Contributions to Zoology, 70 (I) 1-22 (2001) SPB Academic Publishing bv, The Hague Molecules, morphology, and the monophyly of diapsid reptiles Michael+S.Y. Lee Department of Zoology, University of Queensland Brisbane, QLD 4072, Australia, e-mail: [email protected] Keywords: Reptilia, testudines, cladistics, osteology, fossils, molecular phylogenetics Abstract Contents The morphological and molecular evidence forhigher-level reptile Introduction 1 relationships is reassessed, A combined analysis of 176 osteo- Morphological and molecular data sets 4 logical, 40 soft anatomical, and 2903 (1783 aligned) molecular Data analyses 6 characters in 28 amniote taxa yields the traditional reptile tree. Phylogenetic results 7 Synapsids the sister taxon to all other 7 (including mammals) are Separate analyses ' amniotes, all Turtles with 8 including extant reptiles. group anapsid Homogeneity tests parareptiles and fall outside a monophyletic Diapsida. Within Combined analyses 9 diapsids, squamates and Sphenodon form a monophyletic Molecules, morphology and the importance of fossils 11 Lepidosauria, and crocodiles plus birds form a monophyletic Acknowledgements 16 Archosauria. This tree is identical to the tree strongly supported References 16 by the data included. 18 osteological alone when fossils are In a Appendix combined analysis the strong osteological signal linking turtles with is anapsids sufficient to override a soft anatomical signal placing turtles next to a heterodox archosaur-mammal clade, Introduction and a weaker molecular signal linking turtles with archosaurs. However, the turtle-archosaur clade cannot be statistically rejected. When fossils of the are ignored, the signal in the osteological Despite over a century research, relationships data set disappears and, in a combined analysis of morphology of amniotes between the major groups living (mam- and molecules, the molecular (turtle-archosaur) signal prevails. mals, turtles, squamates, Sphenodon and archosaurs) These results highlight the importance of fossils, not just in remain contentious. There is now a consensus that osteological studies, but even in “combined” analyses where all form i.e. that turtles they cannot be living reptiles a clade, are scored for the majority ofcharacters (soft anatomy and molecules). Although the total number of molecular traits more closely related to other living reptiles than is much (2903) than the total number of greater morphological to mammals. This arrangement has now been sup- taits (216), when only characters informative at the relevant ported compelling evidence from both morphol- levels by are considered, the two data sets are approximatelyequal (Gauthier et ah, 1988;Laurin and Reisz, 1995; size. The ogy in partition homogeneity test yields unreliable results de and unless uninformative Lee, 1997; Braga Rieppel, 1997) and mol- (invariant and autapomorphic) characters are excluded. ecules Eernisse and Analyses ofthe mitochondrial data suggest that (Marshall, 1992; Kluge, 1993; recent evidence from nuclear for heterodox turtle- genes a Van de Peeret ah, 1993; Caspers et ah, 1996; Zar- crocodile clade be artefact of inade- (excluding birds) might an doya and Meyer, 1998). However, relationships quate of sampling a diverse outgroup (mammals) and thus, within the clade remain controversial, due problems with reptile rooting the reptile tree. mainly to disagreement over the affinities of turtles with the remaining reptiles, which are all conven- termed the basis of their tionally “diapsids” on pos- session oftwo temporal fenestrae(Williston, 1917; Romer, 1966; Carroll, 1988; Benton, 1996). There Downloaded from Brill.com10/08/2021 07:17:21AM via free access 2 M.S. Y. Lee - Phylogenetics of diapsid reptiles archosaurs are two major clades of living diapsids, (crocodiles and birds) and lepidosaurs (Sphenodon and and squamates). The monophyly of archosaurs, of lepidosaurs, is extremely strongly corroborated (e.g. Gauthier, 1984; Benton, 1984,, 1985; Evans, 1984, 1988; Gauthier et ah, 1988; Kemp, 1988; but Rieppel and de Braga, 1996; Dilkes, 1998; see Hedges and Poling, 1999 and comments below). If these two diapsid clades are provisionally ac- in which turtles cepted, there are three ways might outside be related to diapsids: they might fall a monophyletic Diapsida, or might render diapsids (as currently construed) paraphyletic by being the Fig. I. The three proposed positions for turtles with respect to sister to either or archosaurs. group lepidosaurs living diapsids. 1 is the “anapsid”hypothesis, 2 is the “lepidosaur” These three possibilities are here called the “anap- hypothesis, and 3 is the “archosaur” hypothesis. sid”, “lepidosaur” and “archosaur” hypotheses re- of diapsid mono- spectively (Fig. I). Each has some recent charac- The widespread assumption and or was challenged by Rieppel ter support from either morphology molecules. phyly recently and 1997), who The anapsid hypothesis is the traditional arrange- deBraga (1996; deBraga Rieppel, the of detailed of ment. It implies that turtles retain primitively on basis a analysis osteological the second - that turtles unfenestrated (anapsid) skulls, and that the presence traits suggested arrangement nested within as the sister to of two temporal openings is one of the diagnostic are diapsids group and con- derived characters uniting archosaurs and lepido- extant lepidosaurs. DeBraga Rieppel’s clusions were based on the first saurs as a monophyletic Diapsida (e.g. Williston, comprehensive of basal amniotes (including 1917; Gaffney, 1980; Reisz, 1981; Gauthier, 1984; phylogenetic analysis assumed to be Benton, 1985, 1996; Gaffney and Meylan, 1988; turtles) where diapsids were not The results when Carroll, 1988). Under the most recent interpreta- monophyletic. were unexpected: into all its lin- tions of this hypothesis, the nearest relatives to the Diapsida was split component these lineages did not form a clade. Rather, turtles are primitive anapsid reptiles, in particular, eages, the cluster of procolophonoids (Laurin and Reisz, 1995) or turtles nested within diapsid lineages, marine as the sister to sauropterygians, a pareiasaurs (Gregory, 1946; Ivachnenko, 1987; group of and Lee, 1997). The hypothesis of diapsid monophyly radiation consisting placodonts, plesiosaurs, relatives. were the nearest has been supported largely by morphological, and their Lepidosaurs extant the clade. This in particular skeletal, traits. However, even the relatives of sauropterygian-turtle also that turtles have skeletal traits are relatively few, and some are implies secondarily correlated with temporal fenestration (e.g. Reisz, closed their temporal fenestra. However, a reanalysis 1981; Gauthier, 1984; Benton, 1985; Evans, 1988; of deBraga and RieppeTs data showed that, even ifall their character Laurin and Reisz, 1995). Furthermore, soft ana- codings were accepted as cor- tomical for their over traits support a radically different arrange- rect, support preferred phylogeny not ment of amniotes (Gauthier et ah, 1988; Gardiner, the traditional phylogeny was significant the Further where 1982, 1993; see below). Thus, morphological (Wilkinson et ah, 1997). reanalysis, evidence for diapsid monophyly is not strong. Nev- some miscoded characters in the data set were with turtles ertheless, (his arrangement was widely accepted, rescored, yielded the traditional tree, and outside and analyses of basal amniote relationships con- grouping with anapsid reptiles falling However, as tinued to assume diapsids were monophyletic, and a monophyletic Diapsida. relatively thus were to cause this represented them using only one or two pre- few characters changed topo- tree not sumably basal forms (e.g. Laurin and Reisz, 1995; logical shift, the revised (traditional) was Lee, 1995). strongly supported over the heterodox topology Downloaded from Brill.com10/08/2021 07:17:21AM via free access Contributions to Zoology, 70 (1) - 2001 3 (Lee, 1997). Thus, although deBraga and Rieppel’s sizeable portions of the mitochondrialand nuclear analysis did not demonstrate that turtles were molecular data support the archosaur hypothesis, modified diapsids, it certainly highlighted that the the results are not yet conclusive. traditional anapsid hypothesis is far weaker than Given the disagreement over reptile relationships generally assumed, and that the lepidosaur hypo- and the observation that both the morphological thesis deserves serious consideration.A more recent and molecular evidence is ambiguous, it was de- combined study (Rieppel and Reisz, 1999) has again argued cided to integrate these datainto a phylo- for a turtle-sauropterygian clade, and discussed a genetic analysis. This study thus revisits an impor- scenario for an aquatic origin of turtles. tant earlier study of amniote relationships (Eernisse It was suggested that molecular sequences would and Kluge, 1993) that included an earlier version provide independent data to resolve the issue (Wil- of the osteological data (Gauthier et ah, 1988) and kinson et the molecular data the short and ah, 1997). However, only relatively poorly sampled gene have then surprisingly supported neither of the two sequences that were available. That analysis included arrangements proposed on the basis ofmorphology. also only two “anapsid” groups (turtles molecular results seemed and thus could the Instead, early most consis- captorhinids), and not test mono- tent - with the third possibility that turtles are most phyly of diapsids with respect to other anapsids, closely related