Evolution and Zoogeography of Australian freshwater turtles

13 Pages·1996·3.6 MB·English

by  Ben Manning

EVOLUTION ANDZOOGEOGRAPHY OFAUSTRALIANFRESHWATERTURTLES
BEN MANNING ANDCHRISTOPHER KOFRON
P.
Manning, B. & Kofron, C.P. 1996 07 20: Evolution and zoogeography of Australian
freshwater turtles. Memoirs ofthe QueenslandMuseum 39 (2): 319-331. Brisbane. ISSN
0079-8835.
ThecontemporaryturtlefaunaofAustraliacomprisesfreshwaterandmarinespeciesbutno
terrestrial taxa. The literatureon evolution and zoogeography ofthe Australian freshwater
turtles (Chelidae) is reviewed. Because opposing conclusions were reached in several of
these studies, we critically examine each ofthem. We accept the phylogeny ofGeorges &
Adams (1992); Elseya and Emydura are not synonymous and Elseya consists of the E.
dentata and E. latisiemum species groups (generically distinct). However, Pseudemydura
umbrina shares common ancestry with other Australian short-necked turtles and is their
extant sister taxon. The E. latisternum group is the sister taxon ofthe non-Pseudemydura
Australian short-neckedturtlegenera.
Understandingpre-Cretaceousdispersaloftheancestral lineageontheSouthAmerican-An-
tarctic-Australian supercontinent is important for ascertaining relationships among
Australian freshwater turtles. Increasing aridification during the Pleistocene, resulting in
isolationwithin riverdrainages,appearstohaveaugmented allopatric speciation.
Werecognize23extantspeciesofAustralianchelidturtles in7genera: Chelodinaexpansa,
C. longicollis, C. novaeguineae, C. oblonga, C. rugosa, C. steindachneri and one un-
described species; Elseyadentata and 3 undescribed species; Elseya latisternum and three
undescribed species; Elusormacrurus; Emydura macquarrii (inclusive ofE. australis, E.
krefftiiimdE.signata),E.subglobosa,E. victoriaeand2undescribedspecies;Pseudemydura
umbrina; and Rheodytesleukops. Relationshipsamonggeneraremainenigmatic.
Australia, evolution, turtles, zoogeography,
BenManning & ChristopherP. Kofron, DepartmentofZoology, UniversityofQueensland,
Brisbane, Queensland4072,Australia. (PresentaddressCPK:DepartmentofEnvironment,
Queensland National Parks and Wildlife Service, Ear Northern Region, PO Box 2066,
Cairns, Queensland4870, Australia); received2November 1995.
The reptilian order Testudines contains about notwithdrawtheheadbecauseofnon-specialised
257 living species of turtles (Ernst & Barbour, cervical vertebrae (Gaffney, 1977).
1989) in two suborders, Cryptodira and Pleurodira contains the Pelomedusidae and
Pleurodira. Cryptodira contains 10 families and Chelidae. Pelomedusids inhabit Africa, Mada-
about 203 species, and Pleurodira two families gascar and South America, but previously also
and about 54 species. Cryptodiran turtles with- Europe, Asia and North America (Frair, 1980).
draw theheadby vertical flexureoftheneckinto They retain mesoplastral bones in the shell anda
an S, and the pelvic girdle does not attach to the squamosalboneintheskull;thus,Pelomedusidae
plastron. Pleurodirans withdrawthehead by hor- is plesiomorphic toChelidae (McDowell 1983),
wmch Iacks mesoplastral and squamosal bones,
izontal flexure ofthe neck, and the pelvic girdle
attachestothe plastron. Each suborderis charac- The contemporary fauna of Australia com-
terised by its own specialised cervical vertebrae. Prises freshwaterand marineturtles, witha com-
Although cryptodirans are more derived than P,ete absence of terrestrial species. The
^
plaltetuerro(dEirrnasnts,&thBearfboorumre,r1d9i8d9)n.otThseteemarlfiresotmftohse- ^*ucsChe,l,1iadnaef,rewsihtwhateexrcetpurttiloens oafreopnleeucrryopdtiordainrsano,f
si.l,sofrcryptod,i.ran turt.,les are fMciydt-dnl,eJiurassi•c. 5r0n dCiadreatetocThneely'slatitnesrcusl£pteacieost tihnehabCialrsetntoorcthheelryJ-n
mfo.slshilosn(yLeeagrlsero&ldeGreotrhgaens,the19e9a3rah)e.stHopwleevureord,.rtahne eAxutsatnrtalmiaemabnedrNGefwits£fuaimnielya.^ aAnsdrekceinstltyheasstohlee
fossil record is poor, and pleurodiran turtles Eocene, Carettochelydidae had a broad distribu-
p&rLoebgalbelry,o1r9i8g0i;naEtrendstea&rliBearrb(Goaufrf,ne1y9,891)9.7T9;heBtuwlol tgiCoSn,j1n99t3hae)N.oTrhtheelrinviHnegmmiasrpihneerteur(tLleesgloefrA&ustGreaolri-a
subordersareprobablyderivedfrom separatean- are cryptodirans of the Cheloniidae and Der-
cestral lineages,each ofwhich presumablycould mochelyidae.
1
320 MEMOIRS OFTHEQUEENSLAND MUSEUM
TABLE 1. Recentprimary researchpapersthatrecog- Pseudemydura and Rheodytes are short-necked
nize the two nominal turtle genera Elseya and active foragers (Legler& Georges, 1993b; Cann
Emyduraas separate, orascongeneric & Legler, 1994). These have 5 claws on the
forelimbs, and a neck less than 1/3 the length of
Separaterecognition Congenericrecognition the shell (Cogger, 1992).
Cann& Legler(1994) Gaffney(1991) SouthAmericanchelidsalsoseparateintolong-
White&Archer(1994) Gaffney,Archer&White necked sit-and-wait predators and short-necked
(1989) &
active foragers (Ernst Barbour, 1989). Like
Georges(1993) McDowell(1983) Australian long-necked Chelodina, South Am-
Georges& Adams(1992) Gaffney(1981) erican long-necked Chelus and Hydromedusa
Legler(1985) Frair(1980) have 4 claws on the forelimbs. Also, like Aust-
Legler(1981) Gaffney(1979) ralian short-necked genera, South American
Legler&Cann(1980) Gaffney(1977) short-necked Phrynops and Platemys have 5
Burbidge.(1K9i7r4s)ch&Main cHloawwsaornetthheefogreeolgirmabpsh.icTahlelyq-udeissttiaonntsAaursetproasleida:n
and South American chelids related? Are the
Australia was inhabited also by freshwater
similargroupsaresultofecologicalconvergence,
cryptodiran&softhelivingTrionychidae(Pliocene ordo they share common ancestry?
- Gaffney Bartholomai, 1979; Willis, 1993),
and possibly Emydidae (middle Miocene; Mur- Australianchelidsaremorphologicallyconser-
ray & Megirian, 1992). A primitive non-marine vative (Cogger, 1992), which has hindered the
recognition ofvalid taxa and obscured their true
cryptodiran was reported from the Early Cret-
aceousofVictoria(Gaffneyetal., 1992),andalso rneelwatisopneschiiepss.(DLiesgcloevre&rieCsaonfn,3n1e9w80g;enGeeroaragneds1&
primitive terrestrial turtles that a&re neither cryp- Adams, 1992; Cann & Legler, 1994), ofwhich 1
todiran or pleurodiran (Rich Rich, 1989). genusand9speciesawaitdescription,havemade
Australia was inhabited as recently as the Pleis- their phylogeny even moreenigmatic. Ecologies
tocene by giant horned terrestrial turtles (shells ofmost species arc poorly known. There is con-
1.5m length) of the extinct eucryptodiran
troversy regarding relationships and validity of
Meiolaniidae, which also occurred in South taxa, especially the nominal genera and species
NAmeewriCcaal,edMoandiaag(aMsictatre,rmeLioerrd, 1H9o84w;eArIcshleanrdetaaln.d, oisfiEnlasesytaataenodfEflmuyx.duCrogag(eTrab(l1e9912).)Trheecotganxizoendom1y5
1991). The fossil turtles of Australia were species in 5 genera; Georges & Adams (1992)
reviewed by Gaffney (1981, 1991) and Molnar recognized 23 species in 7 genera.
(1991).
This paper has 2 purposes; a review of the
Outside Australia and New Guinea, the Chel- literature on evolution and zoogeography of
idaeinhabitonly freshwaterseastofthe Andes in Australian freshwaterturtles and a review ofop-
South America (Iverson, 1992), but a fossil posing published conclusions. We recommend
specimen was found in Antarctica (Pritchard & acceptance ofone ofthe proposed phylogenies.
Trebbau, 1984). Although chelids are not the
dominantturtlefamilyinSouthAmerica,theyare
EVOLUTION OFTHECHELIDAE
comparable in diversity to those in Australia.
Consequently,constructingaphylogenyoffresh-
waterturtlesrequiresunderstandingofevolution- The most important evolutionary and
aryrelationshipsbetween the twogeographically zoogeographic studies of Australian freshwater
distantgroups, togetherwith the Pelomedusidae. turtles are by Burbidge et al. (1974), Gaffney
Australianchelidsseparateintotwobroadmor- (1977), Legler & Cann (1980), Legler (1981),
phological and ecological groups (Goode, 1967; McDowell (1983), Georges & Adams (1992),
Cann, 1978), long-necked species and short- and Cann & Legler (1994). The zoogeographic
necked species. Chelodina contains long-necked background was established by Burbidge et al.
species that spear or ambush their food (1974), McDowell (1983), and Pritchard&Treb-
(Pritchard, 1984; Legler & Georges, 1993b). bau (1984). The systematic studies by Legler &
They have 4 claws on the forelimbs, and a neck Cann (1980), Legler (1981), McDowell (1983),
atleast2/3 the lengthoftheshell (Cogger, 1992), and Cann & Legler (1994) approach an ideal
with which they strike out rapidly and capture sampling strategy. Georges & Adams (1992)
prey (Legler, 1993). Elseya, Elusor, Emydura, sampled extensively and used sensitive bio-
AUSTRALIAN FRESHWATERTURTLES 321
ed Phrynops and Platemys were positioned be-
tween ChelodinaandtheAustralianshort-necked
genera.
Although Gaffney (1977) recognized Elseya
and Emydura as congeneric (as did Frair [1980]
andMcDowell [1983&]),thisviewisnotcommon-
ly followed (White Archer, 1994). However,
Gaffneyetal. (1989)claimedthatthetwogenera
are not adequately diagnosed.
Pseudemydura shares no derived skull charac-
ters with either the Australian and South
American long-necked turtles or Australian
short-necked turtles, but instead posseses 7 uni-
que skull characters (Gaffney, 1977). Gaffney
(1977) erected the monogeneric Pseudemy-
durinae and considered Pseudemydura as sister
taxon to other chelids. Pseudemydura is autapo-
morphic among chelids in lacking temporal
emarginations ofthe skull (Gaffney etal., 1989),
and P. umbrina is unique among turtles by its
FIG. 1. Relationshipswithin theChelidae, basedupon behavior of excavating the egg chamber with
26 cranial characters (Gaffney, 1977). Black dots forelimbs rather than hindlimbs (Kuchling,
indicate Australian genera, open circles South 1993). However, several derived characters are
Americangenera. shared between Pseudemydura and other
Australian short-necked turtles, such as reduced
chemical techniques capable ofdetecting cryptic neural bones and cornified head scutes (Mc-
relationships. Dowell, 1983). The intergular scute extending
Among the 3 types of systematic methods posteriorly to separate the humeral scutes was
(evolutionary,phenetic,cladistic),thelasthasthe considered unique and derived in Pseudemydura
strongest Popperian scientific basis. Cladistics (Burbidge et al., 1974), but McDowell (1983)
attempts to test hypotheses by designating char- observed it also in Elseya dentata and Emydura
acters as either plesiomorphic (ancestral) or australis. Thus there may be no need for the
derived (descendent). Derived characters shared Pseudemydurinae.
betweentaxaareconsideredevidenceofcommon Because changes in chromosome morphology
ancestryunlesshomoplasyisdemonstrated(Hen- are quantum changes (King, 1985), cladistic
nig,1966; Brooks & McLennan, 1991). The in- analysesusingkaryotypesarefreefromproblems
group is the taxon whose evolutionary relation- associated with measuring continuous variables
ship is being tested. It is compared toa reference (King, 1985). Mutations detected by G- and C-
taxon, the outgroup, which is usually considered banding can be treated as analogues of mor-
the sister taxon of the ingroup (Brooks & Mc- phological character states (King, 1985). Evol-
aLecnrnuacnia,l1a9n9d1)o.fDteetnermmoisntatsiuobnjeocftitvhee odeuctigsrioounpiins usttuidoineadryirneltahtiisonmsahinpnserofbtyheBcircykphtoadmir&ansBwaekreer
cladistics (Hennig, 1966; Maddisonet al., 1984). (1976) and Bickham & Carr (1983). Also using
&
Gaffney (1977) undertook a cladistic analysis G- and C-banding, Bull Legler (1980) con-
oflivingchelidsusing 26cranial characters (Fig. ducted chromosomal studies of 13 of the 14
1). On the basis ofskull simplification, he desig- pleurodiran genera. However, chelids do not
nated Emydura (inclusive of Elseya) as the out- respond well to this staining technique and a
group of all other chelid genera. By assigning cladistic analysis was not done.
equal weight to all character states, Gaffney Even so, Bull & Legler (1980) gave valuable
(1977) assured that all changes had equal prob- insightintoevolutionofchelids. Forexample,the
ability. The Australian long-necked Chelodina ancestral karyotype is thought to be the one that
were shown to be more closely related to South mostspecies share(King, 1985). The majorityof
Americanlong-neckedChelusandHydromedusa chelids are uniform in number (2n = 50) and
than to Australian short- necked Emydura and bands oftheirchromosomes, which supports the
Pseudemydura. TheSouth American short-neck- outgroup designation of Emydura (and Elseya)
322 MEMOIRS OFTHEQUEENSLANDMUSEUM
Chelidae occurred in both South America and
Africawhenjoined 115 million yearsago.Erym-
nochelys madagascariensis is a relict species
&
(Pritchard Trebbau, 1984)andisprobablymost
like the ancestral condition.
&
Pritchard Trebbau (1984) reviewed earlier
zoogeographic hypotheses of the Chelidae, as
well as Pelomedusidae and other living and ex-
tinct South American turtles. In light of recent
geological andevolutionary studies, alongwitha
fossil chelid from Antarctica of Miocene age,
they suggested the Chelidae could have
&
originated in Antarctica. Pritchard Trebbau
(1984) argued that passive dispersal across vast
stretches ofocean is possible only for terrestrial
turtles. Thus an Antarctic landbridge or narrow
sea should be considered in zoogeographic
hypotheses of Australian and South American
chelids (Fig. 2).
ThehypothesisthattheAustralianlong-necked
turtles (Chelodina) did not originate from
Australian short-necked turtles (Gaffney, 1977)
FItGhe.m2.iTdhCereptoasciteioounso(fafAtuesrtRrailciha&anRdiAcnht[a1r9c8t9i]c)a.dBulraicnkg &issuApdpaormtsed(b1y9r9e2c)entstbuidoicehdemitchaeledavtoal.uGteioorngaerys
areas in Australia indicate the seaway extant during
relationships of Australian chelids using al-
theCretaceous.
lozyme electrophoresis (Hillis, 1987). Their ex-
by Gaffney (1977). The Cryptodira existed tensive data set included 54 independent loci of
277 specimens of all species (except
duringtheEarlyCretaceous. Becausethelevelof
Pseudemydura umbrina) from 76 populations
chromosomal variation among cryptodiran fam- through Australia and New Guinea.
ilies is lessthanthatbetween thetwo pleurodiran Georges & Adams (1992) employed 3 sys-
families, pleurodirans probably originated prior
& tematic methods. Their phenetic analysis incor-
to the Cretaceous (Bull Legler, 1980).
McDowell (1983) used osteological and soft porated principal co-ordinates, which gave
representations ofgenetic distance. Thecladistic
organ characters to study relationships of the analysisusing PAUP(Swofford, 1985)produced
Chelidae and Pelomedusidae. He concluded that unresolvedcladogramsbecauseonly 14of54loci
ElseyaandEmydurawerethemostpeiomedusid- couldbeusedascharacters. Aparsimonymethod
like and, thus, plesiomorphic of the Chelidae, (Farris, 1972; Baverstock & Schwaner, 1985)
which further supported Gaffney's (1977) out- using distance-Wagner and Fitch-Margoliash
group designation. Specifically, Elseya dentata procedures, which are not widely accepted, also
has a longitudinal ridge on the maxilla, a charac- produced unresolved cladograms. Although the
ter it shares only with the pelomedusid Podoc- several generated phylogenies varied, relation-
nemisofSouth America. ships ofsome taxa remained constant.
Frair (1980) conducted a biochemical study of Georges & Adams* (1992) results suggest that
pleurodirans using total protein electrophoresis divergence between the Australian long- and
and Immunoelectrophoresis with phenetic anal- short-necked turtles is about twice that within
yses, sampling all Chelidae and Pelomedusidae. either of the two groups; and the level ofdiver-
He confirmed that each family is monophyletic, genceamongtheshort-neckedspeciesissubstan-
as first proposed by Gaffney (1975) and sup- tially greater than that among the long-necked.
ported the arrangement ofGaffney (1979). Frair These results support Gaffney's (1977)
(1980) also showed the Chelidae to be more phylogeny in which South American genera are
closely related to the Madagascan pelomedusid intermediates, and short-necked turtles are
Erymnochelys than to the South American pelo- probably more primitive.
medusid Podocnemis and Peltocephalus. Thus it McDowell (1983) presented a simple dispersal
appears that the lineage giving rise to the theory to explain Gaffney's (1977) results. The
AUSTRALIAN FRESHWATERTURTLES 323
ancestor of Emydura entered South America ment. During theEarly Cretaceous, the ancestral
from Australia. The lineagediverged, givingrise lineageincluding Chelycarapookusinhabitedthe
to South American short-necked turtles, from southern supercontinent. DuringthelateEocene,
whichevolvedalong-neckedformthatdispersed Australia was isolated by sea, with members of
back into Australia. theChelycarapookuslineageleftinAustraliaand
Neuralbones are considered an important tax- in the Antarctic-South American landmass. The
onomiccharacterforchelids.They werereported trend in neural bones in the Australian lineage
absent in all living Australian chelids, except was forreduction and loss, giving rise to Elseya,
Chelodina oblonga (Burbidge et al., 1974; Mc- Elusor, Emydura, Pseudemyduraand Rheodytes.
Dowell, 1983). Neural bones were reported in The South American lineage retained well-
South American chelids, which McDowell developed neural bones, evolving into the South
(1983) considered evidence for convergence of American short-necked and long-necked turtle
the long-necked condition. However, Rhodin & genera. However, the ancestral lineage ofSouth
Mittermeier (1977) reported low frequencies of American long-necked turtles also dispersed
neural bones incertain populationsofbothshort- back into Australia, giving rise to Australian
and long-necked Australian chelids. long-necked turtles. The general trend forneural
The Australian long-necked Chelodina oblon- bonesinthisgroupwasforreductionandloss,but
ga and South American chelids possess well- well-developedneuralboneswereretainedinthe
developed neural bones, which in C. oblonga immediate lineage to Chelodina oblonga. Mor-
vary from 5-8 (Burbidge et al., 1974; Rhodin^ phologically and biochemically, C. oblonga is
Mittermeier, 1977). Neural bones (although
plesiomorphic in itsgenus(Burbidgeetal., 1974;
reduced in size) persist at low frequencies in Georges & Adams, 1992).
certain populations ofElseya latisternum, an un- Why did only the ancestral South American
described Elseya sp. from the Manning River
(NSW), C. longicollis, C. novaeguineae and C. long-necked turtle lineage disperse back to
siebenrocki. Oftwofossil specimensofEmydura Aluasntdrmaalsisa, farnodmnotthealsAonttahrectaincce-sStoruatlhSoAumtehriAcma-n
from mid-TertiarydepositsinTasmania,onehas
onereducedneuralboneandtheothernone(War- erican short-necked turtle lineage? The answer
might be elucidated by examining ecologies of
ren, 1969a). Apparently the character should be
assessed as absent/reduced in size, or well the two groups. Unlike any South American
developed (Pritchard, 1984). short-necked turtle genus, in addition to inhabit-
inTthheefoEsasrillyreCcroertdaocfeocuhseliodfs Visicptooorri,abe(gWianrrneinn,g Hinygdfrroemsehwdautsears,altsheoSionuhtahbiAtmseersitcuaarnielson(gE-rnnesctke&d
1969b) with a specimen identified first as extant Barbour, 1989). Australian long-necked Chelo-
Emydura macquarrii but later as Chelycara- dina(C. longicollis, C. expansa)hasgreaterresis-
pookus arcuatus. It has well-developed neural tance to evaporative water loss than Australian
bones (Burbidge et al., 1974), and is considered short-necked Emydura (E. macquarrii; Chess-
ancestral to short-necked turtles in which neural man, 1984). Chelodina inhabits permanent and
bones are absent or reduced. Early Cretaceous temporary freshwaters (Chessman, 1984), and
chclid fossils have been found in NSW and in can also tolerate saline waters. During drought,
western QLD (a Chelycarapookus-hke form: Chelodina walksoverland and findsotherwater,
R.E. Molnar, pers. comm.). or buries into substrate and aestivates (Goode,
Chelycarapookusarcuatuswasariver-dweller 1967;Cann, 1978; Griggetal., 1986).Chelodina
(Molnar, 1991) and probably had a broad dis- rugosa can inhabit highly saline ephemeral
tribution in eastern Australia. With vulcanism swamps (Ehmann, 1992), and C. longicollis can
forming the Great Dividing Range later during tolerate brackish water for prolonged periods
the Cretaceous (Galloway & Kemp, 1984), and (Smith, 1993). On the other hand, all Australian
with increasing aridification during the Pleis- andSouthAmericanshort-neckedchelidsinhabit
tocene, the contemporary turtle fauna may have only permanent freshwaters, except Pseud-
evolvedby allopatric speciation. emydura which inhabits ephemeral swamps and
aestivates. Thus itappearsthatthelineagegiving
McDowell's (1983) zoogeographic hypothesis rise to Australian long-necked turtleshad greater
explains certain phylogenetic relationships. The potentialfordispersalacrosslandand/oranarrow
following model isadaptedfrom his,considering sea between Australia and the Antarctic-South
that at times Antarctica had a warmer environ- American landmass.
324 MEMOIRSOFTHEQUEENSLAND MUSEUM
EVOLUTIONARY RELATIONSHIPS karyotypes (2n=50 and gross chromosome mor-
AMONG AUSTRALIAN FRESHWATER phology), and Pseudemydura differed only
TURTLES slightly by having chromosome pairs 6 and 10
acrocentric. Pelomedusidturtlesdifferedgreatly,
InastudyofevolutonaryrelationshipsofAust- with 2n=26-36 and 5 large macrosomes.
ralianchelids, the ideal sample would includeall Using total protein electrophoresis, Frair
species; with specimens from throughout a (1980)observedonlyonetypeofelectrophoretic
species' range, ofvarious sizes and both sexes to pattern among two species of Elseya and two
include the variations of geography, ontogeny Emydura. The electrophoregrams ofElseya lati-
and sexual dimorphism. Legler & Cann (1980), sternum and Elseya novaeguineae were similar,
Legler (1981), McDowell (1983), Georges & but the former longer and with its cathodal line
Adams (1992), and Cann & Legler (1994) ap- closer to the negative pole. The electrophore-
proach this ideal sampling strategy. grams of Emydura signata and Emydura sub-
Using cladistic analysis of cranial characters, globosaweresimilartoElseyanovaeguineae,but
Gaffney (1977) diagnosed distant chelid genera, the latter slightly more negative. Among
but he could not separate Elseya from Emydura. Chelodina, there were two electrophoretic pat-
Becausetheskull andjawofchelidsare probably terns: 3 anodal lines for C. longicollis and C.
C
correlated conservative characters, the data set novaeguineae; and 1 for C. expansa, rugosa
appearsinadequatefordiagnosingcloselyrelated and C. siebenrocki. Observing little variation be-
genera. In addition, the data were from small tweenElseyaand Emydurarelativetothatwithin
samples, often single specimens. Chelodina, Frair (1980) recognized Elseya and
Burbidge et al. (1974) diagnosed Elseya from Emydura as synonymous.
Emyduraon nuchal and interguiarscutes: nuchal Frair (1980) concluded the results of Bull &
scute usually absent in Elseya, present in Emy- Legler(1980)(similarchromosomenumbersand
dura', interguiar scute smaller and more rectan- bands of Elseya and Emydura) supported their
gular in Elseya, larger and less rectangular in synonymy. However the chromosomes of
Emydura. Elseya was reported to have an elon- Pseudemydura umbrina and Rheodytes leukops
gate snout. However, data for Elseya and Emy- are also similar in number and bands to Elseya
dura were from single specimens of Elseya and Emydura, but Pseudemydura and Rheodytes
latisternum and Emydura macquarrii, their sole are separate genera because oftheirunique mor-
representatives ofthe two genera. phologies.
Burbidge et al. (1974) also used phenetics to Frair (1980) also used the observations of
analyze serological data of Australian chelids. Webb (1978) on basking of captive Australian
The first 3 dimensions ofthe principal compon- turtles to support synonymy of Elseya and
ents described 85% ofthe variation. Among the Emydura: Elseyaand Emydurabasked regularly,
short-necked turtles, Elseya dentata clustered whereas Chelodina never basked. However,
with Elseya novaeguineae', Emydura australis, baskingisprevalentintheSouthAmericanshort-
Emydura krefftii and Emydura subglobosa necked chelids Phrynops and Platemys (Lacher
clustered; and the two groups together formed a et al., 1986; Monteiro & Diefenbach, 1987); in
closely associated group. Elseya latisternum, Pelomedusidae (short-necked turtles) (Miller,
Emydura macquarrii and Pseudemydura 1979); and catholically in the suborder Cryp-
umbrina were outliers; however, after omitting todira (short-necked turtles), especially the
the unU-Elseya latisternum data, Elseya latister- Emydidae (Ernst & Barbour, 1989). Because
num clustered with Elseya dentata and Elseya basking behavior is widespread in short-necked
novaeguineae. The long-necked turtle species turtles, ithas little value in supporting synonymy
C
(Chelodina) clustered together, but with ofElseya and Emydura. Rather, it probably indi-
oblonga an outlier within the genus. catesecological differencesbetween short-neck-
These results showing P. umbrina and C. ed and long-necked turtles.
oblongaas outliers to the short-necked and long- Legler (1981) examined more than 3000
necked turtles, respectively, are accepted. Their specimens of Australian chelids and considered
sera react strongly and they have different mor- geographic distributions to elucidate relation-
phologies. However, the designation of Elseya ships. He stated that Elseya dentata (and at least
latisternum as an outlier is treated with caution 3 undescribed species) and Elseya latisternum
because&its serum reacted weakly. (and at least 1 undescribed species) belonged
Bull Legler(1980)reportedtheshort-necked to separate genera, but no diagnoses were
Elseya, Emyduraand Rheodvtestohave identical presented.
1
AUSTRALIANFRESHWATERTURTLES 325
macquarrii, E. krefftii (including Cooper Creek
Pseudemydura and FraserIslandpopulations) andE. signata, all
C. steindachneri of which they considered to be E. macquarrii.
C. novaeguineae Georges (1993) regarded E. macquarrii as
— C. longicollis
C.oblanga polytypic or a species complex, with unresolved
C. (Mann) systematic positionsofdistinctivepopulations in
1 C. rugosa Cooper Creek and Fraser Island and the taxa
C. expansa
referred to as E. krefftii and E. signata.
EEll..(lGatwiysdtierr)num Although Elseya latisternum was an outlier to
El. (Bellingen) otherElseyaspecies(andEmydura),neitherBur-
^ El. (Manning) bidgeetal.(1974)norFrair(1980)suggestedthat
Em. subglobosa Elseya was paraphyletic. In serological studies
Em. (Sleisbeck) like Frair's (1980), thevariation ofelectrophore-
Em. victoriae grams is misconstrued by multiple protein
1 EEmm..m(aDcaqluyarMiiission) analysis (Baverstock & Schwaner, 1985). How-
ever, monovalent techniques with antisera for
EEll..d(eSntthatAalligator) single proteins are more amenable to genetic
El.novaeguineae interpretation because the strength ofcross reac-
' El, (Johnstone) tion is directly proportional to differences in the
1 El. (Burnett) gene locus encoding the protein (Baverstock &
Short-necked alpha Schwaner, 1985). Georges& Adams(1992)con-
Rheodytes sidered single locus protein variation and conse-
quently has the same objectivity as monovalent
immunological methods.
FIG.3.ThephylogenyofAustralian freshwaterturtles
(and Elseya novaeguineae), based on allozyme Georges & Adams' (1992) principal co-or-
electrophoresis (Georges & Adams, 1992). Short- dinates analysis of the first 3 dimensions des-
necked alpha isElusormacrurus. cribed 63% of the variation. Australian chelids
clustered into 5 groups, which were recognised
McDowell (1983) examined at least 5 spec- by Legler (1981). Elseya novaeguineae, which
imens from various localities of all Australian McDowell (1983) placed in synonymy ofElseya
short-necked turtle species, except Emydura sig- dentata, had a divergence level similar to that of
nata. Becausethere were no significantdifferen- Elusor and Rheodytes to the Elseya dentata
ces among the characters he tested, McDowell group. Thisspeciesalsolost itsaffinitieswiththe
(1983)placedElseyanovaeguineaein synonymy Elseyadentata group in higherdimensions.
of Elseya dentata; and Emydura albertisii, The highest levels of divergence were within
Emydura krefftii and Emydura subglobosa in Elseya and Chelodina; that within Emydura was
synonymyofEmyduraaustralis.Also,hisresults relatively small, suggesting its5 speciesarerela-
showed Elseya dentata (and Elseya novae- tively young. Divergence between Elseya and
guineae) to be more closely related to Emydura Emydura was less than that within Elseya. The
australisthan toElseyalatisternum, sohe recog- ElseyadentatagroupwasclosertoEmydurathan
nized Elseya and Emydura as synonyms. How- totheElseyalatisternumgroup,whichwasestab-
ever, the biochemical results of Georges & lished also by McDowell (1983). However, in-
Adams (1992) argue against these synonymies. stead of recognizing Elseya and Emydura as
cInesadadmitoinogn,tthheerneoamrienaplrosnpoeucniceesdocfoElmorydduifrfae.reInn- tcohnegeEnlesreiyca,Gleaotrigsteesr&numAdgarmosup(19t9o2)cocomnpsriidseereda
life Emydura subglobosa, commonly called the separate genus, their rationale that synonymiza-
painted turtle, has acrimson plastron and yellow tion would have to include also Elusor and
on the face (Legler, 1981). Living Emydura sp. Rheodytes. The genetic distances between the
(Daly, Nicholson and Roper Rivers) has red on provisional genera(Elseyadentatagroup, Elseya
the face and legs, and Emydura krefftii yellow latisternumgroup,Emydura)aresimilarto, ifnot
across the head (Cann, 1978; Cogger, 1992). But greater than, that between many of the cryp-
the significance of color pattern in diagnosing todiranemydidgenera,especiallythebatagurines
short-necked turtles remains unknown. (Sites etal., 1984).
On allozyme characters, Georges & Adams InregardtoAustralianlong-neckedChelodina,
(1992) could not distinguish between Emydura the phylogenetic results of Georges & Adams
326 MEMOIRS OFTHEQUEENSLAND MUSEUM
TABLE 2. Distribution ofAustralian freshwater turtles (Georges & Adams 1992) and Elseya novaeguineae.
FromCogger(1992), Ehmann (1992),Georges& Adams (1992), Iverson (1992), Legler&Georges (1993b),
Georges(1993). ?=systematicpositionofthepopulation isunresolved.Zoogeographic regions from Fig. 4
GROUP SOUTH- WESTERN CENTRAL SOUTHEASTERN EASTERN NORTHERN NEW
WESTERN GUINEA
WESTERN EASTERN WESTERN EASTERN
Chetodina expansa expansa rugosa
expama
sp.(Liverpool
+MannR.i
Chetodina steindachneri longicollis longicolus
longicollis
novaeguineae
Chetodina obtonga
obtonga
Elseya sp.(Burnell+ sp.(South sp.(Johnstone novae-
dentala FitzroyR.) AlligatorR.) R.1 guineae
dentata
Elseya sp.(Gwydir tatislernum
tatislernum R.)
sp.(Bellingen
R-)
sp.(Manning
R.)
Elinor macrurus
Emydura macquarrii ' krefftii? krefftii?
(CooperC.+ victoriae
DlamantinaR.) signatat subglobosa
1 1 sp.(Daly,Mitchell+South
I (FrascrI.) AlligatorR.)
1
i
i sp.(Daly,Nicholson+RoperR.)
Pseudemydura umbrina
Rheodytes leukops
(1992) aligned with the species groups estab- ters from large samples (n=41-1000) of the El-
lished by their prinicpal components analysis. In seya dentata group, Elseya latisternum group,
general, their interpretation of relationships Elusor, Emydura and Rheodytes. The data were
agreed with Burbidge et al. (1974) and Legler treated as percentages ofcarapace length. In the
(1981); however, contrary to Legler (1981), the resulting 3D scatter plots, the groups separated
level ofdivergence between the C. expansa and equidistant from each other but with Emydura
C. longicollis groups did not warrant separate separate from all. Consequently, Cann & Legler
generic recognition. (1994) recognized the Elseya dentata group, El-
Georges & Adams (1992) tentatively placed seya latisternum group, Elusorand Rheodytesas
ElusorandRheodytesoutside theElseyadentata, the Elseyagroup ofgenera.
Elseya latisternum and Emydura radiation (Fig. The primary weakness of Cann & Legler's
3). Although the cladograms of Georges & (1994) multidiscriminate analysis is that a
Adams (1992) were unresolved, their phenetic species' morphology is adaptive to its environ-
analysis suggests both Elusor and Rheodytes ment. For example, extrapolation from Cann &
have their closest affinities with the Elseya den- Legler (1994) would suggest that Australian
tata group. Similarly, the PAUP-generated short-necked chelids are more closely related to
cladogram ofCann & Legler(1994), based upon short-necked emydid turtles than to Australian
35 morphological and 16 aliozyme characters long-necked chelids. Also, sample groups must
treatedequallyandcomparedtoChelodinaasthe beofsimilarsex and ageclasses tonegate sexual
outgroup, suggests Elusor and Rheodytes form
and ontogenetic differences in morphology. For
the sistergroup to the Elseyadentata group. example,90%ofCann& Legler's(1994)sample
Cann & Legler (1994) also performed multi- of Elusormacrurus were smalljuveniles. Ifthis
discriminate analyses with 31 mensural charac- species experiencesontogeneticchanges in mor-
AUSTRALIAN FRESHWATERTURTLES 327
of incubation (Legler, 1985). The tropical
strategy invoked largehard eggs being deposited
during the dry season (winter) with long incuba-
tion; and the temperate strategy, smallerflexible
eggsdepositedinspringorsummerwithrelative-
ly short incubation.
A temperate origin was indicated (Legler,
1985) for the Elseya latisternum group (E. lati-
sternum and at least 3 undescribed species),
Chelodina longicollis group (C longicollis, C.
novaeguineae,Csteindachneri)andEmydura.A
tropical origin was indicated for the Elseya den-
tata group (E. dentata, E. novaeguineaeand two
undescribed species) and C. expansa group (C.
expansa, C. oblonga, C. rugosa).
However, the two patterns are not realistic and
the criterion for egg hardness was only subjec-
tively defined. For example, the C. expansa and
E. dentata groups lay the largest eggs (Legler,
1985),butthesespeciesarethelargestAustralian
120 130 freshwater turtles. Also, larger eggs may neces-
sarily be harderbecauseofstructuralconstraints.
FIG.4.ThezoogeographicregionsofAustralianfresh- In cryptodirans there is positive correlation be-
water turtles (modified from Burbidge el al., 1974). tween adult size and egg size; egg size and hatc-
Species withineach region are listed inTable2. hingsize;and forhard shelledeggs,eggsizeand
shellthickness(Ewert, 1979).Also,sometropical
phology, then comparison against a sample of species lay softshelledeggs, and someboth hard
&
another species except ofsimilar composition is and soft (Moll Legler, 1971; Ewert, 1979).
meaningless. Legler (1985) was selective of data from pre-
The relationship ofPseudemydura umbrina to vious authors, ignoring that which lessened dis-
other Australian chelids remains obscure. Its en- tinction between the two patterns. For example,
dangered status and small population size (Kuc- C. oblonga (proposed tropical origin) nests
hling & DeJose 1989) limit certain avenues of during summer (Burbidge, 1967, 1984) and C.
research(e.g.Georges&Adams 1992).Nonethe- expansa (proposed tropical origin) sometimes
less,P. umbrinasharesseveralderivedcharacters duringautumn(Goode&Russell, 1968;Georges,
with other Australian short-necked turtles, but 1984); and eggs of C. longicollis (proposed
none with South American genera or Australian temperateorigin) undergo long incubation, upto
long-neckedChelodina(McDowell, 1983). Thus 200days (Cann, 1978). Thus itappearsthataddi-
Pseudemydura may be the sister taxon to other tional reproductive patterns exist. Also, if the
Australian short- necked turtles. contrasting tropical and temperate environments
are strong selective pressures on reproduction,
REPRODUCTIVEPATTERNS OF why do some species of each group inhabit the
AUSTRALIAN FRESHWATER TURTLES other environment? Within Emydura (proposed
temperate origin), E. subglobosa, E. victoriae
Examining the ecology of a species can help and 2 undescribed species inhabit only the
elucidate itsevolutionary history: knowing func- tropics; and E. macquarrii (inclusive of E.
tionsofstructuresmakesthegradingofcharacter australis, E. krefftii and E. signata) inhabits both
states more accurate, and it gives insight into temperate and tropical environments (Cogger,
selective pressures for higher or lower rates of 1992: Iverson, 1992). Georges el al. (1993) pre-
speciation (Shine, 1985). Legler (1985) recog- viously questioned the fit of Legless (1985)
nized 2 patterns of reproduction among Aust- reproductive patterns to species groups.
ralian chelids, one having evolved in the tropics However, according to the centre ofdiversity
and the other in temperate environments. rule (Ross, 1974), Emydura probably evolved in
The tworeproductive patterns were defined by thetropics,andalsotheElseyadentata, C. expan-
egg size and hardness, time ofnesting and length sa (C expansa, C. rugosa and an undescribed
328 MEMOIRSOFTHEQUEENSLAND MUSEUM
species in the NT, sensu Georges & Adams speciation, directed by isolation of river drain-
[1992])and C. longicollisgroups. With 3 species ages with increasing aridification.
in the temperate zone and Elseya latisternum Southwestern Australia is inhabited by en-
inhabiting both temperateand tropical zones, the demic Chelodina oblonga and Pseudemydura
E. latisternum group probably evolved in temp- umbrina. This region became isolated by forma-
erate environments. In addition, the groups for tion of a broad inland sea from the GulfofCar-
whichthecentreofdiversityrulepredictstropical pentaria to thecoastofS AUSTduring the Early
origins all have representatives and/or closely Cretaceous(Rich&Rich, 1989;Fig.2).Burbidge
related species in New Guinea, dispersal north- et al. (1974) proposed thattheeastern Chelodina
ward having been possible by Pleistocene land- ancestor evolved into the form lacking neural
bridges (Burbidge et al., 1974). On the other bones, and the eastern Pseudemydura became
hand, the Elseya latisternum group (predicted extinct. Prior to the Cretaceous sea, Elseya and
temperate origin), with E. latisternum ranging Emydura had tropical northern distributions.
fromnorthernNSW(temperate)tothetipofCape LaterduringtheCretaceoustheseareceded,leav-
York Peninsula, has no close relative in New ing a vast waterless desert equally impenetrable.
Guinea. Alternatively, Pritchard & Trebbau (1984) sug-
gested two separate invasions from Antarctica,
ZOOGEOGRAPHY OFAUSTRALIAN one each into eastern and western Australia.
FRESHWATERTURTLES The distributions of freshwater turtle and fish
species groups support the vicariance hypothesis
ofBurbidgeetal.(1974).Thespecieswithineach
Zoogeographic regions for Australian fresh- group are largely allopatric, with sympatry be-
water turtles were defined by Burbidge et al. tween species ofdifferent groups (Legler, 1981;
(1974), which in general corresponded to those Iverson, 1992). In addition, there are fossils and
forfreshwaterfish(Whitley, 1959). However,we arelictpopulationofEmydurainthe aridzoneof
also recognize Central and Eastern zoogeo- SW QLD and adjacent S AUST (Gaffney, 1979;
graphic regions for freshwater turtles (Fig. 4, Ehmann, 1992); and fossils of Pseudemydura
Table 2). The Central Region is in the arid zone (early to middle Miocene) from Riversleigh, on
of interior Australia and contains the Cooper the GulfofCarpentaria (Gaffney etal., 1989).
Creek and Diamantina River population oflarge Burbidgeetal. (1974)showed C. oblongatobe
macrocephalous Emydura, which Ehmann morphologically and serologically an outlier to
&
(1992)andLegler Georges(1993b)reportedto otherChelodina. Similarly, thepheneticanalysis
be an undescribed species. Although Georges & (3D) of Georges & Adams (1992) showed C
Adams (1992) could not distinguish the popula- oblonga to be an outlier within the C. expansa
tion by allozyme electrophoresis from E. mac- group(althoughthiswasnotmaintainedathigher
quarrii, Georges (1993) regarded it as a dis- dimensions), and their phylogenetic analyses
tinctive population with unresolved systematic consistently placed C. oblonga as pleisiomor-
position. The Eastern Region encompasses the phic. These results together with retention of
Fitzroy, Burnett, Mary and Brisbane riverdrain- well- developed neural bones (Burbidge et al.,
ages, to which are endemic only an undescribed 1974) indicate that C. oblonga approaches the
species ofthe Elseya dentata group (Georges & ancestral condition, further supporting Bur-
Adams, 1992), Elusor macrurus and Rheodytes bidge'set al. (1974) vicariance hypothesis.
leukops. Our Central and Eastern regions cor- The hypothesis of Burbidge et al. (1974) re-
respond to Whitley's (1959) Sturtian and Kref- quires the short-necked turtle genera to have a
ftian regions, respectively. Zoogeographic northern distribution prior to formation of the
regionsforfreshwaterturtlesarealmostidentical inland Cretaceous sea. However, the Elseya
tothoseforfreshwaterfish.Also,likesomefresh- latisternum group probably had a southeastern
water fish, some freshwater turtles inhabit both origin,andthuspotential fordispersalintosouth-
northern Australia and southern New Guinea. western Australia. The following hypotheses
The vicariance hypothesis of Burbidge et al. may explain the absence of the Elseya latister-
(1974) suggests that ancestors of each species num group from southwestern Australia. (1) The
group were distributed widely throughout north- E. latisternum lineage previously inhabited also
ern and eastern Australia during the Pleistocene southwestern Australia but has sincebecomeex-
epoch when the climate was cooler and wetter. tinct there. (2) The southern Australian environ-
The existing chelid fauna resulted by allopatric mentduringtheCretaceouswasnotfavorablefor