Maureen Kearney

Assembling the Squamate Tree of Life: Perspectives from the Phenotype and the Fossil Record

Abstract We assembled a dataset of 192 carefully selected species—51 extinct and 141 extant—and 976 apomorphies distributed among 610 phenotypic characters to investigate the phylogeny of Squamata (“lizards,” including snakes and amphisbaenians). These data enabled us to infer a tree much like those derived from previous morphological analyses, but with better support for some key clades. There are also several novel elements, some of which pose striking departures from traditional ideas about lizard evolution (e.g., that mosasaurs and polyglyphanodontians are on the scleroglossan stem, rather than parts of the crown, and related to varanoids and teiids, respectively). Long-bodied, limb-reduced, “snake-like” fossorial lizards—most notably dibamids, amphisbaenians and snakes—have been and continue to be the chief source of character conflict in squamate morphological phylogenetics. Carnivorous lizards (especially snakes, mosasaurs and varanoids) have proven a close second. Genetic data, presumably less burdened by the potential for adaptive convergence related to fossoriality, were expected to resolve these conflicts. Although recent gene phylogenies seem to do so, they also differ radically from any phylogeny based on the phenotype, especially for the most ancient crown squamate divergences that occured during the latter half of the Mesozoic. Our study relied on traditionally prepared specimens as well as high-resolution computed tomography scans that afforded unprecendented access to the cranial anatomy of Squamata. This, along with the inclusion of stem fossils, provided an unparalleled sample of the phenotype enabling us to more fully explore the extreme incongruences between molecular and morphological topologies for the squamate tree of life. Despite this extensive new database, we were unable to find morphological support for the major rearrangement of the deep divergences in Squamata proposed by recent molecular studies. Instead, our data strongly support the same fundamental topology suggested by most previous morphological studies—an Iguania-Scleroglossa basal split, a sister-group relationship between Gekkota and Autarchoglossa, and the divergence between Anguimorpha and Scincomorpha—and documents the extreme degree of morphological homoplasy required by those molecular topologies.

Rejecting “the given” in systematics

How morphology and systematics come together through morphological analysis, homology hypotheses and phylogenetic analysis is a topic of continuing debate. Some contemporary approaches reject biological evaluation of morphological characters and fall back on an atheoretical and putatively objective (but, in fact, phenetic) approach that defers to the test of congruence for homology assessment. We note persistent trends toward an uncritical empiricism (where evidence is believed to be immediately “given” in putatively theory‐free observation) and instrumentalism (where hypotheses of primary homology become mere instruments with little or no empirical foundation for choosing among competing phylogenetic hypotheses). We suggest that this situation is partly a consequence of the fact that the test of congruence and the related concept of total evidence have been inappropriately tied to a Popperian philosophy in modern systematics. Total evidence is a classical principle of inductive inference and does not imply a deductive test of homology. The test of congruence by itself is based philosophically on a coherence theory of truth (coherentism in epistemology), which is unconcerned with empirical foundation. We therefore argue that coherence of character statements (congruence of characters) is a necessary, but not a sufficient, condition to support or refute hypotheses of homology or phylogenetic relationship. There should be at least some causal grounding for homology hypotheses beyond mere congruence. Such causal grounding may be achieved, for example, through empirical investigations of comparative anatomy, developmental biology, functional morphology and secondary structure.

An Investigation into the Occurrence of Plicidentine in the Teeth of Squamate Reptiles

Abstract Despite its use as a diagnostic taxonomic feature, the occurrence and distribution of plicidentine in the teeth of squamate reptiles is unclear. This appears to be due to several factors: the various kinds of folding, wrinkling, and striation that occur within different dental tissues; difficulty of interpreting conditions in poorly preserved extinct taxa; and incomplete knowledge of tooth development. We investigated tooth development and morphology in extant and fossil squamate reptiles using skeletal preparations, histological sections, and CT-scanning data. Among squamates, we found plicidentine only in the teeth of varanoid lizards and note that much more anatomical complexity exists than previously thought in the dental and attachment tissues of these groups. Degree of development of plicidentine is variable within varanoids, with the strongest development occurring in some species of Varanus. In contrast to some reports, we found no evidence for the occurrence of plicidentine in the teeth of mosasauroid lizards or snakes. Some mosasaurs exhibit raised ridges on the enamel surface that extend from the base to the tip of the tooth, as well as occasional striation of the tooth bases built from bone of attachment; neither of these features is considered homologous to plicidentine infolding. Some snakes exhibit weak wrinkling of the tooth base that corresponds closely to the pattern of wrinkling in the overlying bone of attachment. This condition occurs infrequently in snakes, and details of tooth development and attachment also do not support its homology with plicidentine. Our results indicate that plicidentine is best interpreted as a synapomorphy of Varanoidea.

First Report of a Pectoral Girdle Muscle in Snakes, with Comments on the Snake Cervico-dorsal Boundary

Abstract The morphology, topological relationships, and innervation patterns of m. cervicoquadratus in various snakes, especially basal taxa, were examined. Nearly all aspects of its anatomy suggest that this muscle is homologous with a part of m. cucullaris, most likely m. episternocleidomastoideus, the muscle that connects the skull and pectoral girdle in non-ophidian squamates. This is the first report of any pectoral girdle muscle persisting in snakes. In snakes examined, the most posterior extent of m. cervicoquadratus does not extend more posteriorly than the level of the 11th precloacal vertebra, suggesting that the length of the neck inferred on this characteristic would be quite short compared to the length of the entire precloacal region. However, different anatomical and developmental characteristics are known to indicate various lengths of the neck in snakes. This suggests that such characteristics potentially evolve independently from one another, and thus no single criterion may be sufficient to delimit the neck in snakes.

The Random Cladist: A Review of the Software Package RANDOM CLADISTICS

The software package RANDOM CLADISTICS by M. Siddall (1995a, version 4.0 the Ohio edition) is intended to conduct randomization analyses, data editing functions, and the incongruence length difference test. The program has been updated recently and is available free of charge on this society’s web site (www.zoo.toronto.edu/,mes/hennig/hennig.html). The programs which make up the RANDOM CLADISTICS package work in conjunction with either HENNIG86 (Farris, 1988) or NONA (Goloboff, 1993), both phylogenetic analysis programs for MS-DOS computers. The RANDOM CLADISTICS package could be viewed as a supplement to HENNIG86 and NONA for users who wish to add more statistical tests to their analyses. Since the RANDOM CLADISTICS package closely interfaces with HENNIG86 and NONA, many of the results are dependent on the capabilities of the daughter software. To fully appreciate the interaction of this software, it should be understood that the RANDOM CLADISTICS package is responsible for the random sampling portion of these analyses and the building of new resampled matrices, whereas HENNIG86