o, ee7 by ch",J;Xl?:i::1,';;lii'll,:''"-

The Phylogeny of Cheloniid Sea Revisited Jauns F. Pnnnlur2 AND Dlvrn E. Fmrovsxvr

tDepartmentof Geologt, Universitl,of Rhode Island, Kingston, Rhode Island0288l USA; 2Present Address: Department of Integrative Biolog1,, Universi4' of Califumia, Berkeley, Califurnia 94720 USA I F ax : 5 1 0 -642 - I 822 ; E-mail : P arham@ s oc rate s.be rke let'.e dn ] Ansrnacr. - Thirteen chelonioid taxa are examined cladistically using 24 cranial and post-cranial osteological characters. Shell characters are found to be useful in elucidating phylogenetic relation- ships within the . The results of this study support a monophyletic Osteopyginae. The mutual affinities of the stem cheloniines are not resolved. Syllomus is the sister taxon to all the Cheloniinae (not just to Natator) and, Eretmochelys is the sister taxon to the Carettini.

Kny Wonos.-Reptilia; Testudines; Cheloniidae; ;CtenochelyslOsteopygis;Erquelinnesia; Argillochelysi ; Puppigerus; Syllomus; Natatorl Chelonia; Eretmochelysl Caretta; Lepidochely s; systematics ; paleontology; phylogeny

Cryptodiran turtles have invaded the marine realm Eretmochelys, Caretta, Lepidochelys) and numel ous extinct twice: once in the and once in the early taxa (see Pritchard and Trebbau, 1984). Excluding the prob- (Gaffney andMeylan, 1988;Hirayaffia, 1994). By Campanian lematical extinct generaToxochelv s and Ctenochebs (whose time, the Cretaceous radiation (Chelonioidea) had already affinities are discussed below), the Cheloniidae can be diversified into three major clades: Cheloniidae, subdivided into three, not necessarily monophyletic, groups: , and . The latter two families 1) Osteopygin ae (Zangerl, 1953) ;2)Eocene stem cheloniines form the monophyletic Dermochelyoidea (Gaffney and ("Eochelyinae" of Moody, 1968); and 3) Cheloniinae (de- Meylan, 1988), the sister-taxon to Cheloniidae. Pelagic fined here as the crown-group of the Cheloniidae, i.e., those adaptations such as the loss of moveable articulations in the taxa that share the most recent common ancestor of the manus to form a paddle, straightening of the humerus, and extant cheloniids; see de Quetroz and Gauthier, 1990). reduction of the plastral bones were all apparently indepen- Toxochelys and Ctenochelvs () are dently acquired in dermochelyids, protostegids, and among the earliest chelonioids represented by ample cranial cheloniids (Hirayaffi&, 1994). Of the three clades, Cheloniidae and post-cranial material. They were considered to be mem- seems to have retained the primitive condition the longest; bers of the Toxochelyidae (Zangerl, 1953) until Fastovsky even Recent cheloniids (Caretta, Chelonia, Eretmochelys, ( 1985) demonstrated that Ctenochelys (referred to as FMNH Lepiclochelys,, Ncrtcrtor) have not developed the level of PR444; Gaffney and Meylan [ 1988] have since referred this pelagic specialization (e.g., broad flat humeri, extremely specimen to Ctenochelys) is more closely related to Recent reduced shells, greatly elongated forelimbs) exhibited by cheloniids than to Toxocltelys. As a result, the family dernrochelyoids (epitomized in Recent Dennochelys). Toxochelyidae (sensu Zangerl,, 1953) was rendered Classification within the Cheloniidae has traditionally paraphyletic. been based on the abundant shell material (Zangerl, 1953, These findings have been incorporated into two hypoth- 1958). While the appendicular skeleton was given some eses. Gaffney and Meylan ( 1988) suggested that Ctenochelys consideration, cranial material was essentially ignored. More should be considered a sister taxon to all chelonioids except recently, de scriptions of new and neglected cranial material Toxochelys (Fig. I a). Hirayama (1994), however, argued (Moody, 1974; Gaffney, 1979; Foster, 1980; Fastovsky, that Toxochelys and Ctenochelys should be considered as I 985) combined with cladistic methodology have resulted in members of an expanded Cheloniidae (Fig. I b). The latter hypotheses based almost entirely on cranial characters, hypothesis is based on an analysis comprised of more taxa particularly from the basicranium (Fastovsky, 1985; Gaffney and more characters and is adopted here. However, we have and Meylan, 1988). Hirayama (1994) was the first to attempt included neither Dermochelyidae nor Protostegidae in our data to treat cranial and post-cranial characters equally. set and therefore cannot comment further on this problem. This study pools the data of previous workers who have We use a stem-based definition (de QueirozandGauthier, studied the evolutionary relationships of cheloniids and 1990) of Cheloniidae as those turtles that share a more recent expands them with additional shell characters. Combination common ancestor with extant marine turtles (exclusive of of all the available data should produce the most defendable Dermochelvs) than with Dermochelys or Protostegct This hypothesis of cheloniid phylogeny to date. definition agrees with Hirayama (1994) and differs only slightly from Gaffney and Meylan ( 1988) who recognized SYSTEMATICS an Osteopygidae (Osteopyginae here and rn Zangerl, 1 953,, l9l l, Fastovsky, 1985). The three major groupings of Chel,oniidae. This is a well-known family of marine Cheloniidae (Osteopyginae, stem cheloniines, Cheloniinae) turtles comprised -of five living genera (Natator,, Chelonia, are discussed below in detail. Pnnunla AND FasrovsKy Cheloniid Phylogeny - 549

a.) (6 H Perhaps the most interesting attribute of these stem qJ OJ€ OJ o.o 6 c6 6 "r{ cheloniines is the disparate palatal and dentary morpholo- (D 0.) rd q) (6 .35 (6 !^E-} 'r, }: .r< >, 3- 'n qJ \H' ro bD 'Fo bo gies that they exhibit. Prior --r q) : to the appearance of these taxa ^H!F\ x 91 u X \.t t"E" { t P-,. tr O* s -f)Ll in the fossil record, all well-known (To.uochelt,s, - '-r O o R-baU cheloniids H o OF o .v* OJ .t-r >( 0.) S d9 E 4, q) G) .L' /^ L< OJ *HXk *.i V) Ctertochells, and the osteopygines) exhibit flat palates and Su si A &A 5 dentaries of varying lengths. Each of the Eocene stem cheloniines possess a different morphology. The symphysis Cheloniidae of Puppigerus does not extend beyond the forarnina dentofaciale majus, as it does in osteopygines (Zangerl,, (a) (b) l9l l), but Puppigerus does have a secondary palate equal to Figure 1. Hypotheses of the phylogenetic relationships of sea the extent of that in osteopygines . E,oclrelorte. however. turtles (Chelonioidea) showing alternative relationships of lacks a secondary palate (Casier, 1968), and this has been (a) To.rochelys and Ctenocltelys to Cheloniidae: Gaffney and interpreted as a reversal (Gaffney, 1979; Hirayamzl, 1994). Meylan, 1988: (b) Hirayama,, 1994. Finally . Argillochelys retains the secondary palate, but with a Osteoplginae This subfamily is composed of two groove to receive a markedly ridged dentary (the ridge occur- taxa, Osteopl,gis (Late Cretaceous) and Ercluelinnesicr (Early ring along the syrnphysis). A ridged dentary is not restricted to Eocene)., both known from cranial and post-cranial material. Argillrtchehs; some lowerjaws from the Late Cretaceous of The skulls of osteopygines are easily identifiable due to their New Jersey (AMNH 14205), not referable to any known extremely long secondary palates and broad, flat, dentaries. taxon, also have an elevated symphysis. Fr"rrthermore, some Like Toxocltel_r,s and Ctenochelts, osteopygines were once cheloniine taxa have a symphyseal ridge on the dentary. considered to be members of the Toxochelyidae (Zangerl, Chelortiinae This subfamily is defined here as those 1953,l97l). This assignment was based on the presence of turtles that share the most recent common ancestor of the wide plastra and curved humeri, characters that are now living cheloniids. S.r'//o t't't"Lt,s, a cheloniid with considered to be prirnitive for cheloniids (Gaffney and pseudodont dentition, has recently been assigned to this Meylan, 1988). Hirayama (1994) inferred that, like group by Hirayama (1994). He hypothesized that Syllomus Toxocltelys and Ctenochells,, osteopygines retain movable and Natator form a monophyletic sister group to the rest ol' articulations in the first and second digits (suggesting that the cheloniines. they lack the pelagic adaptation of a well-developed paddle); Some morphological analyses have divided the however, osteopygine specimens which retain these ele- Cheloniinae into two groups, Chelonini (Cheloniu and ments are not known to us. Eretmochely,s; Zangerl, 1958) and Carettini (Carerra and Fastovsky ( 1985) used cranial characters, such as a Lepiclochelys; Gray, 1825; as Carettidae). A rnorrophyletic basioccipital depression and the presence of a secondary Chelonini has been based primarily on characters which are palate, to ally the osteopygines with Recent cheloniids now considered to be primitive for cheloniines (e.g.', Al1 (Cheloniinae). Because the taxa included both have long elevated symphysis, a wide angle fonned by the scapular palates and the coinciding extensive symphyses, the processes, and the number of peripherals and pleural scutes; monophyly of osteopygines has never been challenged. Zangerl, 1958; Gaffney and Meylan, 1988). However, a However, the dentaries of turtles are notoriously plastic (Gaffney and Meylan, 1988) and the shells of the two taxa are quite different; Osteop.ygis possesses an ex- tremely ankylosed shell with no post-nuchal or costo- peripheral fontanelles whereas Ercluelinnesia has both (Fig. 2). For these reasons the monophyly of the Osteopyginae requires testing. Eocene Stem Cheloniines The subfamily "Eochelyinae" was established by Moody ( 1968) to include three taxa from the Eocene of Europe (Argilloc'hells, Eochelone, Puppigerus), each with character distributions intermediate between the osteopygines and cheloniines. Eocene stem cheloniines possess some of the presumed pelagic adaptations of the Cheloniinae (the loss of movable articulations in the manus, straight humeri, elongate and/or reduced plastra), but an unambiguous character has never been proffered which unites even two of these taxa into a monophyletic group (Gaffney and Meylan, 1988;Hirayama, 1994). Even Moody's (1914) emended diagnosis of the a character be subfamily did not include single that could Figure 2. Carupaces of two osteopygine turtles: (a) Osteopt,gis considered uniquely "eochelyine." ( after Zangerl. I 95 3 ); (b) Ercluelirtrte,sict (afier Zangerl,, 197 l). 5:tt CsEr-oNrnN CoNSERVATToN AND BroLoGy, Volmne 2, Ntuttber 4 - 1997 phe netic analysis by Carr (1942) suggested that the Chelonini (Hirayaffi?, 1994). Toxochelys lacks a second- is paraphyletic, with Eretmochely's being more similar to the ary palate, while Ctenochelys possesses a less extensive Carettini than to Chelonia. This is supported by serological shelving than any of the members of the ingroup except for (Frair, 1977 ) and molecular evidence (Bowen et al., 1993; Eocltelone, which lacks a secondary palate (Casier, 1968). Dutton et al., 1996). OsteopJ'gis, Ercluelinnesia, and Puppigerus have extensive The monophyly of the carettines, on the other hand, has secondary palates. long been well established. Caretta and Lepidochelys share 2. Foramen palatinum posterius: (0) wide; ( 1) greatly a unique shell morphology characterized by the presence of reduced; (2) lost (ordered). The presence of a foramen additional carapace elements. The presence of a dentary palatinum posterius is primitive for chelonioids. Foramina without ridges is also a presumed synapomorphy for are present in Allopleuron, an undescribed basal protostegid carettines. (Hirayaffi?, 1994), and the basal cheloniids Toxochel-r-'s and Ctenochelys, but are lost in all advanced cheloniids and CHARACTER ANALYSIS dermochelyoids. 3. Contact of vomer with premaxillae: (0) broad; (1) Our data set consists of 24 osteological characters ( l0 reduced. A broad contact of the vomer with the premaxillae cranial, 9 non-shell post-cranial, 5 shell) for 13 taxa. Char- is primitive for chelonioids. Fastovsky ( 1985) noticed that a acter polarity is established by reference to To.roc:helys and reduced contact is present in most of the Recent taxa (all Ctenochelys as a paraphyletic outgroup for the rest of except Notcttor'). Caretta has lost the contact entirely and the Cheloniidae. The utility of these taxa for the problem inves- maxillae contact each other in palatal view. tigated here is corroborated by Fastovsky ( I 985 ) and Gaffney 4. Foramen caroticum laterale much larger than fora- and Meylan (1988). Through his inclusion of the problem- men anterius canalis caroticus interni: (0) absent; ( 1) present. atical Allopleuron within the Cheloniidae, Hirayama (1994) This character is absent in basal dermochelyoids rejected To.rochely-s and Ctenochelys as a paraphyletic (Desnmtochelys, Allopleuron, and ; Gaffney, outgroup of the remaining cheloniids. Allopleuron is ex- l9l5; Hirayama, 1994) and protostegids, but seems to have cluded here. It exhibits presumably cheloniid features such been derived independently in Cheloniidae and as a secondary palate (which protostegids and dermochelyids Dermochelyidae. Within the Cheloniidae, the genera lack), but as noted by Hirayama (1994), "it shows a complex Toxocltelys, Ctenochelvs, and Eochelone lack this character combination of presumed primitive and apomorphic charac- (Gaffney, l9l5). ters often shared by the protostegids and dermochelyids." It 5. Orientation and position of the foramina orbito- would be an aberrant form in any of the present clades, and nasale: (0) with a vertical component; ( I ) horizontally trend- only an outgroup comparison, for all chelonioids, can im- ing and in extensive contact with the prefrontals. In most prove our understanding of its affinities. cheloniids the foramina orbito-nasale are vertically trending Analysis ofthe data was performed using the branch and ound and bounded by the prefrontals, maxillae, and palatines. algorithm ofPAUP 3. I . I (Swoffbrd, I 990). Characterdescriptions Osteopl:gis and Ercluelinnesia possess a unique morphol- and distributions are listed below, and shown in Table l. ogy; in these taxa the foramina are horizontally trending, are bounded on either side by the prefrontals (Fastovsky, 1985), Cranial and contact with the maxillae is lost. This character is probably related to the extension of the palatal region seen I . Secondary palate involving vomer: (0) absent; ( 1) in these turtles. present; (2) extensive (ordered). A secondary palate occurs 6. Dentary: (0) flat triturating surface; ( 1) lingual ridges twice within the chelonioids. Once in the Cheloniidae present; (2) ridge along length of symphysis (ordered). The (Fastovsky, 1985) and once in the basal dermochelyoid primitive condition for cheloniids is to have a relatively

Table 1. Matrix of character states and taxa. The primitive (plesiomorphic) state (0) for each character state is based on the assumption that Tottochelx-s and Ctenochel_rs are a paraphyletic outgroup to the rest of the Cheloniidae.

l0 ll T2 t3 T4 r5 t6 t1 l8 t9 2A 2l 22 23 24

To-rochelys 0 0 0 0 000 0 0 ? 00 00 0 0 0 000 0 0 00 Cterrcclre /.]rs I I 0 0 000 0 0 ? 00 00 0 0 0 000 I 0 00 2 2 0 I l0l 0 0 0 00 0t 0 0 ? 0?0 0 I 01 Ercluelirutesict 2 2 0 I l0l 0 0 00 01 0 0 ? 000 I 0 0? Argillochel\,,s I 2 0 I ?20 0 0 00 ?0 0 ? ? 0l Eocltelone 0 2 0 0 ?00 0 0 0? 00 0 I I 0l Pttpltigeru,s 2 2 0 000 0 0 00 00 ? I 00 Svllomus 2 , ?20 0 ? l0 ) ? 00 Ncttcttor 2 010 0 l0 2 00 Cltelonict 2 020 0 20 0 I 00 E retrnochehs 2 010 I 20 0 2 00 Carettct 2 000 I 0 20 0 2 t0 Lepiclochelys 2 000 I 0 20 0 2 l0 PnnHnu AND FnsrovsKy Cheloniid Phylogeny - 551

short, flat symphysis. The ingroup has a longer symphysis Primitively the femoral trochanters are separated by a dis- and many taxa possess an elevated symphysis (symphyseal cernible intertrochanteric fossa which extends onto the ridge). When it occurs, the syrnphyseal ridge occludes ventral surface of the bone (zangerl, 1953). In most with a corresponding groove in the palate. Hirayarna Recent taxa the trochanters are connected by a ridge, (1994) recorded Natator and Eretmochelys as having a obliterating the ventral extent of the intertrochanteric symphyseal ridge that is visible from lateral view, but fossa. Natator and Syllomus possess the ridge, but a this is just an extension of the lingual ridge which distinct notch remains between the trochanters . Zangerl extends partially onto the dentary. This is different from et al. ( 1988) hypothesized that the notch is intermediate the symphyses of Syllomus and Chelonia which are between the condition seen in primitive cheloniids and elevated along their entire length. Juvenile specimens of the ridged condition seen in the Recent taxa (and Lepiclocltelys exhibit elevated symphyses, but this con- S),llontus). This notch is also seen in juvenile specimens dition is lost in adults. of the remaining cheloniines. The symphyses of cheloniids vary in lengths from Hirayama (1994) stated that in ,S_1,//otnlts and Nrfiator very short in Toxochelys to extremely long in the trochanter major is much higher than the trochanter Ercluelinnesia. Puppigerus has an extensive symphysis, minor. This is definitely the case in S),llontas (see Weems, but the triturating surface does not extend beyond a line 1974), but Natstor has subequal trochanters and does not connecting the foramina dentofaciale majus as it does in appear to be significantly different from the other Recent Osteo1tygis and Ercluelinnesia (Zangerl, l9l l). In all taxa (see Zangerl et al., 1988). cases the extent of the symphysis closely matches the 14. Ventral knob on nuchal for articulation with the extent of the secondary palate (above) and it would be eighth cervical vertebra: (0) present; ( l) absent. A ventral redundant to code for both. knob on the nuchal is present in both members of the 7. Tomial ridge: (0) pronounced; ( I ) low. The cutting outgroup. Hirayama (1994)coded it as absent for Ctenochelys., surface of all cheloniids is pronounced with the exception of butZangerl ( 1953) indicated that the character is present in Osteopl;gi.r and Ercluelinnesia (Fastovsky, 1985). This is this taxon. Within the Cheloniidae, the only taxa which lack consistent with the large, flat jaws of these two genera which the knob are the osteopygines. are presumably indicative of durophagy. I 5. Tibial pit for pubotibialis and flexor tibialis internus 8. Surangular extending anteriorly onto dentary: (0) muscles (Zangerl, 1988; Hirayama, 1994): (0) absent; (l) absent; ( I ) present. Hirayama ( 1994) coded this character as present. This character is only found in the Eocene stem absent for Natotot', but Zangerl et al. ( 1988) stated that the cheloniines and Natator. It occurs as a deep pit on the surangular actually extends as far forward as the foramen ventromedial side of the tibia. dentofaciale majus. 16. Centra of the seventh cervical vertebra: (0) 9. Shape of vomer: (0) constant width; ( I ) variable procoelous; (l) platycoelous. Williams (l 950) recognrzed width. Toxocltely5 and Ctenochelvs both possess a vomer the platycoelous condition for modern taxa and Moody which, in palatal view, is of constant width. Therefore, the (1974) described its occuffence rn Puppigerus. presence of a vomer of variable width is probably a derived 17. Articulations of first and second digits: (0) movable; feature within the Cheloniidae. ( I ) immovable. 10. Prefrontal scutes: (0) one pair; (l) two pairs. The I 8. Straight humerus with V-shaped or triangular lateral preservation of sulci on skulls allow for the coding of this process: (0) absent; (l) present. Hirayama (1994) recog- character for some fossil taxa. ntzedfive humeral morphologies among chelonioids.' two of which occur in cheloniids. The "toxochelyid" humerus is Non-Shell Post-Cranial characterized by a nalrow, curved shaft. The "cheloniid" humerus is straighter and the V-shaped or triangular lateral I l. Dorsal process of scapula forming relatively wide process is more distal. The "cheloniid" morphology is shared angle with acromion: (0) absent; ( I ) present. The primi- by stem cheloniines, Cheloniinae, and the dermochelyoid tive condition is to have a roughly perpendicular angle AllopleLtron. between the scapular prong and the acromion. Wider 19. Coracoid length in relation to humerus: (0) angles, upwards of I 10", occur in S1://ptt'tLts and the non- shorter; ( 1) longer. carettine Recent taxa. 12.Metischial processes: (0) pronounced; (l) reduced. Shell In Recent taxa the metischial processes are significantly reduced. This is exemplified by Cltelonia in which they 20. Elongated plastron with a broad bridge: (0) absenr: are almost absent. Of all the Recent tax a, Natalor has the ( I ) present. The outgroup condition is a cruciforrn plastron largest metischial processes, but they are much reduced that is wider than long. This primitive condition was once from the condition seen in Toxochelys and Ctenochelys used to unite the Toxochelyidae (Zangerl, 1953), now con- (Zangerl., 1953). sidered paraphyletic. 13. Femoral trochanters: (0) separated by a fossa; (1) 21. Rib-free peripherals: (0) anterior and posterior ro separated by a notch1' (2) connected by a ridge (ordered). ribs; ( I ) between seventh and eighth ribs; (2) between sixth and 552 CHeI-oNtnN CoNSERVATIoN AND Brorocy, Volume 2, Number 4 - 1997

Osteopyginae \ l.,l (n (nT _ (n\ $s\) t+. \ ri.r- .s$ qlFS q) F F * (a S)qJ ooS !q)(J "S >).= .Pp $r "s\Jh s b SEs E t8 H Fil-'r \, *- dS o *bs*r'ri r\XqJ +r bs E $9" >, FU Oru t{la.{ EsSS T


Figure 4. Phylogenetic relationships of 13 cheloniid sea genera based on 24 characters (see Table l). This cladogram represents a consensus of 12 equally parsimonious hypotheses (49 steps, CI = 0.75).

phylogenetic relationships are outlined with these problem- atical affinities left unresolved. A monophyletic Osteopyginae is based chiefly on char- acters of the palate and dentary. Acceptance of this hypoth- esis requires that, at least within the osteopygine clade, there , 20cm r is the potential for variation in the shell. Despite their putative plasticity, shell characters apparently do not Figure 3. Carapaces of three cheloniine taxa exhibiting points of hinder rib insertion; (a) Chelonia; (b) Eretmochelys; (c) Careita (after the elucidation of phylogenetic relationships suggested by Zangerl, 1958). Arrows indicate the rib-free peripheral bone. cranial and non-shell post-cranial characters in cheloniids. - Ultimately, our hypotheses on the relationships of seventh ribs (unordered). With the exception of Chelonia, Osteopyginae and Cheloniinae to each other and to the cheloniines possess a unique shell morphology. In Chelonia Eocene stem cheloniines are in agreement with Hirayama the rib-free peripheral lies between the seventh and eighth (1994) and Gaffney and Meylan (1988), phylogenies in costal ribs (Fig. 3a). In Natator, Eretmochelys, Caretta, and which skull characters weigh heavily. Some shell characters I'epidochelys there is a rib-free peripheral between the inser- presented here support cranial and appendicular-based hy- tion points of the sixth and seventh costal ribs (Figs. 3b, 3c). potheses of cheloniid phylogeny. For example, the elonga- 22. Post-nuchal fontanelles: (0) present; (l) absent. tion of the plastron unites stem cheloniines and Cheloniinae Post-nuchal fontanelles are present in the outgroup and and the additional carapace elements are a synapomorphy Erquelinnesia (Zangerl, 1971). Adult specimens of the re- for a monophyletic Carettini. maining taxa lack these fontanelles, but they are present in Hiray ama ( I 99 4) hypothe s i zed that A r g ill o c helys s hou I d juveniles of the Recent taxa. be considered a sister taxon to the cheloniines based upon the 23. Additional carapace elements: (0) absent; ( 1) present. presence of a ridged dentary, but Puppigerus and the The primitive condition for cheloniids is to have eleven cheloniines both possess a platycoelous seventh cervical peripheral bones and four pleural scutes on each side of the vertebra. Thus, we cannot support an Argillochelys carapace. Caretta and Lepidochelys have at least twelve Cheloniinae clade over a Puppigerus - Cheloniinae clade. peripheral bones and five pleural scutes on each side. As a result, the Eocene stem cheloniines remain completely Lepidochelys also exhibits extensive fragmentation of the unresolved. neural bones (Zangerl and Turnbull, 1955). There is no evidence to suggestthatNatatorandSyllomus 24. Shape of pygal: (0) notched posteriorly; (l) form a monophyletic group. Hirayama (1994) allied these not notched. taxa on the relative positions of the femoral trochanters, but the condition rn Natator rs not different from the rest of the RESULTS Cheloniinae. Additionally, our analysis suggests that

Syllomus cannotbe considered acheloniine. Instead , Syllomus Using the 24 characters, no fewer than 12 equally is the sister taxon to Cheloniinae (Fig. 4). parsimonious cladograms can be produced. The multiple The hypothesis presented here also differs from those of cladograms result from an inability to resolve the phyloge- previous workers using morphology in terms of the phyloge- netic relationships of the Eocene stem cheloniines and the netic interrelationships of the cheloniine taxa. In the past positions of Natator and Chelonia. In Fig. 4, cheloniid decade, many different views of cheloniine phylogeny have Pnnueu AND FnsrovsKy - Cheloniid Phylogeny s53

( taxon to the rest of the Cheloniinae (Fig. 5a). (A >j Dutton et al. >, .so (n ('J E ( 1996) examined the >, LS q) .s* ND4-leucine IRNA region of mtDNA o H S Sg qJ \-J -- U sds ct *r\ o -o\SE and combined their data with the cytochrome b data from EF\FiFT EI' ucs *.r SS P U tat S +rr cs\. N'ilqJ z .sp q) r-J cs Bowen et al. (1993). This "total molecular evidence" Rt z6' L '\\J U Fl- hl Q).\ r{ (Fig. )_l \,/ 5c) approach suggested that Chelortia rs the sister taxon to a clade comprised of Natator, Eretmochelys, and the Carettini. In this case, the rib-free peripheral pattern of Natator, Eretmochelys, and the Carettini would have only evolved (D (r) ( 5 5 L.S ('I >, once. Becaus e Natator retains characters considered primi- q) q) 5 q) cs qJ L U qJ tive for Cheloniinae (an intertrochanteric notch and a tibial FI !q) o o €sH (.) Hcs o S+r o F \) *r ts N o .H A\ pit) the osteological data presented here are unable to resolve a.) v qs qJ zo \ l\\L L q) \)s which taxon, Natator or Chelonia,, is the sister taxon to the U z ru L t-l U ra rest of the Cheloniinae. Because our results are equivocal, we did not combine the osteological data with the molecular data. Our results agree with Bowen et al . (1993) and Dutton (d) et al . (1996) in suggesting that neither Natator nor Chelonia forms a monophyletic group with each other or any other Figure 5. Recent hypotheses of the phylogenetic relationships of Recent unless all cheloniines are included. Cheloniinae: (a) based on morphology (Zangerl et al., 1988); (b) based on morphology (Hirayama, 1994); (c) based on cytochrome CONCLUSIONS b and ND4-leucine sequences of mtDNA (Dutton et al., 1996); (d) based on morphology (present study). Our data support a monophyletic Osteopyginae, but like been presented (Figs. 5a-d).As noted above, the monophyly Gaffney and Meylan (1988) and Hirayama (1994), chal- of the Chelonini is based on morphological characters (ridged lenge the monophyly of the "Eochelyinae." Until further dentary, wide scapular angle) which are now considered to analysis of these taxacan be performed, it is best to continue be primitive for Cheloniinae, but the polarity of these to refer to the "Eochelyinae" as Eocene stem cheloniines. In characters cannot be recogntzed unless Syllomus and our study, the inclusion of shell characters did not hinder, but Natator are included in the analysis. Hirayama (1994) facilitated, the elucidation of phylogenies established using was the first to include these taxa and in doing So, cranial and non-shell post-cranial characters. questioned the monophyly of Chelonini. He suggested Cheloniines share a most recent common ancestor as that Eretmochelys is the sister-taxon to a Chelonia early as the Middle Miocene. However, whether as a result Carettini clade (Fig. 5b). of their long generation times (Ehrhart and Witham , 1992: Our hypothesis (Fig. 5d) suggests that Eretmochelys rs Zug et al., 1997; Parham and Zug, in press) or relative lack the sister taxon to the Carettini, based on but one character, of isolation in their marine environment, cheloniines have the presence of two prefrontal scutes. It should be noted, evolved into an assemblage of turtles characterized by few however, that an Eretmochelys - Carettini clade is supported morphological differences. Consequently, many of the os- by non-osteological evidence. Serological (Frair,, l9l9) and teological characters used to establish relationships among molecular (Bowen et al., 1993; Dutton et al., 1996) studies fossil taxa are too stable to be useful in the estimation of a of living taxa corroborate a paraphyletic Chelonini by indi- cheloniine phylogeny. Osteological characters useful in cating that, genetically, Eretmochelys is more similar to phylogeny reconstruction (e.g., in the establishment of carettines than to Chelonia (Fig. 5c). Carettini) generally are derived from the dentary and cara- Bowen et al. (1993) proposed that Eretmochelys be pace, due to the plasticity of these regions. A detailed considered a carettine, corroborating the earlier conclusions cladistic analysis of non-osteological characters (e.g., scutes, of Carr (1942) and Pritchard and Trebbau (1984). If we behavior, additional genes) for living taxa may be necessary define the Carettini as those taxa that are more closely in order to establish well-defined clades within the related to Caretta than to Chelonia or Natator, then Cheloniinae. Eretmochelys should be considered a carettine. Given the unique features shared by Lepidochelvs and Caretta and Acknowledgments absent in Eretmochelys (low scapular angle, additional carapacial elements, and broad, flat, dentaries), it might be This study would not have been possible without the best to define Carettini as those taxawhich are more closely assistance of the Brooks family, C.R. Shoop, and P.D. related to Caretta and Lepidochelys than to any of the other White. Visits to the New Jersey State Museum and Ameri- Recent genera (a node-based definition; de Queiroz and can Museum of Natural History were facilitated by D.C. Gauthier, 1990). Pamis and E.S. Gaffney, respectively. C. Kishinami of the Previous morphological analyses (Zangerl et al., 1988; Bernice P. Bishop Museum (Honolulu), J. Rosado of the Hirayaffi&, 1994) have suggested that Natator is the sister Museum of Comparative Zoology (Harvard), and P. Couper 554 CUEI-oxtaN CoNSERVATIoN AND Brolocy, Volume 2, Number 4 - 1997 of the Queensland Museum (Brisbane) were helpful regard- Spec. Vol 35A: 157 -219. ing information on specimens of Natatortntheir collections. GnRy, J.E. 1825. A synopsis of the generaof and amphibia, with E.S. Gaffney and P.A. Meylan provided helpful comments a description of some new . Ann. Philos. (2) l0: 193-217 . HtnRvauR, 1994. on an earlier draft of this manuscript. Much of the research R. Phylogenetic systematics of chelonioid sea turtles. Island Arc 3:210-284. was done at the Smithsonian Natural Museum of Natural Lrveus, C.J., GvuRrs, E., RNo MnLeR, J.D. 1988. Reassessment of the History with the generous assistance of G.R .Zug.This study taxonomic status of the genus NatatorMcCulloch, 1908, was partly supported by a grant from Molson Breweries with a redescription of the genus and species. Trans. Roy. Soc. USA, Inc., a university of Rhode Island Undergraduate South Aust. ll2:l-9. Research Award, and by the Department of Geology, Uni- Moony, R.T.J. 1968. A turtle,, Eochelys crassicostata (Owen) fiom versity of Rhode Island. the London Clay of the Isle of Sheppey. Proc. Geol. Assoc. Lond. 79:129-140. LITERATURE CITED Mooov, R.T.J. 1914. The and morphology of Pr,tppigerus camperi (Gray), an Eocene sea-turtle from northern Europe. Bull. BowEN, 8.W., NElsoN, W.S., AND Avrsg, J.C. 1993. A molecular Brit. Mus. Nat. Hist. Geol. 25(2):153-186. phylogeny for marine turtles: trait mapping, rate assessment, and PnnHnna, J.F., AND ZuG, G.R. In press. Age and growth of loggerhead conservation relevance. Proc. Natl. Acad. Sci. USA 90:5514-5511. sea turtles (Caretta carettct) of coastal Georgia: an assessment of CRRR, A.F., Jn. 1942. Notes on sea turtles. Proc. New England Zool. skeletochronological age estimates. Bull. Mar. Sci. Clurb 2l:l- 16. PnrrcHnno, P.C.H., RNn Tnpeenu, P. 1984.The Turtles of Venezuela. CRstER, E. 1968. La squelette cephalique de Eochelone brabanticctL. Soc. Stud. Amph. Rept. Contrib. Herpetol. No. 2,403 pp. Dollo, du Bruxellien (Lutetian Inferieur) de Belgique, et sa Sworrono, D.L. I 990. PAUP: Phylogenetic Analysis Using Par simony, comparaison avec celui de Chelonia mydas Linne. Bull. Inst. Roy. Version 3.1.1. Champaign, IL: Illinois Natural History Survey. Sci. Nat. Belgique 44:l-22. WpEvrs, R.E. 1974. Middle Miocene sea turtles (Syllontus, nE QuerRoz, K., AND GRurureR, J. 1990. Phylogeny as a central , Psephoplrcrus ) from the Calvert Formation. J. principle in taxonomy: phylogenetic definitions of taxon names. Paleont ol. 48(2) :27 8-303 . Syst. Zool. 39:307 -322. WIt-ltnvs, E.E. 1950. Variation and selection in the cervical central Durroru. P.H.. DAVrs, s.K., Gunnnn, T., AND owENs, D.1996. Molecu- articulations of living turtles. Bull. Amer. Mus. Nat. Hist. 94:505-562. lar phylogeny for marine turtles based on sequences of the ND4- ZnNcEnL, R. 1953. The vertebrate fauna of the Selma Forrnation of leucine IRNA and control regions of mitochondrial DNA. Molec. Alabama. Part IV. The tuftles of the family Toxochelyidae.

Phylo. Evol. 5(3):51 I -521 . Fieldiana Geol. Mem. 3:135-271 . Ennsnnnr, N.M., AND WrrHRvr, R. 1992. Analysis of growth of the ZnrucEnl, R. 1958. Die oligozdnen Meerschildkroten von Glarus. green sea tuftle (Chelonia nryclcts) in the western central Atlantic. Schweiz. Palaeontol. Abh. 7 3:l-56. Bull. Mar. Sci. 50:275-281. ZnNcEnL, R. 1971. Two toxochelyid sea turtles from the Landenian Fn,srovsrv, D.E. 1985. A skull of the Cretaceous chelonioid turtle sands of Erquelinnes (Hainaut) of Belgium. Mem. Insr. Roy. Sci. O,steopt'gis and the classification of the Osteopyginae. Invest. New Nat. Belgiqu e 169:l-32. Jersey State Mus. 3:l-23. ZnNcEnL, R., AND TunNeur-1, W.D. 1955 . Procolpochelys grcutclaeva FostER,, D.E. 1980. Osteopl,gis sp., a marine turtle from the Late (Leidy), an early carettine sea turtle. FieldianaZool.3T .345-382. Cretaceous of California. PaleoBios 34, l5 pp. Znrucnnl, R., HeNoRrcKSoN, L.P., AND HENpnrcKSoN, J.R. 1988. A FRntR' W. I 979. Taxonomic relations among sea turtles elucidated by redescription of the Australian flatback sea tu r1tle,Natcttor cleprerrur. serological tests. Herpetologica 35(3):23 9-244. Bishop Mus. Bull. ZooL 1: I -69. GnppNev, E.S. 1975. A phylogeny and classification of the higher Zuc, G.R., KRLS, H.J., AND LuzaR, S.J. 1997 . Age and growth in wild categories of tuftles. Bull. Amer. Mus. Nat. Hist. 155:387-436. Kemp's ridley sea turtles Lepidochelys kempii from GappNEv, E.S. 1919. Comparative cranial morphology of recent and skeletochronological data. Biol. Conserv. 80:26 I-268. fossil turtles. Bull. Amer. Mus. Nat. Hist. 164:65-316. Receivecl: 27 February 1997 GnnnruEy, 8.S., AND MEvlnN, P.A. 1988. A phylogeny of turtles. In: Reviewecl: 3l August 1997 Benton, M.J. (Ed.). The Phylogeny and Classificarion of the Revisecl ancl Acceptecl: 24 September 1997 Tetrapods, volnme I: Amphibians, Reptiles, Birds. Syst. Assoc.