A Molecular Phylogeny of Reptiles

See allHide authors and affiliations

Science  12 Feb 1999:
Vol. 283, Issue 5404, pp. 998-1001
DOI: 10.1126/science.283.5404.998


The classical phylogeny of living reptiles pairs crocodilians with birds, tuataras with squamates, and places turtles at the base of the tree. New evidence from two nuclear genes, and analyses of mitochondrial DNA and 22 additional nuclear genes, join crocodilians with turtles and place squamates at the base of the tree. Morphological and paleontological evidence for this molecular phylogeny is unclear. Molecular time estimates support a Triassic origin for the major groups of living reptiles.

The number of temporal openings in the skull has long been viewed as a key character in the classification of reptiles (1, 2). A single opening (synapsid condition) is found in mammals and their reptilian ancestors. Turtles and several late Paleozoic and early Mesozoic groups lack a temporal opening (anapsid condition). Most other living and fossil reptiles belong to a clade in which the ancestral condition was the presence of two temporal openings (diapsid condition). Among the living diapsid reptiles, the crocodilians and birds form one group, the Archosauria, and the tuataras and squamates (lizards, snakes, and amphisbaenians) form another group, the Lepidosauria. Other morphological characters (2–7) support this traditional phylogeny of living amniote vertebrates (Fig. 1A). An alternative phylogeny is that turtles are diapsid reptiles and most closely related to the lepidosaurs (8, 9); however, this lepidosaur connection has been contested (10,11).

Figure 1

Relationships among the major groups of living reptiles. (A) The classical phylogeny based on morphology and the fossil record (1, 2). (B) Maximum likelihood phylogeny of combined sequences from 11 nuclear proteins (1943 amino acids). Scale bar indicates amino acid substitutions per site. (C) Consensus phylogeny of combined sequences from four nuclear protein-coding genes for which sequences of tuatara are available (785 amino acids). For the molecular trees, confidence values (%) supporting the nodes are separated by slash marks and based on the following four methods: interior-branch test, and bootstrap analyses of neighbor-joining, maximum likelihood, and maximum parsimony, respectively. In (B) it was determined that myoglobin had the greatest effect in lowering confidence values; removal of that gene did not change significance of turtle-crocodilian node but raised support for bird-turtle-crocodilian node to 94/90/94/88.

Molecular phylogenies have been unable to corroborate the traditional phylogeny of living amniotes. In most studies turtles have clustered with archosaurs (12–14), but the number of genes or sites available generally has been small. When only crocodilians, birds, and mammals were considered, combined sequence data from 15 nuclear genes significantly supported a bird-crocodilian relationship (15), but the position of turtles and other reptiles was left unresolved. Here we address relationships among the major groups of living reptiles (turtles, tuataras, squamates, crocodilians, and birds) with new evidence from two nuclear genes and analyses of other available molecular sequence data bearing on this question.

We sequenced the coding region of the gene for alpha enolase (phosphopyruvate hydratase) in five species of reptiles and the gene for 18S ribosomal RNA (rRNA) in a tuatara to compare with other reptilian sequences of those genes (16). For phylogenetic and timing analyses, these new sequences were added to 340 available protein and DNA sequences representing 24 nuclear and 9 mitochondrial genes (17). The sequences were aligned and subjected to several phylogenetic analyses (18).

Turtles clustered with one or both archosaurs in individual analyses of all 15 genes having representatives of squamates, birds, crocodilians, and turtles (Table 1). In nine of those genes, turtles joined crocodilians in either nucleotide or amino acid analyses. The classical phylogeny (Fig. 1A) was not supported in any comparison, and only one gene analysis resulted in a turtle-squamate relationship. A combined analysis of sequences from all nuclear proteins resulted in significant support (>97% confidence) for a turtle-crocodilian relationship on the basis of an interior branch test and bootstrapping of neighbor-joining, maximum likelihood, and maximum parsimony trees (Fig. 1B). The maximum likelihood tree (Fig. 1B) (lnL = −11405) was significantly better (Δln L> 2 SE) than the classical tree (Fig. 1A) (ln L = −11481 ± 22.1) or the tree grouping turtles with lepidosaurs (squamates + tuataras; ln L = −11477 ± 22.6). A four-cluster analysis (19) also supported (>98% better than both alternative hypotheses) a turtle-crocodilian relation when combined protein sequence data (1943 sites) and combined nuclear rRNA sequence data (2299 sites) were analyzed separately.

Table 1

Molecular evidence for the closest living relative of turtles.

View this table:

Fewer genes are available to address the position of the tuataras, but in six of eight comparisons they cluster with archosaurs or turtles rather than squamates (Table 2). Tuataras are excluded from a close relation with squamates in all analyses of combined proteins, and this is significant with the interior branch test (96%) and bootstrap analysis of maximum likelihood (95%) (Fig. 1C). However, the difference between the maximum likelihood tree (ln L = −5516) and the classical phylogeny (Fig. 1C) (ln L = −5540 ± 16.5) is not significant (<2 SE). Although the remaining nodes in the combined protein analysis are not well resolved, the combined nucleic acid analysis (Table 2) defines a cluster (turtles, crocodilians, and birds) that excludes tuataras. If 18S rRNA, which favors a bird-mammal grouping (13), is removed, support for that cluster is significant (100%, interior branch; 95% neighbor-joining, 97% maximum likelihood). Thus the classic grouping of squamates with tuataras is not supported by available molecular data, and there is support in most analyses for tuataras as closest relatives of a group containing turtles, crocodilians, and birds.

Table 2

Molecular evidence for the closest living relative of tuataras.

View this table:

We analyzed a total of 23 nuclear genes and the two mitochondrial regions (9 genes) to estimate divergence times (20). The time estimates indicate that squamates diverged from the other reptiles at 245 ± 12.2 million years ago (Ma) (9 genes), birds diverged from the lineage leading to turtles and crocodilians at 228 ± 10.3 Ma (17 genes), and that turtles diverged from crocodilians at 207 ± 20.5 Ma (7 genes). These divergence times are close to when the first turtles (223 to 210 Ma) and crocodilians (210 to 208 Ma) appear in the fossil record (21) and earlier than the first birds (152 to 146 Ma) and first squamates (157 to 155 Ma). These results support the notion that many key innovations in tetrapod evolution occurred during the Triassic (251 to 208 Ma) (2).

In light of this phylogeny of reptiles, early molecular analyses that clustered birds with mammals (13, 22) now are more easily explained. When there are no lepidosaurs in an analysis, birds become the basal lineage of reptiles. Thus, birds are closer to mammals in a network and may join together more easily in some analyses, especially if rates of evolution vary among sites or lineages. Typical proteins comprise only 200 to 300 amino acids, and therefore strong statistical support for any phylogeny usually requires multiple genes.

The use of different outgroups also can result in different topologies. For example, a previous study of mitochondrial rRNA genes, in which a more distant (amphibian) root was used, yielded a bird-crocodilian grouping that excluded turtles (15). In this study, mammals were used to root the trees because they were the closest available outgroup and yielded the best alignments. However, the use of more distant roots did not affect the topology of most nuclear genes and did not change the conclusions of this study.

The molecular support for a turtle-crocodilian clade is surprising considering that it seems to have virtually no support from morphology. Even recent studies showing diapsid affinities of turtles did not find a close relation between turtles and archosaurs (8,9). It has been reported that turtles are most similar to crocodilians in sperm morphology (23), but phylogenetic analysis of sperm characters did not support that proposition (24).

Turtles and crocodilians are the only living tetrapod groups with dorsal and ventral bony plates of armor. Some lizards have superficial dermal ossifications, but there is no evidence of dermal armor in primitive squamates, sphenodontians, or synapsids (25). Ventral armor is known in extinct diapsids, including sauropterygians and archosauromorphs (25), but the significance of this character will be unclear until the position of turtles and crocodilians among those groups is established.

Crocodilians have large heads, long snouts, and well-developed teeth. However, some Triassic suchians (archosaurs), such as the aetosaurs (2), have small heads with beaklike jaws and greatly reduced teeth. Body armor was well developed, and their ventral plating has been described as a plastron (2, 25). In one aetosaur (25), the neck spines resembled those of an early turtle (26). Some or all of these similarities may be the result of convergence, but they show that characteristics of turtles and crocodilians can be found together in some extinct reptiles of the Triassic.

The finding in most analyses of the combined molecular data (Table 2and Fig. 1C) that tuataras and squamates are not closely related also is unconventional. Recently, an analysis of characters from sperm morphology argued for a dismantling of the Lepidosauria (24), but otherwise the phylogeny in that study does not agree with the results of this study.

These results highlight a significant discordance between morphological and molecular estimates of phylogeny for a major group of organisms. The consistent pattern of a turtle-crocodilian relationship across independent nuclear genes stands in contrast to the traditional phylogeny of amniote vertebrates. Determining how the many groups of extinct reptiles of the late Paleozoic and early Mesozoic fit into this molecular phylogeny will be a challenge to paleontologists.

  • * To whom correspondence should be addressed at 208 Mueller Laboratory, Pennsylvania State University, University Park, PA 16802, USA. E-mail: sbh1{at}


View Abstract

Stay Connected to Science

Navigate This Article