Thylacinus (Marsupialia: Thylacinidae) from the Mio-Pliocene boundary and the diversity of Late Neogene thylacinids in Australia

Description

The holotype includes the incisor alveoli, the canine, all three premolars, M₂, M₃ and the alveolus for M₁ (Figs. 1–2 and Table 1). Posteriorly the dentary has broken off at the level of the posterior end of M₃, so that M₄ is missing. The dentary is transversely compressed, as it is in Th. cynocephalus but unlike the transversely broad, robust dentaries of Th. megiriani and some Th. potens. The dentary of Th. yorkellus differs further from that of Th. megiriani and Th. potens in lacking a thickened torus along the ventrobuccal margin, below the mental foramina. The dentary depth below the anterior root of M₂ (the posteriormost level at which this measurement can be made) is 24.5 mm which lies within the range of Th. cynocephalus. Given that the length of the lower molar row and the heights of the individual molars lie below the range seen in Th. cynocephalus, I propose that the holotype Th. yorkellus has a deeper jaw relative to tooth size in comparison to the former species. A relatively deep dentary is also seen in some, but not all, specimens of Th. potens (Yates, 2014). The buccal surface of the dentary is pierced by three mental foramina set approximately at the mid height of the dentary, although it is possible that a fourth was present in what is now a large area of missing bone below P₃ (Fig. 2). The anterior mental foramen is a large anteriorly facing opening with a diameter of 3.2 mm located below the anterior root of P₂. The two close-set posterior mental foramina are placed below M₂ and the anterior margin of M₃, respectively. These openings are smaller, with anteroposterior diameters of 2.1 and 1.4 mm, respectively. In buccal view the alveolar margin is concave with P₃ set lower than the rest of the preserved tooth row. The symphyseal surface extends posteriorly to the level of the anterior margin of P₃.

The canine has a procumbent root and a vertically projecting crown. The tip is gently recurved so that the anterior margin in lateral view is convex and the posterior margin is concave. In anterior view the crown has a weak lingual curvature. The cross-section of the crown is a slightly buccolingually compressed oval. A weakly developed carina extends along the anterolingual margin of the crown, from base to apex. The posterior surface is rounded and smooth. The measurements of the canine are given in Table 1.

A diastema of 3.3 mm separates the canine alveolus from that of P₁. P₁ is a smaller, lower-crowned tooth than all of the other postcanine teeth. In occlusal view it is elongately ovoid, with the long axis oriented anteroposteriorly. Although the tip of the protoconid is well worn, it is clear that the unworn tooth would have had a height less than its anterior-posterior length. The apex of the protocone is located in the anterior half with an elongate heel extending posterior to it. The heel has a sharp posterior corner which forms an incipient cuspidule. In buccal view the posterior margin of the protoconid curves gently down to the posterior heel. A weakly developed cristid extends along the anterior margin of the protoconid. The tooth has no other cristids or cuspidules.

A diastema of 2.8 mm sepparates P₁ from P₂. P₂ differs from P₁ in being larger (Table 1) and proportionately taller. The apex of the triangular protoconid is less skewed to the anterior side of the tooth, being placed slightly mesial to the mid length. A weak cristid extend up the anterior side of the protoconid. There is a minute tubercle-like cuspidule developed at the base of this anterior cristid. A posterior cristid extends from the apex of the protoconid to the posterior corner of the basal heel. A distinct inflection in the buccal profile marks the change from the heel to the posterior margin of the protoconid.

A long diastema of 6.4 mm sepparates P₂ from P₃. P₃ is larger than, P₂ and differs from it in a number of subtle features. P₃ shows a greater development of the anterior basal cuspidule than in P₂. The large protoconid is centrally located and triangular in buccal view. Both anterior and posterior margins bear cristids. In occlusal view the crown is widest posteriorly, where the heel forms an incipient talonid basin. A shallow notch in the buccal profile separates the posterior heel from the protoconid, and the posterior heel forms a protrusive cuspidule with a worn tip.

M₁ is missing, and the lingual alveolar margin of the dentary is missing from the level of the posterior root of P₃, posteriorly. Nevertheless, the buccal side of the M₁ alveolus is separated from P₃ by a diastema of 3.0 mm, whereas there is no diastema between the alveolus of M₁ and M₂.

M₂ consists of two large trigonid cusps (paraconid and protoconid) and three far smaller cusps (hypoconid, hypoconulid and entoconid) surrounding the talonid basin. The most remarkable feature of M₂ (and M₃) is the strength of the precingulid, which forms a sharp-edged shelf that terminates in a well-developed cuspidule on the anterobuccal face of the base of the paraconid. The precingulid slopes steeply posteroventrally from this cuspidule and terminates near the base of the crown level with the anterior margin of the protoconid. The paraconid is well developed and triangular in occlusal view. A minute cuspidule is developed on the anterior slope of the paraconid, near the base of the crown, at about the same level as the anterobuccal cuspidule of the precingulid. The protoconid forms the largest cusp of M₂. In buccal view the anterior margin of the protoconid curves posteriorly towards the apex, however this shape has been created by wear. A distinct, tear-drop shaped wear facet is developed at the tip of the protoconid and extends down its anterior side, along the course of the paracristid. A deep carnassial notch is developed between the paraconid and protoconid, dividing the paracristid into two sections. There is no trace of a protocristid or a metaconid on the posterior side of the protoconid, although the cristid obliqua extends approximately half way up this surface. The cristid obliqua is also divided into two sections by a deep carnassial notch, this time separating the protoconid from the hypoconid. In buccal and lingual view the talonid is set distinctly lower than the trigonid. In occlusal view its buccolingual width is slightly greater than that of the trigonid. All of the cusps of the talonid are highly worn and flat-topped. The largest of these and the most strongly projecting is the buccally placed hypoconid. The posteriorly located hypoconulid has been virtually obliterated by wear and is only recognisable as a slight thickening of the enamel around the posterior rim of the wear facet. The entoconid is placed on the lingual side of the talonid, slightly inset from the lingual margin of the crown. A short ridge extends posterobuccally from its apex towards the hypoconulid. A postcingulid extends obliquely up the posterior surface of the hypoconid from a basally located point on the buccal side to a more apical point on the lingual side.

M₃ is very much like M₂, differing in the following respects. The minute anterior cuspidule at the base of the paraconid is less well-developed. The protoconid is taller, and the cristid obliqua extends for less than half its height up the posterior surface. The talonid is slightly narrower than the trigonid and the hypoconulid forms a low conical cuspid. The posterobuccal ridge extending from the apex of the entoconid joins the anterior base of the hypoconulid, thus creating an oval basin that is separated from, and much smaller than, the main talonid basin. A low but sharp postcristid between the hypoconulid and the entoconid forms the posterolingual margin of this oval basin. The postcingulid is very reduced, and is little more than a shallow depression on the posterior surface of the hypoconid.

The M₃ of SAM P38799 shows some minor differences from the holotype but the main features are the same, including the distinctive precingulid terminating in an anterobuccal cuspidule (Figs. 3–4 and Table 1). In this specimen, the anterior cuspidule at the base of the paraconid is better developed than in either of the two molars of the holotype, and has an incipient preparacristid leading from it towards the apex of the paraconid. The talonid is as wide as the trigonid and the entoconid appears to lack the posterobuccal ridge that links it to the hypoconulid.

Body mass estimate

Calculating from the third lower molar length regression formula, derived from the dasyuromorphian-only data set of Myers (2001), a body mass estimate of 17.8 kg is obtained for the holotype and 15.9 kg for SAM P38799. These estimates are well below the average body mass of 29.5 kg for recent Tasmanian Th. cynocephalus (Paddle, 2000) but are not outside the range estimated for mainland Holocene samples of this species (Letnic, Fillios & Crowther, 2012). However, it is clear that Th. yorkellus was a far smaller species than either Th. potens or Th. megiriani which have estimated body masses ranging between 38.7 and 57.3 kg (Wroe, 2001; Yates, 2014).

Thylacinus megiriani (Murray, 1997)

Holotype. Museum and Art Gallery of the Northern Territory (hereafter designated NTM) P9618, fragmentary left maxilla with P¹⁻³ and M¹⁻³

Referred specimens. NTM P4376, anterior fragment of a right dentary containing the alveoli with broken roots for P₂, P₃ and M₁ and the empty alveolus for M₂ (Figs. 5–7). NTM P4377, posterior fragment of a right dentary containing an incomplete M₄ (Figs. 8–10).

Locality and stratigraphic age. ‘South Quarry,’ an excavation on the south-western side of ‘Hill 1’ (Woodburne, 1967) an erosional remnant of the upper Waite Formation on the Alcoota Fossil Reserve. Ongeva Local Fauna, latest Miocene or earliest Pliocene in age.

Diagnosis. Th. megiriani differs from all other thylacinids by the presence of the following two autapomorphies. There is a a short, lobe-like postcingulum between the metastyle and protocone of M². A hypertrophied torus is developed along the ventrobuccal margin of the dentary so that its buccolingual width at the level of P₃ is greater than 75% of its depth at the same level. The presence of a small stylar cusp E on M² and M³ might also be an autapomorphic reversal within the genus Thylacinus but the optimisation of the character is ambiguous due to its presence in the closely related outgroup, Wabulacinus ridei, and the absence of a preserved M² or M³ for Th. macknessi. Similarly, the presence of a vestigial metaconid and metacristid on M₄ (and presumably the other lower molars) is ambiguous due to the presence of metaconids in Th. macknessi and their absence in Th. potens. This may represent an autapomorphic reversal in Th. megiriani or a case of convergent loss in Th. potens and the Th. yorkellus + Th. cynocephalus clade. Th. megiriani can be further distinguished from Th. macknessi and all non-Thylacinus thylacinids by: its greater size; the great reduction of the size of the paracone relative to the metacone; elongation of the postmetacrista, so that it is greater than 52% the length of the tooth in M² and M³. It can be further distinguished from Th. potens by: the long axis of P¹ aligned with those of the other upper premolars; the length of M¹ is greater than its width; the complete absence of a precingulum on M¹ and M³; the absence of a metaconule on all upper molars; M³ much longer (>5%) than M²; M³ longer than wide; presence of a diastema between P₃ and M₁.

Remarks. The new dentary specimens cannot be referred to Th. megiriani on the basis of autapomorphic characteristics due to their fragmentary nature and lack of overlap with the holotype maxilla. However, the apparent reversal to the presence of an, albeit vestigial, metaconid in NTM P4377 parallels the reappearance of a tiny tubercular stylar cusp E in the upper molars of the holotype, indicating that the two features may be not be independent in thylacinids and provides further support for the referral of the dentary fragment NTM P4377 to Th. megiriani. It is interesting to note that the expression of the metaconid and stylar cusp E consistently covary in all known thylacinids, ie. taxa either possess both cusps or neither, never one or the other.

Furthermore, the new dentary fragments derive from the same level in the same quarry as the holotype, are of an appropriately large size, and can be distinguished from Th. potens, which is the only other large-bodied thylacinid known from a similar time and place. For these reasons it is reasonable to refer them to Th. megiriani.

Description of the new material

The anterior dentary fragment (NTM P4376; Figs. 5–7, Table 2) has only broken roots and empty alveoli and does not contain any tooth crowns. This coupled with poor preservation means that it is not immediately obvious exactly which teeth the roots and alveoli belong to and therefore the disposition of the teeth in the jaw. However, several lines of evidence indicate that the posterior two empty alveoli are for the anterior and posterior roots of M₂, respectively, while the posterior two projecting root stumps are the anterior and posterior roots of M₁. Anterior to the molar alveoli there are the broken roots and alveoli of P₂ and P₃ (Fig. 7). Evidence for this interpretation comes from a partial bridge of finished bone between the second and third preserved alveoli, indicating that there was a diastema between two different teeth in this position, rather than these alveoli representing the anterior and posterior alveoli of the same tooth. The posterior limit of symphyseal surface is level with the anterior margin of the fourth alveolus. As the posterior end of the symphysis draws level with P₃, or at least the diastema between P₂ and P₃ in all thylacinid specimens examined, the third and fourth alveoli in NTM P4376 should represent the anterior and posterior alveoli for P₃. From the alveoli it can be seen that there was no diastema between M₁ and M₂, but there was a short diastema of 3.6 mm between P₃ and M₁, and a slightly longer diastema of 4.0 mm between P₂ and P₃.

The dentary is dorsoventrally shallow but buccolingually thickened, giving it a distinctly robust appearance. Adding to the thickness of the jaw is a strong torus developed along the ventrobuccal margin of the dentary from a point level with the midlength of P₂ to one level with the distal end of M₁. At the maximum thickness of the torus (level with the posterior root of P₃) the dentary measures 20.8 mm wide, which approaches the dorsoventral depth of the dentary at the same level (27.0 mm). The buccal surface of the dentary, above the ventrobuccal torus, is pierced by three mental foramina. The anterior-most mental foramen lies anterior to the alveoli for P₂, the middle foramen lies level with the anterior end of P₃, and the posterior foramen lies level with the posterior end of M₁.

The second specimen (NTM P4377; Figs. 8–10 and Table 2) is a small fragment from the posterior end of a right dentary including a partial M₄ and the anterior base of the coronoid process. The lingual surface of the dentary is gently convex dorsoventrally, whereas the buccal surface bears a ridge that slopes posterodorsally to merge with the leading margin of the coronoid process. The buccal ridge forms the lower border of a flat dorsobuccally facing surface that occurs below the fourth lower molar. The ventral part of the dentary is missing so the depth to height ratio cannot be measured. The incomplete M₄ includes the talonid and the posterior side of the protoconid. The protoconid is far taller than the talonid with an almost vertically oriented posterior margin. The protocristid extends from the apical tip of protoconid to a tubercle about half-way down the posterior face of the protoconid (Fig. 10). This tiny tubercle appears to be a vestigial remnant of the metaconid which is usually absent in derived species of Thylacinus. Two weakly developed cristids extend ventrally from the tubercle. The two cristids diverge at a highly acute angle. The more buccally placed cristid is the cristid obliqua. It extends to the base of the protoconid and up the anterior face of the hypoconid. A shallow, weakly developed carnassial notch is created between the hypoconid and the protoconid. This notch is similar to the weak notches developed on the lower molars of Th. potens and unlike the deep clefts that can be observed in the carnassial notches of the lower molars of Th. yorkellus and Th. cynocephalus. The lingual cristid that branches from the tubercle on the posterior face of the protoconid is interpreted as a metacristid. It terminates at the base of the protoconid without extending onto the talonid. Its presence strengthens the interpretation of the tubercle as a vestigial metaconid. The moderately large, conical hypoconid comprises the talonid which bears no other cusps. The anterobuccal face of the hypoconid is planed off by an oblique wear facet that faces buccodorsally. The weak cristid obliqua extends directly anteriorly from the apex of the hypoconid. The anterolingual rim of the hypoconid is rounded, lacking a postcristid.