DISCUSSION

Sytematic Position and Taxonomy

The specialized morphology of S. crocodilurus was recognized in its original description in 1930, and for over two decades it was classified as the only known species of the family Shinisauridae. General similarities in cranial and scale characters led McDowell and Bogert (1954) to unite S. crocodilurus with Central American Xenosaurus in an expanded Xenosauridae. The differences in masticatory muscles between S. crocodilurus and Xenosaurus were discussed by Haas (1960). In an expanded treatment of the cranial myology and osteology Rieppel (1980) concluded that retention of these two taxa within Xenosauridae was the most parsimonious explanation of the available data, but that his data provided evidence of a very early split between the two.

Subsequent comparative study of the skeletal system of S. crocodilurus and Xenosaurus resulted in a proposed resurrection of Shinisauridae, and a suggestion that its phylogenetic affinities may lie more closely with Anguidae (Hu et al. 1984). This suggestion was endorsed by Wu and Huang (1986) based on comparison of body proportions as well as similarities in the skull, mandible, vertebral column and girdles.

Many phylogenetic analyses of squamates or anguimorphs in recent years (including the influential Estes et al. [1988] paper) followed the McDowell and Bogert classification and did not analyze S. crocodilurus as a distinct terminal taxon. Estes et al. (1988) listed 12 synapomorphies uniting the two taxa into Xenosauridae, but some of these characters are now interpreted to be plesiomorphic or to apply to more inclusive groups of anguimorphs (Gao and Norell 1998; see also Hu et al. 1984). No consensus has emerged from the recent phylogenetic analyses that did treat the two as separate terminal taxa. Gao and Norell (1998) recovered a tree in which Xenosaurus and S. crocodilurus are each other’s closest living relatives, as did Lee and Caldwell (2000). Phylogenetic analyses by Caldwell (1999) did not recover them as sister taxa, and his trees suggested a closer affinity with Anguidae (see also Hu et al. 1984). A molecular phylogeny by Macey et al. (1999) also did not recover S. crocodilurus and Xenosaurus as sister taxa, and those authors reverted to separate familial-level designation for S. crocodilurus (Shinisauridae). More recently, analysis of mtDNA sequence data by Wiens and Slingluff (2001) yielded a maximum-likelihood tree in which S. crocodilurus is basal to {Xenosaurus + Anguidae}. Given the recognition of phylogenetic uncertainty for S. crocodilurus, and the morphological disparity indicating an ancient split from Xenosaurus (Rieppel 1980; Wu and Huang 1986; Conrad 2004), the continued unification of the two taxa in a single family-level grouping seems ill advised and does not “make a lot of sense” (Pianka and Vitt 2003:273). At a minimum, the two taxa should be evaluated independently in future phylogenetic analyses, and it is clear that Xenosaurus cannot be used as a convenient morphological proxy for S. crocodilurus (Conrad 2004).

Observations on Ontogenetic Change

Some ontogenetic differences in the morphology and behavior of S. crocodilurus were noted previously by several authors, but only minimal data are available. Although both adults and juveniles will seek shelter in water when disturbed, juveniles appear to spend less time under water than adults (less than 5 minutes as opposed to from 10 to 20 minutes for adults; Zhao et al. 1999). Ontogenetic changes in coloration were discussed by Fan (1931) and Zhao et al. (1999; see also color photos in Zhang et al. 1996). The acoustic membrane is distinct in hatchlings but is obscured by scales in older individuals. This is shown in the photos of Sprackland (1995, p. 9) and Mägdefrau (1997, p. 59). The acoustic membrane is already largely obscured in the 59.1 mm SVL individual we scanned. Shinisaurus crocodilurus may be unique in the transformations that take place in scale surface morphology through ontogeny (see discussion by Harvey 1993).

Preliminary observations of ontogenetic differences in the shape of the skull and of individual dermatocranial elements in S. crocodilurus were provided by Conrad (2004) and were based on examination of 11 specimens. Our observations of ontogenetic change in S. crocodilurus complement those of Conrad (2004), were derived from the examination of only four specimens (two juveniles and two adults), and are limited to the anatomical systems we investigated. There are no known previous discussions of ontogenetic changes in the braincase elements of S. crocodilurus, and there is a paucity of this type of information for lizards generally. Some of our observations of morphological features of the braincase in S. crocodilurus that undergo ontogenetic change mirror those reported for other lizards (e.g., Hikida 1978; Barahona and Barbadillo 1997, 1998; Bell et al. 2003). These features include closure of the basicranial fontanelle, increasingly tight sutural contact between elements (initially the elements are separated from one another by cartilaginous growth zones, but these subsequently form sutural contacts and eventually may fuse in large adults), and changes in the relative prominence or size of foramina and canals (e.g., anterior abducens foramen and semicircular canals).

In at least some lizards, the orbitosphenoids are not ossified in early stages of postnatal ontogeny (Barahona and Barbadillo 1998). They are present and ossified as flattened ovoid elements in TNHC 62987 but are elongated bones in FMNH 215541. In the adult, some of the surrounding cartilages are also heavily calcified and, thus, appear in the CT data sets. In iguanines the orbitosphenoid bones become increasingly complex as ossification proceeds along orbital cartilages (de Queiroz 1987), but the range of morphological variation through ontogeny in S. crocodilurus is unknown (the two CT specimens are the only specimens in which the orbitosphenoids are described).

As a result of increasing ossification, the overall shape of the sphenoid changes through ontogeny in S. crocodilurus. This is most evident in views of the height of the dorsum sella (Figure 3G, H) and in lateral view where the ventral margin becomes more strongly curved in the adult (Figure 3E, F). The finger-like secondary processes on the basioccipital processes of the juvenile (Figure 4) are reduced or lost in the adult. The basipterygoid processes show a greater expansion at their distal ends in the adult, similar to that documented for some lacertid and gymnophthalmid species (Barahona and Barbadillo 1997, 1998; Bell et al. 2003). Distinct mineralized distal tips of those processes are evident in the adult CT images (Figure 8) but are absent in the juvenile data set, which reveals that the distal ends are poorly ossified. Increased calcification/ossification at the distal ends of these processes was also reported for lacertid lizards (Barahona and Barbadillo 1998).

The most pronounced ontogenetic change in the basioccipital is the increased development of the basal tubera. In the adult CT data set distinct ossifications are visible at the ventral margins of the tubera. These changes also were reported for some lacertids and skinks (Hikida 1978; Barahona and Barbadillo 1998). In S. crocodilurus the occipital condyle is less prominent in the juvenile, and the relative contribution of the basioccipital to the condyle increases through ontogeny (Figure 1I, J).

The supraoccipital is anteroposteriorly shortened in the juvenile relative to the adult (Figure 10A-D). As is true for other lizards (Barahona and Barbadillo 1997, 1998), the cartilaginous processus ascendens of the supraoccipital in S. crocodilurus is not calcified in the juvenile. The dorsal crest is not well developed in either specimen we scanned, but it rises higher above the posterior surface of the bone in the adult. The foramen magnum and posterior cranial cavity form a smooth arch with straight vertical sides in the juvenile, but become constricted by ossification of the medial wall of the otic chamber in the adult (Figure 10G, H). Calcified endolymph is much more extensively developed in the juvenile than in the adult, supporting the hypothesis that endolymph reserves are utilized for skeletal growth in early postnatal ontogeny (e.g., Packard et al. 1985; Kluge 1987; see also the discussion by Bauer 1989).

The alar process of the prootic is well developed in the adult, but absent in the juvenile (also reported for lacertids by Barahona and Barbadillo 1997, 1998). As a consequence, the anteroposterior extent of the incisura prootica increases through ontogeny. The crista prootica also is not developed in the juvenile, so the facial foramen is exposed in lateral view (Figure 12H). The acoustic recess in the juvenile contains two foramina, one for the facial foramen and a single large foramen for the eighth cranial nerve (instead of two foramina for VIII seen in the adult). This decreased number of foramina in the acoustic recess contrasts with an increased number (four) reported in juveniles of the lacertid Gallotia galloti (Barahona and Barbadillo 1998).

The paroccipital process of the otooccipital in S. crocodilurus and other lizards is weakly developed in the juvenile (Barahona and Barbadillo 1997, 1998). In the S. crocodilurus adult the lateral edge of the paroccipital process bears a dorsally oriented process that is absent in the juvenile. The perilymphatic foramen in the juvenile is not entirely enclosed by bone as it is in the adult.

Cephalic osteoderms are well developed in the adult (Figure 18) but are absent in the juvenile. Ontogenetic differences in the presence of cephalic osteoderms were also discussed by Barahona and Barbadillo (1997).