Content uploaded by Brice P Noonan
Author content
All content in this area was uploaded by Brice P Noonan on Apr 17, 2018
Content may be subject to copyright.
Journal of Biogeography,
27
, 1245 –1249
© 2000 Blackwell Science Ltd
Blackwell Science, Ltd
Does the phylogeny of pelomedusoid turtles reflect
vicariance due to continental drift?
Brice P. Noonan
The University of Texas at Arlington, Department of Biology, Arlington, TX
76019-0498, USA
Abstract
Aim
The turtle hyperfamily Pelomedusoides has been long regarded as one of the best
reptilian examples of the shared biological history of Africa and South America, and of
vicariance due to continental drift. The currently accepted phylogeny of the Pelomedu-
soides is based on morphological and serological evidence. This study uses molecular data
to examine the relationships within the Pelomedusoides and tests the hypothesis that the
speciation within this group was caused by the separation of the Gondwanan continents.
Methods
Phylogenetic analysis of 921 bp of 12S and 16S mitochondrial rRNA obtained
from GenBank.
Results
Our analysis resulted in a well-supported phylogenetic hypothesis for the extant
Pelomedusoides that does not agree with previous hypotheses based on morphology and
serology. The new hypothesis of pelomedusoid relationships is (
Pelusios
+
Pelomedusa
)
(
Peltocephalus
[
Erymnochelys
+
Podocnemis
]). The molecular evidence supports the
monophyly of the Podocnemidae and Pelomedusidae of de Broin (1988). However, the
relationships within the Podocnemidae differ in that the Madagascan
Erymnochelys
is the
sister taxon of the neotropical
Podocnemis
.
Main conclusion
This result is in conflict with the hypothesis of pelomedusoid speciation
being a result of the continental drift of South America and Africa/Madagascar, with the
separation of those three continents resulting in three separate evolutionary groups. Based
on an extensive fossil record, the Pelomedusoides are known to have been a wide-ranging
group. This latest phylogenetic hypothesis, combined with that of
Meylan (1996)
for the
Pelomedusoides, suggests that current pelomedusoid distributions and relationships could
be the result of large-scale extinctions with extant taxa being relicts of this widespread group.
Keywords
Biogeography, continental drift, Pelomedusidae, Pelomedusoides, Podocnemidae, vicariance.
INTRODUCTION
Considerable attention has been given to groups of organisms
which exhibit a trans-Atlantic distribution (Laurent, 1979;
Goldblatt, 1993). Pelomedusoids (including all non-chelid side-
necked turtles) have been cited widely as a primary example
among the reptilia of biological relationships resulting from the
relatively recent geological connection between Africa and
South America, and of vicariance caused by continental drift
(Laurent, 1979; Kluge, 1991; Maisey, 1993). Additionally, the
Pelomedusoides has been considered a prime example of
phylogenetic congruence with current hypotheses of geolo-
gical area relationships (Bauer, 1993). The single exception in
the pelomedusoid situation is the placement of the mono-
typic Madagascan genus
Erymnochelys
madagascariensis
(Grandidier). Until 1978, this species was considered a mem-
ber of the neotropical genus
Podocnemis
based on a number
of morphological similarities. Frair, Mittermeier & Rhodin
(1978) placed this species in its own genus based on ser-
ological evidence. Gaffney (1988) and Gaffney & Meylan
(1988) provided the first major phylogenetic analysis of
the Pelomedusidae (
sensu lato
) and placed
Erymnochelys
as the sister taxon to the neotropical clade,
Podocnemis
+
Peltocephalus
. Gaffney & Meylan (1988) also suggested recogni-
tion of two subfamilies within the Pelomedusidae (
sensu lato
),
the Pelomedusinae (including
Pelusios
,
Pelomedusa
and the
extinct
Bothremys
and
Taphrosphys
) and the Podocneminae
(including
Erymnochelys
,
Podocnemis
,
Peltocephalus
and
Correspondence: Brice P. Noonan, The University of Texas at Arlington,
Department of Biology, Arlington TX 76019-0498, USA. E-mail:
bpn@exchange.uta.edu
JBI483.fm Page 1245 Tuesday, December 19, 2000 4:55 PM
1246 Brice P. Noonan
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 1245 –1249
the extinct
Shweboemys
and
Stereogenys
). In the same year,
de Broin (1988) presented an alternative classification of the
Pelomedusidae (
sensu
Gaffney & Meylan, 1988), which was
further modified by Meylan (1996). In this revised classifica-
tion scheme, the Pelomedusidae (
sensu lato
) was elevated to
the hyperfamily Pelomedusoides, which included four families.
The Araripemydidae contains two extinct genera,
Araripemys
and one unnamed taxon (FR 4922; where FR refers to
Senckenberg Museum, Frankfurt, Germany), both of which are
known only from the lower Cretaceous of South America. The
Pelomedusidae (
sensu stricto
) is now restricted to the extant
African genera
Pelusios
and
Pelomedusa
. The Bothremydidae
is composed entirely of extinct genera (
Bothremys
,
Rosasia
and
Taphrosphys
). Bothremydid turtles were widespread
throughout the regions comprising most of Western Pangea
during the Cretaceous based on fossil evidence (de Broin,
1980, 1988). The fourth family, the Podocnemidae, is the
only one that includes both extant and extinct genera. The
Podocnemidae includes
Neochelys
; ‘
Roxochelys
’, known
from the upper Cretaceous of South America; the Eurasian
Shweboemys
and
Stereogenys
, known from Eocene to the
Pleistocene; the extant Madagascan
Erymnochelys
; and the
extant South American
Podocnemis
and
Peltocephalus
.
Recent studies of testudine phylogeny (Seddon
et al.
, 1997;
Shaffer, Meylan & McKnight, 1997) have utilized several
molecular data sets to recover well-supported hypotheses
of relationships of various groups of turtles. Consequently,
molecular data are available from representatives of all extant
pelomedusoid genera. Although the relationships within the
Pelomedusoides have not been specifically targeted using
molecular data, one study (Seddon
et al.
, 1997) included all
five extant genera as outgroups for an examination of chelid
relationships using mitochondria 12S rRNA gene sequences.
While Seddon and co-workers’ analysis tentatively resolved
the relationships within the extant Pelomedusoides, no mention
was made of the change in the placement of
Erymnochelys
nor its biogeographical implications.
I used 12S and 16S rRNA mitochondrial DNA sequences
obtained from GenBank to resolve the relationships within the
extant Pelomedusoides and elucidate the biogeographical
history of the hyperfamily.
MATERIALS AND METHODS
Partial sequences of the 12S and 16S mitochondrial genes total-
ling 921 bp were obtained from GenBank (see Appendix 1)
for the following pelomedusoid species:
Erymnochelys
mada-
gascariensis
(Grandidier);
Podocnemis
expansa
(Schweigger);
Peltocephalus
dumeriliana
Schweigger;
Pelusios
sinuatus
(Smith);
Pelusios
williamsi
Laurent;
Pelomedusa
subrufa
(Bonnaterre);
and one chelid outgroup taxon:
Hydromedusa
tectifera
Cope.
Sequences were aligned using the Clustal W algorithm imple-
mented in DNASTAR for Macintosh, with gap and gap
length penalties set at 8 and 10, respectively. Alignment was
generally unambiguous, with very few gaps introduced. Phylo-
genetic analyses were performed using the heuristic search
option with stepwise addition and 100 random-taxon-
addition replicates in all searches. Data were analysed using
both the maximum parsimony (MP) and maximum likelihood
(ML) optimality criteria in PAUP* Version 4.0b2a (Swofford,
1998). Maximum likelihood analyses were performed with
frequencies of transitions and transversions empirically
estimated from the data set, corresponding to the HKY model
of substitution (Hasegawa, Kishino & Yano, 1985) (ti/tv
ratio = 1.3). Sequence data were analysed with gaps treated as
missing and nucleotide substitutions both unweighted and
weighted, according to frequencies of change determined
using MacClade (vers. 3.1, W. P. Maddison & D. R. Maddison
1992, unpublished). A single shortest tree was obtained in
all searches, with weighting having no effect on topology and
only a minor effect on support. Node support was determined
via 1000 heuristic nonparametric bootstrap pseudoreplicates
(Felsenstein, 1985; Hillis & Bull, 1993). Bootstrap values
above 50% for parsimony analysis followed by values from
maximum likelihood analysis in parentheses were reported on
the shortest tree (Fig. 1).
The morphological data of Meylan (1996) were analysed
using the branch and bound search under the criterion of max-
imum parsimony in PAUP*, with node support determined via
1000 nonparametric bootstrap pseudoreplicates. Bootstrap
values above 50% were reported on the shortest tree (Fig. 2).
RESULTS
Across all taxa, 319 of 921 sites in the 12S and 16S data set
were variable, 134 of which were parsimony informative.
All analyses produced the same, single optimal tree illus-
trated in Fig. 1 (– ln likelihood = 3363.8). Below, reports of
bootstrap values are values obtained from maximum parsi-
mony analysis followed by those obtained from maximum
likelihood analysis in parentheses. Monophyly of both the
Pelomedusidae and Podocnemidae (
sensu
de Broin, 1988)
are strongly supported, with bootstrap values of 100% (100%)
and 84% (72%), respectively. The relationships within the
Podocnemidae are also well supported, with the Madagascan
Erymnochelys
as the sister taxon to
Podocnemis
96% (92%)
Figure 1 Hypothesis of relationships of living pelomedusoid
turtles based on 12S and 16S mitochondrial rRNA gene sequences
(recovered in all analyses). Numbers below branches indicate
bootstrap support derived from maximum parsimony (and maximum
likelihood in parentheses) analysis with 1000 nonparametric
pseudoreplicates and 100 random taxon addition replicates.
JBI483.fm Page 1246 Tuesday, December 19, 2000 4:55 PM
Pelomedusoid turtles 1247
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 1245 –1249
and
Peltocephalus
the sister taxon to
Erymnochelys
+
Podoc-
nemis
84% (72%). Bootstrap support varied slightly between
MP/ML and weighted/unweighted analyses, but all analyses
of the molecular data produced an identical topology with
high support (> 84% [72%]) for all intergeneric relationships.
Reanalysis of the morphological data of Meylan (1996)
indicates strong support for monophyly of three of the
families within the Pelomedusoides, with low support for only
the Pelomedusoidea (51%), and the relationships within the
Podocnemoidae (Fig. 2). Support was moderate (65%) for the
suprafamilial groups Podocnemoidae and Pelomedusoides.
The Podocnemoidae was very strongly supported by a bootstrap
value of 99%, and the genera
Stereogenys
and
Shweboemys
were
strongly (95%) supported as sister taxa. The support for the
monophyly of and relationships among the extant podocnemids
is poor. The morphological data weakly support (61%) the sis-
ter relationships of the genera
Erymnochelys
and
Peltocephalus
,
which was not reported by Meylan (1996), and is in opposi-
tion to the findings based on molecular data presented here.
The Bothremydidae was strongly supported by a bootstrap
value of 91%, within which the genera
Bothremys
and
Rosasia
were weakly supported as sister taxa. The Araripemydidae
was moderately supported by a bootstrap value of 67%.
DISCUSSION
Pelomedusoid phylogeny
Maximum parsimony and maximum likelihood analyses of
the 12S and 16S data resulted in a strongly supported hypo-
thesis of extant pelomedusoid relationships (Fig. 1). The
monophyly of de Broin’s (1988) Pelomedusidae, consisting of
Pelomedusa
and
Pelusios
, was supported by a bootstrap value
of 100% in both MP and ML analyses. The monophyly of the
genus
Pelomedusa
was well supported by a bootstrap value of
77% (84%) (the lower values may be due in part to the lack
of 16S data for
P
.
williamsi
). de Broin’s (1988) Podocnemidae
was also strongly supported by a bootstrap value of 84%
(72%). The present study is the first in which relationships
within the Podocnemidae were strongly supported in contrast
to the findings of Gaffney (1988) and Gaffney & Meylan
(1988). The sequence data suggest that previous findings
supporting the monophyly of neotropical pelomedusoids are
incorrect, and that
Erymnochelys
is actually the sister taxon
to
Podocnemis
(bootstrap value 96% [92%]), with
Peltoceph-
alus
the sister taxon to this group (bootstrap value 84%
[72%]). Although the generic classifications of
Podocnemis
and
Erymnochelys
are arbitrary, that they are sister taxa has
a major effect on the choice of the Pelomedusoides as an
example of South American–African plate tectonic vicariance.
When addressing pelomedusoid biogeography, it is import-
ant to consider what is known of the distribution of extinct
taxa via fossil evidence. Conclusions regarding the biogeo-
graphical history of this family based solely on extant taxa
would be extremely difficult to make, and most likely in-
correct. The most recent treatment of the phylogeny of the
Pelomedusoides that includes fossil taxa is that of Meylan
(1996). Meylan (1996) followed de Broin’s (1988) classifica-
tion scheme of the Pelomedusoides, which is equivalent to
the Pelomedusidae of Gaffney (1988) and Gaffney & Meylan
(1988). This classification scheme divides the Pelomedusidae
(
sensu lato
) into four families, two of which are entirely
extinct (Araripemydidae and Bothremydidae). A tree based
on reanalysis of Meylan’s morphological characters is pre-
sented in Fig. 2. This tree differs in topology from the original
only in the arrangement of the extant podocnemid turtles,
as the morphological data weakly support the sister relation-
ship of the genera
Erymnochelys
and
Peltocephalus
, whereas
Meylan (1996) showed
Erymnochelys
,
Peltocephalus
and
Podocnemis
as a trichotomy. Note that the group including
the Bothremydidae and the Podocnemidae is referred to here
as the Podocnemoidae, as illustrated in Fig. 11 of Meylan
(1996), rather than the Pelomedusoidae as is stated on p. 32
of the same paper. Reanalysis of Meylan’s data supports his
findings, and provides support for the four families within
the Pelomedusoides. The significance of this reanalysis, and
the newly presented hypothesis of extant pelomedusoid
turtles based on molecular data, is that a well-supported
phylogenetic hypothesis now exists, and biogeographical
conclusions can be drawn.
Many authors have used the Pelomedusoides to illustrate
cladogenic events caused by the shifting of the continents in
the Cretaceous (Laurent, 1979; Bauer, 1993). Indeed, an
area cladogram of extant pelomedusoid turtles (Fig. 3) based
on Gaffney (1988) and Gaffney & Meylan’s (1988) findings
does appear to fit with current geological thinking (i.e.
Africa, Madagascar and South America share a close geo-
logical history, and the order of separation of these land
Figure 2 Cadogram of the extant and extinct pelomedusoid turtles
based on reanalysis of the 35 morphological characters of Meylan
(1996). Dashed lines indicate extinct lineages. Numbers above
branches indicate bootstrap support from maximum parsimony,
branch and bound reanalysis with 1000 pseudoreplicates. Families
(sensu de Broin, 1988) are indicated on the tree, with suprafamilial
groups to the right.
JBI483.fm Page 1247 Tuesday, December 19, 2000 4:55 PM
1248 Brice P. Noonan
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 1245 –1249
masses often corroborates the branching order of phylo-
genies of shared organisms). On the other hand a similar tree
(differing only in the placement of
Erymnochelys
) based on
the hypothesis supported by molecular data (Fig. 4) is more
difficult to explain.
Maisey’s (1993) statement ‘the present day distribution
of pelomedusids … is clearly a relic of an earlier, more
cosmopolitan, post-Gondwanan distribution’ may provide an
explanation for this seemingly surprising result. Examination
of the distributional patterns of the Pelomedusoides illus-
trated in Fig. 5 (in which fossil taxa are included) indicates
no identifiable correlation of phylogeny and biogeography,
let alone one attributable to continental drift. Thus, the events
affecting the phylogeny of the Pelomedusoides, whether they
be vicariant or not, seem to have occurred before the breakup
of the Gondwanan land mass. This likely resulted in the
cosmopolitan post-Gondwanan distribution of the hyper-
family Pelomedusoides, as commented on by Maisey (1993).
A prime example of this is the distribution of the members
of the Bothremydidae, specifically the genera
Bothremys
and
Taphrosphys
, which inhabited most of Western Pangea.
Alternatively, when examining sequence divergence between
taxa it is interesting that the 16–17% divergence observed
between South American and African taxa corresponds
almost exactly (using the 0.25%/Myr calibration of Avise
et al.
, 1992) to the presumed date of separation of these two
continents (roughly 84 Ma; Pitman
et al
., 1997). Interestingly,
when using the presumed time of separation of the continents
and the sequence divergence between the related turtle taxa on
those continents (African–American Pelomedusoides of this
study and the Australian–American Chelidae of Seddon
et al.
,
1997), rates of sequence divergence were nearly identical at
roughly 0.21%/Myr (Table 1). However, using this same
calibration Erymnochelys and Podocnemis diverged a mere
59 Ma. While this is long after the separation of South
America and Africa, it could be a result of oceanic dispersal
from the land mass which included present day South America,
Antarctica and Australia to Madagascar. This land mass was
contiguous as recently as 43 Ma (McGowran, 1973), though
Madagascar had separated from this land mass
c
. 80 Ma.
Thus, while the sequence variation seems to point to diver-
gence caused by or at least since the break-up of Gondwana,
distributional data and the phylogenetic hypothesis based on
morphological data from both extant and extinct species
of the Pelomedusoides indicate a group that was widely dis-
tributed throughout Gondwana prior to the break up of that
land mass. Though our current knowledge of the evolutionary
Figure 3 Area cladogram of extant pelomedusoid turtles based on
relationships derived from morphological and serological data.
Figure 4 Area cladogram of extant pelomedusoid turtles based on
relationships derived from 12S and 16S mitochondrial rRNA gene
sequences.
Figure 5 Cladogram of the Pelomedusoides based on reanalysis of
the morphological data of Meylan (1996), illustrating geographical
distribution of terminal taxa (branch termini are labelled with the geo-
graphical region inhabited by that taxon; refer to Fig. 2 for taxon names).
JBI483.fm Page 1248 Tuesday, December 19, 2000 4:55 PM
Pelomedusoid turtles 1249
© Blackwell Science Ltd 2000,
Journal of Biogeography
,
27
, 1245 –1249
history of the Pelomedusoides does not unequivocally support
cladogenesis associated with continental drift, the molecular
data do lend support to a new view of drift vicariance in this
group.
ACKNOWLEDGMENTS
Special thanks to P. T. Chippindale, J. A. Campbell and
D. P. Lawson for critically reviewing this manuscript.
REFERENCES
Avise, J. C., Bowen, B. W., Lamb, T., Meylan, A. B. & Bermingham, E.
(1992) Mitochondrial DNA evolution at a turtle’s pace: evidence
for a low genetic variability and reduced microevolutionary
rate in the Testudines.
Molecular and Biological Evolution
,
9
,
457–473.
Bauer, A. M. (1993) African-South American Relationships: A
perspective from the Reptilia.
Biological relationships between
Africa and South America
(ed. by P. Goldblat). pp. 244–288.
Yale University Press, New Haven, CT.
de Broin, F. (1980) Les tortues de Gadoufaoua (Aptien du Niger):
apercu sur la paleobiogeographie des Pelomedusidae (Pleurodira).
Mémoires de la Société Géologique de France, N S
,
139
, 39–46.
de Broin, F. (1988) Les tortues et le Gondwana. Examen des
rapports entre la fractionnement du Gondwana et la dispersion
geographique des tortues pleurodires a partir du Gretace.
Studia Palaeocheloniologica
,
2
, 103–142.
Felsenstein, J. (1985) Confidence limits on phylogenies: an
approach using the bootstrap.
Evolution
,
39
, 783–791.
Frair, W., Mittermeier, R. A. & Rhodin, A. G. J. (1978) Blood
biochemistry and relations among
Podocnemis
turtles
(Pleurodira, Pelomedusidae).
Comp Biochemistry and Physi-
ology
,
61B
, 139–143.
Gaffney, E. S. (1988) A cladogram of the pleurodiran turtles.
Acta
Zoologica Cracoviensia
, Krakow,
30
, 487–492.
Gaffney, E. S. & Meylan, P. A. (1988) A phylogeny of turtles.
The
phylogeny and classification of the tetrapods,
Vol. 1 (ed. by
M. J. Benton), pp. 157–219. Claredon Press, Oxford.
Goldblatt, P., ed. (1993)
Biological Relationships Between Africa
and South America
. Yale University Press, New Haven, CT.
Hasegawa, M., Kishino, H. & Yano, T. (1985) Dating of the
human-ape splitting by a molecular clock of mitochondrial
DNA.
Journal of Molecular Ecology
,
21
, 160–174.
Hillis, D. M. & Bull, J. J. (1993) An empirical test of bootstrapping
as a method for assessing confidence in phylogenetic analysis.
Systematic Biology, 42, 182–192.
Kluge, A. G. (1991) Boine snake phylogeny and research cycles,
58 pp. Miscellaneous Publication from the Museum of Zoology
No. 178. University of Michigan, Ann Arbor, MI.
Laurent, R. F. (1979) Herpetofaunal relationships between Africa
and South America. The South American herpetofauna: its origin,
evolution and dispersal (ed. by W. E. Duellman), pp. 55–71.
Museum of Natural History, University of Kansas, Mono-
graph 7.
Maisey, J. G. (1993) Tectonics, the Santana Lagerstatten, and the
implications for late Gondwanan biogeography. Biological rela-
tionships between Africa and South America (ed. by P. Goldblat),
pp. 435–454. Yale University Press, New Haven, CT.
Meylan, P. A. (1996) Morphology and relationships of the early
cretaceous side-necked turtle, Araripemys barretoi (Testudines:
Pelomedusoides: Araripemydidae) from the Santana Forma-
tion of Brazil. Journal of Vertebrate Palentology, 16(1), 20– 33.
McGowran, B. (1973) Rifting and drift of Australia and the
migration of mammals. Science, 180, 759–761.
Pitman, W. C., Cande, S., LaBrecque, J. & Pindel, J. (1993) Frag-
mentation of Gondwana: the separation of Africa from South
America. Biological relationships between Africa and South
America (ed. by P. Goldblat), pp. 15–34. Yale University Press,
New Haven, CT.
Seddon, J. M., Georges, A., Baverstock, P. R. & McCord, W.
(1997) Phylogenetic relationships of chelid turtles (Pleurodira:
Chelidae) based on mitochondrial 12S rRNA sequence vari-
ation. Molecular Phylogenetic Evolution, 7, 55–61.
Shaffer, H. B., Meylan, P. & McKnight, M. L. (1997) Tests of
turtle phylogeny: molecular, morphological, and palentological
approaches. Systematic Biology, 46, 235 –268.
Swofford, D. L. (1998) PAUP*: phylogenetic analysis using
parsimony (*and other methods). Sinauer, Sunderland, MA.
Appendix 1 Genbank partial sequences.
Table 1 Rate o
f
sequence evo
l
ution
d
etermine
d
f
rom o
b
serve
d
sequence
d
ivergence
b
etween A
f
rican/American taxa (t
h
is stu
d
y) an
d
Austra
l
ian/
American taxa (Seddon et al., 1997), and published times of continental separation (from Pitman et al., 1993 and McGowran, 1973).
Areas compared Taxa compared
Mean sequence
divergence
Estimated time
of separation
Rate of sequence
evolution
South America–Africa Pelomedusoides 0.175 84 Ma 0.208%/Myr
South America–Australia Chelidae 0.097 45 Ma 0.215%/Myr
BIOSKETCH
Brice Noonan received his Bachelors of Science at the
University of Florida, his Masters of Science at the University
of Texas at Arlington, and is currently a PhD candidate at
the University of Texas at Arlington. His research interests
focus on the evolutionary history of South American fauna.
Erymnochelys madagascariensis EMU40641, AF113640
Podocnemis expansa PEU40649, AF113642
Peltocephalus dumeriliana PDU40643, AF113643
Pelusios sinuatus PSU40644, AF113641
Pelusios williamsi 16S only, PWU81324
Pelomedusa subrufa PDU40642, AF113639
Hydromedusa tectifera U62017, AF113638
JBI483.fm Page 1249 Tuesday, December 19, 2000 4:55 PM
