Historical Burden In Systematics And The Interrelationships Of ‘Parareptiles’

Summary The interrelationships within the clade comprised of turtles, pareiasaurs, and procolophonid‐like taxa are investigated via a cladistic analysis incorporating 56 characters. A single most parsimonious tree was found (80 steps, c. i. = 0·8) in which the successive outgroups to turtles are: pareiasaurs, Sclerosaurus , lanthanosuchids, procolophonoids (= Owenetta, Barasaurus and procolophonids), and nyctiphruretians (= nycteroleterids). Thus, as suggested recently by other workers (Reisz, in Fischman, 1993) turtles are the highly modified survivors of a radiation of poorly‐known reptiles commonly called ‘parareptiles’. Pareiasaurs are united with turtles on the basis of twenty unambiguous derived features which are absent in other basal amniotes (=‘primitive reptiles’) and reptiliomorph amphibians: for example, the medially located choana, enlarged foramina palatinum posterius, blunt cultriform process, fully ossified medial wall of the prootic, opisthotic‐squamosal suture, lateral flange of exoccipital, loss of ventral cranial fissure, thickened braincase floor, ‘pleurosphenoid’ ossification, reduced presacral count, acromion process, trochanter major, reduced fifth pedal digit, and presence of transverse processes on most caudals. Recent phylogenetic proposals linking turtles with captorhinids, with dicynodonts, and with procolophonoids are evaluated. None of the proposed traits supporting the first two hypotheses is compelling. The procolophonoid hypotheses is supported by only one synapomorphy (the slender stapes). All other synapomorphies proposed in favour of the above groupings either occur in many other primitive amniotes, or are not primitive for turtles, or are not primitive for the proposed chelonian sister‐group. Nyctiphruretus and Lanthanosuchids and nycteroleterids, often considered to be seymouriamorph amphibians, are demonstrated unequivocally to be amniotes. The ‘rhipaeosaurs’, currently considered to be pareiasaur relatives, are shown to be a heterogenous assemblage of seymouriamorphs, therapsids and nycteroleterids. The phylogeny proposed here indicates that many of the traits of the earliest known turtle, Proganochelys , previously interpreted as unique specialisations, also occur in pareiasaurs and other near outgroups of turtles, and must instead represent the primitive chelonian condition: for example, the wide parietals and the short quadrate flange of the pterygoid. The sequence of acquisition of chelonian traits is discussed: many features once thought to be diagnostic of turtles actually characterize larger groupings of procolophonomorphs, and must have evolved long before the chelonian shell appeared. These traits include most of the chelonian‐pareiasaur synapomorphies listed above, and many others which characterize more inclusive groupings found in this analysis. In putting Proganochelys much closer to the main line of chelonian evolution, in elucidating the sequence of acquisition of chelonian traits, and in reducing greatly the number of differences between turtles and their nearest relatives, this study helps bridge one of the major gaps in the fossil record. The failure of previous cladistic analyses to identify correctly the nearest relatives of turtles is attributed to biased character selection, caused by an over‐reliance on cranial characters deemed ‘important’ by earlier workers, and by a tendency to shoehorn ‘parareptile’ taxa into phylogenies derived from analyses restricted to ‘mainstream’ groups such as synapsids, diapsids, turtles, and ‘captorhinomorphs’. Many of the synapomorphies that resolve turtle origins are postcranial, and the three nearest outgroups to turtles are all highly bizarre groups which were dismissed as ‘too specialized’ by early workers and continued to be inadequately assessed even by workers using a cladistic framework.