A perfectly preserved incisor tooth (cast) from the Wellington Caves.

Among the three pairs of maxillary incisors, the first pair is enormous, curved and caniniform, meeting at the tips and exhibiting wear facets on their neighboring surfaces.  An inclined groove on the posterior surface of each has been created by the lower incisors.  I2/ is small but I3/ possesses a large crown which is oblique anteriorly and lingually.

    The small blunt maxillary canine is located in the palate posterior to I3 and has no resemblance at all to the homologous tooth in most carnivores.  The first two maxillary premolars are small as well and are positioned at the anterior margin of P3/.
 

The third maxillary premolar is a massive structure designed for for cutting which may in some specimens measure up to 57 mm in length and has a lingually oriented occlusal surface.  As this tooth becomes worn down, the buccal side of the root becomes further elongated faster than the lingual surface and therefore the crown 'rotates' lingually in order to maintain a close contact with P/3.  The margin of the maxilla appears to grown downwards at the same time, thereby increasing the vaulting of the palate.  Such a rotation has been recorded by Mellett (1969) in the maxillary carnassial teeth (M1/ and M2/) of Hyenodon but this action does not appear to occur in fissipede carnivores.

Intact crown of a left maxillary 3rd premolar.  Wellington Caves, N.S.W.

    The single maxillary molar is positioned obliquely on the lingual side of the posterior root of P3/.  It displays an anteromedial wear surface symmetrical with its function as a stop for the mandibular first molar.

    In the lower jaw, the single pair of recumbent incisors are positioned with their tips posterior to I1/ when at rest.  Wear occurs at the tip and upon the outer margin of the crown and an interdental facet is present on the contiguous incisor surfaces.

    As is the case with the maxillary dentition, in the mandible there are two small premolars at the anterolingual margin of the lower carnassial.  The latter tooth is inclined buccally to converge with the occlusal surface of P3/ and grew in this direction during life as appears to have been the situation with Hyenodon (Mellett 1969).  The lower sectorial surface bears a facet on the trigonid of M/1 so that both P/3 and M/1 shear past P/3.  M/2 is quite small and is rarely present in fossil material.

    The image provided by Thylacoleo's skull morphology is therefore that of an marsupial that displays adaptations to carnivory which are similar to those seen in placentals except in the form of the condyle and the glenoid fossa, the reduction of the canine teeth, the recumbent I/1 lower, and the amplified I1/.
 

    Reconstruction of the masticatory musculature of Thylacoleo by using a dissection of the structurally similar phalangerid, Phalanger maculatus, as a guide (Finch 1981) demonstrated that the temporalis muscle constituted approximately 56% of the total adductor musculature, while the masseter complex (superficial and deep masseters and zygomaticomandibularis) contributed 34% and the pterygoids a mere 10%.  Figures were given by Turnbull (1970) for a wide variety of mammals which he arranges into four basic categories:  A, generalized or unspecialised type;  B, carnivore-shear or 'scissors' type;  C, ungulate-grinding or 'mill' type;  and D, rodent-gnawing or 'forward-shift' type.  There is only a difference of degree between categories A and B as, in each, the temporalis is dominant and the smallest of the adductor muscle groups is the pterygoid.  The forward-shift category (D) does not correspond with the grinding category (C) in that the predominating masseter complex is notably larger and the temporalis and pterygoid ae inclined to be more reduced than in category C.

    It would seem that Phalanger is intermediate between categories C and D.  Examination of its skull demonstrates a glenoid fossa which is not closed anteriorly and possesses an articulating surface which is ventrally convex.  This suggests a forward shift mechanism to allow occlusion of the incisors as, at rest, the tip of I/1 lies anteriorly to I1/.  To attaining this position, the condyle is moved anteriorly on to the convexity of the glenoid fossa which permits contact of the incisor tips.  Also, this prevents occlusion of the cheek teeth at times when the incisors are in use.

    In difference to Phalanger, however, Thylacoleo exhibits a dominant temporalis and a small pterygoid which conspicuously places the animal in the carnivore-shear category (B).  Van Valen (1969:  114) makes note that the masseter complex exceeds the temporalis in Thylacoleo but provides no supporting information.

    The likely areas of origin and insertion of the adductor muscles were outlined on a precise drawing of the skull and the midpoints of the the two attachments of each muscle were joined in order to obtain further insight into the masticatory function in Thylacoleo.  This displayed the direction of pull of the muscle's 'central' fibre.  Taking this into consideration, along with the distance of the carnassial tooth from the condyle (resistance arm) and the length of the effort lever arm (fulcrum to centre of the muscle insertion area), it is possible to evaluate the useful power (E) of a muscle.  This is executed from Turnbull's formula (1970):

where M represents the percentage of weight of the muscle and r is the ratio of the effort lever arm to the resistance lever arm.  FL and Fx are correction factors which counterbalance the deviation, in the parasagittal and traverse planes, of fiber direction from the plane of jaw closure.  A muscle that is oriented at 90 degrees to the lever arm in the parasagittal plane is regarded as capable of exerting its full force to close the jaw and thus FL = 1.  Any departure from 90 degrees, however, will lessen its performance until a muscle lying parallel to the lever arm has FL = 0.  Likewise, Fx = 1 if the muscle, observed from the posterior aspect of the skull, is vertically oriented (lying parallel to the coronoid process).  Fx tends toward zero as the muscle tends towards a horizontal orientation.

    Of the total useful power (E) produced by the adductor muscles of Thylacoleo, the temporalis exerts 62%, the superficial masseter 10%, the deep masseter 11%, the zygomaticomandibularis 14% and the pterygoid 3% (Finch 1981).  Upon comparing these figures to the percentage weights, only the zygomaticomandibularis and ptrygoid exert less force than would be expected from their relative mass.  Examination of the probable fibre direction reveals the reason.  These two muscles approach a horizontal rather than a vertical direction.  Obviously, their functions are not solely adductor.  Scapino (1965) proposes that in a canid, where the mandibular symphysis is not fused, these two muscles act to regulate the angel at which the mandibular carnassial approaches the maxillary one by slanting the dentary to allow the most effective shear.  This would also appear to be the case in Thylacoleo in which the mandibular symphysis is unfused.  Also, only limited lateral excursion of the condyle is possible at the glenoid fossa in Thylacoleo and, even though the carnassial teeth continue to grow as they become worn down, correct positioning of the cutting surfaces demands some adjustment to the stance of the lower blade.
 

    Upon examination of Thylacoleo's skull structure and reconstructed musculature, it can be proposed that this animal was capable of exerting a powerful slicing action with the carnassial teeth and a forceful puncturing bite with the four convergent first incisors.  Wear facets exhibit that there was occlusion between I/1 and the three maxillary incisors which would allow for the biting and tearing of smaller pieces of food.  Wear present on the posterolateral surface of the canine could only have been produced by occlusion with the anterior buttress of P/3 when the jaw was in the forward position.  The position of M1/ and a facet on the posterolateral wall of M/1 uphold Woods' (1956) belief that the maxillary molar was a stop for M/1 as the jaw moved back into the resting position.

    The fact that Thylacoleo does not possess a battery of grinding molars illustrates that it could not have subsisted on plant material.  The massive dimensions of the carnassial teeth and the concomitant carnivore adaptations indicate a carnivorous habit.  However, Thylacoleo has retained the ancestral forward-shift mechanism along with the reduction of the canines, features which are not carnivore characteristics.  A similar case can be seen in the giant panda, Ailuropoda melanoleuca which has been shown by Davis (1964) to have evolved in the opposite direction.  Descended from an omnivore-carnivore stock, it has become a specialized herbivore while still retaining a number of carnivore characteristics.

    Wells (1982) is in general concurrence with the theory that Thylacoleo was a carnivore but sees the animal as a bone-crusher capable of pivoting its mandible laterally on one condyle, thereby enabling it to occlude the I/1 with the three ipsilateral maxillary incisors.