Megafaunal meiolaniid horned turtles survived until early human settlement in Vanuatu, Southwest Pacific

Systematic Paleontology.

Meiolaniidae Boulenger, 1887; ?Meiolania Owen, 1886 (20); and ?Meiolania damelipi sp. nov.

Holotype is AMF136641, right humerus, collected layer 2, Unit 3.3–3.4, Area 3B, Teouma Lapita site, Efate, Vanuatu, 2006 (Fig. 3). Etymology is for Willie Damelip, originally of Ambrym Island (SI Text). Diagnosis is a meiolaniid differing from other Pleistocene species with more gracile longbones; shoulder girdle with coracoid unfused, well developed dorsal and acromion processes diverge at approximately 105°; minor and major femoral trochanters ventrally enclose a deep intertrochanteric fossa; much smaller than Ninjemys oweni. Measurements of holotype are as follows: total length, 95 mm; maximum proximal width, 39 mm; minimum shaft width, 15 mm; and maximum distal width, 35 mm. Paratypes are all elements shown in Figs. 3 and 4; measurements are provided in Table 2.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Pelvic and carapace elements of ?M. damelipi. (A) Left femur AMF136642 in caudal view; (B and C) Right tibia AMF136651 in (B) ventral and (C) dorsal view. (D and E) Dermal armor of carapace in caudal (D) and ventral (E) views of AMF136646. (F) Marginal fragment of carapace with gutter AMF136649. (G and H) Ungual phal AMF136664 in (G) ventral and (H) dorsal aspects. (pat, patellar tendon attachment not raised in ridge; tm, trochanter major; tub, tuberosity.) See SI Text for associated data. (Scale bars, 50 mm in A–F, 10 mm in G and H.)

Given the absence of the diagnostic material of the skull and the tail, and minimal peripheral material of the carapace or plastron, we only tentatively refer this new species to Meiolania. On biogeographic and temporal grounds, we consider it unlikely that the insular ?M. damelipi was conspecific with the Miocene taxa described from Australia (SI Text and Table S1). Whether it differed from other unnamed taxa from the New Caledonian region cannot be established from the material at hand. The preserved elements allow significant comparison with M. platyceps material in the Australian Museum, described by Gaffney (5).

Humeri (Fig. 3 A–D) have less expanded ends than in M. platyceps. Estimated lengths range from 40 to 140 mm, with a maximum shaft diameter of 30 mm. As in M. platyceps, the proximal articular surface is hemispherical and offset dorsally from the shaft, the medial process is larger than the lateral one, and proximal width is greater than distal width. The ectepicondylar foramen begins as a distinct groove, wider than in M. platyceps, on the dorsal facies of the shaft, before penetrating the ectepicondyle to open ventrally. Humeri differ from M. platyceps and M. mackayi (11) with a less expanded lateral process and a more proximally projecting medial process.

Ulnae (Fig. 3 E, I, and J), as in M. platyceps, have proximally an extremely well developed olecranon process and sigmoid notch, and a well defined radioulnar articulation (5). Radii are more elongate and the rugosity for biceps superficialis on the shaft is smaller than in M. platyceps.

The shoulder girdle (Fig. 3 F–H) is triradiate: as in M. platyceps, the glenoid is not supported by a neck; the well developed dorsal and acromion processes diverge at approximately 105° compared with 120° in M. platyceps and other terrestrial turtles (5); the coracoid (Fig. 3H), unlike M. platyceps, is not fused to the glenoid and is more elongate. A wide scapular angle is usually correlated with a high body profile (5) suggesting that ?M. damelipi had a low body profile, somewhat more like marine turtles.

Femora (Fig. 4A) are stocky and range from 45 to 145 mm in shaft length; the head is large and hemispherical, being wider than long, and directed more dorsally to the shaft than in M. platyceps, such that it does not project proximally past the trochanter major. The minor and major trochanters are distinct from the femoral head, but unlike M. platyceps, have similar proximal extent and are linked ventrally by a bony web to enclose a deep intertrochanteric fossa.

Ankle, wrist, and digit elements are uncommon, but the few unguals are blunt, dorsoventrally thick, and ventrally flattened as in M. platyceps (Fig. 4 G and H).

Shell fragments of ?M. damelipi are similar to those of meiolaniids in having a thin dense outer bone layer and finely cancellous internal structure, but differ with a smoother external texture, and at least part of the carapace margin is concave dorsally. Dermal armor was present on the carapace (Fig. 4 D and E).

The size of ?M. damelipi can be compared with that of M. platyceps from the dimensions of the long bones. Femora and humeri were as long as 145 mm and 140 mm, respectively, similar to those in AMF57984 with a shell length of 1 m. However, larger turtles were present as one section of shoulder girdle preserves a glenoid cavity with a diameter of 40 mm compared with 30 mm in AMF57984.

These data show that the meiolaniid radiation in the southwestern Pacific region was more extensive than previously recognized (5). Dispersal to and between islands of this region would have been easily achieved by meiolaniids. Terrestrial turtles are highly buoyant (21) and some, e.g., Dipsochelys giganteus, are known to have survived in oceans without access to fresh water for many weeks (22). Although incapable of directed swimming, they are thus ideal candidates for ocean drifting, which dispersal mode explains the distribution of extant terrestrial turtles in the Indian Ocean (22, 23). There is nothing unique about Vanuatu to explain why meiolaniids survived there until the advent of humans, but could not have done so in other island groups, such as Fiji and New Caledonia. In the Lord Howe group, postglacial sea level rise greatly reduced land area, which in the absence of any evidence of pre-European occupation (24) could have facilitated the extinction of M. platyceps. For archipelagoes that retained islands of significant size in the Holocene, this cannot be the explanation. A poor or absent fossil record for most islands is the probable reason for a lack of other Holocene meiolaniids so discovery of further populations or taxa should be expected on other southwest Pacific islands where adequate habitat existed. Investigations of first contact human southwest Pacific sites will likely extend the record, and reexamination of bones previously interpreted as marine turtle might reveal that some are in fact those of terrestrial turtles.

The discovery of meiolaniid remains at Teouma provides conclusive evidence that they survived into the late Holocene and that humans encountered them. Relatively large numbers of meiolaniid bones occur, particularly in the basal levels of the Teouma midden, dating to approximately 2,900 or 2,800 cal BP, where they overlay burials dated 3,100 to 3,000 cal BP (14). Some burials were associated with meiolaniid carapace fragments (SI Text). In younger layers, they are absent. Remarkably for a Pacific coastal site, bones of marine turtles are rare in the lower layers. Early colonizers of the western Pacific normally hunted sea turtles and impacted many populations (25). At Teouma, large comparatively heavy bodied and fleshy terrestrial turtles were available and were the preferred prey until their disappearance by approximately 300 y after the initial encounter. Skeletal representation (Table 1) is markedly biased toward legs and associated fleshy parts. We infer that most turtles were killed and butchered elsewhere with mainly the fleshy upper limbs being taken back to the village. It seems probable that the first colonists, who created the cemetery at Teouma, and whose habitation sites have not yet been found, had eliminated proximate populations of turtles.

Hunting undoubtedly contributed to the extinction of ?M. damelipi, but may not have been the only cause. In the Mascarenes (Indian Ocean), where Europeans were the first humans to encounter the terrestrial turtles on Mauritius and Rodrigues, and initial densities were high, intensive exploitation for food rendered all populations extinct in a little more than one century (22). This rapid extirpation of turtles was partly attributed to the introduction of pigs, which prey on young and eggs. Similarly, pigs introduced by Lapita people may have affected the survivorship of Vanuatu meiolaniids. Whatever the exact synergy of factors, meiolaniids were extinct on Efate in Vanuatu within 300 y of the arrival of Lapita people.

The discovery of ?M.damelipi and that it became extinct soon after human arrival fits a recurring pattern across the Pacific. In Vanuatu, similar timed extinctions include a crocodilian and several birds (26, 27). Such losses are widespread and numerous and include crocodilians and large birds in New Caledonia (5, 27), these and large iguanids in Fiji (13), with losses of birds on every island for which a record exists (ref. 27 and references therein). Whether the turtle species known from fossil sites in New Caledonia (12) and Fiji (13) also survived until human arrival has yet to be established. The overall situation is such that in terms of total species, the loss of diversity in the late Holocene for the Pacific exceeds all other areas on Earth.