The three living species of wombat (Vombatus ursinus, Lasiorhinus latifrons and Lasiorhinus krefftii; family Vombatidae) and the koala (Phascolarctos cinereus; family Phascolarctidae) are among the most iconic and unusual of Australia’s Marsupials. They are the only survivors of the clade Vombatiformes, one of the two major subclades within the order Diprotodontia; the other, Phalangerida, includes Possums, Kangaroos, Wallabies and Rat Kangaroos. Fossil Vombatiforms were highly diverse and included the large-bodied, herbivorous Diprotodontids, the possibly Tapir- or Chalicothere-like Palorchestids, and the carnivorous Thylacoleonid 'Marsupial Lions'. Vombatiformes suffered extensive extinctions during the Pleistocene; the last surviving members of several families (Diprotodontidae, Palorchestidae and Thylacoleonidae) went extinct, as did several Vombatid and Phascolarctid species, leaving the modern Wombats and Koala as the only remnants of this former diversity.
Monophyly of Vombatiformes is relatively uncontroversial, but relationships within the clade remain unclear. In part, this is because the best known forms are also the youngest and most derived. They include the Rhino-sized (over 2 tonne) Diprotodontid Diprotodon optatum, the bizarre Palorchestid Palorchestes azael, and the hypercarnivorous Thylacoleonid Thylacoleo carnifex, all of which are known from relatively complete specimens from the Pleistocene. Among modern forms, the Koala exhibits extreme modifications of its masticatory and auditory anatomy (connected with its diet of Eucalyptus and its sedentary lifestyle, respectively), while the skull and skeleton of Wombats is also highly specialised. Another major difficulty in unravelling Vombatiform relationships is the fact that the premolars and molars of modern Wombats are hypselodont (ever-growing) and their unworn occlusal morphology, a fundamental source of information for Mammalian phylogeny, is lost through wear early in post-natal life.
The discovery of well-preserved remains of more plesiomorphic Vombatiforms from Oligo-Miocene fossil sites in Australia has given additional insight into the evolutionary history of the clade. However, a number of key questions remain unresolved, particularly regarding the origin and interfamilial relationships of Vombatids.
In a paper published in the journal Scientific Reports on 25 June 2020, Robin Beck of the Ecosystems and Environment Research Centre at the University of Salford, and the PANGEA Research Centre at the University of New South Wales, Julien Louys of the Australian Research Centre for Human Evolution at Griffith University, Philippa Brewer of the Department of Earth Sciences at the Natural History Museum, Michael Archer and Karen Black, also of the PANGEA Research Centre at the University of New South Wales, and the late Richard Tedford of the Division of Paleontology at the American Museum of Natural History, describe a cranium and associated partial skeleton of a new, highly distinctive fossil Vombatiform from the late Oligocene (approximately 26-25 Ma old) Pinpa Local Fauna of the Namba Formation, in the Lake Eyre Basin of northeastern South Australia. This taxon is one of the oldest Australian Marsupials known from an associated skeleton. It provides key new data for understanding the evolutionary history of Vombatiforms, including the evolution of digging adaptations, homologies of molar structures, phylogenetic relationships, and patterns of dental and body mass evolution within the clade.
The new species is named Mukupirna nambensis, where 'Mukupirna' derives from the words 'muku' meaning 'bones' and pirna meaning 'big' in the Dieri (Diyari) language traditionally spoken in the area around Lake Eyre and refers to the large size of the animal, and 'nambensis' is after the Namba Formation in which the only known specimen was found.
Due to its uniqueness, Mukupirna nambensis is placed in a new family, the Mukupirnidae, which is defined as differing from known members of Wynyardiidae in possessing a P3 that lacks a posterolingual cusp (hypocone), less well-developed selenodonty, a less well-developed masseteric process, palatal vacuities entirely enclosed by the palatines, a proportionately longer deltopectoral crest and broader distal end of the humerus (Epicondylar index 0.44), a proportionately longer olecranon of the ulna (Index of Fossorial Ability 0.42), and a much larger body size (estimated body mass based on postcranial measurements 143–171 kg); differs from Vombatids in lacking bilobate molars (molars are only slightly bilobate in Nimbavombatus, Rhizophascolonus and Warendja, but strongly bilobate in other Vombatids); differs from all Vombatids except Nimbavombatus in retaining three upper incisors and the upper canine; differs from Nimbavombatus in larger size, more bicuspid P3, and palatal vacuities entirely enclosed by the palatines; differs from Vombatids known from postcranial remains in lacking a laterally extensive deltopectoral crest of the humerus; differs from hypselodont Vombatids in having closed premolar and molar roots; differs from known members of Thylacoleonidae in retaining only a single upper premolar (P3), with this tooth not as elongate or bladelike, lacking a marked reduction in molar size posteriorly, having a proportionately longer deltopectoral crest and broader distal end of the humerus, and having a proportionately longer olecranon of the ulna; differs from known members of Phascolarctidae in lacking strongly selenodont molars, having a less well-developed masseteric process, a proportionately longer deltopectoral crest and broader distal end of the humerus, and a proportionately longer olecranon of the ulna; differs from known members of Ilariidae in lacking posterobuccal and lingual cusps on P3, in lacking strongly selenodont molars, and in lacking a well-developed masseteric process; differs from known members of Diprotodontidae and Palorchestidae in lacking a molariform P3, molars not strongly bilophodont, in lacking a well-developed masseteric process, and in retaining palatal vacuities. Mukupirna nambensis cannot be compared directly with Marada arcanum (the only known representative of the Vombatiform family Maradidae), because Mukupirna nambensis is only known from the cranium and upper dentition whereas Marada arcanum is known only from the lower dentition, and it is possible that they represent the same taxon or are closely related.
The holotype, and only, specimen of Mukupirna nambensis is AMNH FM 102646 (previously, QMAM 168) a badly crushed cranium (preserved length 197 mm; dorsal surface not preserved) with left and right P3-M4, and associated partial postcranial skeleton comprising vertebrae, ribs, left and right scapulae, left humerus, left ulna, left radius, left and right femora, left tibia, left fibula, and parts of the autopodia. The adult dentition is fully erupted, except possibly for M4, which does not appear to be in line with the occlusal surfaces of M1-3 (although this may be the result of post-mortem displacement); the molars are only lightly worn. In the postcranium, most fracturing has occurred at the epiphyseal plates. Collectively this suggests that this individual was probably a late subadult or young adult.
This specimen comes from the Namba Formation at the Lake Pinpa Site C, in the Lake Frome area, of South Australia. The Namba Formation has been correlated with the Etadunna Formation, which has been estimated to be 26-24 million years old (i.e. latest Oligocene) on the basis of isotopic, foraminiferal, magnetostratigraphic and radiometric (rubidium-strontium dating of illite) data. More recently, the Etadunna Formation has been proposed to be 26.1-23.6 million years old based on a best-fit age-model of magnetostratigraphic data26. The Pinpa Local Fauna is the oldest of the three distinct faunal units recovered from stratigraphic levels in the Namba Formation, and has been correlated with the oldest faunal zone (Zone A) of the Etadunna Formation, which has been dated as 25.3-24.9 million years old (chrons 7An and 7Ar) based on magnetostratigraphy. In summary, available evidence suggests a probable age of between approximately 26 and 25 million years for the Pinpa Local Fauna.
The preserved craniodental morphology of Mukupirna appears to be approximately intermediate between that of Wynyardiids such as Namilamadeta and Muramura on the one hand, and that of definitive Vombatids on the other. Based on the shapes of the preserved alveoli, the upper first incisor of Mukupirna was proportionately larger than those of Namilamadeta and Muramura, but the second and third incisors were still present. In Vombatids, I1 is very large and (with the probable exception of Nimbavombatus) is the only upper incisor present. Also based on alveolar evidence, Mukupirna retained a large, single-rooted upper canine, which is a plesiomorphic feature seen in Wynyardiids and several other Vombatiforms, including Phascolarctids, Ilariids, Thylacoleonids and some Diprotodontoids. However, an upper canine is absent in all vombatids described to date except Nimbavombatus.
Although it is somewhat worn or damaged apically, the distinctly bicuspid P3 of Mukupirna clearly differs from the more strongly bladed P3 of Phascolarctids, Thylacoleonids and Wynyardiids, and from the relatively molariform P3 of Ilariids, Diprotodontids and Palorchestids. The P3 of Mukupirna shows vertical ridging or fluting (most obvious on its posterior half); among other Vombatiforms, somewhat similar vertical ridging is seen in the P3 of Wynyardiids, but also Phascolarctids. The P3 of Mukupirna also lacks any lingual cusps, unlike the condition seen in most Vombatiforms besides Thylacoleonids.
Interestingly, the unworn P3 of the extant Common Wombat, Vombatus ursinus, (assuming that this is indeed P3 and not a retained dP3) is also bicuspid and without evidence of lingual cusps, although it differs in being proportionately much smaller and more distinctly bicuspid, and in lacking vertical ridging. However, the P3 (again, if it is not dP3) of Lasiorhinus latifrons and of fossil Vombatids is sub-triangular and not bicuspid, and a small posterolingual cusp is typically present. Thus, the bicuspid P3 of Mukupirna and Vombatus may be the result of homoplasy; overall, the P3 of Mukupirna most closely resembles that of the Wynyardiid Namilamadeta. However, the enamel on the labial surface of P3 of Mukupirna extends onto the root of the tooth, which is a distinctive feature of most Vombatids but apparently absent in other Vombatiforms, including Wynyardiids.
The molar morphology of Mukupirna is very similar to that of the Wynyardiids Namilamadeta and Muramura in being selenolophodont, i.e. exhibiting both lophodont and selenodont features. Prominent stylar cusps are present along the labial margin of the molars, with the paracone and metacone located further lingually, but they are not as centrally placed on the tooth crown as they are in Wynyardiids. A weak selenodont pattern is apparent in Mukupirna, at least on M1-2, with identifiable pre- and postparacristae and pre- and postmetacristae extending from the paracone and metacone respectively; these crests are also present, but much better developed, in Wynyardiids. Weak lophs connect the paracone to the protocone and the metacone to the metaconular hypocone in Mukupirna, again as in Wynyardiids. The upper molars of most other Vombatiforms are either fully lophodont (Diprotodontids, Palorchestids), fully selenodont (Ilariids, Phascolarctids), or bunodont-bunolophodont (Thylacoleonids). Intriguingly, however, the same basic occlusal morphology seen in Mukupirna occurs in unworn upper molars of the plesiomorphic fossil Vombatids Nimbavombatus and Rhizophascolonus and juveniles of the living Wombats Vombatus and Lasiorhinus, in which homologues of the paracone and metacone retain traces of a selenodont pattern, but are also connected by weak lophs to the protocone and metaconular hypocone respectively.
In contrast to Wynyardiids and most other Vombatiforms (in which a distinct cervix usually separates the root and crown), there is no clear distinction between the root and crown of the upper molars of Mukupirna. This is another feature also seen in Vombatids, although Mukupirna differs from Vombatids in that the enamel of its molars does not extend down the lingual surface of the roots. The molar roots of Mukupirna are long, and the lingual roots extend ventrally far beyond the molar alveoli, whereas the labial roots are not visible, again as in Vombatids. However, Mukupirna has closed molar roots, and so in this respect is unlike all known Vombatids except the early Miocene Nimbavombatus and Rhizophascolonus. The occlusal surface of early Miocene vombatids was subject to moderate to severe amounts of wear (particularly towards the anterior end of the molar row). The most extreme wear is seen in Rhizophascolonus crowcrofti where the occlusal morphology appears to have been obliterated relatively early in the Animal’s life, leaving an enamel perimeter surrounding a dentine wear surface, as is the case in hypselodont vombatids. By contrast, the molars of Mukupirna evidently retained their original occlusal morphology into at least early adulthood, with no evidence of accelerated wear in the M1 position.
Although the anterior part of the zygomatic arch is poorly preserved in AMNH FM 102646, the masseteric process appears to be weakly developed or absent, as it is in Vombatids; by contrast, this process is distinct in most other Vombatiforms, including Wynyardiids, although it is also absent in most Thylacoleonids. In most vombatids (with the notable exceptions of Warendja and Nimbavombatus), a very large fossa extends across the lateral surface of the maxilla and jugal, at the anterior end of the zygomatic arch; in extant Wombats, this fossa has been shown to house a greatly enlarged superficial masseter, which generates high occlusal forces during the medially-directed power stroke of the lower jaw. The Wynyardiids Muramura and Namilamadeta have a similarly-positioned but much smaller fossa, and the preserved part of the jugal of Mukupirna also preserves a shallow but distinct fossa on its lateral surface. The presence of this fossa in Wynyardiids and Mukupirna might represent a precursor of the much larger fossa seen in most Vombatids, in which case, its absence in Warendja and Nimbavombatus is presumably secondary; alternatively, it may instead reflect the presence of enlarged snout musculature in Wynyardiids and Mukupirna. Palatal vacuities are present in Mukupirna and appear to be entirely enclosed by the palatine bones, as in most Vombatids and also the modern Koala, Phascolarctos cinereus. In Wynyardiids, Thylacoleonids, fossil Phascolarctids and the Vombatid Nimbavombatus, these vacuities are between the maxillae and palatines, which is probably the plesiomorphic condition within Marsupialia. Diprotodontids and palorchestids lack palatal vacuities.
The glenoid fossa of Mukupirna is planar, and the postglenoid process also appears to have been either very weakly developed or entirely absent, although the region may be damaged in AMNH FM 102646. By contrast, in most Vombatiforms (including Wynyardiids), the glenoid fossa has a raised articular eminence anteriorly and distinct mandibular fossa posteriorly, and the postglenoid process is well-developed. The overall morphology of the glenoid region of Mukupirna somewhat resembles that of the plesiomorphic Vombatid Warendja wakefieldi, which is also planar with a very weakly developed postglenoid process. The glenoid region of other Vombatids known from cranial remains is highly specialised, with a mediolaterally broad and convex glenoid fossa that lacks any trace of a postglenoid process. The auditory region of AMNH FM 102646 is also damaged, but it appears that the zygomatic epitympanic sinus of the squamosal was either absent or very shallow; Vombatids are also characterised by a shallow and laterally open zygomatic epitympanic sinus, in contrast to most other Diprotodontians in which this sinus is largely enclosed and invades deep into the zygomatic process of the squamosal.
The postcranium of Mukupirna exhibits a number of features that are indicative of digging behaviour. The humerus is broad distally, giving an Epicondylar Index (humeral epicondylar width/humeral length) of 0.44, similar to that of modern Wombat species Lasiorhinus latifrons and Vombatus ursinus. The olecranon process of the ulna is elongate in Mukupirna, providing increased mechanical advantage to the triceps brachii when extending the forearm, which again is common in digging Mammals. The Index of Fossorial Ability (olecranon length/(total ulnar length-olecranon length)) of Mukupirna is 0.42, which is similar to Vombatus ursinus but somewhat less than in Lasiorhinus latifrons. A distinct third trochanter is present on the femur, which is unusual among marsupials, and which indicates the presence of well-developed gluteal musculature that may be connected with digging behaviour. A distinct third trochanter is present on the femur, which is unusual among Marsupials, and which indicates the presence of well-developed gluteal musculature that may be connected with digging behaviour. The wide distal end of the humerus of Mukupirna and Vombatids is largely due to a prominent medial epicondyle, which reflects the presence of powerful extensors and pronators of the forearm. The manual and pedal phalanges of Mukupirna are strikingly similar to those of Vombatids and also the Ilariid Ilaria, and are strongly indicative of digging behaviour: they are dorsoventrally flattened (especially the unguals), and their distal ends taper dorsoventrally. In addition, manual phalanx V of Mukupirna exhibits a medial twist at its distal end, and lateral buttressing of its proximal end, which is a distinctive feature also seen in Vombatids; this morphology serves to position digit V close to digits I-IV, and in living Wombats allows the manus to be used as a shovel during digging.
However, the humerus of Mukupirna differs from those of Vombatids in lacking a hypertrophied deltopectoral crest that extends laterally beyond the edge of the humeral shaft and forms a tunnel-like fossa for the origin of the brachialis muscle.
Phylogenetic analysis of 79 craniodental and 20 postcranial characters using undated Bayesian inference (using the Mkv model and an eight category lognormal distribution to accommodate rate heterogeneity between characters) places Mukupirna as sister to Vombatidae, with high support (Bayesian posterior probability 0.98), and hence a member of Vombatoidea as defined by Beck et al. Maximum parsimony analysis of the same matrix also recovers a Mukupirna + Vombatidae clade, although with relatively low support (bootstrap 45%). The Bayesian analysis and the maximum parsimony analysis identify the same four features as unambiguous synapomorphies of Mukupirna + Vombatidae: prominent lingual cusp on P3 absent; enamel extending down the buccal surface of P3 and onto the root present; articular eminence of glenoid fossa planar or concave, and mandibular fossa absent or indistinct; postglenoid process absent or weakly developed.
Monophyly of all currently recognised Vombatiform families represented by two or more terminals is supported by a Bayesian posterior probability of over 0.5, except Wynardiidae, with a Bayesian posterior probability of 0.36. Thylacoleonidae is sister to the remaining Vombatiforms, rather than (as in most previous studies) a member of Vombatomorphia; however, this relationship was also found by Anna Gillespie, Michael Archer, and Suzanne Hand in a study published in 2016. Monophyly of Vombatomorphia and Phascolarctidae to the exclusion of Thylacoleonidae receives relatively strong support (Bayesian posterior probability 0.76). Within Vombatomorphia, Vombatoidea and Diprotodontoidea are very weakly supported as sister-taxa (Bayesian posterior probability 0.16), with similarly weak support (Bayesian posterior probability 0.26) for Wynyardiidae as sister to this clade to the exclusion of Ilariids.
Ancestral state reconstructions of Epicondylar Index and Index of Fossorial Ability values on the Bayesian majority rule consensus retaining all compatible partitions with Bayesian posterior probability of less than 0.5 using StableTraits indicate that the high Epicondylar Index and Index of Fossorial Ability values shared by Mukupirna and Vombatids are homologous; median ancestral state reconstructions values for Vombatoidea are 0.43 for Epicondylar Index and 0.41 for Index of Fossorial Ability, which are very similar to the values for Mukupirna and Vombatus ursinus.
Ancestral state reconstructions for presence or absence of P1 and P2 on the same topology and branch lengths using a single rate Mk model in Mesquite provide strong support for the hypothesis that loss of P1 and P2 occurred within Vombatiformes after the divergence of Thylacoleonids (with loss of P2 occurring independently within Thylacoleonidae), but before the split between Vombatomorphia and Phascolarctidae. Ancestral state reconstructions of molar type using the same Mk model indicates that bunolophodonty is ancestral for Vombatiformes, with subsequent acquisition of selenodonty by the Vombatomorphia+Phascolarctidae lineage, which was retained by Phascolarctids and ilariids. After the divergence of Ilariids, the remaining Vombatomorphians evolved selenolophodonty, which was retained by Wynyardiids and Vombatoids (including Mukupirna), with Diprotodontoids subsequently evolving fully lophodont molars. This scenario for the evolution of different molar types within Vombatiformes is attractive in its simplicity, but we acknowledge that it is partially dependent on some very weakly supported relationships, specifically the position of Wynyardiidae and Ilariidae as successive sister taxa to Diprotodontoidea + Vombatoidea.
Estimated body mass of Mukupirna using the 'total skull length' regression equation of Troy Myers is 46 kg, which seems implausibly low given the size of the postcranium. Beck et al. note that the largest specimen used by Myers in calculating his regression equations was a 70 kg Macropus individual (species unspecified), and so use of these equations to estimate body masses of likely larger extinct taxa involves extrapolation beyond the data used to calculate them. The overall proportions of Mukupirna and several other extinct Vombatiforms (e.g. Diprotodontoids, Thylacoleonids) also appear very different from the extant Marsupial species used by Myers to produce the regression equations. Beck et al. have therefore used the postcranial regression equations presented by Hazel Richards, Rod Wells, Alistair Evans, Erich Fitzgerald and Justin Adams, which were produced using a dataset of Mammalian and non-Mammalian taxa with body masses that collectively span from 51 g to 6.4 tonnes. These give body mass estimates for Mukupirna of 143 kg based on femoral circumference only, 160 kg based on combined humeral and femoral circumference, and 171 kg based on humeral circumference only. The humerus of Mukupirna is very robust, and so the estimates that incorporate humeral circumference might be inflated, as Richards et al. also suggested for the Vombatiform Palorchestes; nevertheless, it seems likely that Mukupirna exceeded 100 kg. This is compared to an average weight of 32 kg for the largest living Vombatiform, the northern Hairy-nosed Wombat, Lasiorhinus kreffttii.
Compared to estimated body masses for selected other late Oligocene Vombatimorphians, Mukupirna is much larger than the Wynyardiid Muramura williamsi (16–20 kg), slightly larger than the Diprototodontid Ngapakaldia bonythoni (about 119 kg), but somewhat smaller than the Ilariid Ilaria illumidens (about 215 kg). Ancestral state reconstruction of body mass on the Bayesian majority rule consensus retaining compatible partitions with Bayesian posterior probability of less than 0.5 (in which branch lengths are proportional to the estimated amount of change in the morphological characters used to infer the phylogeny) using StableTraits suggest a median body mass of 5.5 kg for the last common ancestor of Vombatiforms, which is only slightly smaller than the modern Koala, Phascolarctos cinereus. The Thylacoleonid Microleo attenboroughi from the early Miocene of Riversleigh World Heritage Area has not been included by Beck et al. due to its comparative incompleteness; however, it has an estimated body mass of 590 g, and, if it is sister to all other known Thylacoleonids (as found by Gillespie et al.), then it implies an even smaller ancestral body mass for Vombatiformes, possibly under 1 kg. Beck et al.'s StableTraits analysis of body mass indicates independent evolution of very large (over 100 kg) size at least six times within Vombatiformes. For the Thylacoleonid Thylacoleo carnifex, Beck et al. have used the mean of estimates based on postcranial measurements (57.2 kg); however, they note other studies have found that body mass of some Thylacoleo carnifex individuals may have exceeded 100 kg, in which case it would represent a seventh independent evolution of over 100 kg body mass within Vombatiformes.
Beck et al.'s phylogenetic results imply that postcranial digging adaptations, such as large Epicondylar Index and Index of Fossorial Ability, evolved in Vombatoidea in an ancestor that still retained a somewhat Wynyardiid-like, selenolophodont molar dentition (as seen in Mukupirna and also the plesiomorphic Vombatids Nimbavombatus and Rhizophascolonus), rather than the specialised, hypselodont molars characteristic of later Vombatids. However, Mukupirna lacks a laterally extensive, flange-like deltopectoral crest, which is present in all vombatids known from associated postcranial material. This modified deltopectoral crest creates an enclosed, tunnel-like fossa for the origin of the brachialis muscle, and is likely a specialised fossorial adaptation. Absence of this feature in Mukupirna, together with its large size (over 100 kg) means that Mukupirna may not have been capable of the true burrowing behaviour of modern Wombats. Instead, it may have used scratch-digging to access subterranean food items, such as roots and tubers, as has also been proposed for Ilaria and Rhizophascolonus.
It has been suggested, based on the ecology and relatively close relationship of modern Wombats and Koalas, that Vombatiforms (and also other Diprotodontians) have been characterised by 'long-term maintenance of ecological niche differentiation' However, the evidence from the fossil record of Vombatiformes clearly demonstrates that this is an artefact of the relictual nature of the three extant representatives: known fossil Vombatiforms range in size from small (under 5 kg) Oligo-Miocene Phascolarctids and Thylacoleonids, to the Rhino-sized (over 2 tonne) Diprotodon from the Pleistocene, and collectively span a diverse range of morphologies and ecologies, including several that lack obvious modern analogues (at least in the Australian Mammal fauna). The over 100 kg Mukupirna, together with similarly sized contemporaneous taxa such as the Ilariid Ilaria and the Diprotodontid Ngapakaldia, demonstrates that multiple Vombatiform lineages had already evolved very large (over 100 kg) body size by the late Oligocene, and possibly considerably earlier. In total, our ancestral state reconstructions indicate at least six independent origins of body masses over 100 kg within Vombatiformes, from an ancestor that is estimated to have been between 1 and 5.5 kg, the upper bound being slightly smaller than the mean mass of the living Koala, Phascolarctos cinereus. In this respect, Vombatiforms resemble another Mammalian clade with relictual modern diversity but for which far more diverse fossil representatives are known, namely Sloths.
See also...
Online courses in Palaeontology.
Follow Sciency Thoughts on Facebook.
Follow Sciency Thoughts on Twitter.
