123 



arguci-l lioili Biilatiiii iilac ami 1 my- 



didac arc |">n)lialil\' moimphvlcl ic, and llial 

 the latter is the sisici group ol a cladc 

 consisting of ihc former aiiel ihc Icsiu- 

 dinidae. Recently we also lound states 

 of two morphological characters (i.e., 

 well dc\ ckiped foramina i^l axillaiv and 

 inguinal musk duels wiih pcnci i ai ion inio 

 the bi casi clianihcr, and anici olalci all\ 

 flared ihac liladc |l i;'. 2|) lhal seemingly 

 lend I lil t her support lo ihe monophyly 

 of the liataguridae (Vasukawa, 1997; 

 Hirayama, Yasukawa and Aoki, un|-)ub- 

 lished dala), because llie\ oeeui in mosi 

 bataginid luriies Imi noi in liie 1 iu\dii.lae 

 or Tcstudinidae. In oui examinai ions, ilie 

 foiiner ehaiaeiei siale occuiied in all 

 balaguiids e\elusi\e ol Morciuu. Musk 

 ducts and indisiinet musk foramina alsii 

 exist in some non lesi udinoid lamilies of 

 turtles, such as ihe ( 'hel\ di idae, ( hehdae, 

 Kinostcrnidae, and I'laivsieriiklae, and iheii 

 (K'currence is iluis likeK lo reprcseni piimi- 

 li\e condiiious ui tlie lesiudines (Cialine\ 

 and \le\ian, 19SS). However, develop- 

 ment of musk duct foramina lo ihe point 

 ol leachniL' the breast chamber is a condi 

 IKMi umi.|ue lo the Hataguridae lllua\aina. 

 \asuka\\a .mil \oki, unpubhshed dalal. 

 ()n the oiliei liand. an aniei olalei all\ 

 flared iliac IMadc was obsei\ed m ail 

 balagurid. but not m other tuitles inchul- 

 ing ihose beKmging to non-lesiiKlinoid 

 lamilies. \Vc exeludcd these characters 

 from OUI analyses, since these are olni- 

 oiisK not udottnati\e \o\ the iiderence of 

 I clal loiisliips anuMiv the geoemydincs. 

 However, the slates of these characters arc 

 possiblv svnapomorphs of the Halagiiridae. 

 Thus, in the sections below we lake an a 

 prion .issumpiion of ihe bataguritl mono- 

 phyly. 



A/oiio/'/i v/v oj ihc ( icocni vdmac 



Mcl^owell (1964) referred lo two charac- 

 ter states as discriminating Gcoemydinac 

 from Batagurinac — narrow and flat tritu- 

 rating surfaces, and letluction or absence 

 of the c|uadralojugal. 1 lie loiincr probably 

 lepicsents a primitive condition of the 

 I est udinoidea (ilirayama, 1984; Gaffney 

 and Meylan, 1988). In this study, the latter 

 condition is observed in the Geoeinydu 

 group exclusive of (}. spen^leri. Because 

 the Maiiremys group had a relatively large 

 c|uadt atojugal as in KLichw^u. Ocudia, 

 Odsr/iiiii, and Sichciuockiclln . it seems to 

 be more apiM opruite to considei the reduc- 

 or absence of ihe ciuadratojugal as a 

 characteristic of the Geoeniyda group 

 rather than as a synapomorph of the Cieo- 

 emydinac. 



Ilira\ama (I9S4) considered that the 

 subdivision o\ ihc foramen nervi trigemina- 

 lis, oliscr\ed in more than 4()"'o of the spec- 

 imens examined lor each gcoemydine 

 species, is a synapomorph of the Gcoemyd- 

 inac besides the loss or reduction of the 

 quadraiiijugal. However, Gaffney and 

 \Ie\lan (19SS) were concerned about such 

 an ineousisient occurrence of the former 

 characiet stale, and argued that its use at 

 this level uas dubious. Our additional 

 observations aUo confirmed the remarkable 

 variability of this character at the intraspe- 

 cific level (\asukawa. unpublished obser- 

 vation), rhereforc. wo did not include this 

 character in our phylogcnclic analyses. 

 Although Gaffney and Mcylan (19SS) 

 regarded the loss or reduction of the 

 quadratojugal as "relatively consisicni" in 

 geoemydines, this character stale turned 

 out to be confineil to the Geoeniyda group 

 alone (see above). 



Our analyses yielded stales of three char- 

 acters as possibly supiH'>riing monophyly of 



Fin. 2. Dorsal views of pelvic jtirdlcs. I inydulac: (a) rruihcmys Mrif>ia Wc-Kum. H ■> 

 Chinvmys nrvvxii: Cicocniydinac: (c) .Sfaurcniys japoima, (d) M. miiiua miiiua. (c) < 

 ala, (0 Rhinoi U'tnmys pukherrima manni, (g) Cistocletnmys Jflavomatgmata jnavomarginai. 

 Geoeniyda japonica. 



