Shell geometry and habitat determination in extinct and extant turtles (Reptilia: Testudinata)

Abstract

A variety of means, including forelimb proportions and shell bone histology have been used to infer the paleoecology of extinct turtles. However, the height-to-width ratio of the shell (as a one-parameter shell model) has been dismissed because of its unreliability, and more complex aspects of shell geometry have generally been overlooked. Here we use a more reliable, three-parameter geometric model of the shell outline in anterior view as a means to assess turtle paleoecology. The accuracy of predictions of extant turtle ecology based on our three-parameter shell model is comparable to that derived from forelimb proportions when distinguishing between three ecological classes (terrestrial, semiaquatic, and aquatic). Higher accuracy is obtained when distinguishing between two classes (terrestrial and non-terrestrial), because the contours of aquatic and semiaquatic turtles are often very similar. Our model classifies Proterochersis robusta, a stem turtle from the Late Triassic of Germany, as non-terrestrial, and likely semiaquatic. Our method, combined with inferences based on limb proportions, indicates a diverse range of ecotypes represented by Late Triassic stem turtles. This implies that the ecological diversification of stem-group turtles may have been rapid, or that a substantial period of currently cryptic diversification preceded the first fossil appearance of the turtle stem lineage during the Late Triassic.