Dr. Walter Joyce on turtle origins

A new paper on turtle origins
(Joyce 2015) supports the hypothesis that the Middle Permian Eunotosaurus (Fig. 1) is an intermediate stem turtle (Watson 1914). Joyce sought to clear up a series of misconceptions about amniote systematics and the fossil record, including the remaining ambiguity regarding the phylogenetic position of turtles.

Unfortunately, no phylogenetic analysis was presented.

Fiigure 1. The turtle mimic Eunotosaurus from the Middle Permian was actually closer to Acleistorhinus.

Fiigure 1. The turtle mimic Eunotosaurus from the Middle Permian was actually closer to Acleistorhinus. Note the small number of dorsal vertebrae and the width of the ribs. Note also the ribs are larger  and with more coverage in Eunotosaurus than in the more derived Odontochelys, which is odd considering Eunotosaurus is supposed to be more primitive.

Joyce leads his charge
by undermining Osborn (1903) and Williston (1917) who proposed dividing reptiles into clades based on their temporal fenestrae or lack thereof. Turtles, lacking temporal fenestrae were considered primitive, citing the possible placement of turtles within the Diapsida with secondary loss of the temporal fenestrae (Broom 1924, Goodrich 1930, Rieppel and deBraga 1996, Rieppel 2000a). Note that Eunotosaurus (Fig. 1) has a lateral temporal fenestra lacking a lower temporal bar. It is not a diapsid, but shares this trait with Acleistorhinus and Australothyris, both of which retain the lower temporal bar formed by the jugal + quadratojugal. These are members of a slow, wide, herbivorous clade that also includes Caseasauria (which are not synapsids).

Figure 1. Odontochelys with a newly reconstructed skull.

Figure 2. Odontochelys is a basal turtle close to soft-shelled turtles.

Joyce reviews the DNA evidence
for turtles in the first decade of research as sisters to Aves, Crocodilia, Diapsida, Lepidosauria, Mammals + Archosaurs, or Archosaurs (refs in Joyce 2015). The second decade of research supported a sisterhood to archosaurs with Lyson et al. (2012) as the major exception to this trend. Lyson et al. linked turtles with lepidosaurs. HJoyce notes an expansion of the dataset by subsequent workers retrieved archosaurian affinities.

Joyce notes, “Using one of the largest data sets assembled to date, Lu et al. (2013) recently evaluated the phylogenetic signal contained in 4,584 orthologous genes  separately and arrived at the surprising conclusion that the three primary placements of turtles (i.e., as sister to Diapsida, Lepidosauria, or Archosauria) are supported by roughly the same number of genes, and that an archosaurian signal emerges only through the concatenation of the data. There only is one tree of life and it is apparent that either the molecular or the morphological signal is wrong.”

Joyce then notes that morphology does not support an archosaur relationship to turtles, an observations supported by several studies including the large reptile tree. Citing Lee (2013) Joyce notes “The morphological signal is therefore not able to resolve the placement of turtles withinamniotes for the moment, just like the molecular data, perhaps because of a sampling bias, or perhaps because the three major reptilian lineages diverged from one another very rapidly during the Late Paleozoic, the same two biases effecting molecular data.” Joyce agrees with Lee (2013) that “the interpretation of Eunotosaurus as an intermediate stem turtle is highly robust, relatively immune to perturbances and is independent from the placement of turtles within Amniota.”

Eunotosaurus does have an impressive number of traits otherwise shared only with turtles. Unfortunately, other taxa not listed by Lee or Joyce have more similar traits and the addition of these taxa and traits result in a separation of Eunotosaurus from turtles, as shown in the large reptile tree. Joyce notes that Odontochelys bridges the apparent morphological gap between Eunotosaurus and turtles and discusses their respective anatomies point by point. This is always a problem because Joyce lists no more than a half dozen traits when phylogenetic analysis  examines well over 200 — and a broader range of taxa.

Joynce notes, “In the older literature, turtles were variously allied with Eunotosaurus africanus (Watson, 1914), plesiosaurs (Moodie, 1908), placodonts (Jaeckel, 1902), “cotylosaurs” (Cope, 1896), temnospondyls (Vallén, 1942), pareiasaurs (Gregory, 1946), diadectids (Olson, 1947), and captorhinomorphs (Carroll, 1969), but most of these hypotheses were support by isolated characters and lacked a global perspective. Computer-assisted analyses hypothesized a sister group relationship to “anapsid” procolophonids (Reisz and Laurin, 1991; Laurin and Reisz, 1995) or pareiasaurs (Lee, 1993, 1997), “parapsid” sauropterygians (Rieppel and deBragga, 1996; deBraga and Rieppel, 1997), or “parapsid/diapsid” lepidosaurs (Müller, 2004). Interestingly, all modern analysis excluded E. africanus a priori, even though this taxon had consistently been listed as sister to turtles in standard paleontological textbooks and classifications throughout much of the century (e.g., Huene, 1956; Romer, 1956; Carroll, 1988). The Eunotosaurus hypothesis is different from all previous hypotheses regarding the origin of turtles because it is based on an impressive set of characters that were previously thought to be unique to turtles that originated one by one over the course of tens of millions of years. The most convincing aspect of the Eunotosaurus hypothesis is that these characters do not originate in parallel with the turtle shell, but rather in sequence as exaptations (Gould and Vrba, 1982) millions of years prior to the formation of a full shell and for reasons not related to the formation of the shell.” See Joyce 2015 for refs.

Joyce concludes with, “Future work will therefore have to focus on better understanding the anatomy of E. africanus, in particular its cranial anatomy, and in revising the phylogenetic relationships of Late Paleozoic amniotes.” Indeed, since so many Eunotosaurus sisters in the large reptile tree are known from skulls only, this is the prudent way to go. Also look at the feet, the pectoral and pelvic girdles. Look at everything and add taxa. Convergence will separate from homology as more data flows in. And don’t forget to add Stephanospondylus, Sclerosaurus and Elginia, three taxa typically omitted from turtle studies, but nest close to turtles in the large reptile tree (now up to 504 taxa).

Joyce W 2015. The origin of turtles: a paleontological perspective. Journal of Experimental Zoology Part B Molecular and Developmental Evolution. 00B:1–13.
Lyson TR, Sperling EA, Heimberg AM, et al. 2012. MicroRNAs support a turtle + lizard clade. Biol Lett 8:104–107.
Watson DMS. 1914. Eunotosaurus africanus Seeley, and the ancestry of the Chelonia. Proc Zool Soc Lond 1914:1011–1020.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s