Najash, : A FRESH START FOR THE DEBATE ON THE

Zaher, Hussam, Apesteguía, Sebastián & Scanferla, Carlos Agustín, 2009, The anatomy of the upper cretaceous snake Najash rionegrina Apesteguía & Zaher, 2006, and the evolution of limblessness in snakes, Zoological Journal of the Linnean Society 156, pp. 801-826 : 820-823

NAJASH: A FRESH START FOR THE DEBATE ON THE

EVOLUTION OF LIMBLESSNESS IN SNAKES

Forelimbs are completely absent in all extant snakes, and in the fossil snakes Pachyrhachis , Haasiophis , Eupodophis , Dinilysia , and Najash . On the other hand, hindlimbs persist in several families of extant snakes as rudimentary elements, consisting of a pelvic girdle, a rudimentary femur, and an external claw-like vestige. These elements are completely lost in the derived Caenophidian snakes, and in a number of unrelated alethinophidian snake families (i.e. Xenopeltidae , most Uropeltinae, the genus Tropidophis within Tropidophiidae , Bolyeriidae , and Xenophidiidae ).

The presence of fully formed hindlimbs in the Tethyan marine limbed snakes Haasiophis , Pachyrhachis , and Eupodophis enforced the idea that they were the most primitive snakes known, and that they were perfect transitional taxa linking extant snakes to the marine mosasauroids ( Caldwell & Lee, 1997). However, as previously suggested by Zaher & Rieppel (2000), the appendicular skeleton of the Cenomanian marine limbed snakes corresponds more accurately to a derived ophidian morphology, with free lymphapophyses and a pelvis that is not suspended from the axial skeleton, but rather lies within the ribcage and lacks a differentiated sacral region ( Fig. 13). On the other hand, Najash lacks free lymphapophyses, and retains two sacral vertebrae that anatomically separate the trunk region from the caudal region, with a pelvis that is functionally but loosely connected to the sacral region, and which lies outside of the ribcage: all being plesiomorphic nonophidian conditions ( Fig. 13).

The loss of a sacral region and consequent disconnection of the pelvis from the axial skeleton represents a synapomorphy of extant snakes, shared by Pachyrhachis , Haasiophis , and Eupodophis , whereas the presence of free lymphapophyses is a derived feature shared by extant snakes, Pachyrhachis , and probably Haasiophis , but not by Eupodophis ( Rage & Escuillié, 2000; Rieppel & Head, 2004). The re-interpretation of the postcranial anatomy of Pachyrhachis , Haasiophis , and Eupodophis renders it more closely comparable with the rudimentary hindlimbs of extant snakes than to non-ophidian squamates, a result that is in accordance with a derived position of these fossil taxa within the extant snake clade ( Zaher, 1998; Tchernov et al., 2000; Rieppel et al., 2002; Apesteguía & Zaher, 2006), rather than as the basalmost snakes ( Caldwell & Lee, 1997; Rage & Escuillié, 2000).

Scanlon & Lee (2000) and Rage & Escuillié (2000) reported the presence of true chevron bones (Y-shaped elements that articulate or fuse with paired pedicels located at the posterior margin of the centrum of the caudal vertebrae) in W. naracoortensis and E. descouensi , respectively, another allegedly primitive feature of these snakes. However, Rieppel & Head (2004) argued convincingly that the structures called ‘chevron bones’ in Eupodophis are unique, and not homologuous with the chevron bones of other squamates. The two arms of the Y-shaped element embrace an unpaired, blade-like pedicel that is located at the anterior end of the centrum, a morphology that is not comparable with any condition known in non-ophidian squamates ( Rieppel & Head, 2004: 21). Similarly, the assignment of a single disarticulated caudal vertebra with a chevron bone to W. naracoortensis by Scanlon & Lee (2000) was questioned by Rieppel et al. (2002), as this vertebra might not belong to the specimen described by Barrie (1990; but see Scanlon, 2005: 143).

Scanlon (1993) recognized the presence of true haemapophyses (paired pedicels, laminae, or rod-like projections that do not fuse with each other distally, located at the posterior margin of the centrum of the caudal vertebrae) in the caudal vertebrae of an undetermined species of Patagoniophis from the Eocene of Tingamarra (Queensland), and Rage (1998) described haemapophyses in the posterior caudal vertebrae of Madtsoia camposi Rage, 1998 from the Paleocene of Itaboraí ( Brazil). The presence of true haemapophyses in Patagoniophis and Madtsoia suggest that most ‘Madtsoiids’ did retain the derived condition known for all snakes, and that the chevron bone reported in Wonambi by Scanlon & Lee (2000) should be viewed with caution.

It can be concluded that there is no unquestionable record of true chevron bones in snakes, and the presence of well-preserved haemapophyses in the posterior caudal vertebrae of N. rionegrina seems to support the view that chevron bones are absent in snakes ( Fig. 12).

The recent report by Cohn & Tickle (1999) on Hox gene expression patterns in Python represents an important piece of evidence from a distinct source, which helps in the better understanding of the evolution of limblessness in snakes. However, some of their conclusions regarding the steps that led to the acquisition of limblessness in snakes need to be re-evaluated, according to the new morphological and phylogenetic implications brought by the subsequent discovery of N. rionegrina .

Cohn & Tickle (1999) suggested that the progressive expansion of Hox gene expression domains along the body axis is the main factor responsible for the evolution of limblessness in snakes, and would account for the loss of forelimbs, hindlimbs, and regional identity along the axial skeleton in snakes (i.e. distinct cervical, thoracic, lumbar, and sacral regions). These authors used Caldwell & Lee’s (1997) phylogenetic hypothesis, in which Pr. problematicus represents the most primitive snake, and an intermediate form between mosasaurs and extant snakes, as a backbone hypothesis for their developmental scenario of snake evolution ( Cohn & Tickle, 1999: fig. 5). According to Cohn & Tickle’s (1999) model, the expansion of Hox domains that led to the transformation of the entire axial skeleton towards thoracic identity, and to the reduction of hindlimb development by the elimination of the ectodermal competence, to form an apical ridge expansion, appeared only in the ancestor of extant snakes, excluding Pachyrhachis and, by extension, the remaining Tethyan marine legged snakes, Haasiophis and Eupodophis . Alternatively, if Najash represents the sister taxon to all known extant and extinct snakes, including Pachyrhachis , Haasiophis , and Eupodophis , and if the latter three are deeply nested within alethinophidian snakes as derived macrostomatans, then, according to the model advanced by Cohn and Tickle, the reduction of regional differentiation in the axial skeleton must have occurred in the ancestor of the clade formed by Najash and all other snakes, with the former retaining a sacral region posterior to a uniform presacral region (i.e. there was a loss of a distinct cervical region; phase 1 of Cohn and Tickle; Fig. 14). The loss of the sacral region through the transformation of the entire axial skeleton towards thoracic identity (phase 2 of Cohn and Tickle) appeared only posteriorly, in the ancestor of the clade formed by extant snakes, including the Tethyan marine legged snakes ( Fig. 14). The loss of the sacral region also caused the disconnection of the pelvis from the axial skeleton and the consequent anteroventral dislocation of the entire appendicular skeleton to a position internal to the ribcage. The loss of the apical ridge might also have occurred concomitantly in phase 2 described above ( Fig. 14). However, the potential of the hindlimb bud mesenchyme to act as a polarizing region, and to coordinate limb bud outgrowth, as shown by Cohn & Tickle (1999) in Python , is retained not only in the latter genus, but also in most scolecophidians, anilioids, and macrostomatans. Such potential plasticity explains the strikingly distinct rudimentary patterns of the appendicular skeleton shown by scolecophidians, anilioids, and macrostomatans ( Duerden & Essex, 1923; Bellairs, 1950; List, 1955, 1966; Mlynarski & Madej, 1961; Underwood, 1977). It also helps understand the presence of well-developed hindlimbs in the macrostomatan fossil snakes Pachyrhachis , Haasiophis , and Eupodophis .

A possible explanation for the presence of hindlimbs in the Tethyan marine snakes is that the reduction of hindlimb elements occurred independently in the major basal clades of extant snakes (i.e. in Scolecophidia, Anilioidea, and Macrostomata) ( Zaher & Rieppel, 1999a; Greene & Cundall, 2000). Under such a scenario (our preferred hypothesis), basal representatives of most major modern snake lineages would have retained somewhat well-developed hindlimb morphologies during the Cretaceous. The process of independent reductions to vestigial posterior hindlimbs during the late Cretaceous, and possibly early Cenozoic, also resulted in independent losses within these lineages (e.g. in Xenopeltidae , most Uropeltinae, the genus Tropidophis within Tropidophiidae , Bolyeriidae , Xenophidiidae , and Caenophidia). The hypothesis of independent losses is directly open to testing by the fossil record, as it presupposes that basal fossil representatives of all major extinct and extant lineages of modern snakes are likely to have retained well-developed hindlimb morphologies during at least the Cretaceous period ( Zaher & Rieppel, 1999a). Although the latter hypothesis is less parsimonious than the alternative re-development of complete hindlimbs in the Tethyan marine snakes, it fits more accurately with the new evolutionary scenario revealed by the intermediate morphology of Najash , which suggests that the loss of the sacral region occurred prior to the reduction of hindlimbs.