Molecular phylogenetics and evolution of turtles

Turtles are one of Earth's most instantly recognizable life forms, distinguished for over 200 million years in the fossil record. Even so, key nodes in the phylogeny of turtles remain uncertain. To address this issue, we sequenced >90% of the nuclear recombination activase gene 1 (RAG-1) for 24 species representing all modern turtle families. RAG-1 exhibited negligible saturation and base composition bias, and extensive base composition homogeneity. Most of the relationships suggested by prior phylogenetic analyses were also supported by RAG-1 and, for at least two critical nodes, with a much higher level of support. RAG-1 also indicates that the enigmatic Platysternidae and Chelydridae, often considered sister taxa based on morphological evidence, are not closely related, although their precise phylogenetic placement in the turtle tree is still unresolved. Although RAG-1 is phylogenetically informative, our research revealed fundamental conflicts among analytical methods for estimating phylogenetic hypotheses. Maximum parsimony analyses of RAG-1 alone and in combination with two mitochondrial genes suggest the earliest phylogenetic splits separating into three basal branches, the pig-nosed turtles (Carettochelyidae), the softshell turtles (Trionychidae), and a clade comprising all remaining extant turtles. Maximum likelihood and Bayesian analyses group Carettochelyidae and Trionychidae (=Trionychoidae) in their more traditional location as the sister taxon to all other hidden-necked turtles, collectively forming the Cryptodira. Our research highlights the utility of molecular data in identifying issues of character homology in morphological datasets, while shedding valuable light on the biodiversity of a globally imperiled taxon.