A new Early–Middle Miocene phalangerid (Marsupialia: Phalangeridae) from the Riversleigh World Heritage Area, Boodjamulla (Lawn Hill) National Park, northwestern Queensland

Abstract

Archerus johntoniae represents a new genus and species of phalangerid possum from the Early to Middle Miocene of the Riversleigh World Heritage Area, northwestern Queensland. The new taxon is the tenth extinct phalangerid species to be described and is the seventh species recognized from the diverse Riversleigh fossil assemblages. A. johntoniae has its closest morphometric affinities (such as relatively short molars, narrow premolars, an anteriorly narrow upper first molar, and posteriorly narrow fourth molars) with trichosurine and stem phalangerids. Morphologically, A. johntoniae has similarities to phalangerines and trichosurines, but is most dissimilar to other fossil phalangerids. The new species is one of the smallest (ca 1.3 kg) phalangerids known and may have occupied Early and Middle Miocene frugivorous/insectivorous niches demarcated from other phalangerids by body size and diet. A. johntoniae is one of a minority of Riversleigh species to have survived a significant decline in marsupial diversity associated with a rapid spike in global temperature at the Middle Miocene Climatic Optimum.

Troy Myers [[email protected]], Earth and Sustainability Science Research Centre, School of Biological, Earth & Environmental Sciences, UNSW Sydney, NSW 2052, Australia; Kirsten Crosby [[email protected]], GHD Group Pty Ltd, Level 15, 133 Castlereagh St, Sydney, NSW, 2000 (currently), School of Biological, Earth & Environmental Sciences, UNSW Sydney, NSW 2052, Australia.

Keywords:

• Neogene
• possums
• brushtail
• Trichosurinae
• Middle Miocene Climatic Optimum (MMCO)

THE PHALANGERIDAE comprises at least 29 extant species in six genera (Helgen & Jackson 2015, Kealy et al. 2020). The three taxa proposed by Ruedas & Morales (2005): the Trichosurinae (Trichosurus Lesson, 1828 and Wyulda Alexander, 1919), Phalangerinae (Phalanger Storr, 1780 and Spilocuscus Gray, 1862) and Ailuropinae (Strigocuscus Gray, 1862 and Ailurops Wagler, 1830), are supported by molecular and total-evidence studies (Kirsch & Wolman 2001, Raterman et al. 2006, Meredith et al. 2009, Beck et al. 2022). Phalangerids, including the commonly named ‘cuscuses’, ‘scaly-tailed’, and ‘brush-tailed’ marsupial possums, are native to the rainforests, sclerophyllous forests, and woodlands of Australia (Trichosurus, Wyulda, Phalanger and Spilocuscus spp.), New Guinea, Timor, Eastern Indonesia, and surrounding islands (Ailurops, Strigocuscus, Phalanger and Spilocuscus spp.). Most species are medium-sized (ca 1–3 kg), arboreal and folivorous, although some species are more frugivorous to carnivorous, or can be more terrestrial (e.g., some Trichosurus and Ailurops spp.) or have large body sizes (e.g., up to 7 kg in some Spilocuscus spp., and up to 10 kg in some Ailurops spp.) (Flannery 1994, Helgen & Jackson 2015).

Nine fossil phalangerid species in four genera (Crosby (2002) concluded that ‘Trichosurus’ dicksoni Flannery & Archer, 1987 should be referred to a new, but as yet undescribed, genus), are currently recognized (Table 1). Here we describe another Early to Middle Miocene phalangerid from the Riversleigh World Heritage Area, Boodjamulla (Lawn Hill) National Park, in Waanyi country, northwestern Queensland (Fig. 1), Archerus johntoniae gen. et. sp. nov. The new taxon can be morphologically and morphometrically distinguished from all other phalangerids. Its closest morphometric affinities are with extant trichosurines, fossil species of Onirocuscus Crosby, 2007 and ‘Trichosurus’ dicksoni, with the greatest differentiation being from extant phalangerines and ailuropines. A. johntoniae shares some morphological characters with trichosurines and phalangerines but is most dissimilar to other fossil species.

Figure 1. Location of the Riversleigh World Heritage Area, northwestern Queensland, and localities where Archerus johntoniae has been recovered (modified from Myers et al. 2017).

Table 1. Described fossil phalangerid species.

Genus	Species	Species author	Age	Location	Faunal zone^a
Trichosurus	hamiltonensis	Flannery et al., 1987	Early Pliocene	Hamilton, Vic.	–
‘Trichosurus’	dicksoni	Flannery & Archer, 1987	Middle-late Middle Miocene	Riversleigh, Qld	C & D
Archerus	johntoniae	Present study	Early-Middle Miocene	Riversleigh, Qld	B & C
Onirocuscus	notialis	(Flannery et al., 1987)	Early Pliocene	Hamilton, Vic.	–
	rupina	Crosby, 2007	Middle Miocene	Riversleigh, Qld	C
	reidi	(Flannery & Archer, 1987)	Middle Miocene	Riversleigh, Qld	C
	inversus	Crosby, 2007	Early Miocene	Riversleigh, Qld	B
	silvacultrix	Crosby, 2007	Early Miocene	Riversleigh, Qld	B
Wyulda	asherjoeli	Crosby, Nagy & Archer, 2001	Early Miocene	Riversleigh, Qld	B
Eocuscus	sarastamppi	Case, Meredith & Person, 2008	Late Oligocene	Lake Palankarinna, S.A	Etadunna Fmn B

^aRiversleigh Faunal Zone (A–D) as defined by Arena et al. 2016, unless otherwise stated.

Materials and methods

Definitions and abbreviations

Dental nomenclature follows Beck et al. (2022), where premolars are numbered 1–3, molars 1–4, and there is a deciduous P3 and p3. Molar cusp and blade terminology follow Archer (1984). QM F, catalogue number in the Queensland Museum Geosciences fossil collection; AR, catalogue number in the Vertebrate Palaeontology mammal collection at UNSW Sydney; AMNH, catalogue number in the American Museum of Natural History collection; AM M, catalogue number in the mammal collection of the Australian Museum; MV C and DTC, catalogue number in the Museums Victoria mammal collection; ANWC CM and M, catalogue number in the Australian National Wildlife Collection (CSIRO); WA M, catalogue number in the Western Australian Museum mammal collection.

Morphometric analysis

Thirty dental variables were measured, using digital callipers, including upper and lower: molar row lengths (UMRL, LMRL), third premolar length and maximum width (P3L, P3W, p3L, p3W), molar lengths (M1–4L, m1–4L), and maximum anterior and posterior widths on all molars (M1–4A, M1–4P, m1–4A, m1–4P). The dataset included 103 extant phalangerid specimens: Trichosurus vulpecula vulpecula (Kerr, 1792) (n = 14); Trichosurus vulpecula arnhemensis Collett, 1897 (16); Trichosurus vulpecula fuliginosus (Ogilby, 1831) (12); Trichosurus johnstonii (Ramsay, 1888) (13); Trichosurus caninus (Ogilby, 1836) (16; five of these are from Victoria and may therefore be Trichosurus cunninghami Lindenmayer, Dubach & Viggers, 2002 but have not yet been reidentified as such); Wyulda squamicaudata Alexander, 1919 (6); Spilocuscus maculatus (Geoffroy Saint-Hilaire, 1803) (9); Phalanger gymnotis (Peters & Doria, 1875) (9); Phalanger orientalis (Pallas, 1766) (3); Phalanger vestitus (Milne-Edwards, 1877) (3); Strigocuscus celebensis (Gray, 1858) (1) and Ailurops ursinus (Temminck 1827 [1824]–1841) (1). Thirty-five (35) fossil phalangerids were also included: Archerus johntoniae nov. gen. et. sp. (6); ‘Trichosurus’ dicksoni (13); Onirocuscus inversus Crosby, 2007 (3); Onirocuscus silvacultrix Crosby, 2007 (3); Onirocuscus rupina Crosby, 2007 (2) and Onirocuscus reidi (Flannery & Archer, 1987) (8). All specimens examined and their measurements (mm) are listed in the Supplemental Material. Morphometric data were log transformed to reduce variation resulting from size and allometry, and to normalize residuals (Hammer & Harper 2006, Marramà & Kriwet 2017), before being subjected to Canonical Variates Analysis using PAST software (version 4.09; Hammer et al. 2001). Data for Trichosurus and Onirocuscus species were combined into their respective genera to examine generic variation.

Systematic palaeontology

ORDER DIPROTODONTIA Owen, 1877

SUBORDER PHALANGERIDA Aplin & Archer, 1987

FAMILY PHALANGERIDAE Thomas, 1888

SUBFAMILY ?TRICHOSURINAE Kirsch & Wolman, 2001

Archerus gen. nov.

LSID. urn:lsid:zoobank.org:act:AAE6685A-E251-4D2A-89F3-172307CACA7B

Type species

Archerus johntoniae sp. nov.

LSID. urn:lsid:zoobank.org:act:584F138B-06C4-4229-806F-639A7C7826BB

Diagnosis

As for the holotype and only known species.

Etymology

In honour of Professor Michael Archer, for his remarkable contribution to Australian palaeontology over 50 years, and for his ongoing mentorship.

Archerus johntoniae sp. nov.

(Figs 2, 3)

Figure 2. A, A’, Stereophotograph of the holotype right dentary of Archerus johntoniae (QM F40103). Teeth present are p3, m1–4. B, Lingual view. C, Buccal view.

Figure 3. A, Buccal view of the paratype right maxilla of Archerus johntoniae (QM F50418). Teeth present are P1, P3, M1–3. B, B’, Stereophotograph of occlusal view.

Diagnosis

Archerus johntoniae is distinguished from other phalangerids by: a large difference between third premolar width and molar length (i.e., A. johntoniae has a wide P3 and p3 combined with relatively short upper and lower molars; except Wyulda squamicaudata); a crest between cuspules on the P3 that is approximately 40% of the total tooth length along the longest axis (whereas it is greater than 50% in other phalangerids, except W. squamicaudata and some Phalanger vestitus); a strongly buccally convex postparacrista on M2 (as opposed to being straight); a strongly buccally convex postmetacrista forming a continuous curve with the posterior cingulum on M1 (except some O. spp. and P. vestitus; as opposed to a straight postmetacrista, sometimes forming an angular junction with the posterior cingulum, in other phalangerids), M2 and M3 (except in some Spilocuscus maculatus); a small basin present medially on the buccal side of the crown in line with the interloph valley on M3 (absent or larger in other phalangerids); a moderately large fossa formed with the protolophid posterolingual to the protoconid on m2 (absent to very small in other phalangerids); a large, rectangular fossa originating anterior to the protoconid and directed towards the centre of the protolophid on m3 and m4 (absent to medium-sized in other phalangerids, except for ‘Trichosurus’ dicksoni); a very fine fissure divides the cristid obliqua on m4 (much longer and wider cleft in other phalangerids, or absent in some P. spp.); a near continuous lingual ridge on m4 that is formed by the postmetacristid, pre- and postentocristid, with only a fine lingual break medially (in other phalangerids the cristids descend medially into a larger cleft, except in some Trichosurus specimens); well developed, relatively straight lophids on m4 (in other phalangerids at least one lophid not as well developed and with greater medial kink).

Archerus johntoniae is differentiated from Trichosurus spp. by: shorter anterior upper molars; shorter lower molars (particularly m2–4); a narrower m4; a posteriorly narrower m3; a shorter upper molar row; presence of a well-developed parastyle on M1; absence of a protoloph on M1; a metaloph with a slight kink on M1; the presence of a medial basin on the buccal side of the crown on M1 and M2 (separated from the interloph valley by the postparacrista and premetacrista); more extensive crenulations; a less well developed protoloph on M2; a straighter lower molar row; a symphysis that is more deflected from the main axis of the dentary (about 155° as opposed to ca 165°); a fine fissure dividing the cristid obliqua on m3; a small cleft present buccal to the cristid obliqua on m4 (as opposed to a much wider and deeper fossa).

Archerus johntoniae is distinguished from W. squamicaudata by: longer molars, particularly anterior upper molars, and posterior lower molars; a wider m4; a posteriorly wider m3; longer molar row lengths; the absence of a protoloph on M1; a poorly developed buccal basin on M1 and M2 (as opposed to none); the presence of crenulations; less curvature of the anterobuccal margin of M2; a better-developed fossa adjacent to the protoloph and anterior cingulum on M2; a more inflated posterolingual corner of the crown on M2; a better-developed metaloph on M3; a straighter lower molar row; a more ovate p3; a cristid obliqua on m2 and m3 that is divided by a fine fissure (as opposed to the absence of a fissure); a smaller buccal cleft on m2, m3 and m4; the presence of a small fossa posterolingual to the protoconid on m2 and m3; a better developed and straighter protolophid, as well as the presence of a hypolophid, on m4.

Archerus johntoniae is distinguished from Phalanger spp. by: a narrower p3 and P3; an anteriorly narrower M1 and m4; a posteriorly narrower M4 and m4; shorter molars; shorter total upper molar row; a slightly curved anterobuccal crown margin on M1 (rather than straight or very slightly curved); the presence of a very small medial basin on the buccal side of the crown of M2 (either absent or on lingual side of postparacrista in other Phalanger spp.; except for some Phalanger orientalis); presence of a fossa of uniform width between the protoloph and anterior cingulum on M2; a better-developed hypolophid with only a slight kink on m4.

Archerus johntoniae is distinguished from S. maculatus by: much shorter lower and upper molars; a narrower m3, m4 and M4 posteriorly; a narrower M4 anteriorly; a much shorter total upper molar row; a larger basin between the lingual cusps and extending to the lingual edge on M1; a more curved anterobuccal margin on M1, M2 and M3; fewer and less well-developed crenulations on all molars; presence of a fossa of uniform width between the protoloph and anterior cingulum on M2; presence of a small parastylid on m1; a better developed hypolophid with only a slight kink on m4.

Archerus johntoniae is distinguished, morphometrically, from Strigocuscus celebensis by: a narrower p3 and P3; an anteriorly narrower M1 and M2; a posteriorly narrower M4; a shorter M3, M4, m1, m2, m3 and m4; and a shorter total upper molar row.

Archerus johntoniae is differentiated, morphometrically, from Ailurops ursinus by: very much shorter lower and upper molars; a much narrower m3, m4 and M4 posteriorly; a much narrower M4 anteriorly; and a very much shorter total upper molar row.

Archerus johntoniae is distinguished from Onirocuscus spp. by: an occlusal surface on P3 that aligns with the occlusal surfaces of the molars, rather than being much higher; shorter lower and upper molars; a narrower m4; a posteriorly narrower M4; a shorter total upper molar row; the absence of a protoloph on M1; a lingual basin between the lingual cusps on M1 that is not closed at its distal extremity; a more curved anterobuccal crown margin on M1, M2 and M3; a less well-developed protoloph on M2; a wider lingual basin between the lingual cusps on M2 and M3; a metaloph that has a kink on M2; a postmetacrista that has a slight convex curve on M2 rather than being posteriorly directed; a fossa between the protoloph and anterior cingulum on M2 that is of uniform width, rather than being wider buccally; a less rounded posterolingual crown below the metaconule on M2; a less well-developed metaloph on M3; a p3 that is more ovate and without a buccally bent anterior end; a slightly larger parastylid on m1 (if present in Onirocuscus spp.); a cristid obliqua that is divided by a much finer fissure on m3 (for some Onirocuscus spp.); a slightly wider buccal cleft between the lingual cuspids that extends to distal margin on m3; a postmetacristid and pre-entocristid that do not descend medially into a buccal cleft on m4; much better-developed, and less kinked, lophids on m4.

Archerus johntoniae is distinguished from ‘T.’ dicksoni by: shorter lower and upper molars; a shorter total upper molar row; a narrower m4; a posteriorly narrower M4; a cusplet blade on P3 that is not buccally offset relative to the long axis; a P3 that is anteriorly triangular, rather than near hemispherical; a smaller parastyle on M1; a posterior cingulum that is much better developed on all molars (variably absent in ‘T.’ dicksoni); the absence of at least a partially developed protoloph on M1; a metaloph with a slight kink on M1; a narrower and shallower lingual medial basin between the lingual cusps on M1 and M2; a more buccally convex anterobuccal crown margin on M1; fewer crenulations on all molars, particularly in the central basin; a less well-developed protoloph on M2; a fossa between the protoloph and anterior cingulum on M2 that is of near-uniform width for its length, rather than much wider buccally; a posterolingual corner on M2, below the metaconule, that is sub-rounded and partly inflated, rather than being poorly developed and deflated; a much better-developed metaloph on M3; the presence of a very small medial basin on the buccal side of the postparacrista and premetacrista on M3; a more curved, less angular, anterobuccal margin on M3; a lower molar row that is relatively straight, without buccal curve; a symphysis that is more deflected from the axis of the dentary (i.e., about 155°, rather than 165°); a p3 that is more ovate; the posterobuccal margin of the p3 not being swollen; the absence of a large kink in the anterior half of the cristid obliqua on m1; a wider posterior cingulum; a more buccally positioned parastylid on m1; a very fine fissure dividing the cristid obliqua on m2; a larger medial cleft on the buccal margin between the buccal cuspids on m2 and m3; a longer posterior cingulum on m2 and m3; a postmetacristid and pre-entocristid that do not descend medially into a buccal cleft on m4; cuspids smaller on m4; lophids on m4 that are less kinked medially.

Etymology

For John and Toni Myers and their unwavering support of a palaeontological career.

Holotype

QM F40103, right dentary with p3–m4 and alveoli for i1–p2 (Fig. 2).

Referred material

Paratype: QM F50418, right maxilla with P1, P3–M3 and alveoli for P2 and M4 (Fig. 3). QM F41222, a partial skull including a partial left dentary with p3–m4 (anterior to p3 missing; ascending ramus, articular condyle and angular process missing; p3 mostly missing; m1 missing anterobuccal corner; m4 missing posterior cingulum), a partial right dentary with p3–4 (anterior to p3 missing; tip of coronoid process missing; lingual side of p3 missing), basicranial fragment containing periotic, a left maxillary fragment with P3, M1–3 (exhibiting moderate wear) and alveoli for M4, a right maxillary fragment with P3, M1–3 and alveoli for M4 (lingual half of P3 missing; all but posterolingual corner of M1 missing; anterobuccal corner of M2 missing; some wear to M2 and M3), left premaxillary fragment with the canine and alveoli for I1–3; QM F23088, a right dentary containing p3–m4 (some wear); QM F42705, a partial right dentary containing p3–m2 and the alveoli for i3, p2, m3–4 (moderate wear to molars); QM F24047 a partial right dentary with m2–4.

Type locality, unit and age

All specimens were discovered in the Oligo-Miocene freshwater carbonate deposits of the Riversleigh World Heritage Area, Boodjamulla (Lawn Hill) National Park, in Waanyi country, northwestern Queensland (Fig. 1). The holotype (QM F40103), paratype (QM F50418), and QM F41222 were recovered from micritic limestone from AL90 site, on the central Gag plateau (Fig. 1), interpreted as a pitfall-trap cave (Arena et al. 2014) and radiometrically (U/Pb) dated at approximately 14.7 Ma (Langhian Stage; Woodhead et al. 2016). QM F23088 and QM F24047 are from, respectively, Last Minute and Group Site, on the northern Gag plateau (Fig. 1). Arena et al. (2016) also allocated the Last Minute and Group Site LFs to the middle Miocene (ca 16−11.6 Ma) Faunal Zone C. This was supported by assignment of the Last Minute LF to the Nimbadon & Trichosurus palaeocommunity (Myers et al. 2017), which included Ringtail LF with a radiometric date of approximately 13.6 Ma (Serravallian Stage; Woodhead et al. 2016). QM F42705 is from Cadbury’s Kingdom site, near Godthelp’s Hill on the southern D-site plateau (Fig. 1). Arena et al. (2016) assigned the Cadbury’s Kingdom LF to biostratigraphic Faunal Zone B (interval B3). Neville’s Garden LF and Camel Sputum LF, also from interval B3, were dated at approximately 18 Ma (Burdigalian Stage; Woodhead et al. 2016), suggesting that Cadbury’s Kingdom LF may be of a similar age. The chronological range of Archerus johntoniae, as currently understood, is therefore approximately 18–12 Ma (Burdigalian to Serravallian Stages). Precise locality coordinates may be obtained by request to the Queensland Museum, Geosciences.

Description

Dentary

The dentary (Fig. 2) is relatively straighter than in Trichosurus spp., with less deflection of the symphysis. The maximum point of curvature is below m1 ventrally. The origin of the ascending ramus is ventral to the border between m3 and m4. Alveoli are present for i1–3 and a very small single-rooted p2, this latter being at the anterolingual base of p3. The articular condyle (QM F41222) is near cylindrical and as long as many specimens of larger Trichosurus spp. The molar row has only a slight convex buccal curvature, and the premolar has an angle of deflection from the molar row of approximately 35–45°. The angular process of the dentary appears to be relatively larger than in other phalangerids, extending further lingually and posteroventrally. A large mental foramen, directed anterodorsally, is located ventral to the anterobuccal crown of p3.

Lower third premolar (p3)

The occlusally ovate premolar barely reaches above the height of the m1. Two ridges descend from the first anterior cuspule, one to the anterior of the tooth, and the other on the buccal face. A convex face is formed between the ridges. There is only a very slight lingual swelling around the crown base, which develops into a weak cingulid anterolingually where it forms the base of a shallow concavity descending from the apex. A similarly slight cingulid, and associated concavity, is found on the posterobuccal side. A heart-shaped wear facet, caused by the P3, is present on the posterobuccal surface, extending to the m1 contact. Occlusally, the anterolingually convex crest begins about 40–50% from the anterior of the tooth and descends posteriorly, ending just lingual to the anterior apex of m1. The crest is approximately midline along the longest axis, and not offset. At least seven cuspules are evident on the crest with, anteriorly, the first or second being the highest.

Lower first molar (m1)

The lower first molar is sub-rectangular posteriorly and sub-triangular anteriorly, typical for phalangerids. The protoconid is lingually displaced, compared to the other lower molars, such that it is positioned slightly lingual to the transverse midpoint. A paracristid runs anteriorly from this cusp, with a slight medial notch before the small parastylid. The anterior cingulid is better defined on the anterobuccal side (only weakly developed on QM F23088 but extending almost to below the protoconid on QM F40103). Where the metaconid is distinct, it is located just posterolingual to the protoconid with a small notch between the two cusps. The postmetacristid drops steeply, curving posteriorly to either meet the similarly descending pre-entocristid in a valley, or is truncated by a cleft dividing both cristids that forms a small basin lingually. A semi-lophate hypolophid runs directly between the entoconid and hypoconid, descending more steeply from the entoconid. The posterior cingulid is posteriorly convex, formed from the curving postentocristid and posthypocristid, and is much lower than the hypolophid. About half of the posterior cingulid contacts the anterior cingulid of m2. There is a slight medial kink in the cristid obliqua¹, such that the anterolingual running prehypocristid portion turns anteriorly and becomes the postprotocristid. A moderately sized, but shallow, fossa is present on the face buccal to the cristid obliqua. A similarly sized fossa is variably present posterior to the hypoconid and buccal to the posterior cingulid. The protoconid and metaconid are the highest cuspids, followed by the entoconid and much shorter hypoconid.

Lower second molar (m2)

The second lower molar is semi-lophate and sub-rectangular, with slight constrictions medially on the buccal and lingual surfaces. The anterior surface is slightly concave medially, where it abuts the m1. There are well-developed anterior and posterior cingula, which curve strongly convexly. The metaconid is the tallest cuspid, followed by the entoconid, which is about twice the height of the sub-equal protoconid and hypoconid. The lophids are nearly straight occlusally, with slight posterior flexure in the buccal half that is more evident in the protolophid. The lophids are better developed on the lingual side and descend to the centre before slightly rising again. Immediately anterolingual to the protoconid is a small tear-shaped fossa, with an anterior end that cuts into the paracristid without quite bisecting it. Beyond this point the buccal anterior cingulid is a wide shelf extending anterolingually, before narrowing lingually. A second, variably present, shallow fossa extends posterolingually from the protoconid, between the protolophid and postprotocristid. A third small fossa, extending slightly transversely, is found anterior to the hypoconid. The buccal end of this latter fossa pinches the posterior portion of the cristid obliqua (prehypocristid) without bisecting it. A small basin, and associated continuation of the anterior cingulid, is variably present on the anterobuccal face of the tooth. Faint crenulations are variably present, particularly in the central basin. The cristid obliqua is kinked at the interlophid valley, initially running lingually from the prehypocrisitid before turning sharply anterobuccally. A very fine fissure divides the cristid obliqua where it kinks. The posterior cingulid is much lower than the hypolophid. A small to moderately sized cleft truncates the joining of the postmetacristid and pre-entocristid at the lingual end of the interlophid valley, extending lingually almost to the distal extremity of the tooth. The buccal surface, between the protoconid and hypoconid, forms either a narrow cleft descending posteroventrally to the crown base, or a shallow basin extending ventrally about 2/3 towards the crown base.

Lower third molar (m3)

The morphology of the m3 is similar to m2 except in the following: the anterior half of the tooth is relatively narrower, with less swelling of the crown bases below the anterior cuspids; the anterior surface is flat to slightly convex where it abuts the m2; the metaconid remains the highest cuspid, but is relatively shorter, as well as more anteriorly displaced; the hypolophid remains nearly straight, but the protolophid has a greater posterior flexure; the small fossa anterolingual to the protoconid is slightly larger and deeper, as are the fossae posterolingual to the protoconid and anterior to the hypolophid; the basin and cingulum on the anterobuccal face are smaller; the fissure dividing the cristid obliqua, in line with the interlophid valley, is wider; the buccal cleft, truncating the postmetacristid and pre-entocristid, is slightly wider and extends to the distal surface; the buccal cleft or basin, between the protoconid and hypoconid, is narrower.

Lower fourth molar (m4)

The m4 is similar to the m3 and m2 except in the following: it is the smallest of the lower molars and, although sub-rectangular like the m2 and m3, the posterior half is relatively much narrower; the posterior cingulid is less well-developed; the posterior face, between the posterior cingulid and hypolophid, is less inclined; the heights of the lingual cuspids are reduced relative to the m3, and the entoconid is equal in height to the protoconid, which is slightly taller than the hypoconid; the protolophid has slightly less posterior flexure than in m3; the hypolophid is very weakly developed; while the fossa anterior to the protoconid is moderately well-developed, no fossae posterolingual to the protoconid or anterior to the hypoconid are evident; there are no crenulations; the kink in the cristid obliqua is reduced, as the prehypocristid is directed more anteriorly; the fissure dividing the cristid obliqua at the kink is as narrow as that in the m2, and joins the buccal cleft; the posterior cingulid is at the same height as the hypolophid; there is only a very small cleft dividing the lingual ridge, with the postmetacristid and pre-entocristid descending gently and without the truncation resulting in a near-vertical alignment, as seen in m2 and m3; the buccal cleft, between the protoconid and hypoconid, curves slightly posteroventrally, descending only half way down the buccal surface.

Maxilla

The maxilla (QM F50418; Fig. 3) contains P1 and P3 and M1–3, with the alveolus for P2 (lying directly below the infraorbital foramen), alveoli for M4 and most of the alveolus for the canine. The posterior and half of the lateral walls of the canine crypt are also preserved. The P1 is caniniform, double-rooted, about half the height and one quarter the width and length of the P3. A very small diastema separates the canine and P1 while small diastemas, about twice the length of the latter, separate P1 from P2 and P2 from P3. The molar row has a slightly convex buccal curvature.

Upper third premolar (P3)

The third premolar has a near triangular anterior half and is hemispherical posteriorly, being widest about three quarters of the way towards the posterior end. The longest axis is oriented buccally from the molar row by approximately 55°. The crest, with five cupules, follows the main axis until the widest point when it turns posteriorly and meets the parastylar ridge on M1. The length of the crest between the cuspules is approximately 40% of the tooth length along the longest axis (it is more than 50% of the length in all other phalangerids, except Wyulda squamicaudata and some Phalanger vestitus). The most anterior cuspule is positioned near the centre of the tooth. The second cuspule is the highest and the crest descends gently to the fifth cuspule, before stepping down and descending at a great decline. Two vertical crests run between the anterior cuspule and the coronal base, one anteriorly and one buccally. A ridge also runs vertically anterolingually from the fourth cuspule, turning more horizontal as it runs over a slight posterolingual bulge. Weakly defined vertical ridglets are evident on the lingual side of the first four cuspules. A thin cingulum is present at the base of the anterolingual face. A large heart-shaped wear facet, from the m1, occupies much of the posterobuccal face. The occlusal surface roughly aligns with the occlusal surface of the molar row (unlike in species of Onirocuscus, where it is much taller than the molars).

Upper first molar (M1)

This is an anteroposteriorly sub-rectangular, bunolophodont and quadritubercular molar. The anterior half is wider than the posterior, due primarily to a ventrolingual swelling of the coronal base of the protocone. The buccal cusps are pyramidal, while the lingual cusps are more bunodont. The protoloph is partially developed, descending to the halfway point between the anterior cusps. No such partial loph from the protocone is evident (although a short ridglet, possibly a crenulation, emerges posterobuccal to the midpoint of the preprotocrista). The metaloph is well defined, descending at about 45° from the metacone to a point about ¾ across, before rising very slightly to the metaconule. The metaloph is slightly S-shaped in occlusal view, with a posterior flexure just buccal to the midpoint. The uniformly curving posterior cingulum is a well-defined shelf, being about half the total posterior tooth width and formed as the continuation of the postmetacrista and postmetaconulecrista. The anterior cingulum is partially well developed as a much smaller shelf, descending transversely from the parastyle to about 1/3 of the anterior width, before joining the preprotocrista. A small parastyle (stylar cusp A) is present at the end of the slightly buccally convex preparacrista. The paracone² and metacone are approximately the same height, slightly taller than the metaconule, which is higher than the metacone. The metaconule is more buccally positioned than the protocone. Three very small cuspules are evident along the postparacrista: the first is nearly twinned with the paracone and as high; the second, almost imperceptible, lies just anterior to the point where the postparacrista turns; and the third forms the apex from which the postparacrista plunges into the buccal cleft. Any of these three cuspules could, nominally, be a stylar cusp C. A neocuspule (which could be an artefact of crenulation) lies anterolingual to the paracone, separated from the partial protoloph by a thin crevice. The postparacrista and premetacrista are truncated by a small valley. A fine fissure extends buccally from this valley, over the ectoflexus to a weakly developed buccal cingulum. A moderately sized basin descends ¾ of the lingual tooth face between the protocone and metaconule. The postprotocrista and premetaconulecrista meet at an angle of about 90°, in line with the interloph valley. The postmetacrista runs posterobucally before curving to join the posterior cingulum. The tooth has minor crenulations, most restricted to the faces of the interloph valley.

Upper second molar (M2)

As for the M1 except in the following: the anterior cusps are more buccally positioned, so that the buccal cusps are longitudinally aligned and the protocone lies slightly lingual to the metaconule; the buccal faces of the buccal cusps are more inclined; there is less inflation of the lingual surface of the protocone; both lophs are well-developed (but still semi-lophate); the protoloph is straight, crossing transversely between the paracone and protocone; the paracone contribution to the protoloph is much longer than in M1, with the lowest point on the loph being just buccal to the protocone; the metaloph has a similar S-shape as on the M1, but with less posterior flexure near the metacone and more anterior flexure towards the metaconule; the buccal half of the posterior cingulum is straighter; the anterior cingulum is much better developed than in M1, running transversely and parallel to the protoloph for ¾ of its length before curving as the preprotocrista; a uniformly thin and shallow fossa is present between the anterior cingulum and the protoloph; the preparacrista is about the same length as its counterpart on M1 but has greater lingual curvature; no parastyle is evident on the preparacrista; the paracone is taller than the metacone, which is slightly taller than the equal lingual cusps; the two most anterior cuspules on the postparacrista of M1 are not present here, although the most posterior is present at the point where the postparacrista turns posterolingually; the buccal valley is not as well developed; the fissure extending from the buccal valley ends in very small basin on the ectoflexus, which is blocked anterobuccally by a cuspule on the weakly developed buccal cingulum; the neocuspule anterolingual to the paracone on M1 is not present; the lingual basin, between the protocone and metaconule, is narrower; the postmetacrista runs posteriorly to join the posterior cingulum.

Upper third molar (M3)

As for the M2 except in the following: it is smaller and sub-triangular, due to a narrower posterior half; the paracone is more buccally positioned; the buccal faces of the buccal cusps are more inclined (near vertical on the metacone); both lophs are more poorly developed, being thinner and not as high; the metaloph runs anterolingually from the metacone to metaconule, and is less S-shaped; the posterior cingulum is relatively shorter, due to the postmetacrista curving posterolingually rather than being transversely directed from the metacone; the anterior and posterior cingula are less well-developed, thinner shelves; the fossa between the protoloph and anterior cingulum is wider and shallower; the posterior basin is relatively smaller; the paracone is relatively taller; the buccal valley is larger but does not appear to have the ventral fissure; the buccal basin on the ectoflexus is much reduced and blocked anterobuccally by a much smaller cuspule; the lingual basin between the protocone and metaconule is narrower; the premetaconulecrista is almost non-existent and, as such, the angle between this and the postprotocrista is far more acute.

Upper fourth molar (M4)

Missing, but the alveoli suggest a triangular tooth about ¾ the length of the M3 and almost as wide anteriorly, but perhaps only half as wide posteriorly.

Results

Morphometric analyses complement mostly qualitative morphological studies, providing means to quantitate morphospecies that can be compared to established and proposed species, and to determine the influence of variables on multivariate dispersion (Hammer & Harper 2006, Harper & Owen 1999, Marramà & Kriwet 2017). Herein, phalangerid taxa (fossil and extant) are well differentiated by CVA (>93% of all specimens are correctly classified and >98% when fossil and extant taxa are separated; Fig. 4). Axis 1, explaining 52.01% of total variation, primarily reflects individual molar lengths, upper molar row length, and premolar widths. The largest contributors to Axis 2, which explains 21.43% of variation, are anterior upper molar lengths, posterior lower molar lengths, lower fourth molar widths, lower third molar posterior width and, inversely, premolar widths. Axis 3 (Fig. 5), explaining 9.74% of total variation, is mostly influenced by lower molar row length, anterior upper molar posterior widths, lower premolar length and width, and lower first molar length and widths.

Figure 4. Canonical variates analysis of extant and extinct phalangerids. Axes 1 and 2 represent 52.01% and 21.43% of total variation respectively. ■ Trichosurus spp.; ▲ Ailurops ursinus; ● Spilocuscus maculatus; × Phalanger spp.; + Onirocuscus spp.; □ ‘Trichosurus’ dicksoni; ○ Archerus johntoniae; ◆ Wyulda squamicaudata; ❋ Strigocuscus celebensis. Biplot variable abbreviations are explained in Materials and Methods and listed in Supplemental material.

Figure 5. Canonical variates analysis of extant and extinct phalangerids. Axes 2 and 3 represent 21.43% and 9.74% of total variation respectively. ■ Trichosurus spp.; ▲ Ailurops ursinus; ● Spilocuscus maculatus; × Phalanger spp.; + Onirocuscus spp.; □ ‘Trichosurus’ dicksoni; ○ Archerus johntoniae; ◆ Wyulda squamicaudata; ❋ Strigocuscus celebensis. Biplot variable abbreviations are explained in Materials and Methods and listed in Supplemental material.

Except for Onirocuscus spp. and ‘Trichosurus’ dicksoni, all taxa are clearly differentiated by the first two axes in the CVA morphospace (Fig. 4). Ailurops ursinus is separated from all other phalangerids on Axis 1, reflecting its overall greater size, particularly for molar lengths and premolar widths. Spilocuscus maculatus is also separated from other taxa on Axis 1, but with smaller dimensions for most variables compared to A. ursinus, except for its marginally wider upper and lower premolars. Relative differences in molar widths and premolar lengths are less pronounced. Species of the other phalangerine genus, Phalanger, occupy morphospace close to S. maculatus, separated mostly by its shorter molar row lengths and shorter individual molars, combined with a narrower m4. Strigocuscus celebensis is differentiated from P. spp. primarily along Axis 2 due to wider lower and upper premolars, as well as a wider anterior on M1.

The convex hulls for the fossil phalangerids, O. spp. and ‘T.’ dicksoni, are positioned near the origin of both CVA axes, although the centroids are lower on Axis 2 compared to T. spp. This separation is essentially due to wider lower and upper premolars in the fossil taxa, as well as a longer M1, M2, m3, m4, a posteriorly wider m3 and a wider anterior on m4 in T. spp. Although the convex hulls of Onirocuscus spp. and ‘T.’ dicksoni overlap significantly on axes 1 and 2, they are mostly separated by Axis 3. The length of the total lower molar row and p3, as well as the widths of the posterior half of M1, M2 and m1, are major contributors to this difference.

Trichosurines occupy morphospace with low eigenvalues on Axis 1, clearly separated from the ailuropines and phalangerines. The biplot indicates that this is mostly due to the trichosurines having shorter lower molars, upper molar row length, M3 and M4, as well as a narrower p3. Within the trichosurines, Wyulda squamicaudata separates from T. spp. primarily along Axis 2. This separation corresponds, primarily, to the length of M1, M2, m2, m3, m4, m3 posterior width and m4 width, as well as the difference between these and premolar widths. W. squamicaudata therefore has relatively wide p3 and P3, short anterior upper and posterior lower molars, a posteriorly narrow m3 and a narrow m4 compared to the inverse characteristics for T. spp.

The centroid for the Archerus johntoniae convex hull is positioned midway between W. squamicaudata and T. spp. in morphospace. The difference in the characteristics separating the trichosurines is therefore less marked in A. johntoniae, with the latter having relatively shorter anterior upper and posterior lower molars, a narrower posterior on m3, narrower m4, proportionately shorter total molar row lengths, and slightly wider premolars than Trichosurus spp. The opposite morphometric relationship is true for A. johntoniae compared to W. squamicaudata.

The other fossil taxa, Onirocuscus spp. and ‘T.’ dicksoni, are differentiated from A. johntoniae principally by longer upper and lower molars, longer total upper molar row length, a wider m4 (anteriorly and posteriorly), and a wider p3. The difference in these characteristics is exaggerated between A. johntoniae and P. spp., magnified further between A. johntoniae and S. maculatus, and then to a much greater extent still between A. johntoniae and A. ursinus. Separation of S. celebensis and A. johntoniae is mostly a result of a wider p3 and P3, longer M3, M4 and m1, a wider anterior on M1, and a wider posterior on M4.

The CVA correctly classified 93.5% of specimens, and 74.6% after jackknifing. No specimens of extant phalangerid were incorrectly classified as other extant taxa, and only one specimen of ‘T.’ dicksoni was classified as another fossil taxon (Onirocuscus spp.), in the original classification.

Discussion

Archerus johntoniae increases the diversity of extinct Australian phalangerids to 10 species. Four genera (Onirocuscus, ‘Trichosurus’, Wyulda, and Archerus) and seven species (Onirocuscus reidi, Onirocuscus inversus, Onirocuscus silvacultrix, Onirocuscus rupina, ‘T.’ dicksoni, Wyulda asherjoeli Crosby, Nagy & Archer, 2001 and A. johntoniae) are from the species-rich fossil deposits of the Riversleigh World Heritage Area, northwestern Queensland. Several more species, and at least two genera, from Riversleigh remain to be described (Crosby 2002). At least three recognized phalangerid species (O. silvacultrix, O. inversus and W. asherjoeli) occur together in the early Miocene of Riversleigh (Archer et al. 2006, Crosby, Nagy & Archer 2001, Crosby 2007). Archerus johntoniae is known from Cadbury’s Kingdom site, which was assigned to Faunal Zone B (interval B3) by Arena et al. (2016). Phalangerid species from the early Miocene of Riversleigh are also from LFs assigned to this interval zone (Arena et al. 2016) and that belong to the same ‘Bulungamaya & Nimiokoala’ palaeocommunity (sensu Myers et al. 2017), suggesting that A. johntoniae occurred sympatrically with these three species. Similarly, A. johntoniae is present in the Middle Miocene Last Minute LF, which is a member of the ‘Trichosurus & Nimbadon’ palaeocommunity (sensu Myers et al. 2017), indicating possible sympatry of A. johntoniae and at least two phalangerid species from this palaeocommunity (O. reidi and ‘T.’ dicksoni).

Flannery (1987) observed sympatry in up to five species (Strigocuscus and Phalanger spp.) of extant phalangerid in montane New Guinea. Niche differentiation within phalangerid guilds appears to be based primarily on body size and altitude, with larger species tending to be more folivorous and found at higher altitudes or over a greater clinal range, while smaller species tend more towards frugivory and faunivory at lower altitudes (Crosby 2007, Helgen & Jackson 2015). In Trichosurus vulpecula, body size is also related to climate, with smaller individuals (1–2 kg) being found in the warmer regions of northern Australia, and larger individuals (4–5 kg) being found in cooler southern areas (Helgen & Jackson 2015).

The four Early Miocene phalangerids from Riversleigh range from 0.7 to 3.6 kg (W. asherjoeli—0.7 kg, A. johntoniae—1.3 kg, O. silvacultrix—1.9 kg, O. inversus—3.6 kg; based on body mass equations of Myers 2001 in Crosby 2002, Crosby 2007, Crosby, Nagy & Archer 2001). Each species is at least approximately 40% larger than the next smallest. Similarly, the three Middle Miocene species that occur together have substantial size differences (A. johntoniae—1.3 kg, ‘T.’ dicksoni—1.6 kg, O. reidi—2.5 kg). At 1.3 kg A. johntoniae was about the size of the extant Wyulda squamicaudata (0.9−2 kg) and, like the latter, may have incorporated more nuts, fruit, and insects into its diet (Helgen & Jackson 2015) than its larger sympatric species. The Early Miocene palaeoguild includes small to medium-sized phalangerids, possibly suggesting climatic conditions similar to present-day low-mid montane Papua New Guinea. The loss of the largest phalangerid species from the Middle Miocene palaeoguild may indicate warmer prevailing conditions.

Archerus johntoniae can be discriminated from other phalangerids by morphology as well as size. It occupies morphospace that is distant from extant phalangerines and ailuropines but closer to, and approximately equidistant, from extant trichosurines and other fossil species (Onirocusus spp. and ‘T.’ dicksoni). Archerus johntoniae has comparatively smaller cheekteeth than the other fossil species although differences in molar widths are generally less marked than length. Archerus johntoniae also forms a morphocline with extant trichosurines, such that it is intermediate between W. squamicaudata with relatively wide premolars, a narrow m4, and short upper (notably anterior) and lower molars (especially m2–4), and T. spp. with the inverse characteristics.

Archerus johntoniae has its closest morphological affinities with extant phalangerines and T. spp. These similarities are due, predominantly, to features of the lower dentition anterior to m4. Archerus johntoniae shares the fewest morphological characteristics with other fossil phalangerids, and the lowest number with ‘T.’ dicksoni. Of these, only the presence of a moderately large fossa anterior to the protoconid, on m3–4, is shared exclusively with ‘T.’ dicksoni. Other morphological features variably shared with other phalangerids, and therefore also potentially phylogenetically informative, include: (1) the presence of a protoloph on M1; (2) the direction and curvature of the postmetacrista on M1 and M3; (3) the development of the buccal basin on M1; (4) the curvature of the anterobuccal margin of the crown on M1; (5) the development of the protoloph on M2; (6) the development of the lingual basin on M2 and M3; (7) the presence of a fossa of uniform width between the protoloph and anterior cingulum on M2; (8) the presence of a large fossa anterior to the protoconid on m3 and m4; (9) the presence of a very small buccal cleft on m4; and (10) the presence of a lingual ridge, formed from by the postmetacristid and pre-entocristid, without medial valley on m4. None of these has previously been identified as potentially phylogenetically informative (e.g., Flannery, Archer & Maynes 1987, Case, Meredith & Person 2008).

Beck et al. (2022) found that crown phalangerids (with ‘T.’ dicksoni and Onirocuscus excluded) dated to approximately 11 Ma (8.0−13.1 Ma with 95% limits). This is younger than other proposed median divergence dates, ranging from 17.3 Ma (Meredith, Westerman & Springer 2009) to approximately 26.5 Ma (Ruedas and Morales 2005, Raterman et al. 2006, Kealy et al. 2020) With the reassignment of Strigocuscus reidi to Onirocuscus, and the identification of ‘T.’ dicksoni as a new genus, only the presence of W. asherjoeli supports the divergence of crown Phalangeridae prior to the late Middle Miocene (an undescribed species of Trichosurus has been identified from Riversleigh’s ca 12 Ma Encore LF: Crosby 2002). We believe a taxonomic reassessment of W. asherjoeli is warranted to determine if it should be assigned to a new genus and, consequently, substantiate the more recent divergence date of Beck et al. (2022).

The temporal range for A. johntoniae (18–12 Ma) indicates that, while it could have overlapped with crown phalangerids, it is more likely to be a stem phalangerid, with an origin predating the divergence of crown phalangerids. Any morphometric/morphological affinities between A. johntoniae and extant trichosurines are therefore probably due to retention of plesiomorphic features rather than synapomorphies, or due to similar ecomorphology. A preliminary phylogenetic analysis of fossil phalangerids by Crosby (2002) determined that A. johntoniae (then undescribed as ‘AL90 species’) may be more closely related to an unnamed genus from the Early Miocene of Riversleigh, than it is to the Middle Miocene ‘T.’ dicksoni and a second unnamed species in this genus. No current phylogenetic analysis has been made due to the number of undescribed species still to be published.

Archerus johntoniae is one of the rare mammalian species (<10 of 200+ spp.) that is present in both Riversleigh palaeocommunities framing the Middle Miocene Climatic Optimum (MMCO) (‘Bulungamaya & Nimiokoala’ palaeocommunity 17–18 Ma; and the ‘Trichosurus & Nimbadon’ palaeocommunity 13.5 Ma), as well as in the peak-MMCO AL90 LF (14.5 Ma). Archerus johntoniae persisted at Riversleigh during a period that witnessed a dramatic decline in mammalian family and species level diversity, followed by a partial recovery (Myers et al. 2017, Myers 2019), including the apparent demise of the smallest (W. asherjoeli—0.7 kg) and largest members (O. inversus—3.6 kg) of the phalangerid palaeoguild. Future palaeoecological studies incorporating A. johntoniae will be important for understanding how some species (particularly small (<5 kg) mammals) survived such periods of contrasting environmental change and associated extinctions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Support for our research has come from the Australian Research Council [DP170101420 grants to Professors M. Archer and S. J. Hand]; the University of New South Wales; P. Creaser and the CREATE Fund; the Queensland Parks and Wildlife Service; Environment Australia; the Queensland Museum; the Riversleigh Society Inc.; Outback at Isa; Mount Isa City Council; the Waanyi people of northwestern Queensland; and private supporters including D. and A. Jeanes, K. and M. Pettit, E. Clark, M. Beavis, M. Dickson and the Rackham family including Alan Snr, Alan Jnr, Kerry and Dale. Highly skilled fossil preparation was undertaken by Dr Anna Gillespie. Assistance in the field has come from many hundreds of volunteers as well as staff and postgraduate students at the University of New South Wales.

Notes

1 Beck et al. (2022) postulate that the kinked buccal crest in phalangeroids may not be homologous with the cristid obliqua. The specimens examined herein provide only equivocal data (i.e., rarely present, very minor ridglets extending from the buccal cuspids, that may be crenulations) regarding a plesiomorphic, more lingually placed crest. We therefore follow Beck et al. (2022) in referring to the kinked crest as the cristid obliqua, in the absence of more compelling data.

2 Beck et al. (2022) proposed that the buccal cusps on the protoloph and metaloph are stylar cusps B and D respectively. Consequently, the neoparaconule and neometaconule present in some phalangerids (e.g., Eocuscus sarastamppi Case, Meredith & Person, 2008) would be the reduced paracone and metacone, which were eventually incorporated into the lophs of more derived taxa. Although the argument is compelling, we have decided herein to adopt the traditional nomenclature and assume that the buccal cusps are homologous to the paracone and metacone, until further evidence is forthcoming.