A combined phylogenetic analysis of molecular and morphological data
for 23 genera of living turtles and seven key fossil taxa (Shaffer,
Meylan and McKnight, 1997) corroborates much of the hypothesis of higher
relationships suggested by studies of morphological data alone, but
also leads to some serious questions about relationships within the
side-neck family, Chelidae and about the higher relationships among
the living families of hidden-necked (cryptodiran) turtles. Shaffer
and his colleagues studied nearly 1 kilobase of cytochrome b sequence,
325 base pairs of 12s ribosomal DNA and 115 morphological characters.
They found that these data sets contained similar phylogenetic information,
but that each data set contributed unique information on different nodes
of chelonian history. The authors combined their data in order
to estimate a final phylogeny of the major lineages of living turtles.
========================================== Proganochelys 
|
| ======================================= Pleurodira
| |
| | ====== Trionychidae
| | ==================|
| | | ====== Carettochelyidae
| | |
| | |======================= Chelydridae
| | |
<<===| | |======================= Chelonioidea
===| |
===Cryptodira==| ====== Dermatemydidae
|=================|
| ====== Kinosternidae
|
| ====== Emydidae
| |
===Testudinoidea==| === "Bataguridae"
===|
=== Testudinidae
Tree from Shaffer et al. (1997)
The result of their combined anlysis is the tree shown above. They were able to make defendable arguments about the relationships among most of the higher groups of turtles. Only a few of the deepest branches within the Cryptodira (hidden-necked turtles) could not be determined. They added six relatively complete fossils to try to resolve this basal polytomy but this did not improve the resolution of these deep splits. The very small number of characters that are available to determine relationships in this part of the tree and independent dates from the fossil record suggest that these unresolved nodes may represent a rapid radiation of the major cryptodiran lineages 90-120 million years ago.
In spite of this lack of resolution of turtle relationships, parsimony analysis of the total data set results in a tree that includes most of the family level taxa that have traditionally been recognized, and reflects considerable agreement between the morphological and molecular trees. Sixteen of 21 higher groups proposed for living turtles by Gaffney and Meylan (1988--the basis for many of the TOL turtle pages) are included in the tree based on combined evidence.
Within the Cryptodira, the Testudinoidea (sensu Gaffney, 1984; Gaffney, and Meylan, 1988) is only weakly supported which is surprising given the virtually universal acceptance of this group (Williams, 1950; Bickham and Carr, 1983; for more see Gaffney, 1984). However, the independent support for the Testudinoidea by both the morphological and molecular data sets strengthens the interpretation of the monophyly of this group. The Testudinoidea contains two well supported groups: a restricted Emydidae (sensu Gaffney, 1975), and the currently unnamed group that includes the Bataguridae and Testudinidae for which Shaffer et al. (1997) propose the name Testudinoidae. Within the Testudinoidae strong support for the monophyly of the Bataguridae rests entirely on the molecular data set. Hirayama (1985) suggested that the Bataguridae may be paraphyletic with respect to the monophyletic Testudinidae, whereas Lamb and Lydeard (1994) found weak support for a paraphyletic Testudinidae with respect to batagurids. The resolution of this problem clearly awaits additional data from more complete sampling of both groups rather than the sparse sampling available at present. This problem is now being studied in Shaffer's lab and others.
There is a series of other cryptodiran groups for which strong and unambiguous support is now in hand. These include the monophyly of the Kinosternidae (mud turtles). The naturalness of this group had been questioned on the basis of karyological data by Bickham and Carr (1983). The close relationship of mud turtles (Kinosternidae) and the monotypic family Dermatemydidae, as the Kinosternoidae, is supported as is the close relationships between softshells (Trionychidae) and the Fly River or Pignosed Turtle (Carettochelyidae) as the Trionychoidae. The two marine turtle families (Dermochelyidae+Cheloniidae) are clearly a natural group, the Chelonioidea. Also, the Asian big-headed turtle (Platysternon) is clearly shown to be a member of the snapping turtle family Chelydridae (Gaffney, 1975).
On major conflict between the molecular evidence and previous morphological studies is the purported monophyly of the large superfamily Trionychoidea (Trionychidae, Carettochelyidae, Kinosternidae and Dermatemydidae). Bootstrap analysis of the morphological data set demonstrates moderate support based on morphological characters, but it is not supported by the molecular data. Although no other alternative set of relationships is strongly supported, the only likely alternative indicated by the cytB plus 12S data set is one that was advocated by Bickham and Carr (1983) on the basis of karyology. This hypothesis is that softshells and carettochelyids (the Trionychoidae) are the sister group to all other living cryptodires. The higher relationships among cryptodires is an area that will require more study.
A second conflict between the combined molecular and morphological
data set of Shaffer et al. and previous morphological studies by Gaffney
and coworkers, is the relationships among the genera of the side-neck
turtle family Chelidae. The fundamental difference is that the morphological
view (based on Gaffney, 1977) suggests that South American and Australian
genera are closely interrelated, that is, neither the South American
nor the Australian chelids are monophyletic. The molecular data of Shaffer
et al. (1997) disagrees. It weakly supports the monophyly of the Australian
genera and the monophyly of the South American genera. However, this
study utilized a limited number of genera that did not include Hydromedusa,
a genus which another study (Seddon et al,
