Redescription of the Skull of the Extinct Madagascan Giant Tortoise, Dipsochelys abrupta
Abstract
The only known skull of the extinct Madagacan giant tortoise Dipsochelys abrupta is redescribed. It is similar to the other extinct Madagascar tortoise, Dipsochelys grandidieri, differing in the width of the vomer, in having a ridge on the occipital condyle, and in lacking diverging articular condyles of the quadrates.
The extinct giant tortoise Dipsochelys abrupta is known only from subfossil remains from the southwest of Madagascar. It was a large (approximately 100 cm straight carapace length) dome-shaped species, partially overlapping in range with the more flattened Dipsochelys grandidieri. Both species appear to have died out during the Pleistocene extinction of large animals on Madagascar; the most recent specimens of D. abrupta date to 750 years ago and of D. grandidieri to 1050 years ago (Bour 1994). These species are believed to have evolved in Madagascar as part of the endemic tortoise radiation of the western Indian Ocean. One tortoise species probably dispersed to the granitic Seychelles islands within the last 500,000 years, giving rise to the extant Dipsochelys species of those islands (Gerlach 2004). On geographical grounds, D. grandidieri may have been the most likely ancestor of the Seychelles tortoises (being found in the east of the island), but this is only speculative. Although some DNA has been extracted from bones of D. grandidieri (Austin and Arnold 2002), too few samples of D. abrupta are currently known to allow attempts at DNA extraction. Morphological comparisons have considered both species to be similar but distinct species (Bour 1994; Gerlach and Canning 1998). In one study, D. abrupta was suggested to be conspecific with the Aldabran Dipsochelys dussumieri (Arnold, 1979); although, this was based only on a superficial resemblance of the domed shell. It has since been demonstrated that numerous skeletal characters support a close relationship between D. abrupta and D. grandidieri, with relatively little similarity to D. dussumieri (Bour 1994). Morphological analysis has been hindered in part by the paucity of D. abrupta material, particularly of the more diagnostic bones of the skull. The single skull attributed to this species was figured by Bour (1994) and interpreted based on photographs taken in 1981 and 1985. This interpretation formed the basis of subsequent accounts (Gerlach and Canning 1998; Gerlach 2004). In 2006, an opportunity was taken to reexamine the skull, and this allowed a redescription. This is presented below.
Material. — The single skull is preserved in the collection of the Academie Malgache (specimen Academie Malgache ‘Ampasambazimba, 1909’) (Fig. 1). It is reconstructed (Fig. 2) to measure approximately 133 mm from premaxilla to occipital condyle. This specimen is part of a collection obtained from the rich deposits in a marsh near a thermal spring between Tsarazza and Amparaky, Madagascar, near Lake Itasy, at 1150 m above sea level (Bour 1994). This site has preserved a diverse vertebrate fauna, including marsh-associated species (hippopotami and crocodiles) as well as forest animals. Jully and Standing (1904) reported finding 2 tortoise skulls and numerous skull fragments as well as several bones, 2 plastra, and fragments of carapaces in 1904. All ischia, ilia, and pubes were considered by Bour (1994) to be referable to D. abrupta. One of the skulls is identifiable as ‘Ampasambazimba, 1909’ which was first mentioned (and identified as ‘Testudo abupta’) by Mahé (1965) and measured at 143 mm. The second skull has not been located.
Citation: Chelonian Conservation and Biology 7, 2; 10.2744/CCB-0694.1
Citation: Chelonian Conservation and Biology 7, 2; 10.2744/CCB-0694.1
In addition to the skull, there is a partial dentary from Ankevo, Madagascar, referred to this species (Museum Nationale d'Histoire Naturelle, Paris MAD 3508) (Bour 1994); however, although the Ankevo site contains numerous bones of D. abrupta, there is a tibia referred to D. grandidieri (Bour 1994), consequently there is some doubt as to the association. Bour (1994) admitted that the referral of the tibia to D. grandidieri was somewhat arbitrary, and it remains possible that all bones from Ankevo should be considered to belong to D. abrupta.
The date of the Ampasambazimba and Ankevo subfossil deposits has been estimated at 4000–8000 years old (Waard and Straka 1961).
Description. — The postorbitals are not preserved; reconstructions suggest that they were broad, sloping downwards at a gradual angle based on the angle of their attachment to the parietals—their preserved contact with the skull roof is almost horizontal. The processus frontalis circumolfactorius is large and asymmetrical with a prominent ridge. The foramen orbitonasale is large. The upper margin of the nasal aperture is not preserved, but appears to extend to near the upper margin of the orbits. The quadrate has a notch on the upper posterior margin of the tympanic aperture. The quadrates do not diverge. The stapes and the posterior portion of the squamosal are not preserved. The processus trochlearis oticum is moderately developed. There is a vertical ridge on the lower half of the descending process of the parietals, originating at the level of, and medial to, the trigeminal nerve foramina. The foramen nervi trigemini is not divided. The processus vomerinus dorsalis is not preserved, but a broad, rough area indicates the presence of a large process. The alveolar ridges on the maxillae are well developed, strongly dentate on the lingual surface and extending onto the premaxillae. There is a depression on the basisphenoid extending onto the anterior margin of the basioccipital. The tuberculum basioccipitale is well developed, and the occipital condyle bears a ventral ridge. The palatal surface is extensively damaged; the vomer is not preserved, but a relatively wide attachment with the premaxillae is evident, and the complete vomer is visible as a wide bone in the photograph from 1980 (Fig. 1).
Discussion. — Bour (1994) considered the skull of D. abrupta to be similar to that of D. grandidieri, differing in the vomer being relatively wide and with a deep depression on the occipital condyle and in not having diverging articular condyles of the quadrates. The dentary thought to belong to this species differs from D. grandidieri in having the lateral ridge restricted to the anterior part and having the lingual ridges converging and fusing at the symphysis (in D. grandidieri the ridges terminate before the symphysis). These interpretations were followed by Gerlach and Canning (1998) and Gerlach (2004).
Examination of the material confirms the distinctiveness of the D. abrupta specimen in having a relatively wide vomer. The depression on the occipital condyle could not be substantiated, and a distinct ridge was identified instead. The articular condyles of the quadrates are not clearly divergent, being typical of all Dipsochelys taxa other than D. grandidieri.
The skull of D. abrupta is clearly most similar to that of D. grandidieri and Dipsochelys hololissa, differing in only 2 and 3 of the preserved characters, respectively (Table 1). There is less similarity with Dipsochelys arnoldi or D. dussumieri, with 8 and 10 differences, respectively. A close morphological similarity between D. abrupta, D. grandidieri, and D. hololissa has been noted previously (Gerlach and Canning 1998), and this has been suggested to indicate that D. hololissa is the most plesiomorphic of the recent descendents of the Madagascar giant tortoises.
The material of this species is very limited, being restricted to 10 partial or complete carapaces and disassociated bones, including the single referred skull. The skull is from a deposit referred to D. abrupta; although, none of the bones are in association, and it is therefore possible that at least some of the material may be confused with D. grandidieri specimens as there are suggestions that the 2 species were sympatric (Bour 1994). The limited material also does not permit any evaluation of variation in the taxon. It has been noted that some giant tortoise taxa are variable in cranial features (Van Denburgh 1914); however, this refers to quantitative characters (morphometrics and the relative size of different processes) in a single lineage. Very limited intraspecific variation has been found in the skulls of Dipsochelys and Cylindrapsis giant tortoises (Günther 1897; Bour 1985; Gerlach and Canning 1998). Accordingly, it is considered improbable that individual variation can explain the differences between the single D. abrupta skull and the 2 skulls of D. grandidieri. This supports the view that D. abrupta and D. grandidieri were at least partly sympatric.
This is unusual in that few living congeneric tortoise species are sympatric; of the larger tortoise taxa there are no extant sympatric giant tortoises (either in the Galapagos or Seychelles); Manouria species are disjunct, as are Gopherus. Chelonoidis denticulata and Chelonoidis carbonaria overlap; although, only C. denticulata is a particularly large species. There are exceptions in extinct taxa with 2 sympatric Cylindrasips species on Mauritius and Rodrigues, but possibly only one on Reunion Island (although the identity of Reunion tortoises is particularly unclear), and suggested sympatry in the recent populations of Dipsochelys taxa in the Seychelles islands (Gerlach 2004). The few definite examples of sympatric large tortoises and the suggested cases are all believed to have been broadly sympatric but locally separated by habitat preferences, with ecological separation between browsers and between grazers and browsers and/or between dense forest and edge habitat species. This may have been the case with the giant tortoises of Madagascar—unfortunately, with the extinction of all the island's giant taxa (the endemic giant tortoises, lemurs, and birds, and the more widespread hippopotamus) the natural ecological structure of the ecosystems and the precise distribution of these taxa remains speculative.
Skull of Dipsochelys abrupta (Academie Malgache ‘Ampasambazimba, 1909’), photo from 1980 (provided by R. Bour).
Reconstruction of the skull of Dipsochelys abrupta: a) dorsal, b) anterior, c) ventral, d) posterior, e) lateral, f) anterior view of section through skull, g) ventral view of processus circumolfactorius. Shading represents areas not preserved. Scale bar, 1 cm.
