First record of Protemnodon (Macropodidae: Marsupialia) from Pleistocene lowland New Guinea

Abstract

An isolated molar from the archaeological deposit of Lachitu Cave (RIQ), on the north coast of Sandaun Province, Papua New Guinea represents the first Pleistocene occurrence of the genus Protemnodon below 1700 m on the island of New Guinea. The specimen is identified as a deciduous second lower premolar (dp₃) from a juvenile individual. It is distinguished from other New Guinean Protemnodon by its large size, proportions, and some morphological features. The find is the first evidence of New Guinean Protemnodon in Pleistocene coastal lowland rainforest environments and likely represents an undescribed species.

Loukas G. Koungoulos [[email protected]] Archaeology and Natural History, School of Culture, History and Language, Australian National University, Canberra, Australian Capital Territory, Australia; Timothy F. Flannery [[email protected]] Australian Museum, 1 William Street, Sydney 2000, New South Wales, Australia; Sue O’Connor [[email protected]] Archaeology and Natural History, School of Culture, History and Language, Australian National University, Canberra, Australian Capital Territory, Australia.

Keywords:

• Protemnodon
• Macropodidae
• New Guinea
• Lachitu
• archaeology
• North Coastal Range

THE GENUS PROTEMNODON Owen, 1874 is one of just eight genera of macropodids recorded from the island of New Guinea, where it is represented by two Pliocene (5.33–2.58 million years ago [Ma]) species (Protemnodon otibandus Plane, 1967, and Protemnodon buloloensis Plane, 1967) and two Pleistocene (2.58 Ma–11.7 thousand years ago [ka]) species (Protemnodon tumbuna Flannery, Mountain & Aplin, 1983 and Protemnodon hopei Flannery 1992). Protemnodon nombe Flannery, Mountain & Aplin, 1983 has recently been reallocated to the genus Nombe (Kerr & Prideaux 2022).

Lachitu (site code RIQ) is a limestone karst cave located within equatorial rainforest at the foot of the Oenake Range, between the northern coastline of Papua New Guinea and the Bewani Ranges (Fig. 1). The cave is today located approximately 150 m from the shoreline and 20–25 m above sea level, but the immediate area is tectonically active and has probably been uplifted several metres since the onset of the Holocene; it was never less than 12 m above sea level. Between 2004 and 2005, three 1 × 1 m archaeological excavations were carried out in the eastern (A), western (B) and southern (C) parts of the cave, from which a moderate quantity of very fragmented but otherwise well-preserved bone—approximately 100 gm per spit—was obtained (O’Connor et al. 2011).

Figure 1. Location of Lachitu Cave (RIQ) on the northern Coast of Papua New Guinea. Image prepared by CartoGIS Services, Scholarly Information Services, the Australian National University.

The material of Square B included the isolated macropodid tooth (PM 26591), provisionally identified as Protemnodon by the late Ken Aplin (Aplin pers. comm.), and separated from the other material before subsequent analysis by L.K. and T.F. Although numerous Australian species are known from low elevations, the tooth marks the first Pleistocene occurrence of a species of Protemnodon from a lowland (<1700 m) setting in New Guinea. Here, we describe the tooth and compare it with other fossil New Guinean macropodid dental material, with commentary regarding its affiliations.

Institutional abbreviations

AM, Australian Museum, Sydney, Australia. CPC, Australian Commonwealth Palaeontological Collection, Geosciences Australia, Canberra, Australia. PM, Papua New Guinea National Museum & Art Gallery, Port Moresby, Papua New Guinea. UCMP, University of California, Museum of Paleontology, Berkeley, CA, USA.

Geological setting

PM 26591 was retrieved from Spit 9 of Square B at Lachitu. A specific chronology is not available from Square B as no material from this square has yet been submitted for radiocarbon dating. However, the chronology of the other excavated Squares (A and C) is well understood from a series of marine shell dates. The uppermost sedimentary units at Lachitu, as dated from Spits are very recent (<200–300 years BP) as attested by dates from Spits 2 (Wk-16532) and 4 (Wk-16533) in Square A. Immediately below the uppermost unit(s) in both Square A and C the underlying strata date to the mid-Holocene, suggesting a period of sediment erosion and/or reduced cultural utilization of Lachitu occurred during the Mid–Late Holocene. Similar erosional episodes or depositional hiatuses appear to have occurred at least once earlier during the cave’s late Pleistocene to early Holocene history (Gorecki et al. 1991, O’Connor et al. 2011).

A date of 6399 ± 45BP (Wk-16524) was obtained from Spit 7 of Square A, and dates of 6519 ± 46BP (Wk-16523) and 6842 ± 48BP (Wk-16525) for upper and lower constituent units of Spit 10 respectively (O’Connor et al. 2011). O’Connor et al. (2011, p. 9) report that the stratigraphy of squares B and C both vary only to a minor degree from the well-dated sequence from Square A. Although no dates from the uppermost units are available for Square C, here dates returned from marine shell of 6062 ± 46BP (Wk-17536), 6013 ± 50BP (Wk-17535) and 5932 ± 45BP (Wk-17534) from Spits 6, 8 and 14 respectively indicate a similar Mid-Holocene timeframe for the formation of deposit at commensurate depths between Squares A and C. Therefore, it seems a reasonable preliminary assumption that the sedimentary context from which PM 26591 was excavated from is of a similar broadly Mid-Holocene antiquity.

There is, however, strong reason to suspect that the original (primary) deposition of PM 26591 occurred considerably earlier than the mid-Holocene. The tooth itself has cemented carbonate infilling most of the interlophid valley and the hollowed base, indicating it most likely was integrated into breccia formed during an earlier point in time, before it eroded out and was redeposited into the floor sediments during the Mid-Holocene. Dating of marine shell from the breccia deposits incorporating cultural materials and bone adhering to different speleothems at Lachitu has provided dates of 15,357 ± 109 BP (Wk-17978), 15,325 ± 97 BP (Wk-17979) and 6680 ± 90BP (OZI282), respectively, indicating that the cave has undergone a number of deposit gutting and filling episodes (O’Connor et al. 2011). Much older dates from the lower levels of Square A of 24,140 ± 155 BP (Wk-16527) from Spit 18 and 25436 ± 282 BP (Wk-16528) from Spit 20, read together with a date of 35,410 ± 1400 BP (ANU-7610) obtained from the earlier excavation of a Square X1 at Lachitu, located approximately equidistant between the 2004–2005 Squares A and C (Gorecki et al. 1991), indicate that bone/shell deposition has occurred at Lachitu prior to and during the Late Glacial Maximum (O’Connor et al. 2011).

We are therefore inclined to believe that the retrieval of PM 26591 from a mid-Holocene stratum at Lachitu is not evidence of the late survival of Protemnodon in New Guinea, but a consequence of the cave deposit’s complex sedimentary history involving ongoing reworking of deposited materials. This is in line with current understanding of the final extinction of New Guinean macropodid megafauna occurring prior to 22 ka (Prideaux et al. 2022). Although human utilization of the site may well overlap with the original depositional timeframe of PM 26591, there is no particular reason indicated by examination of the specimen or its stratigraphic context to suspect this individual was brought into the cave via human hunting activity.

Systematic palaeontology

Infraclass MARSUPIALIA Illiger, 1811

Order DIPROTODONTIA Owen, 1866

Family MACROPODIDAE Gray, 1821

Protemnodon sp.

(Fig. 2)

Figure 2. PM 26591, in views clockwise from top left: anterior, dorsal, posterior, buccal, ventral, lingual. Abbreviations as follows: acm: anterior cingulid; ecd: entoconid; hcd: hypoconid; mcd: metaconid; pad: paracristid; pcd: protoconid; pcm: posterior cingulid; phd: posthypocristid; pmd: premetacristid.

Description

Our description follows the dental anatomic terminology used by Kerr & Prideaux (2022). The lophid occlusal surfaces are very minimally to not visibly worn, and the base of the tooth shows no sign of root development, suggesting that the tooth belonged to a pouch young and was likely unerupted. Although the enamel surface in many areas has been corroded, the main diagnostic features of the crown appear to be intact. The interlophid valley is however obscured by cemented breccia carbonate sediment infill, as is the hollowed base of the crown.

Viewed occlusally, the tooth has a subobovoid shape, rounded anteriorly at the termination of the cingulid and somewhat flattened posteriorly. A well-defined paracristid progresses anterolingually from the protoconid, at a moderately ventrally steep slope, to a meet a strongly anteriorly convex anterior cingulid. The paracristid terminates at a junction with the cingulid just slightly on the buccal side of the midpoint of the tooth. The anterior cingulid is well developed buccally, meeting the base of the protoconid in the anterobuccal corner of the tooth. The anterior cingulid is less defined or developed lingually and is not connected to the paraconid by any visible link. The premetacristid is well developed and does not connect to the buccal extension of the anterior cingulid, rather recurving from the metaconid to progress medially and terminate before the midline.

Both the anterior (protolophid) and posterior (hypolophid) ridges are crescentic and posteriorly convex. The points of greatest convexity appears to sit slightly closer to the metaconid and entoconid respectively, i.e., lingually. The protoconid and hypoconid apexes rise sharply from the approximate midpoint of the lophid ridges, and sit higher than the metaconid and entoconid, respectively. The mediolateral profiles of the lophid ridges thus appear to be straight to slightly ventrally convex on the protolophid, and essentially straight on the hypolophid, until the mediolateral (lingual–buccal) midpoint at which point both lophid ridges rise into sharp peaks at the protoconid and metaconid apexes. Viewed mediolaterally, the lingual crown base rises straight vertically to the metaconid and entoconid; whereas the buccal crown bases exhibits a strong slope as the crown narrows to reach the more lingual position of the protoconid and hypoconid.

A computerized tomography (CT) scan of the tooth did not produce the clarity required for assessment of the interlophid area, but it was sufficient to demonstrate the presence of some morphological features. PM 26591 has an unusually well-developed preentocristid which either splits to meet the protoloph in two places (taking a “Y”-shape), or broadens substantially at its junction with the hypoloph but has been hollowed by damage. It is not possible to determine between these options until further preparation of the tooth can be performed to remove the carbonate overlaying this feature. The cristid obliqua is weakly developed; its enamel spurs sit low, do not proceed far from the protolophid and hypolophid respectively, and do not meet.

A well-developed posterior cingulid crosses the buccal-most 2/3 of the posterior crown face, sitting (at its highest point) just barely less than halfway to the height of the hypolophid ridge. The lingual end of the cingulid sits higher than the buccal end, presenting a clearly sloped appearance when the tooth is viewed posteriorly. A very weak posthypocristid connects the hypoconid with the buccal edge of the posterior cingulid. At their interface, the cingulid rises into a sharp dorsal “peak” and falls again to meet the posthypocristid. No equivalent cristid structure linking the lingual margin of the cingulid with the entoconid is evident.

Comparative analysis

Measurements of PM 26591 are listed in Table 1. Comparison with teeth of adult and juvenile members of both Australian and New Guinean specimens of Protemnodon indicated PM 26591 is best identified as the deciduous lower right second premolar—dp₃ in the terminology of Luckett (1993) and following van Nievelt & Smith (2005) of a very young individual, almost certainly still in the pouch and not taking solid food. We refer to the following materials in identification of the tooth position, and in all following comparative analysis of PM 26591.

Table 1. Dimensions of PM 26591 morphological features.

Feature or measurement	Measurement (mm)
Maximum length	11.36
Anterior width, protoconid–metaconid apexes	4.49
Posterior width, entoconid–hypoconid apexes	4.50
Anterior width, buccal face of protoconid–lingual face of paraconid	5.57
Posterior width, buccal face of metaconid–lingual face of entoconid	5.77
Anterior width, crown base	8.25
Posterior width, crown base	8.70
Paracristid length, anterior cingulid–protolophid junction	2.37
Height of junction between anterior cingulid and paracristid, from base of crown	4.93
Interlophid distance, at midpoints	4.51
Crown height, protoconid apex to crown base	7.15
Crown height, metaconid apex to crown base	5.69
Crown height, entoconid apex to crown base	5.65
Crown height, hypoconid apex to crown base	8.33
Posterior cingulid length	3.97
Height of posterior cingulid midpoint from base of crown	3.43
Displacement of posterior cingulid (cingulid) from posterior crown face	0.68

• Protemnodon tumbuna: AM F153579: a partial right mandible with dp₁, dp₃, m_1-3 (m₃ in crypt), and freestanding p₃ removed from crypt. Replica of type specimen PN 82/40/9 described by Flannery et al. (1983). AM F153579: a freestanding left dp₃ from a different individual to the above (registration number collectively refers to a group of AM plaster casts of the Flannery et al. (1983) P. tumbuna type material). Replica of type specimen PN 82/40/13 described by Flannery et al. (1983).

• Protemnodon hopei: AM F83614: a partial left mandible with i₁, dp₁, dp₃, m₁.

• Protemnodon otibandus: UCMP 45246: a partial right mandible with i₁, dp₁, dp₃, m_1-2. Material not sighted by authors; comparative information derived from description, figure and measurements provided by Plane (1967). CPC 6772: a partial right mandible with dp₁, dp₃, p₃. Material not sighted by authors; comparative information derived from measurements provided by Plane (1967).

• Protemnodon anak Owen, 1874: AM F18905, a partial right mandible with dp_1, dp₃, m_1-2. AMF30578, a partial right mandible with p₃ (in crypt), dp₃, m_1-2. AM F30772, a partial right mandible with i₁, dp₁, dp₃, m₁.

• Protemnodon brehus Bartholomai, 1973: AM F7684: a partial right mandible with i₁, dp₁, dp₃, m₁.

• Protemnodon chinchillaensis Bartholomai, 1973: AM F30693: a partial left mandible with i₁, dp₁, dp₃.

It must be acknowledged that PM 26591 is unusually robust for a dp₃ and exhibits a more rounded occlusal plan, particularly at the anteroposterior margin of the base which is slightly more convex than in comparative materials. This may give some impression of a resemblance to m₁ and the other permanent molars. However, a reasoning for identification as dp₃, as opposed to m₁, is based on the combination of the following characteristics which are present in the dp₃ specimen most relevant in terms of geographic location and also individual age with respect to tooth development and wear.

The general occlusal plan of the tooth PM 26591 is sub-ovoid and exhibits an anterior narrowing, compared to the generally broader impression of the anterior region of m₁ and more posterior molars; specifically, the anterior edge of the anterior cingulid is rounded and relatively pointed compared to the flatter, broader anterior margin of the corresponding area in the m₁. A strongly sloping and relatively straight buccal profile is observed for the protocone and hypocone, rather than relatively upright and outwardly convex or bowing as in the m₁. There is a prominent premetacristid which sits slightly lower than the metaconid and recurves from the metaconid to progress medially and terminate just before the midline, which is not present in comparative m₁ material. The lophid ridges have a distinctive anteriorly concave curvature, whereas they tend to be straighter to almost totally straight in the m₁. The point of greatest curvature in the protoloph and hypoloph is lingually positioned, producing a slight posterolingual bulge. Finally, when viewed anteriorly, the dorsal edge of the lingual portion of the anterior cingulid is convex and sits high, whilst on the buccal portion it is concave and sits low, whereas in m₁ these tend to be closer to level.

Very few examples of the dp₃ exist in an unworn state amongst the available fossil material for Protemnodon from either Australian or New Guinean species, with none available for “Protemnodon” buloloensis, nor the only material of related genus Nombe nombe.

In terms of shape and gross morphology, PM 26591 very closely resembles the freestanding P. tumbuna dp₃ from Nombe Rockshelter (PNG 82/40/13: Flannery et al. 1983). The only substantial differences between the two is that the protoloph and hypoloph are subequal in width rather than the latter being considerably wider; the lophid ridges appear somewhat higher and noticeably narrower relative to the base, and the buccal profile protoconid/hypoconid to the crown base is more strongly sloped with the dorsal-most portion presenting mildly convex whilst the ventral-most appears mildly concave (this pattern being more pronounced for the hypoconid).

In these respects, PM 26591 also differs from all other compared Protemnodon dp₃ specimens. In P. tumbuna and P. hopei a sloping buccal profile is observed to be similar in angle relative to the base, but without a mid-height change in concavity. It is possible that in PM 26591 this has been exaggerated by damage visible on the protocone base, but not the hypocone. Corrosion of enamel could conceivably play some role, but this appears to be evenly applied to the crown, not concentrated in this region. In the rest of the (Australian) comparative series, the buccal profile of dp₃ is either near-totally straight or bowed outwards (convex) to varying degrees.

The occlusal plan of PM 26591 is very similar to the three Pleistocene PNG specimens but is differentiated from the Australian species and P. otibandus. The former group tend to be more elongate anteroposteriorly, and are more inwardly pinched in the middle region between the protoloph and hypoloph than observed in PM 26591 or the other Pleistocene PNG species. PM 26591 is additionally distinguished from P. otibandus by the fact that the latter’s hypoconid is notably displaced lingually relative to the protoconid.

Four features were chosen for metric comparisons of PM 26591 with comparative Protemnodon dp₃ material, revealing further differences with described species (Table 2). With respect to maximum length, PM 26591 is appreciably larger than all of the compared Protemnodon specimens, although the list shown here is not exhaustive for Australian material. It is clearly closer in size to the Australia taxa, for all of which dp₃ is >10.40 mm long, than to either the Pleistocene or Pliocene New Guinean representatives, for all of which dp₃ is <9.60 mm long. Of the former group, it is very close to the length recorded for P. anak AM F30578.

Table 2. Metric comparisons of PM 26591 with the dp₃ of other Protemnodon spp.

Referred material	Taxon	Max. length	PL/HL width	PL/ACB width	HL/PCB width
PM 26591	Protemnodon sp. nov.	11.36	0.997	0.544	0.517
AM F153579	Protemnodon tumbuna	9.25	0.958	0.676	0.618
AM F153579*	Protemnodon tumbuna	8.19	0.974	0.656	0.610
AM F83614	Protemnodon hopei	9.01	0.982	0.826	0.696
UCMP 45246	Protemnodon otibandus	9.60	–	–	–
CPC 6772	Protemnodon otibandus	8.80	–	–	–
AM F18905	Protemnodon anak	10.46	0.930	0.785	0.731
AM F30578	Protemnodon anak	11.30	0.967	0.957	0.928
AM F30772	Protemnodon anak	10.53	0.910	0.785	0.731
AM F7684	Protemnodon brehus	10.99	0.801	0.709	0.680
AM F30693	Protemnodon chinchillaensis	10.47	0.776	0.674	0.747

All measurements are in mm and given to three decimal places. Maximum length data for P. otibandus follow Plane (1967). *PL & HL measurements from this specimen are approximate due to a casting imperfection affecting the lingual lophid apexes.

The protolophid and hypolophid widths of PM 26591 are closer to equal than observed in any of the comparative series. Although this index is quite variable, it appears that P. anak (particularly AM F30578) and the Pleistocene New Guinean species are closest to PM 26591 in this regard.

The width of the protolophid ridge, and of the hypolophid ridge, in relation to that of their respective crown bases are both markedly lower in PM 26591 compared to all other Protemnodon species, a result consistent with the highly sloped appearance of the buccal faces. No particular phylogenetic pattern is notable in the distribution of these indices for the comparative taxa, although the scores for P. tumbuna are consistently among the nearest to PM 26591.

Discussion and conclusions

A fossil macropodid tooth (PM 26591) retrieved from the coastal Lachitu Cave, Vanimo region, Sandaun Province, Papua New Guinea is identified within the genus Protemnodon, marking the first record of this genus from Pleistocene lowland New Guinea. The specimen is also the northernmost record of the genus Protemnodon, and indeed the northernmost record of any megafaunal marsupial. Of the Protemnodon material available for comparison, it is most closely allied in gross morphology to Protemnodon tumbuna, and shows other similarities to the Pleistocene specimens of this genus from PNG, but is markedly larger, being closer in size to Australian species such as Protemnodon anak. In addition to its large size, the specimen PM 26591 exhibits unique proportions between the lengths of its lophid ridges, and between these and the base of the crown. A strongly sloping buccal profile which is weakly convex above the dorsoventral midline and concave below it is also identified as a distinguishing feature between PM 26591 and all congeners.

The described size and morphological differences in addition to the thus far unique occurrence of Protemnodon in this altitudinal and biogeographic context suggests that this individual likely belongs to a new and otherwise undescribed species. As it is unfeasible to erect a new species based on a single tooth without further material, future excavations in the Vanimo area and other lowland or foothill regions of northern New Guinea are anticipated to retrieve additional specimens which will aid in formal description of the probably novel taxon represented by PM 26591.

The two known New Guinean species of Pleistocene Protemnodon are known exclusively from montane environmental settings (Flannery et al. 1983, Flannery 1992, Menzies and Ballard 1994). The Pliocene species (Protemnodon otibandus and Protemnodon buloloensis) are represented in the Otibanda Formation, a fossil-bearing deposit currently >800 m above sea level, but which formed at lower elevation. The taxonomic composition of species present in this formation (known as the Awe Local Fauna) are suggestive of a palaeoenvironment closer to lowland rainforest rather than montane forest (Plane 1967). The other faunal remains produced by the 2004–2005 Lachitu excavations are presently being formally analysed. Preliminary assessments have already found that the lower (Pleistocene) levels contain a number of mammalian taxa which are not known to be extant in the lowlands and foothills in the nearby vicinity of Lachitu, but which are restricted to higher-elevation montane rainforest in the closest mountains, the Bewani and Torricelli ranges (O’Connor et al. 2011, p.9).

Presumably, the same phenomena of ecological change, including anthropogenic pressures, may also account for the subsequent disappearance of a Protemnodon species from the low-elevation settings of the Vanimo district in northern coastal New Guinea. The lack of prior discovery of Protemnodon from other coastal and lowland areas in New Guinea is probably attributable to the extreme rarity of sites with well-preserved Pleistocene materials in these types of settings, compared to those at higher elevations (Sutton et al. 2009), rather than to a genuine absence of the genus from lowland environments.

Acknowledgements

We thank Matthew McCurry, Abram Powell and Thomas Peachey (AM) for access to comparative material, photography, and CT scanning. The late Baiva Ivuyo and Herman Mandui (PM) assisted with excavations in 2004 and 2005. The chiefs, councillors and community of Fichin village granted permission and contributed to our fieldwork. The Alcheringa Editorial Board and two reviewers contributed editing and constructive comments on our manuscript.

Disclosure statement

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

Additional information

Funding

Bulisa Iova (PM) provided collections support. Field excavations were funded by Australian Research Council Discovery Grant [DP0451043] with permission from PM and the Sandaun Provincial Government. Analysis funding was provided by the Australian Research Council Centre of Excellence for Australian Biodiversity and Heritage [CE170100015].