Turtle genes upset reptilian family tree.

Paleontologists have long viewed turtles as evolutionary slowpokes, the sole survivors of an ancient group that later gave rise to other reptiles, birds, and mammals. A new genetic analysis, however, dramatically redraws the evolutionary tree of vertebrates and challenges conventional wisdom on turtle origins.

"The slow-moving turtles that everyone thinks are slow in terms of evolution turn out to be fast evolvers," says Axel Meyer of the University of Konstanz in Germany, who collaborated in the study with Rafael Zardoya of the National Museum of Natural Sciences in Madrid.

Turtles have always stood apart from lizards, snakes, crocodiles, and other reptiles because of their skulls. Most reptiles, together with birds, have two holes on each side of their skull, to the rear of their eyes. Paleontologists label these animals diapsids. Turtles lack any such holes and so are the only living anapsids, a group that includes fossils of the most primitive vertebrates capable of living entirely on land. Mammals are termed synapsids because they evolved from animals with one skull opening behind each eye.

According to the standard evolutionary story, turtles retain some characteristics of the ancient anapsids. As such, biologists have regarded them as an example of the stock from which reptiles, birds, and mammals later evolved.

Zardoya and Meyer explored this hypothesis by comparing the sequences of two mitochondrial genes from turtles to those of iguanas, tuataras, alligators, chickens, and mammals. Turtles fell squarely within the modern diapsids rather than in their expected position on a branch outside the group, the scientists report in the Nov. 24 Proceedings of the National Academy of Sciences.

Their results lend support to a 1996 analysis of reptile fossils, which also suggested that turtles evolved from diapsids, says Olivier C. Rieppel of the Field Museum of Natural History in Chicago. The two studies do not agree in their details, though. Rieppel and a colleague found that turtles were closest to snakes and lizards, whereas Zardoya and Meyer place turtles nearest to alligators and birds.

Both cases suggest that turtles evolved from diapsids and then lost the characteristic skull holes. If so, turtles would not make a good model for early land-dwelling vertebrates.

Critics counter, however, that both studies have major flaws. The new gene analysis, for example, included only a few types of animals. "If you sample only a limited number, you tend to get very, very wrong [evolutionary] trees," says Michael S.Y. Lee of Monash University in Melbourne, Australia.