A Fossil Snake with Limbs

See allHide authors and affiliations

Science  17 Mar 2000:
Vol. 287, Issue 5460, pp. 2010-2012
DOI: 10.1126/science.287.5460.2010


A 95-million-year-old fossil snake from the Middle East documents the most extreme hindlimb development of any known member of that group, as it preserves the tibia, fibula, tarsals, metatarsals, and phalanges. It is more complete than Pachyrhachis, a second fossil snake with hindlimbs that was recently portrayed to be basal to all other snakes. Phylogenetic analysis of the relationships of the new taxon, as well as reanalysis of Pachyrhachis, shows both to be related to macrostomatans, a group that includes relatively advanced snakes such as pythons, boas, and colubroids to the exclusion of more primitive snakes such as blindsnakes and pipesnakes.

The lower to middle Cenomanian (basal Upper Cretaceous) carbonates of ‘Ein Yabrud near Jerusalem, deposited in a low-energy shallow marine platform environment (1), have yielded two species of fossil snakes,Pachyrhachis problematicus (2–4) and the new taxon reported here. Because of the presence of relatively well-developed hindlimbs and a supposedly primitive skull structure, a series of recent publications (5–7) have interpreted Pachyrhachis to be basal to all other snakes, indeed to represent “an excellent example of a transitional taxon” (8) linking snakes to an extinct group of “lizards,” the mosasauroids. On the basis of this pattern of phylogenetic relationships, it was claimed that snakes had a marine origin (8) and that the mosasauroid jaws provided the starting point for the evolution of the ophidian feeding mechanism (9). The transitional position of Pachyrhachisinfluenced a scenario explaining the origin and evolution of limblessness in snakes, based on the analysis of underlying developmental mechanisms as revealed by patterns of Hox gene expression in Python (10). The basal position ofPachyrhachis and the putative relationships of snakes to mosasauroids were tested by a review of the character evidence and the methods of phylogenetic analysis used, and were found to be refuted by the position of Pachyrhachis as the sister taxon of relatively advanced (i.e., macrostomatan) snakes (11–15).

Here, we describe the second snake from ‘Ein Yabrud, which is better preserved than Pachyrhachis in the skull and hindlimb, and which highly corroborates the macrostomatan affinities of these fossil snakes. Haasiophis, gen. nov.

Genotypical species: Haasiophis terrasanctus, sp. nov.

Diagnosis: A snake with a snout-vent length of 717 mm; premaxilla small and narrow, edentulous; 24 tooth positions on the maxilla, 8 on the palatine, 15 to 17 on the pterygoid, and 26 on the dentary; enamel surface of teeth distinctly striated; mandibular nerve foramen underlapped by distinct prootic process; quadrate slender and vertically oriented; coronoid process on mandible small, formed by coronoid bone only; 155 precloacal vertebrae; at least 12 proximal caudal vertebrae with distally expanded and bifurcated lymphapophyses; expanded hemapophyses on posterior tail vertebrae.

Distribution: Early Upper Cretaceous, Middle East. Haasiophis terrasanctus, sp. nov.

Holotype: Hebrew University of Jerusalem, Paleontological Collections, HUJ-Pal. EJ 695.

Stratum typicum: Aminadav Formation or the slightly younger Bet-Meir Formation, middle part of the Judea Group, early to middle Cenomanian, basal Upper Cretaceous.

Locus typicus: Limestone quarries of ‘Ein Yabrud, Judean hills, 20 km north of Jerusalem.

Diagnosis: Same as for genus, of which this is the only known species (specimen).

Etymology: Haasiophis, in honor of Prof. G. Haas, who initiated research on vertebrate fossils from ‘Ein Yabrud;ophis (Greek, snake); terrasanctus (Latin, Holy Land).

This specimen is identified as a fossil snake on the basis of its highly kinetic skull with anteriorly free ending maxillae and dentaries, slender and elongate tooth-bearing palatines and pterygoids, single mental foramen in the dentary, high number of presacral vertebrae, and the presence of hypapophyses or hemal ridges throughout the trunk, distally bifurcated lymphapophyses in the cloacal and proximal tail region, and paired hemapophyses on the tail vertebrae. Its cranial structure (Fig. 1) (16) displays relatively primitive characters, such as are present in anilioids (pipesnakes) with advanced macrostomatan features. The extended contact between the anteroventrally sloping prefrontal and the ascending process of the maxilla is plesiomorphic, as is the coronoid process on the lower jaw formed by the coronoid bone only. Advanced features include an elongate preorbital region recalling the condition seen in Python; a nearly complete postorbital arch; highly mobile connections among the elements of the dermal palate and upper jaw (vomer, palatine, pterygoid, ectopterygoid, maxilla, and premaxilla); the presence of well-developed [neomorph (17)] basipterygoid processes as revealed by radiographs; a slender, elongate, and vertically oriented quadrate suspended from a posteriorly free-ending supratemporal; the development of longitudinal crests for muscle attachment both on the skull roof (parietal, supraoccipital) and on the skull base (parabasisphenoid and basioccipital); and the anterior extent of the splenial and development of a deep fossa for the insertion of jaw adductor muscles on the lower jaw. The exoccipitals appear not to meet above the foramen magnum, but this may well be an artifact of preservation.

Figure 1

The skull of H. terrasanctus in dorsal (above) and ventral (below) views. Abbreviations: ang, angular; bo, basioccipital; bs, basisphenoid (parabasisphenoid); c, coronoid; com, compound bone; d, dentary; ec, ectopterygoid; eo, exoccipital; f, frontal; m, maxilla; mp, medial process of palatine; n, nasal; p, parietal; pl, palatine; pm, premaxilla; po, postorbital; prf, prefrontal; pro, prootic; ps, parasphenoid rostrum; pt, pterygoid; q, quadrate; so, supraoccipital; sp, splenial; st, supratemporal; and v, vomer.

Except for pachyostosis and the well-developed hindlimb (Fig. 2), the postcranial skeleton ofHaasiophis is typically snake-like. Pachyostosis of vertebrae and ribs occurs between the 45th to 48th and the 105th to 108th vertebrae, with a distinct hypertrophy of the parapophysis separated by a furrow from the smaller, dorsal diapophyseal component of the rib articulation. Anterior hypapophyses are gradually transformed to distinct hemal ridges along the trunk. Broad and plate-like hemapophyses add to the lateral compression of the tail, which must have served as a propulsive organ.

Figure 2

The limb of H. terrasanctus. Abbreviations: as, astragalus; ca, calcaneum; dt4, fourth distal tarsal; fe, femur; fi, fibula; ph, phalanx; mt, metatarsal; r, rib (of 154th vertebra); ti, tibia.

The last dorsal rib is associated with the 154th vertebra. There is no evidence for the suspension of rudimentary pelvic elements from the axial skeleton. Two poorly preserved, obliquely oriented, delicate rods of bone, located near the 155th vertebra, may represent the pubis and ilium of a rudimentary, originally triradiate pelvis. The left femur (Fig. 2) is a small (7.2 mm long), straight, slender element with moderately expanded proximal and distal ends, which emerges from below the last dorsal rib. The tibia (3.3 mm), characterized by a relatively broad proximal end, has been flipped across the fibula (3.1 mm) during fossilization. Three tarsal ossifications are identified as astragalus, calcaneum, and the fourth distal tarsal. The straight metatarsals of digits two through five are at least partially preserved, as are two partial phalanges.

Phylogenetic analysis (Fig. 3) (18) shows Haasiophis to be the sister taxon of Pachyrhachis, both nested within basal macrostomatans (i.e., near pythons and boids). Statistical support for the position of Haasiophis and Pachyrhachiswithin Alethinophidia (Fig. 3), and for their relationship to Macrostomata, is strong (18–20). By contrast, the position of these taxa within basal macrostomatans, as well as the sister-group relationship of Haasiophis and Pachyrhachis, remain weakly supported, probably because of their divergent specialization. Boine characters of Haasiophis are the laterally projecting process of the prootic (underlapping the mandibular nerve foramen) and the posteriorly dilated free-ending process of the supratemporal. Pythonine characters ofPachyrhachis are the straight frontoparietal suture and the nature of the postorbital-parietal contact. Haasiophis andPachyrhachis differ in other respects as well, such as tooth counts, shape and relative size of the coronoid process and quadrate, size of the neural spines on the anterior (“cervical”) vertebrae, differentiation of the ribs, and relative proportions of the stylopodial and zeugopodial limb elements. Haasiophistherefore cannot represent a juvenile specimen of the largerPachyrhachis.

Figure 3

The phylogenetic relationships of H. terrasanctus. A strict consensus tree of two equally parsimonious trees is shown [(18); for list of apomorphies, see (16)].

Given the relationships of Pachyrhachis andHaasiophis to macrostomatans, the presence of well-developed hindlimbs optimizes unequivocally as a reversal (Fig. 3). Implicit weight can be added to the (plesiomorphic) presence of limbs by splitting those into discrete characters numerous enough to pull the fossils to the base of the ophidian tree. The number of limb characters required to break Haasiophis and Pachyrhachisaway from macrostomatans is 14, and 15 limb characters are required to pull these fossils to a basal position. Loss of resolution throughout the cladogram, caused by the addition of more than 13 limb characters, is significant, indicating that the overall data set matches the prediction of a redevelopment of the hindlimbs better than it matches the assumption that the skulls of Pachyrhachis andHaasiophis are convergent on macrostomatans.

As macrostomatan snakes, Haasiophis andPachyrhachis have no particular bearing on snake-mosasauroid relationships or snake origins. Instead, they represent the first unequivocal documentation of the incursion of macrostomatan snakes into the sea. Basal snakes—including basal macrostomatans—retain rudimentary hindlimbs, which, however, remain much more incomplete than those of Haasiophis. With Haasiophis andPachyrhachis related to basal macrostomatans, the conclusion based on parsimony must be that these limbs redeveloped from rudiments such as those present in Python (10). The assumption of a multiple loss of hindlimbs among basal snakes is less parsimonious but remains a possibility, given the incompleteness of the fossil record of snakes (21) and the recognition of multiple loss of limbs among squamates in general (22).

  • * To whom correspondence should be addressed. E-mail: rieppel{at}


View Abstract

Stay Connected to Science

Navigate This Article