus capitis cranialis in the giant panda (as figured but not mentioned by Davis, 1964: fig. 92) and in the dog (where it is described as having a separate insertion in the skull; Evans & Christensen, 1979) would point to the existence of an extensor function of the muscle besides that of head-flexion. By contrast, its reduction or absence among the reviewed members of the Feliformia points to an almost exclusive flexor action. The lack of comparative data led Akersten to say that modern felids and canids possess tiny mastoid processes with only very minute muscles extending between them and the atlas wings (Akersten, 1985: 5). Our results show that generalization to be wrong. Although it is true that modern canids have relatively small mastoid processes (Fig. 7), those of tigers and lions are relatively much larger, providing attachment areas for powerful atlanto-mastoid muscles. A further difference between large canids and larger felids is that the mastoid process of the latter projects further downwards than in the former, not only enlarging the muscle attachment area but also changing the action of the atlanto-mastoid muscles. Effectively, the fibres of the m. obliquus capitis cranialis in larger felids are located more clearly below the level of the occipital condyles, and as a result of this change, the contraction of that muscle is more effective at flexing the head, whereas in canids the head-flexing action would be less efficient and the muscle would contribute to stabilize the atlanto-cranial articulation when contracting bilaterally, while unilateral contraction rotates the head laterally (Barone, 1989). In the domestic cat, there is little ventral projection of the mastoid process, and the atlanto-mastoid muscles are said to generate little torque on the sagittal plane (Selbie et al., 1993). So we can say that modern larger felids show an incipient stage of the implication of the atlanto-mastoid musculature in head flexion during the bite, which would become far more emphasized in machairodontines. Contribution of the neck musculature to vertical motions of the head associated with mastication is already known in domestic cats (Gorniak & Gans, 1980), so that the adaptation of

8 214 M. ANTÓN ET AL. A Figure 4. (A) Photograph and schematic representation of deep muscles of the neck in male tiger. 5, m. obliquus capitis caudalis; 6, m. obliquus capitis cranialis; 8, m. digastricus; At, lateral border of the atlas wings; Ax, dorsal border of axis. (B) Photograph and schematic representation of deep muscles in a male puma. f, additional superficial fibres of m. obliquus capitis cranialis, dorsal to the atlas wing.

9 MASTOID ANATOMY IN FELIDS 215 B Figure 4. Continued

10 216 M. ANTÓN ET AL. A Figure 5. (A) Photograph and schematic representation of deep muscles of the neck in a lioness. The posterior portion of the temporalis muscle has been removed to make visible the nuchal region of skull and neck muscles attaching to it. 7, deep extensors of the neck, including rectus capitis dorsalis major and minor; Am, auditory meatus; Mp, mastoid process; Nc, nuchal crest. (B) Photograph and schematic representation of deep muscles of the neck of a male puma in ventral view. 9, m. rectus capitis lateralis.

11 MASTOID ANATOMY IN FELIDS 217 B Figure 5. Continued

12 218 M. ANTÓN ET AL. M P M P Figure 6. Photographs of the mastoid region of skull in female lion, Panthera leo (top) and scimitar-toothed cat, Homotherium latidens (bottom) from Incarcal, Spain (IN-I 929). Note that the back of the skull is broken in the fossil. Muscle insertion areas are marked; muscle numbering as in Figs 1 5. M, mastoid process; P, paroccipital process.

13 MASTOID ANATOMY IN FELIDS 219 Figure 7. Drawing of the skull and anterior cervicals of Panthera tigris (top) and Homotherium latidens (bottom) with fibres of selected muscles. Muscle numbering as in Figs 1 5. A black circle in the condylar area represents the position of the rotation centre of the atlanto-occipital articulation. Notice how, in Homotherium, most fibres of the obliquus capitis cranialis extend well below that centre of rotation, and would therefore have a stronger head-flexing action. Notice also how the greater distance between the posterior tip of the atlas wings and the tip of the mastoid process in Homotherium makes for longer inferior fibres of the obliquus capitis cranialis muscle. sabretooths would simply elaborate on an existing biomechanical system (Bryant, 1996). All this is in contrast with the interpretations of Matthew, who considered that the most relevant muscles for head flexion in machairodonts where the m. brachiocephalicus (his cleidomastoids) and m. sternomastoideus, and showed these muscles as occupying with their attachments the full extent of the mastoid process (Matthew, 1910: figure 9). Our results clearly show that the insertion of the m. brachiocephalicus and the more superior portions of the sterno-mastoids, although broadly coinciding with the position indicated in Matthew s figure, are actually aponeurotic and restricted to a thin band in the outer margin of the mastoid process, i.e. in the mastoid crest, whereas the attachment of the m. sternomastoideus is restricted to the process tip. The insertion of the m. obliquus capitis cranialis, by contrast, is fleshy and extensive, and the action of this muscle is likely to have had a stronger effect on the flexion of the head. These findings support the interpretation that the action of the atlanto-mastoid musculature (i.e. the m. obliquus capitis cranialis) was the most relevant for head-flexing motions aiding the penetration of the upper canines into the flesh of prey. The immediate implication of this is that the relevant action involved was rotation of the head around the atlanto-occipital joint (an action associated with the canine shear-bite), rather than rotation of the whole neck and head

14 220 M. ANTÓN ET AL. around the thoraco-cervical joint (an action associated with the stabbing bite). In the canine shear-bite, muscular action from a static start provides the main force for penetration, whereas in the stabbing interpretation, the momentum gathered during rotation of the head along a wide arc, or even during the leap toward prey, is invoked to explain canine penetration. It is true nonetheless that the downward projection of the tip of the mastoid process would improve the efficiency of the sternomastoid muscles as head flexors, an action which would be incorporated to the canine shear-bite mechanics as discussed by Akersten (1985) and Antón & Galobart (1999). The reduction in size and in ventral projection of the paroccipital process in sabretooths implies changes in the action of the m. digastricus, which is to depress the mandible. In spite of the variation observed in the extent of the area of origin of this muscle in the extant felids studied, the main attachment is always on the tip of the paroccipital, and the lesser the ventral projection of the process tip, the greater the distance between the origin and insertion of the fibres of the m. digastricus. Such an increase in the vertical span of the fibres would be advantageous in terms of the increased gape required by the canine shear-bite (Antón & Galobart, 1999). CLOSING COMMENT: ANATOMICAL DESCRIPTIONS, SOFT TISSUE RECONSTRUCTION AND PALAEOBIOLOGICAL INFERENCE The importance of soft tissue reconstruction in fossil vertebrates is being more and more widely recognized among evolutionists. As summarized by Witmer (1995: 20) (1) Soft tissues largely are responsible for the existence, maintenance and form of bones, (2) judgements about the form and actions of soft tissues are (implicitly if not explicitly) the basis for a host of palaeobiological inferences; and (3), with regard to systematics, soft tissue relationships may provide testable hypotheses on independence or nonindependence of phylogenetic characters. However, accurate soft tissue reconstruction is impossible without a sound database about the relationships between bone and soft tissue in extant taxa. Unfortunately, it appears that descriptive monographs about the gross anatomy of animals have lost the favour of current zoological science, due at least in part to a prejudice among modern scientists against the descriptive aspect of natural history (Gould, : ). However, the field of palaeobiological inference provides many occasions to regret the paucity of accurate descriptions of modern organisms, and of the musculo-skeletal systems of modern vertebrates in particular, the case of the mastoid region in carnivores being just one example of the problem. Whenever palaeobiologists attempt to reconstruct soft tissues in fossil mammals, the descriptive anatomical references cited are often remarkably ancient (Spoor & Badoux, 1986; Witmer, Sampson & Solounias, 1999; Naples & Martin, 2000). While preparing his monograph about the giant panda, Davis commented about the lack of comparative information in plaintive terms: The muscles of the Carnivora are comparatively well known, but even for this order our knowledge is at a primitive level. Descriptions are incomplete and inaccurate, often doing little more than establish the fact that a given muscle is present in species dissected. Even for domestic carnivores the dog and the cat the standard reference works are full of inaccuracies and are inadequately illustrated. Most of the genera of Carnivora have never been dissected at all. (Davis, 1964: 146). He goes on to comment on the importance of muscular differences for the overall problem of evolutionary mechanisms, but then asks: How can such differences be detected and evaluated?. Certainly not on the basis of existing descriptions and illustrations. (Davis, 1964: 146). Most of the descriptions that we used as reference to discern the pattern of muscle insertions in the mastoid region in carnivores simply did not exist when Davis wrote his monograph, but even those references are not enough, either in detail or range of species treated, to solve the problem, and in general the situation today remains much as it was in With our results in hand, and given the insufficient references available to Davis, there is now reason to question the interpretations that he made of atlanto-mastoid musculature in Ailuropoda, but in order to add to the discussion, it would be necessary to describe additional specimens of the giant panda, and of other ursids. Such investigations would be especially pertinent to the subject of atlanto-mastoid muscle function in sabretooths, because the giant panda and other ursids share a mastoid morphology different from most modern carnivores and intriguingly similar to that of machairodontines (Akersten, 1985). As Witmer commented, the accuracy of soft tissue reconstructions is vital because mistakes in soft-tissue inference cascade upwards through the inverted pyramid of paleobiological inference, amplifying the error. (Witmer, 1995: 22). If the soft tissue inferences are based on an insufficient or inaccurate database about the anatomy of modern taxa, then all the subsequent steps will be questionable. CONCLUSIONS The results of our study show that the most extensive muscular insertions in the mastoid process of modern

15 MASTOID ANATOMY IN FELIDS 221 big cats are those of the obliquus capitis cranialis. The sternomastoid muscle inserts on the tip of the mastoid process, and the m. brachiocephalicus inserts through a thin aponeurotic band along the mastoid crest. The m. digastricus inserts on the tip of the paroccipital process but can extend to adjacent areas depending on individual variation. These results allow interpretation, in terms of muscle function, of the differences in mastoid morphology between modern pantherines and machairodontines. The enlarged, ventrally projecting mastoid process of the latter implies that the insertion area of the atlantomastoid muscles was larger and that the more inferior fibres of these muscles had an enhanced action as head flexors, being able to rotate the head along a large arc around the atlanto-occipital articulation. The head-flexing action of the portions of the sternomastoid inserting on the mastoid process tip would also be improved by the ventral projection of the latter, whereas the function of the m. brachiocephalicus would be little altered. The fibres of the m. digastricus were probably longer in machairodontines as a result of the reduction of the paroccipital process, which implied that the process tip was higher and thus further from the insertion of the m. digastricus in the mandibular ramus. ACKNOWLEDGEMENTS This paper was supported by project MCT-BTE We are grateful to Angel Galobart (Instituto de Palaeontología Miquel Crusafont) and Josep Tarrús (Museu Arqueológic Comarcal de Banyoles) for the opportunity to study the fossils of Homotherium latidens from Incarcal. REFERENCES Akersten W Canine function in Smilodon (Mammalia, Felidae, Machairodontinae). Los Angeles County Museum Contributions in Science 356: Antón M, Galobart A Neck function and predatory behavior in the scimitar toothed cat Homotherium latidens (Owen). Journal of Vertebrate Paleontology 19: Barone R La myologie du lion (Panthera leo). Mammalia 31: Barone R Anatomie comparée des mammifères domestiques. Tome 1 (Ostéologie), Tome 2 (Arthrologie et myologie). Paris: Vigot. Biknevicius AR, van Valkenburgh B Design for killing: craniodental adaptations of predators. In: Gittleman JL, ed. Carnivore behavior, ecology, and evolution, Vol. 2. New York: Cornell University Press, Bohlin B The sabre-toothed tigers once more. Bulletin of the Geological Institute of the University of Uppsala 32: Bryant HN Force generation by the jaw adductor musculature at different gapes in the Pleistocene sabretoothed felid Smilodon. In: Stewart KM, Seymour KL, eds. Paleoecology and paleoenvironments of late Cenozoic mammals. Toronto: University of Toronto Press, Davis DD The giant panda. A morphological study of evolutionary mechanisms. Fieldiana, Zoology Memoirs 3. Chicago Natural History Museum. Done SH, Goody PC, Evans SA, Stickland NC Color atlas of veterinary anatomy, Vol. 3. The dog and the cat. Barcelona: Mosby-Year Book Inc. Harcourt Brace de España S.A. Evans HE, Christensen GC Miller s anatomy of the dog, 2nd edn. Philadelphia: W.B Saunders Company. Gorniak GC, Gans C Quantitative assay of electromyograms during mastication in domestic cats (Felis catus). Journal of Morphology 163: Gould SJ La vida maravillosa. Barcelona: Editorial Crítica. Matthew WD The phylogeny of the Felidae. Bulletin of American Museum of Natural History 28: Naples VL, Martin LD Evolution of hystricomorphy in the Nimravidae (Carnivora; Barbourofelinae): evidence for complex character convergence with rodents. Historical Biology 14: Reinhard J, Jennings HS Anatomy of the cat. New York: Henry Holt & Company. Selbie WS, Thomson DB, Richmond FJR Suboccipital muscles in the cat neck: morphometry and histochemistry of the rectus capitis muscle complex. Journal of Morphology 216: Simpson GG The function of saber-like canines in carnivorous mammals. American Museum Novitates 130: Spoor CF, Badoux DM Descriptive and functional myology of the neck and forelimb of the striped hyena (Hyaena hyaena, L. 1758). Anatomischer Anzeiger 161: Witmer LM The extant phylogenetic bracket and the importance of reconstructing soft tissues in fossils. In: Thomason JJ, ed. Functional morphology in vertebrate paleontology. Cambridge: University of Cambridge Press, Witmer LM, Sampson SD, Solounias N The proboscis of tapirs (Mammalia: Perissodactyla): a case study in novel narial anatomy. Journal of Zoology, London 249: