The origin of turtles and their unusual body plan has fascinated scientists for the last two centuries. Over the course of the last decades, a broad sample of molecular analyses have favored a sister group relationship of turtles with archosaurs, but recent studies reveal that this signal may be the result of systematic biases affecting molecular approaches, in particular sampling, non-randomly distributed rate heterogeneity among taxa, and the use of concatenated data sets. Morphological studies, by contrast, disfavor archosaurian relationships for turtles, but the proposed alternative topologies are poorly supported as well. The recently revived paleontological hypothesis that the Middle Permian Eunotosaurus africanus is an intermediate stem turtle is now robustly supported by numerous characters that were previously thought to be unique to turtles and that are now shown to have originated over the course of tens of millions of years unrelated to the origin of the turtle shell. Although E. africanus does not solve the placement of turtles within Amniota, it successfully extends the stem lineage of turtles to the Permian and helps resolve some questions associated with the origin of turtles, in particular the non-composite origin of the shell, the slow origin of the shell, and the terrestrial setting for the origin of turtles.
© 2015 Wiley Periodicals, Inc.