Biology vs. Taphonomy

The cornerstone of this study is our consideration that the preserved anatomy of Adriosaurus microbrachis is not an artifact of preservation/taphonomy, but rather is the result of biologic processes, in this case, limb reduction. Our rationale for concluding that the preserved anatomy is not taphonomic follows.

The head and tail were lost due to breakage along the margins of the block during quarrying, while the poor preservation of the axial skeleton and hindlimbs is due to splitting of the block, not pre-burial taphonomy. The pectoral girdle and forelimbs were not exposed at the time of collection. Preparation of the specimen (Figs. 2A, B; 3B) revealed tiny, perfectly symmetrical, right and left humeri, both of which taper distally instead of being expanded distally as are the femora (Fig. 2A) or other typical adriosaur humeri (Fig. 3A–D). The bone surface is so well preserved that very fine wrinkles are visible on the periosteum. Preparation distal to the humeri did not recover a single limb element from either side. Additionally, more proximal elements are present, that is, the very small, unfused coracoids and scapulae, but very much reduced in size. Dermal elements such as the clavicles and interclavicle are absent, as are the costal cartilages and calcified sternum (Fig. 2A). Finally, the anteriormost two to three ribs and vertebrae show reduced pachyostosis as compared to the remaining ribs (compare Fig. 2A, B with Fig. 3A–D). The significance of this latter phenomenon is difficult to interpret by comparison to extant limb-reduced squamates because none of them are known to display skeletal pachyostosis.

Form and Function

Based on these observations, we conclude that the anatomy of Adriosaurus microbrachis preserves evidence of evolutionary developmental loss of forelimb and girdle elements. This is the first known example of extreme forelimb reduction in a fossil squamate where only a vestige of the humerus is retained. Two additional specimens of Adriosaurus sp. (MCSNT 9400 and MCSNT 7793) confirm the presence of significant limb element variation within the genus (see Fig. 3). The humeri of Adriosaurus sp. (MCSNT 7793; Fig. 3C) are stouter, larger, and more pachyostotic than those of A. suessi (BMNH R2867; Fig. 3A); the largest individual, MCSNT 9400 (Fig. 3D), possesses a gracile humerus similar to BMNH R2867 and thus indicates that humeral robustness is not allometrically related to increasing size. We consider this variation to be demonstrative of species-level differences between closely related forms and relate it to the variation observed in the speciose living scincid genera, Lerista (Greer, 1987; 1990) and Hemiergis (Shapiro, 2002), where closely related species show a full spectrum of limb reduction to limb-lessness.

In terms of their function, the vestigial limbs of Adriosaurus microbrachis appear to have been too small to be of use in typical limb assisted tetrapodal locomotion and so had lost its original function (see Hall, 2003). In modern snakes that possess a remnant pelvic girdle and femoral spur, these anatomies are often overdeveloped in males and used to assist in maintaining cloacal position during intromission (we judge it unlikely that the humeral spikes of A. microbrachis performed a similar function). A. microbrachis likely spent most of its time in the water, locomoting in an angulliform manner with all of limbs pressed close to the body (Caldwell and Dal Sasso, 2004). When out of the water, it likely locomoted in a similar manner to modern scincids such as Lerista (Greer, 1987; 1990), which display a range of forelimb reduction from absent to fully limbed; these animals are cryptic, often use the large hindlimbs as friction points, and locomote using angulliform undulations of the entire body.

Evolutionary Patterns

While the anatomy of Adriosaurus microbrachis does not directly address the complex question of snake origins, the phylogenetic hypothesis finding adriosaurs and snakes to be sister taxa within a monophyletic Pythonomorpha allows inferences on the evolution of axial elongation and limb reduction/limbloss (see Fig. 4). In this context we recognize a unique form of axial elongation shared by ophidiomorph pythonomorphs: an increased number of vertebrae in both the neck and trunk (anterior to the cloaca), and a reduction of the forelimb and girdle leading to complete loss.

While the complete cervical series is not known for Adriosaurus microbrachis, we infer it to have at least 10 (cf. A. suessi BMNH R2867), and thus a precloacal vertebral count of 38 to 39. The fully limbed Cenomanian marine squamate, Dolichosaurus longicollis, possesses 32 dorsals, and 17 to 19 cervicals, for a total of 49–51 presacrals and is approaching the maximum number of 60 presacrals observed among elongate tetrapods still possessing a forelimb and pectoral girdle (Caldwell, 2003). Importantly, for all known fossil and extant snakes, the pectoral girdle is absent, the neck appears to be very long (based on osteological characters), and the total precloacal count is significantly greater than 100 vertebrae. Complete loss of the forelimb and girdle appears to have been essential to the extreme elongation of the precloacal region of snakes.

Again, basal ophidiomorphans provide some insights on the evolution of the forelimb and girdle reduction and loss. In Dolichosaurus longicollis and Pontosaurus kornhuberi (Caldwell, 2006) the calcified sternum only possesses three facets for ribs, while in Adriosaurus microbrachis there is no sternum and thus no sternal ribs. This contrasts with the condition on Carsosaurus marchesetti, a basal mosasauroid, where there are five sternal ribs, and with mosasaurs, where there are as many as seven. Examined phylogenetically (Fig. 4), basal ophidiomorph and snake anatomy strongly suggests that complete loss of the forelimb and girdle appears to have been essential to the evolution of extreme axial elongation.

Molecules and Musculature

It is critical to note that our hypothesis that snakes have a long neck (see also Caldwell, 2000; 2003; Scanlon, 2004) is dependent on osteological criteria describing the presence, number and type of anguimorph-like hypapophyses and peduncles (intercentra). However, as the conventional osteological landmark for demarcating the posterior terminus of the neck is the pectoral girdle (specifically the vertebra bearing the first sternal rib), the neck of snakes remains problematic despite the presence of hypapophyses (Hoffstetter and Gasc, 1969).

In attempting to find other criteria for addressing the evolution of elongate axial skeletons, Cohn and Tickle (1999), using molecular genetic data, proposed that the neck is absent in snakes. They found that in Python, the Hox gene, HoxC8, normally expressed axially just posterior to the pectoral girdle (i.e., in limbed tetrapods), was expressed anteriorly up to the first or second vertebra. In addition, it was noted that anterior vertebrae have ribs, and that a large number of vertebrae, far posterior to the tetrapodal pectoral girdle, possess ventral hypapophyses (a neck character). The authors concluded that anterior ribs indicated that thoracic identity had been overprinted onto the snake ‘neck’, and that the trunk resembles an elongated thorax (Cohn and Tickle, 1999). However, the presence of ribs on anterior vertebrae (cervical ribs and cervical vertebrae) is a primitive character of tetrapods (Caldwell, 2003; Scanlon, 2004). Thus the overprint of regional identity appears to be cervical, not thoracic.

A third, and very recent hypothesis based on data homologizing the lizard m. episternocleidomastoid (ESCM) with the snake m. cervicoquadratus (CQ), is that snakes have a comparatively short neck (no more than 11 vertebrae) (Tsuihiji et al., 2006). This conclusion is interesting but problematic because 11 vertebrae actually constitute a long squamate neck, and is consistent with the value of 10 or more as a synapomorphy for Ophidiomorpha. The origin of the snake CQ is in the dermis or fascia while its insertion is on the quadrate. In contrast, the origin (pectoral girdle elements) and insertion (cranial elements including the quadrate) of the lizard ESCM is very complex. Additionally, the ESCM is, with the exception of amphisbaenids, innervated by cranial nerve XI (spinal accessory), which is absent in snakes and amphisbaenids. The only similarities between CQ and ESCM are the quadrate insertion and innervation by two to three spinal nerves, rendering the hypothesis of homology by Tsuihiji and colleagues (2006) as very problematic. Recognizing this, the authors note that osteological, myological and molecular data present three different patterns for identifying the neck in snakes.

Conclusions

Whether the patterns of limbloss and axial elongation observed in Upper Cretaceous aquatic snakes and lizards are convergent or homologous remains controversial. To properly address this problem, the synergistic efforts developing between evolutionary developmental biology and paleontology must be directed toward addressing the criteria for defining anatomical regions with developmental boundaries. Otherwise, any attempts to understand the evolution of anatomy, such as that displayed by Adriosaurus microbrachis, will be brought up short by the limitations of a static system, that is, the limits of an individual, when the real goal is to understand the transformation of those systems within populations, and between species and other higher taxa.