The Cervical and Caudal Vertebrae of the Cryptodiran Turtle, Melolania Platyceps, from the Pleistocene of Lord Howe Island, Australia

AMERICAN MUSEUM

Nornltates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2805, pp. 1-29, figs. 1-22, tables 1-3 January 30, 1985

The Cervical and Caudal Vertebrae of the Cryptodiran Turtle, Melolania platyceps, from the Pleistocene of Lord Howe Island, Australia

EUGENE S. GAFFNEY'

ABSTRACT Meiolania platyceps, a cryptodiran turtle from platyceps, but a conservative reconstruction based the Pleistocene of Lord Howe Island, Australia, on degree of serial variation and comparison with has cervical and caudal vertebrae exhibiting a Recent turtles, yields a total ofat least ten caudals number of phylogenetically interesting features. plus tail club. The caudals are all opisthocoelus The cervicals have fully formed central articula- and have well-developed haemal spines; charac- tions with a formula of (2( (3( (4) )5) )6) )7) )8), ters that are primitive for cryptodires. Meiolania interpreted as a synapomorphy of eucryptodires. has an ossified tail club consisting ofa conical layer Free ribs are present on cervicals two through six, of bone surrounding and fused to the terminal a retained primitive character in Meiolania, lost caudals. Proganochelys has a similar, but not iden- independently in other eucryptodires, baenids, and tical, terminal ossification, and the presence of a pleurodires. The presence in Meiolania ofcervical tail club is interpreted as a retained primitive fea- ribs articulating with paired intercentra shows that ture in Meiolania, being lost independently in the supposed vestigial ribs identified by previous pleurodires and other cryptodires. The tail of authors in the neck of Recent turtles are probably Meiolania platyceps has a series of ventrally in- intercentra and not rib remnants. The neural spines complete rings, contrasting with the complete rings of cervicals seven and eight articulate with the found in "Meiolania" oweni and Niolamia argen- nuchal bone; this is interpreted as an autapomor- tina. The vertebral features ofMeiolaniaplatyceps phy, occurring independently in chelonioids. are consistent with its hypothesized systematic po- Complete tails are unknown for Meiolania sition as a primitive eucryptodire.

INTRODUCTION The cervical and caudal vertebrae ofMeio- and phylogenetically interesting features. The lania platyceps show a number of unusual well-developed cervical ribs and the tail club

I Curator, Department of Vertebrate Paleontology, American Museum of Natural History.

are found in the Triassic turtle, Proganoche- lys, and suggest the presence ofvery primitive chelonian features in this extinct turtle. The purpose of the present paper is to describe and systematically assess these and other features in the vertebral column of Meiolania platyceps. Further information on Meiolania platyceps, including geographic and geologic occurrence, previous work, and cranial morphology can be found in Gaffihey (1983). Lists of specimens examined are also available in FIG. 1. Meiolaniaplatyceps, AM F:61105. Left, anterior view of atlantal centrum; right, posterior that paper. view of fused atlantal intercentrum and right atlantal neural arch. See figure 5 for views of the ACKNOWLEDGMENTS articulated atlas. Much of this work was undertaken while I was a Visiting Curator at the Australian Mu- seum in 1980. I am indebted to that insti- F:5536 (centrum) and AM F: 18315 (centrum tution for its support and for the cooperation fragment; figured in Anderson, 1925, pl. 37, ofits staff, particularly Dr. Alex Ritchie. Dr. fig. 1). John Pickett, Mining and Geological Mu- As in other turtles (and amniotes generally) seum, Sydney, and Dr. Angela Milner, British the atlas of Meiolania (figs. 1, 5-8) consists Museum (Natural History) both aided me by of four elements: paired neural arches an- permitting me to study their collections. More terodorsally, an intercentrum anteroventral- extensive acknowledgments relating to the ly, and a centrum posteriorly. The Meiolania Meiolania project can be found in Gaffney atlas is generally similar to the generalized (1983). chelonian atlas, as in Proganochelys, but it This paper was supported in part by NSF differs in being somewhat shortened antero- grant DEB 8002885. Ms. Mary Jane Spring posteriorly and broadened laterally. did figures 5, 11, 14, 15, and 21. Most of the The chelonian neural arch (see Kasper, photographs were taken by me. Mr. Frank 1903), can be roughly divided into a dorsal Ippolito greatly enhanced the value of the portion covering the neural cord and having paper with his work on all the figures and photographs. ABBREVIATIONS AM, Australian Museum, Sydney BM(NH), British Museum (Natural History), Lon- don MM, Mining Museum, Sydney SMNS, Staatliches Museum fir Naturkunde, Stuttgart ATLAS The specimens available for the atlas are as follows: AM F:49141 (complete, articulated), MM F: 13825a (complete, articulated; figured in Owen, 1888, pls. 31, 32; Gaffney, 1983, figs. 42, 43), AM F:57984 (left arch, FIG. 2. Meiolania platyceps. Medial views of centrum fragment), AM F:61105 (arch and right atlantal neural arches, anterior to the left. intercentrum), AM F:18835 (arch), AM Left, AM F:57984; right, AM F:18835. Cu 00 o 4 0 0 z3 0 .C C) Cu cn Cu 0' o

03 w GMC Cu ,OU (A o 0 C) I - 0 0 C) C) 10'0. '0 -uCu

C) '0 Cur 0 Cu

Cu C) $- C) 0 - N. 0 C.) ni '0 o gr '0 04 0 ._ oC); m Co *1. -u 0~ '0 C O - Ed C) ~0 .eC) 10 '0 2 Cu (A 3 hIIt C) 83 _ 0

Q o Cu-_ '0 Cu3 0 Cu4 C) 0 0 en .-C.._ nu z C-4 o a toCu'34b (10 CA Cu 6I 0 0 r Cu

Cu~~ ~ ~ ~ ~ ~ ~

(A 0

N~ ~ ~ ~ ~ ~ 0

04 0 t Cd N

0 CAC) C) C)~-. 0 I-4 Z(U 01 4 AMERICAN MUSEUM NOVITATES NO. 2805 a posteriorly directed postzygapophyseal Neural Spine flange articulating with the axis, and a ventral Postzygapophysis portion bearing articular facets for the occipital condyle, the atlantal centrum and the atlantal intercentrum. In Meiolania platyceps Prezygapophysis (figs. 1, 2) the neural arch has a particularly broad dorsal portion that inclines medially to lie almost horizontally above the space for the neural cord. The neural arches do not Diapophysis meet in the midline in known specimens. The postzygapophysis has a well-developed articular facet for the prezygapophysis ofthe axis. The very large proportion ofthe ventral part Parapophysis ofthe neural arch is occupied by the articular facet for the centrum of the atlas. This facet Centrum faces posteromedially and lies approximately at right angles to a smaller facet facing an- FIG. 3. Diagram showing morphologic fea- teromedially that articulates with the occip- tures of cervical vertebrae. Seventh cervical of ital condyle of the skull. Ventral to the facet Meiolania platyceps. with the centrum, the neural arch has a small contact area with the atlantal intercentrum. The lateral surface of the neural arch has a variation in other areas and I interpret this transverse process with a broad base but a as another example. relatively shallow lateral extension. The An unassociated neural arch, AM F: 18835 transverse process extends more than in Pro- (fig. 2) is not so easily interpreted as individ- ganochelys and most cryptodires but less than ual variation, but I see no other alternative. in pleurodires. This is a right neural arch, roughly similar to The transverse process of the neural arch the others but about a third narrower in lat- extends ventrally to a variable extent in the eral view. The dorsal process that extends direction of a short dorsal extension of the over the neural cord is a cylindrical process intercentrum. In MM F: 13825a (see Gaffney, rather than a broad plate as in the other spec- 1983, fig. 42, middle, for a stereophotograph) imens. The ventral portion of the arch has these processes join and enclose a small fo- the central and occipital facets but much of ramen, whereas in AM F:57984 (figs. 2, 5) the surrounding bone is reduced in compar- they do not meet but end in surfaces that ison to other neural arches. The arch is clearly appear to have been finished in cartilage. It Meiolania; it has no particular similarities to is possible that a cervical rib participates in sea turtles or pleurodires. It could be from a the ossification of this area but no evidence juvenile but the dorsoventral dimension is of a discrete element has been seen. the same as in AM F:57984. The contact between the neural arch and The intercentrum ofthe atlas in Meiolania intercentrum is fused on both sides of MM platyceps (figs. 1, 5) is a crescentic element, F:13825a and on the right side of AM connecting the bases ofthe neural arches, and F:61105 (fig. 1). The other specimens (in- forming the ventral third ofthe atlantal ring. cluding AM F:57984 and AM F:49141, which It articulates with the occipital condyle an- are both well preserved) show no indication teriorly and with the atlantal centrum pos- of fusion. Fusion ofthese elements occurs in teriorly. The intercentrum of Meiolania is pleurodires and trionychids and it may be narrow transversely, being distinctly wider interpreted as characteristic of certain taxa than long, in contrast with most turtles which within these groups, but I do not think its have a more equidimensional intercentrum. presence in Meiolania platyceps has system- The intercentrum of Proganochelys, how- atic significance. Meiolania platyceps is char- ever, is similar to Meiolania, suggesting that acterized by a high degree of morphologic this is the primitive chelonian condition. The 1985 GAFFNEY: MEIOLANIA PLATYCEPS 5 intercentrum is fused to the neural arch in Meiolania platyceps has well-developed MM F: 13825a and on the right side of AM cervical ribs (figs. 9, 11) that are freely artic- F:61105 (fig. 1) but it is unfused in the re- ulated on cervicals two to seven. There is no maining specimens. There is a concavity on free rib on cervical eight. The free ribs are the posterolateral corner of the intercentrum double-headed and articulate with a laterally directly below a similar area on the neural projecting transverse process (diapophysis) arch, discussed above. dorsally and a less-pronounced ventral artic- The atlantal centrum (figs. 1, 5) is relatively ulation (parapophysis). The diapophysis is short anteroposteriorly but broad and equi- about equally developed on cervicals two to dimensional laterally and dorsoventrally. The seven and is situated at the junction of the anterior articulation is broadly convex, end- centrum and neural arch midway along the ing in a low midline projection that faces the length of the centrum. The lateral margin of shallow concavity on the condylus occipitalis the transverse process bears a posteroven- of the skull. The articular surface is roughly trally facing articular facet for the tuberculum T-shaped, the two dorsolateral areas articu- ofthe rib. The articular facet is always larger late with the posterior facets of the neural than the ventral parapophyseal articulation arches and the ventral area articulates with with the capitulum of the rib. The diapophy- the posterior facet of the intercentrum. The seal articulation decreases in area posteriorly posterior articular surface ofthe atlantal cen- to a slight extent. trum is concave and articulates with the con- The ventral rib articulation (fig. 9) is with vex centrum ofthe axis. The atlantal centrum an element that I interpret as an intercen- has a paired projection extending postero- trum. This element in Meiolania may be fused ventrolaterally that bears an articular facet to the centrum or it may be separated by a for the capitulum of the first cervical rib (fig. suture. The intercentra occur in pairs and may 11). The tuberculum of this rib articulates be fused on one side and separate on the other with the transverse process (diapophysis) of (as in the fifth cervical of AM F:57984). The the second cervical vertebra. intercentra are associated with the centra in CERVICAL VERTEBRAE a unique manner. The first intercentrum is TWO TO EIGHT single, C-shaped, and part of the atlas (see Figures 3-9 above), whereas the second are paired and attached (fused in all known specimens) to CENTRA: The centra of the cervicals of the atlantal centrum at its posterolateral mar- Meiolania platyceps all have fully formed ar- gins. Paired intercentra are also attached (by ticulations, with cervicals two and three opis- a suture in all known specimens) to the pos- thocoelus, cervical four biconvex, and cer- terolateral margins ofcervicals two and three, vicals five to eight procoelus. This pattern is but cervical four has no intercentra attached common in eucryptodires and also occurs in to it, instead cervical five has the intercentra some baenids (Chisternon). The anterior con- attached to its anterolateral margin rather vexity in the second cervical (axis) is less pro- than the posterolateral margin. This position nounced than the convexity in any of the change continues posteriorly and the sixth, other cervicals. The degree of development seventh, and eighth cervicals also have the ofthe central surface is about the same in the intercentra attached to the anterior margin. other cervicals. The ventral surface of the The intercentral attachment position is cor- centrum has two, low parasagittal ridges in related with the central articulation pattern cervicals two to four, they are indistinct in so that the intercentra always attach to the cervical five, and they are absent in cervicals margin of the centrum that has the concave six to eight, which have a curved ventral sur- articular surface. Thus cervical four, which face. The centrum main body is slightly con- is biconvex, lacks intercentral attachments. I stricted in the cervicals of Meiolania but not see no particular explanation for this corre- to the extent seen in baenids and pleurodires. lation. Another aspect of this serial change There is no indication ofdouble articulations in intercentra attachments are the rib artic- or saddle-shaped surfaces. ulations. In cervicals five through seven the 6 AMERICAN MUSEUM NOVITATES NO. 2805

FIG. 4. Meiolania platyceps. Right lateral view of articulated cervicals from atlas to eighth plus first thoracic. Anterior on right. Reconstructed neck based primarily on AM F:57984. tuberculum and capitulum of each rib artic- transverse process than any of the other cer- ulates with the diapophysis and parapophysis vicals, about three times in length. The pro- of the same vertebra, but for cervicals two cess curves posterolaterally. There is a par- and three the rib articulates with the diapoph- apophysis that is smaller than in any of the ysis of one vertebra and parapophysis of the other cervicals. The cervical rib appears to vertebra anterior to it, thus the two anterior be absent, the long transverse process bears ribs each articulate with two vertebrae and no evidence of being a fused rib. Progan- cervical four has no rib articulation. ochelys has fused ribs on cervicals six through Although the morphology ofthe transverse eight and these retain the dorsal and ventral process (diapophysis) is relatively stable from attachments with a foramen between them. cervical two to cervical seven, the par- In Meiolania there is no evidence of a per- apophysis (intercentrum) shows some vari- sistent rib shaft, only an elongate diapophy- ation along the column. Anteriorly, the par- sis. apophyses ofthe atlantal centrum, second and NEURAL ARCH: The neural arch is de- third cervicals, project posterolaterally and scribed in three general headings: prezyg- bear the rib articulation facet on the postero- apophyses, postzygapophyses, and neural lateral surface. The articulation facets are of spines (fig. 3). The prezygapophysis of the about the same size. The fourth cervical com- second cervical is small, much smaller than pletely lacks a parapophysis. Cervicals five the prezygapophysis ofany other cervical, and through seven have their parapophyses ex- articulates with the postzygapophysis of the tending anterolaterally from the anterior cen- atlantal neural arch. The prezygapophysis is tral margin (as described above) and the ca- missing on available specimens except AM pitulum articulation facet lies on the F:49141 where it can be seen preserved on posterolateral edge of the parapophysis. The both sides. It lies on a short projection from parapophyseal facet is smaller on the fifth the neural arch about midway up the arch. than on the third cervical and is more re- The prezygapophysis of the second cervical duced on the sixth and seventh cervicals differs from all other cervicals in facing dor- where it may only be a rugosity (as in AM solaterally rather than dorsomedially. This is F:57984). the usual situation for the axis in other turtles, The eighth cervical has a much longer including Proganochelys, as well as other am- 1985 GAFFNEY: MEIOLANIA PLATYCEPS 7 niotes. The prezygapophyses ofcervicals three margin. The third cervical has a slight boss to seven are similar to each other, they face for a neural spine and this is a larger, discrete dorsomedially with the facets on cervicals knob in the fourth cervical. The neural spine three and four facing slightly more anteriorly of the fifth cervical begins the development than in the others. The prezygapophyses are of a posterodorsal extension which becomes separated from each other to a greater extent best developed in cervical six. All the neural in cervicals three to six than in the other cer- spines have rugose dorsal surfaces, suggesting vicals, the prezygapophyses begin to approx- the presence of well-developed ligaments. imate each other in cervical seven, and they Cervicals five and six have posterior concav- are only barely separated from each other in ities, most extensive in cervical six, that are cervical eight. The surface area of the pre- also rugose and may have been involved in zygapophyseal articulation facet is relatively ligament attachment. The concavity on the small in the second cervical but uniformly neural spine of cervical six is divided down larger in the remaining cervicals. the midline by a longitudinal ridge in AM The postzygapophyses generally parallel the F:49141 but not in AM F:57984. The neural orientation and size of the prezygapophyses, spine of the seventh vertebra (best seen in they all face ventrolaterally with the farther AM F:49141) is high and compressed later- anterior ones more separated from each other ally in contrast to the anterior cervicals. Most than the farther posterior postzygapophyses. of the neural spine of the eighth cervical is The postzygapophyses ofcervical seven near- missing in available specimens, but from the lyjoin at their base. Unfortunately, the neural preserved material a thin, compressed spine spine and postzygapophyses ofcervical eight is indicated. are not preserved in any available specimen. The nuchal bone of Meiolania platyceps Whereas the prezygapophyses are borne on has two oval articular facets on the midline short processes extending anteriorly from the of its ventral surface; the top of the neural neural arch, the postzygapophyses are nearly spine of cervical seven articulates with the integral elements of the neural spine and are anterior facet and it seems likely that the not as discrete. eighth cervical neural spine articulates with One ofthe differences between living cryp- the posterior facet. These facets indicate a todires and pleurodires lies in the wide sep- movable joint rather than a sutured one, as aration of the zygapophyses in cryptodires in Proganochelys, which has a sutured artic- and their close approximation in pleurodires. ulation between the neural spine ofthe eighth Although there are exceptions due to the wide cervical but no articulation between the sev- diversity of vertebral morphology in turtles, enth cervical and the nuchal. Some turtles, it is possible to make broad comparisons. for example, chelonioids, have a close liga- Meiolania and Proganochelys both have zyg- mentous association ofthe eighth cervical and apophyseal positions that are intermediate the nuchal, but turtles which have well-de- between the two extremes seen in the living veloped neck retraction mechanisms seem to turtle groups and are similar to baenids. lack them. The neural spines ofthe cervicals in Meio- lania platyceps exhibit considerable serial variation. The neural arch of the second cer- CERVICAL RIBS vical virtually lacks a neural spine, instead it Figures 6-9, 11 is a broad, flat platform, that seems to artic- Meiolania platyceps is unusual in having ulate with the underside ofthe skull roof. The well-developed cervical ribs, elements absent skull roof of Meiolania has a flat area just in nearly all turtles. There are five free cer- posterior to the crista supraoccipitalis and vical ribs in Meiolania with a very small sixth anterior to the posterior skull margin (Gaff- that may be partially fused. The atlas may ney, 1983, figs. 38, 40). The second cervical have a small rib element associated with it fits in this area when the neck is articulated (see above) but it is not freely articulating nor (as seen in MM F:13825a; Gaffney, 1983, discrete in the available material. The first figs. 42, 43). The postzygapophyses of the well-developed rib belongs to the second second cervical extend just beyond the skull cervical (the ribs will be referred to by the 8 AMERICAN MUSEUM NOVITATES NO. 2805

atlas 2 3 4

ci) (41 @

FIG. 5. Meiolaniaplatyceps. Cervical vertebrae, AM F:57984, except for atlas centrum (AM F:6 1105) and second cervical (AM F:5547). 1985 GAFFNEY: MEIOLANIA PLATYCEPS 9

5 6 7 8

1.I

FIG. 5. Continued. 10 AMERICAN MUSEUM NOVITATES NO. 2805

FIG. 6. Meiolania platyceps. Left lateral view ofAM F:49 141, cervicals one through seven, in natural articulation. cervical they articulate with). This rib, and extends past the limits ofits cervical. The rib the two behind it, are the largest in the series, is acuminate and curves dorsally. The rib of extending posteriorly well onto the posterior the sixth cervical is shorter than any anterior cervicals. The rib of the second cervical dif- to it and is a bit more irregular in shape, fers from the others in being relatively broad though still triangular and tapering, its acu- and spatulate in shape rather than tapering minate point is relatively blunt. The seventh as in the more posterior ribs. The free cervical cervical ofAM F:57984 has a transverse pro- ribs of Proganochelys are also broad and flat cess (diapophysis) with an articular facet at rather than acutely tapering but they are not its end, whereas in AM F:49141 there ap- as broad as the second cervical rib of Meio- pears to be a small conical knob ofbone sep- lania. The free cervical ribs ofProganochelys arated from the diapophysis by a partial su- are also much shorter, not extending very far ture. All ofthis indicates that there is a cervical beyond the centrum. The third and fourth rib attached to the seventh cervical but that cervicals in Meiolania are very similar in size it is small and placed on the tip of the trans- and shape and would be very difficult to tell verse process (as restored in fig. 9). As I in- apart, one reason why the disarticulated ribs terpret the eighth cervical, it lacks a rib and in the collection are questionably identified has a large transverse process. It is possible as to cervical. that part ofthe transverse process consists of The rib of the fifth cervical is nearly half a fused rib. In Proganochelys the cervicals the length of the anterior three ribs but still two through five have free ribs, whereas the 1985 GAFFNEY: MEIOLANIA PLATYCEPS I1I

FIG. 7. Meiolania platyceps. Dorsal view ofAM F:49141, cervicals one through seven, in articulation as found. First four ribs on right side have been removed. more posterior ones do not. Cervicals six and cartilaginous intercentra but they were prob- seven ofProganochelys clearly have fused ribs, ably smaller than the ossifications in Meio- the tuberculum and capitulum are separated lania. by an opening. But the eighth cervical ofPro- Most ofthe cervical ribs (except the rib on ganochelys is similar to that in Meiolania in cervical two) have a small unfinished surface the very long transverse process that has no at the distal tip suggesting continuation in sign of distinct rib heads or foramen. Thus, cartilage. the long transverse process in the eighth cervical of both Meiolania and Proganochelys may not have a rib component. PHYLOGENETIC SIGNIFICANCE OF The articulations of the cervical ribs in CERVICAL VERTEBRAL CHARACTERS Meiolania are described above but they might Free cervical ribs are present throughout be compared here with Proganochelys. Both the Amniota as a primitive feature of the turtles have double-headed ribs but in Pro- group and their absence or fusion in turtles ganochelys the ventral articulation does not is an advanced or derived condition. The articulate with anything. There are facets, al- presence of free cervical ribs in Proganoche- though small in comparison with Meiolania, lys is consistent with the hypothesis that it is where cartilaginous intercentra might be the sister group of all other turtles (Gaffney present on either side of the centra in Pro- and Meeker, 1983), but their presence in ganochelys. It is likely that Proganochelys had Meiolania contradicts the assertion that 12 AMERICAN MUSEUM NOVITATES NO. 2805

FIG. 8. Meiolaniaplatyceps. Ventral view ofAM F:49 141, cervicals one through seven, in articulation as found. First four ribs on right side have been removed.

Meiolania is a eucryptodire (fig. 22). The sys- on the intercentrum and diapophysis of the tematic position of Meiolania is of some in- centrum are also primitive for amniote cer- terest because it involves questions of the vical ribs. All turtles, including Proganoche- relationships of the major groups of turtles. lys, lack the atlantal rib (or may have a small, I have discussed the history of phylogenetic fused one at best). Proganochelys has well- work on Meiolania, developed a series ofsys- developed double-headed ribs on cervicals tematic hypotheses, and tested them with two through five but is advanced over the cranial characters in Gaffhey (1983). The ver- primitive amniote condition by having fused tebral characters described here are discussed ribs on cervicals six, seven, with the rib fused in the context of this earlier work and the or absent on eight. In terms of cervical rib reader should have access to that paper. One fusion, Meiolania is more generalized than problem with this extension of comparisons Proganochelys because it has free ribs on cer- from skulls to vertebrae is that the vertebral vical six as well as on cervicals two through structure of many extinct taxa is less known five. There are other turtles with free cervical than the skull. ribs, although the distribution of this feature The generalized amniote condition ap- is not well understood so far. Pleurosternon pears to consist ofa cervical rib for each cen- (described as Mesochelys by Evans and Kemp, trum, including the axis. Presumably the 1975), has associated with it a cervical having double-headed condition with articulations well-developed diapophyseal and parapoph- 1985 GAFFNEY: MEIOLANIA PLATYCEPS 13

8 FIG. 10. Pseudemys scripta. Ventral view of cervicals one through four (part), intercentra indicated as IC. Compare with figure 9. The ele- FIG. 9. Meiolania platyceps. Ventral view of ments identified here as intercentra are identified cervicals one through eight, intercentra shown in as rib remnants by Williams (1959). stipple. Partially restored, based on AM F:49141. ribs in Meiolania is most compatible with yseal processes and facets, although an actual alternative C (in Gaffney, 1983, fig. 65) in rib is not preserved. A similar cervical, but which Meiolania is hypothesized as a casi- also lacking preserved ribs, has been found chelydian but as the sister group of all other with the as yet undescribed Early Jurassic cryptodires and pleurodires. This would still cryptodire from the Kayenta Formation of require rib loss/fusion once in pleurodires and Arizona. The cervicals of such primitive once in cryptodires due to the presence of cryptodires as baenids and Glyptops, how- ribs in Pleurosternon (Mesochelys). Ifmy pre- ever, clearly lack well-developed free cervical ferred hypothesis of Meiolania as an eucryp- ribs. It is thus the case that postulation of todire is chosen (alternative A in fig. 65, Gaff- one-time loss of free cervical ribs in turtles ney, 1983), loss/fusion of free cervical ribs is inconsistent with nearly all other available would be required at least four times: pleu- derived characters. It is also possible that the rodires, Glyptops, baenids, and living eu- presence of embryologic anlagen of cervical cryptodires. Considering that even Progan- ribs (see below) would allow the development ochelys has some degree of autapomorphic of free ribs, but this is extremely speculative. fusion of cervical ribs, a case can be made In any case, the presence of free cervical for the multiple development of this feature. 14 AMERICAN MUSEUM NOVITATES NO. 2805

tuberculum capitulum

4' (.1 5

V. E 0* V 6

anterior medial posterior FIG. 11. Meiolania platyceps, AM F:49141. Cer from right side, shown in three views. Numbers on left indicate vertebra rib articulates wit}

The identification of intercentra in Meio- have been noted by earlier authors (see Wil- lania also has ramifications for our under- liams for references) and I have seen them in standing of some poorly known features in both cryptodires (fig. 10) and pleurodires. Bo- Recent turtle cervicals. Williams (1959) de- janus (1819, pl. 8, fig. 18 and pl. 14, fig. 51) scribed small nodules found in the interver- figured three pairs in the neck of Emys or- tebral tissue ofRecent turtles. These nodules bicularis. Williams identified the nodules as 1985 GAFFNEY: MEIOLANIA PL,ATYCEPS 15

TABLE 2 Comparison of Cervical Vertebrae of Meiolania With Other Turtles Progano- Eucrypto- chelys Meiolania Baenids dires Pleurodires Formed central artic- No Yes Yes and no Yes and no Yes ulations Biconvex 4th cervical No Yes Yes and no Yes No Well-developed free Yes Yes No No No cervical ribs Intercentra Apparently Paired, well Apparently Present as Present as unossified, developed, absent loosely ossi- loosely ossi- but a wedge- variably fied nodules fied nodules shaped area fused to is present centra Ventral median ridge Absent Absent Present Generally Generally pres- present ent Articulation of 8th Yes Yes No Present only No cervical neural in chelo- spine with cara- nioids pace Transverse processes Present and Present and Present, usual- Usually ab- Usually present, well devel- well devel- ly well de- sent, ante- middle of oped, mid- oped, mid- veloped, rior edge of centrum dle of cen- dle of cen- middle of centrum trum trum centrum Centrum Short Short Intermediate Elongate Elongate

cervical rib rudiments and based his identi- 10). The gamma elements of Williams are in fication on comparisons with early develop- the same morphologic position as the inter- mental stages of cryptodiran embryos. He centra of Meiolania. The fact that Meiolania thought he could identify tissue concentra- also has well-developed free cervical ribs sub- tions that were homologous with the tuber- stantiates the identification of intercentra in culum, capitulum, and main body ofcervical Meiolania. Also, the paired intercentra found ribs in these early developmental stages. In in other amniotes are similar to and in the the adult ofRecent turtles Williams identified same position as the gamma elements of tur- the commonly occurring intervertebral nod- tles. Therefore, the best interpretation of the ules (termed the gamma elements by Wil- paired, intervertebral ossicles commonly liams) as the rudiments ofrib capitula or ven- found in the neck of Recent turtles would be tral rib heads. Williams also found less as intercentra. It is also likely that the less common nodules, beta elements, slightly lat- common beta elements are rib rudiments, but eral to the gamma elements, which he iden- whether main body, capitulum, or tubercu- tified as the cervical rib main body. The de- lum cannot be determined. But the previous velopmental evidence seems ambiguous to idea that cervical rib elements are common me; the various cell concentrations are con- throughout turtles must be questioned. sistent with a number of possible identifica- The central articulation pattern of Meio- tions. The discovery that Meiolania has a lania may shed some light on the distribution series of well-developed paired intercentra ofthis condition in turtles. It is apparent from suggests comparison with these and the in- Proganochelys and pleurosternids (Hay, 1908, tervertebral ossicles of living turtles (figs. 9, cf. Glyptops) that amphicoelus cervicals are 16 AMERICAN MUSEUM NOVITATES NO. 2805

FIG. 12. Proganochelys quenstedti (Triassic, Germany), SMNS 17204. Nearly complete tail consisting of caudals 4 through 13, tail club at end, and series of osteoderms preserved in articulation. Left lateral view, anterior to left. primitive for cryptodires, but the distribution specimen with as many as two articulated of formed central patterns (Williams, 1950) caudals is AM F:9051 (see discussion under are complex within the known extinct cryp- Tail Rings). The reconstruction of a tail (fig. todires, such as baenids. Trinitichelys, a 13) has been entirely speculative and based baenid, has amphicoelus cervicals and Plesi- on analogy with chelydrids and baenids and ochelys (Braim, 1965), a chelonioid, also has on the degree and kind of variation seen in amphicoelus cervicals. Other baenids and the disarticulated Meioiania caudals. The re- chelonioids, however, have formed central stored skeleton (Burke et al., 1983, fig. 18) articulations (although different from each has what I believe are the fewest number of other) suggesting that formed articulations caudals given the amount of serial variation have arisen more than once within crypto- seen in the available caudals, and that is 10. dires. It is extremely likely that they arose It is quite possible that more caudals were independently in pleurodires as well. If the present but fewer caudals would be unlikely. appearance of a biconvex fourth cervical in The number of caudals in Meiolania is hard some, but not all, baenids (the distribution to compare with other turtles because the dis- of central patterns in this family is still not tal caudals in Meiolania are fused into the well known, Macrobaena and possibly Neu- tail club (which may represent four to five rankylus may not be baenids) is interpreted vertebrae), whereas the distal caudals ofmost as an independent acquisition, then a case other turtles are often greatly reduced and can be made for the (2( (3( (4) )5) )6) )7) )8) more numerous for a given length than the pattern as a eucryptodiran synapomorphy. anterior ones. The total tail length, as re- This pattern occurs in chelydrids, testudi- stored for Meiolania, is similar to that in noids, chelonioids, and trionychoids and may chelydrids, which is relatively long compared be hypothesized as the primitive condition with other Recent turtles. for each group. IfMeiolania is the sister group In the absence of articulated material, the of remaining eucryptodires then the occur- relative position ofcaudal vertebrae has been rence of this feature in it is consistent with determined by comparison with chelydrid and the hypothesis that this pattern is a eucryp- baenid caudals, which are most similar to the todiran synapomorphy. caudals of Meiolania, and by trial articulation of Meiolania caudals. The restored tail CAUDAL VERTEBRAE consists of the following caudals, beginning There are no articulated tails in the collec- with the most anterior: AM F:61409, AM tions of Meiolania platyceps and the only F:57984, AM F:18704, AM F:57984, AM 1985 GAFFNEY: MEIOLANIA PLATYCEPS 17

FIG. 13. Meiolania platyceps. Reconstructed tail consisting of ten caudals, tail club, and two tail rings. Upper, left lateral view; lower, dorsal view. Anterior to left. Compare with figure 12.

F:18716, AM F:18710, and AM F:18706. zygapophyses are comparable with those on Two casts of the same caudal (AM F: 18706) the other caudals. The centrum is short in were used to bring the number to 10 as the comparison to posterior caudals and has its last two caudals are largely obscured in our longitudinal axis inclined so that the poste- reconstruction by tail rings. Two caudals, the rior central articulation is lower than the an- first and the fourth are from the same spec- terior central articulation. imen, AM F:57984, that makes up most of The second caudal chosen for the recon- the restored skeleton. Nonetheless, they were struction is a vertebra from AM F:57984. found disarticulated and their position in the This centrum has a somewhat shallow and tail is based on analogy as with other caudals. broad anterior central articulation but it is The identification of sacral vertebrae in more convex than in AM F:61409. The zyg- Meiolania shows that the anterior central ar- apophyses and most of the transverse pro- ticulation ofthe first caudal (fig. 14) is convex cesses are broken off this caudal and it is but very shallow and wider than high. AM possible that it is a first rather than a second F:61409 meets these conditions and is hy- caudal. However, the processes in the posi- pothesized as the first caudal in the recon- tion ofthe haemal arch attachment are larger struction. This vertebra has wide transverse than in AM F:61409, suggesing that AM processes, wider than in any other preserved F:57984 is a more posterior vertebra. caudal, and they curve anteriorly to a slight The third caudal in the restoration is AM extent rather than extend laterally at right F: 18704, and it bridges the morphologic gap angles to the centrum or curve posteriorly as between the extreme anterior caudals which in the other caudals. There are only low, blunt lack haemal arches and the more "typical" processes at the posteroventral margin ofthe caudals which have well-developed haemal centrum, rather than a haemal arch. The pre- arches. In AM F: 18704 a haemal arch is pres- 18 AMERICAN MUSEUM NOVITATES NO. 2805

C E

cms . FIG. 14. Meiolania platyceps. First caudal vertebra, AM F:61409. A, dorsal; B, ventral; C, left lateral; D, posterior; E, anterior. ent with paired basal attachments, but the mination in the anterior caudals while pos- arch is relatively small, inclined posteriorly, teriorly the spines become short and blunt. and comes to a point distally rather than a The zygapophyses are closer together poste- blunt termination as in the more posterior riorly. The haemal spines change slightly caudals. along the column, they are straight anteriorly The remaining caudals, four through nine but acquire a slight posteroventral bend in in the reconstruction, may be described as a the posterior caudals. series ofrelatively similar vertebrae differing within the series from the more anterior to the more posterior ones. Owen figured a typ- PHYLOGENETIC SIGNIFICANCE OF ical caudal of Meiolania platyceps (Owen, CAUDAL CHARACTERS 1888, pl. 35, figs. 5, 6; the specimen has not Table 3 been identified in a collection) which should Determination of the systematic signifi- be used for comparison with those illustrated cance of caudal morphology in Meiolania here (fig. 15). The centrum of the more an- platyceps is made difficult by the absence of terior caudals is relatively short, only slightly comparative literature on turtle caudals. Al- longer than wide but posteriorly the centrum though the cervical vertebrae of turtles have becomes more elongate and narrower in the been described and compared in a number middle of the central body. The transverse of papers, caudals are rarely dealt with. processes, well developed on all caudals, are Among Recent specimens, I am aware only longer and narrower anteriorly and become ofthe following descriptions: Emys (Bojanus, shorter and broader in the more posterior 1819); Dermochelys (Gervais, 1872; V6lker, caudals. The broadest but shortest transverse 1913); Geochelone (Gunther, 1877); Trionyx processes are in AM F: 18706 where they (Ogushi, 1911); Carettochelys (Walther, closely approximate the interior diameter of 1922); and Chelydra (Newman, 1906, see be- the tail ring of AM F:50635. The position of low). attachment ofthe transverse processes moves In 1887, Huxley described and figured a from the center of the centrum more poste- caudal of "Ceratochelys sthenurus" (=Meio- riorly in the posterior caudals. The neural lania playceps, specimen in BM(NH) but not spines are tall and have an acuminate ter- identified) which he compared with Chely- 1985 GAFFNEY: MEIOLANIA PLATYCEPS 19

FIG. 15. Meiolania platyceps. Caudal vertebrae in left lateral views. A, AM F: 18704 (3); B, AM F: 57984 (4); C, AM F:18715 (7); D, AM F:18706 (9 and 10). Numbers in parentheses indicate position in restored tail (fig. 13). dra. He concluded on the basis of the caudal coelus and having very small or absent hae- and skull morhpology (see Gaffney, 1983, for mal arches (see also Gaffney, 1984). This discussion and figure ofskull used) that "Cer- group would consist ofthe chelonioids, trion- atochelys" was a "Chelydroid Chelonian." ychoids, emydids, and testudinids. The pro- Specifically (Huxley, 1887, p. 237): "that the coelus to platycoelus caudals of pleurodires remains of crania and caudal sheaths from would be an independent acquisition of this Australia, hitherto referred to Saurian rep- character. Baur (1889, p. 62) reports two tiles [that is, lizards, by Owen], under the specimens of Clemmys insculpta with opis- names of Megalania and Meiolania apper- thocoelian caudals as an individual variation. tain to a hitherto unknown species of Che- The Chelydridae have a caudal condition lonian, Ceratochelys sthenurus, closely allied that may be hypothesized as a synapomorphy to the living Chelydra, Gypochelys [=Macro- for that group. Two or three of the most an- clemys], and 'Platysternum'." As far as the terior caudals are procoelus, following them caudal vertebrae are concerned, Huxley's ar- is a biconcave caudal and the remainder are gument linking Meiolania with the chely- opisthocoelus. In Chelydra and Macroclemys drids (including Platysternon in the sense of it is the third caudal that is biconcave, where- Agassiz, 1857, and Gaffney, 1975) is based as in Platysternon and Chelydropsis (Mly- on the central articulation pattern and the narski, 1980) it is the fourth caudal. The re- general close agreement in morphology. Hux- maining caudals are opisthocoelus. The ley referred to Baur's (1886) vertebral study biconvex caudal is a rare situation, there is in which Baur showed that the chelydrids no evidence ofit (or ofany procoelus caudals) were unusual among Recent turtles in pos- in Meiolania or baenids, and it has not been sessing opisthocoelus caudals, all other living reported in the remaining living cryptodires turtles being procoelus. This pattern has sys- or pleurodires. tematic significance for cryptodiran relation- It would appear that Huxley's use of opis- ships. The primitive chelonian caudal pat- thocoelus caudals to link Meiolania and che- tern is platycoelus, as seen in Proganochelys, lydrids is unsatisfactory because that feature but the presence of opisthocoelus caudals in is plesiomorphic for cryptodires. Procoelus baenids strongly suggests that that condition caudals do seem to define a natural group, is primitive for cryptodires. The procoelus however, which excludes chelydrids and cryptodires also have reduced or absent hae- Meiolania (fig. 22). mal arches, which are well developed in The caudals of Meiolania platyceps are baenids and chelydrids. Therefore, it would most similar to those in the chelydrids and be useful to propose a group, the Procoelo- baenids. This similarity appears to be the cryptodira, having nearly all the caudals pro- result of shared primitive features for the U, 0 - Cu

o0~~~C - U

PCu

00rn~~~~~~~~~~~~~~~~~~~~~~r

0 00

'0~~~~~~~~~~~~~~~~~~~~'

~~~0 a ~ ~ ~~~C

Cd~~~~~C a~~~ i-'I~C.) ~~) a~z~~~ Cu '0 1985 GAFFNEY: MEIOLANIA PLATYCEPS 21

Cryptodira. In addition to opisthocoely as the and ridges on the tail club that has variable dominant articulation pattern, Meiolania, development among the available speci- chelydrids, and baenids have well-developed mens. haemal arches on most of the caudals, rela- For the purposes of description, the spines tively high and narrow centra, well-devel- may be grouped in pairs aligned as a series oped neural spines, and a relatively long tail; of segments along the length of the tail club. features generally lacking in Procoelocryp- Each segment consists of a dorsolateral pair todira and pleurodires. of spines and a lateral pair. The spines begin Baenid caudal morphology is known from as a low ridge anteriorly and increase in size only a few specimens. Hay (1908, p. 79) com- posteriorly, becoming acuminate to end in a pares caudals of Baena with Chelydra, projection that points posteriorly and away whereas Case (1939) figures disarticulated from the main axis of the tail club. The dor- caudals of Plesiobaena. Russell (1934), how- solateral spines are distinctly smaller. These ever, has published the only figure of what spine pairs are arranged in four segments. appears to be a complete baenid tail, as well The spines vary in size and shape among the as figuring some individual centra. Russell's available specimens of tail clubs (about two specimen is Thescelus (Gaffney, 1972), a ge- dozen, not counting small fragments). In AM nus that seems to have an unusually long tail F: 18721 (fig. 17) the spines are particularly even for baenids. Thescelus also has most of acuminate, more so than in many specimens the caudals nearly platycoelus, although it is such as MM F: 1383 la (figured in Owen, 1888, best described as opisthocoelus. Russell also pl.37). Rounded, blunt spines are seen in AM figures a partially articulated tail of Plesio- F:64435 and AM F:61404. Other types oftail baena. Case (1939) figures and describes dis- club variation also exist. In AM F:61404 the articulated caudals of Plesiobaena. Based on dorsolateral spines are closer together at the re-examination ofall these specimens as well midline than in the other specimens and the as other material in the AMNH collections, lateral spines are comparatively large in con- I have been unable to find consistent features trast to tail clubs such as AM F: 18721. In all differentiating individual caudals ofbaenids, complete tail clubs the dorsolateral and lat- Meiolania, and chelydrids (except in the case eral spines are arranged in four segments along of the procoelus and biconvex anterior cau- the length of the tail. In some, such as MM dals in chelydrids). Regional variation and F:13831a (Owen, 1888, pl. 37) the spines de- numbers of caudals offer more possibilities crease in size posteriorly but in others, such but as entire tails are known in only one bae- as in AM F:18721 and a club in the Lord nid and not at all in Meiolania further pursuit Howe Island Museum, the second spine pair of this avenue must await better material. is the largest, with the first, third, and fourth being smaller. The ventral surface of the tail club lacks TAIL CLUB spines and has a pattern of elongate hexag- One ofthe most bizarre attributes ofMeio- onal scales down the midline. There are four lania is the prominent ossification at the end ofthese scales corresponding to the segments of the tail. The presence of a somewhat sim- formed by the spine sets, but some clubs (AM ilar structure in the Triassic turtle, Progan- F: 18721, for example), show indications of a ochelys, makes the feature even more inter- fifth scale at the very end. In most tail clubs, esting. such as AM F:64435, the scale sets are clear The tail club or tail sheath of Meiolania and well defined and the rugose bone surface platyceps (figs. 16-19) is a hollow, elongate preserves all the details of the pits and fo- cone of bone with vertebrae attached down ramina. But in some tail clubs, such as AM its center by the transverse processes, neural F: 18721 (fig. 17) the ventral surface shows spines, and haemal arches. The external sur- signs of mechanical erosion and abrasion, face is very rugose, with numerous pits and particularly anteriorly, that has resulted in foramina for nutrient vessels, suggesting that thinning ofthe bone and perforation in some the bone was covered in life by a horny scale areas. This erosion does not seem to have or scales. There is a regular pattern of spines been postmortem and may have occurred 22 AMERICAN MUSEUM NOVITATES NO. 2805

1983, p. 474) described and figured two such specimens and Etheridge (1889, pl. 25, fig. 1 A and pl. 26, fig. 1) also figured a broken tail club. Further study of the Australian Mu- seum collection resulted in the discovery of another piece belonging to the tail club figured by Etheridge so that the specimen now consists ofthe figured portion (AM F: 16867) B and a terminal portion (AM F:228, but this is a group number containing other specimens). The newly discovered piece allows all four spine segments to be identified on this specimen expediting comparisons. This tail club has also been acid prepared and now serves as an illustration of the internal morphology of a Meiolania platyceps tail club (fig. 18; see also sections of a tail club, AM F: 18725, fig. 19). The cone of bone forming the external part of the club or sheath has a string of vertebrae that run down its center all the way to the tip. At least three centra are recognizable anteriorly and these corre- spond to the first three segments formed by the spine pairs. Each vertebra is recognizable cms as a posterior caudal because they have an elongate centrum, broad transverse process- FIG. 16. Meiolania platyceps. Partial tail club, es, and a haemal arch positioned at the pos- AM F: 18863. This tail club is one ofthe few show- terior end ofthe centrum. The most anterior ing sutures. vertebra to be attached within the tail club has a centrum with a convex articulation on its anterior surface and a platycoelus syn- chondrosis posteriorly that is open in most during life when dragging the tail abraded specimens but is fused in some, such as AM through the scale covering. There are no other F:61404. All the other centra in the tail club indications in any specimens of tail club articulate by means of a platycoelus syn- damage during life. The tail clubs are not chondrosis but the contact is tighter poste- precisely symmetrical, they may curve slight- riorly to become virtually indistinguishable ly to one side, and the development of the between the third and fourth centrum. The spines may not be precisely symmetrical in centra are constricted in the middle giving extent or position. them an hourglass shape. The neural spine The terminal portion ofthe tail club varies attaches to the tail sheath distally for most from acuminate in AM F: 18721 to blunt in of its length and the neural canal retains its AM F:64435. The terminal segment may be integrity, at least anteriorly (fig. 19; Owen, rugose and featureless as in AM F:64435 but 1888, pl. 37, fig. 3). Each transverse process it often has a variably developed midline ridge also attaches to the bony cone ofthe tail club that is particularly apparent in AM F: 18724 at its distal termination, as does the haemal and AM F:18720. arch of each centrum so that the centra are The internal morphology of the tail club suspended down the middle of the tail club. (figs. 18, 19) is determinable by examination Posteriorly, however, as the cone of the tail of damaged specimens in which the outer club narrows, the centra do not narrow and bone layers have been removed. Owen (1886, the relatively large space enclosed by the tail pl. 31, fig. 3-BMNH R679, fig. 4-BMNH club anteriorly (fig. 19, left) becomes greatly R680; error in figure references in Gaffney, reduced posteriorly, as can be seen in AM F: 1985 GAFFNEY: MEIOLANIA PLATYCEPS 23

18725 (fig. 19, right) and AM F:18361. In separate ossifications, and as some of these these specimens it can be seen that the ter- are partially fused, it is possible that addi- minal portion ofthe club is nearly solid bone tional ossifications were present earlier in de- with only a thin space separating centrum and velopment. The three ventral ossifications external bone. seem to be symmetrical on the midline but The positioning of the centra within the the others are not. The pair of dorsal spikes club corresponds with the segmental arrange- are each made of two ossifications, the right ment ofthe spikes. The centra tend to overlap one, at the tip of the spike, extends over the the segments slightly so that the anterior third midline but the left one does not. The ante- ofthe centrum extends into the posterior part rior margin of the tail club segment is a su- of the preceding spike segment. This char- tural surface that is in the correct position for acterization is a bit subjective because the a tail ring anterior margin and it is possible separation between spike segments would ap- that the terminal segment of a tail club does pear to be on a transverse plane that is tilted ossify in numerous centers ofgrowth, where- posteriorly at the top of the club and ante- as the more anterior portions are more like riorly at the bottom. tail rings. It is also possible that this specimen The tail clubs vary in size but because only is anomalous and does not represent the usu- a few are complete enough to measure the al ossification pattern. range is difficult to determine. The smallest A tail club is also known in "Meiolania" complete club (AM F: 18721) is about 60 per- oweni from the Pleistocene of Queensland, cent the size ofthe largest complete club (MM described and figured by Owen (1882). This F:13831a). As most clubs lack sutures it is tail club is very similar to the one in Meio- difficult to see how these could represent laniaplatyceps, differing primarily in size and growth stages. There are no tail clubs found the number of spike segments. The "Meio- associated with shells, skulls, or other major lania" oweni club is about four times the di- skeletal elements and AM F: 18721 was cho- ameter and much more massive in propor- sen for the AMNH skeletal reconstruction tions than Meiolania platyceps. Meiolania because AM F:57984, the specimen that pro- platyceps has four spike segments consisting vided about 40 percent ofthe reconstruction, of two spike pairs, whereas "Meiolania" ow- is one of the smaller Meiolania platyceps eni has only two spike segments with the spike specimens. projections being more obtuse and thicker Owen (1888, p. 188) suggested, on the basis but the orientation, relative size, and position of MM F: 1383la, that the tail club in Meio- of the spikes being the same in both species lania consisted of "five anchylosed seg- (see Owen, 1882, pl. 65). ments" comparable to the tail rings. He in- The only other turtle to have a tail club cluded the four spike sets plus the terminal fully enclosing caudal vertebrae is Progan- cone. The close similarity between the tail ochelys (fig. 12). The club in Proganochelys rings and the tail club in the position of the is similar to that in Meiolania in consisting spike sets and the vertebrae support Owen's ofa series of spikes fused together. The three contention that the tail club may be inter- Proganochelys tail clubs available show some preted as serially homologous with at least variation indicating some asymmetry rather four fused tail rings. It is, therefore, surprising than the strict bilateral symmetry seen in to find that the only tail club with sutures (fig. meiolaniids. The Proganochelys club has 16) is inconsistent with this hypothesis. This spike sets that are arranged in triads with a specimen, AM F: 18863, is a partial tail club median spike on the midline in contrast to consisting ofthe last set of spikes but lacking the spike pairs of meiolaniids. The ventral the terminal tip. Although sutures are pres- and lateral parts of the Proganochelys are ent, AM F: 18863 is larger than AM F: 18721 covered by short, wide plates with no indi- and seems to be comparable in size to MM cation of hexagonal scales in meiolaniids. In F:13831a. The sutures preserved on AM Proganochelys, as in Meiolaniaplatyceps and F:18863 do not appear to be related to its "Meiolania" oweni, the morphology of the supposed serial homology with the tail rings. tail club is comparable to the dermal ossifi- The preserved portion consists of at least 12 cations found more anteriorly on the tail. 24 AMERICAN MUSEUM NOVITATES NO. 2805

FIG. 17. Meiolania platyceps. Tail club, AM F: 18721. Top, left lateral view; middle, dorsal view; bottom, ventral view. Anterior to left.

Is the tail club of meiolaniids and Pro- meiolaniid relationships (fig. 22) indicates that ganochelys strictly homologous or is it a the tail club would have to have been lost at synapomorphy for meiolaniids which was in- least three times (pleurodires, pleurostemids, dependently present in Proganochelys? Ex- baenids) if this hypothesis is correct and if amination of the preferred hypothesis of the tail clubs are homologous. The tail clubs 1985 GAFFNEY: MEIOLANIA PLA TYCEPS 25

FIG. 18. Meiolania platyceps. Tail club, AM F:16867 and AM F:228 (in part). Ventral view of damaged specimen showing internal features.

FIG. 19. Meiolania platyceps. Tail club, AM F: 18725. Left, anterior view of anterior (natural) end of tail club; right, posterior view of broken posterior end of tail club. are similar but not identical so the alternative are tail clubs, being found in chelydrids and hypothesis ofindependent origin is quite pos- testudinids as well as in meiolaniids and Pro- sible. Nonetheless, I think that the degree of ganochelys; and it might be argued that the similarity is close enough to require the struc- ossification potential exists or existed in many tures to be homologous in spite of the mul- turtle groups, allowing the multiple evolution tiple loss made necessary by accepting this of tail clubs. The presence in testudinids of hypothesis. well-developed "anal bucklers" (Hay, 1908) A "compromise" hypothesis is also pos- is consistent with this hypothesis, which, al- sible making independent acquisition seem though not favored here, should be kept in plausible. The presence of caudal ossifica- mind. tions is more widespread among turtles than In summary, I think that the particular form 26 AMERICAN MUSEUM NOVITATES NO. 2805

FIG. 2 1. Meiolania platyceps. Tail ring, AM F: 50635. A, anterior view; B, posterior view; C, left lateral view (anterior to left).

is no sign of fusion or sutural attachment between tail ring and any vertebral element. In contrast to the tail club, the tail rings in Meiolania platyceps are incomplete ventrally, also in contrast to the tail rings in "Meio- lania" oweni (Owen, 1882) and Niolamia (Woodward, 1901) in which they are complete ventrally. The anterior margin is in- FIG. 20. Meiolania platyceps. Tail club with dented for articulation with adjacent rings. two articulated tail rings, AM F:905 1. It seems likely that the tail rings, definitely known only for the posterior portion of the tail, extend anteriorly for much of the length of the tail, probably one for each caudal. Al- of the tail club with two pairs of spikes is though there are no complete anterior tail the pres- rings it is likely that some fragments belong synapomorphous for meiolaniids, to tail rings from the anterior series. One of ence ofa caudal ossification surrounding cau- these (MM F: 13829) was figured by Owen dal vertebrae is primitive for turtles and is a (1888, pl. 36, figs. 7-9) as a "sternal arch." synapomorphy for the whole group. Examination of the specimen indicates that ofthe three knobs or spikes seen in posterior TAIL RINGS view, the two left-hand ones are equivalent The tail of Meiolania platyceps is covered to the dorsolateral spikes on the posterior tail with a segmented dermal armor composed of rings. If the circumference of this ring were tail rings (fig. 21). Only one specimen, AM restored it would have a diameter of about F:905 1, preserves these rings in articulation three times that of a tail club and might cor- with other elements (fig. 20) and this speci- respond to the base of the tail. Another ring men shows that, at least posteriorly, the rings ofthis sort is AM F: 1 196, but it has a smaller articulate with one another and that there is restored diameter and might be from a more one per vertebra. The tail rings in this spec- posterior section of the tail. If this identifi- imen, which is not well preserved, are bent cation is correct, the anterior tail rings would dorsally and lie at an angle to the tail club, a differ from the posterior ones in being thinner condition that is the result of postmortem anteroposteriorly, lacking articulation with disturbance. These tail rings are similar in each other, having a larger diameter, and curvature and spike morphology to the tail having much lower spikes but the same pat- club. A disarticulated tail ring, AM F:50635 tern. There are many fragments in collections (fig. 21) shows this morphology more clearly. (for example, AM F:193, AM F:5754) that The dorsolateral spike pair extend postero- can be interpreted as belonging to tail ring dorsolaterally, as on the tail club, and are fragments intermediate between the ones de- much larger than the lateral spike pair. There scribed above, but their fragmentary nature 1985 GAFFNEY: MEIOLANIA PLATYCEPS 27

FIG. 22. Cladogram ofturtles showing hypothesized position ofmeiolaniids. See also Gaffney, 1983, figure 65. Characters in this cladogram are described and discussed in Gaffney (1984). precludes useful description. It might be worth the tail ring of "Meiolania" oweni is homol- mentioning here that it is difficult to distin- ogous with a portion ofthe tail club in Meio- guish fragments of tail rings from skull frag- lania platyceps rather than with the tail rings ments in the supraoccipital-squamosal area themselves. of the "A" horn and scale area. A tail ring is known for Niolamia argentina Among the presumed tail ring fragments (Woodward, 1901, pl. 18, fig. 2), but whether are a few (AM F:374, AM F:61 110) that are or not a tail club was also present is not known. bilaterally symmetrical, seem to be from the This tail ring is similar to that in the other dorsal section forming part of the dorsolat- meiolaniids in having two spine pairs with eral spikes, and are limited laterally by su- the dorsolateral one being the largest. It dif- tures. This suggests that the tail rings may fers from the other meiolaniids in having the ossify from a median section and have two spines relatively flat in one plane rather than or more lateral ossification centers. circular in cross-section. As in "Meiolania" Tail rings occur in other meiolaniids and oweni but in contrast to Meiolania platyceps, Proganochelys has caudal ossifications that the tail ring of Niolamia argentina forms a may be compared with Meiolania. "Meio- complete circle. lania" oweni has a tail ring (Owen, 1882, pl. The only other turtles known with ossified 65) that is quite similar to that of Meiolania caudal plates are some ofthe testudinids. Hay platyceps, with a large dorsolateral spine pair (1908, figs. 560, 603) describes caudal ossi- and a smaller lateral spine pair, both with fications in some extinct species of Geoche- their apices trending posteriorly. Although the lone ("Testudo" osborniana and "Testudo" specimen is broken, it is likely that the ring orthopygia) that consist of loose ossicles and formed a complete circle, in contrast to the a flat plate offused ossicles lying on the dorsal known tail rings of Meiolania platyceps. The surface of the last caudal vertebrae. Auffen- "Meiolania" oweni tail ring, furthermore, is berg (1963) also describes a similar ossifica- tightly articulated with the tail club which tion in a subgenus of Geochelone which he consists of only two spine sets rather than named Caudochelys. The caudal vertebrae in four in "Meiolania" oweni. It is possible that these forms are short, with very wide trans- 28 AMERICAN MUSEUM NOVITATES NO. 2805 verse processes and the plate is borne on the Case, E. C. dorsal surface of the caudal amongst many 1939. A nearly complete turtle skeleton from separate dermal ossicles. Auffenberg's hy- the Upper Cretaceous ofMontana. Con- pothesis that this plate, termed by him a su- trib. Mus. Paleont., Univ. Michigan, vol. pracaudal anal buckler, served the purpose 6, no. 1, pp. 1-19. Etheridge, R., Jr. of protecting the anus and surrounding area, 1889. On the occurrence of the genus Meio- seems reasonable. There are no particular lania in the Pliocene Deep Lead at Ca- similarities with meiolaniid tail rings or tail nadian, near Gulgong. Rec. Geol. Surv. clubs. NewSouth Wales, vol. 1,pp. 149-152. Proganochelys has many dermal ossicles in 1893. On further traces of Meiolania in New the tail. On the dorsal surface are a series of South Wales. Records Australian Mus., broad ossicles (fig. 12) bearing three spikes vol. 11, pp. 39-41. that seem to be serially homologous with the Gaffhey, E. S. spikes in the tail club. These ossicles are 1972. The systematics ofthe North American roughly comparable with tail rings in meio- family Baenidae (Reptilia, Cryptodira). laniids but do not extend down the side of Bull. Amer. Mus. Nat. Hist., vol. 147, art. 5, pp. 241-320. the tail and do not have spikes arranged in 1983. Cranial morphology of the extinct pairs. horned turtle, Meiolania platyceps, from the Pleistocene of Lord Howe Island. LITERATURE CITED Ibid., vol. 175, art. 4, pp. 361-480. Agassiz, L. 1984. Historical analysis of theories of che- 1857. North American Testudinata. In Con- lonian relationship. Syst. Zool., vol. 33, tributions to the natural history of the pp.283-301. United States, vol. 1, part 2, pp. 233- Gaffney, E. S., and L. J. Meeker 245. Boston, Little Brown and Co. 1983. Skull morphology of the oldest turtles: Anderson, C. A preliminary description of Progan- 1925. Notes on the extinct chelonian Meio- ochelys quenstedti. Jour. Vert. Paleo., lania, with a record ofa new occurrence. vol. 3, no. 1, pp. 25-28. Records Australian Mus., vol. 14, pp. Gervais, M. P. 223-242. 1872. Osteologie du Sphargis luth. Nouv. Auffenberg, W. Arch. Mus. Mem., vol. 8, pp. 199-228. 1963. Fossil testudinine turtles ofFlorida gen- Gunther, A. era Geochelone and Floridemys. Bull. 1877. The gigantic land-tortoises (living and Florida State Museum, vol. 7, no. 2, pp. extinct) in the collection of the British 53-97. Museum. London, British Museum, Baur, G. 1877, 96 pp. 1886. Osteologische Notizen uiber Reptilien. Hay, 0. P. Zool. Anz., vol. 9, pp. 733-743. 1908. The fossil turtles of North America. 1889. The systematic position of Meiolania, Carnegie Inst. Washington Publ., no. 75, Owen. Ann. & Mag. Nat. Hist., ser. 6, pp. 1-568. vol. 3, pp. 54-62. Kasper, A. Bojanus, L. H. 1903. Ueber den Atlas und Epistropheus bei 1819. Anatome Testudinis europaeae. Soc. den pleurodiren Schildkr6ten. Arbeiten Study Amphibians and Reptiles (fac- Zool. Inst. Univ. Wien, vol. 14, pp. 137- simile reprints in herpetology, reprinted 172. 1970), no. 26, 178 pp. Mlynarski, M. Bram, H. 1980. Die tertiaren Wirbeltiere des Steinhei- 1965. Die Schildkroten aus dem oberen Jura mer Beckens. Teil III. Die Schildkr6ten (Malm) der Gegend von Solothurn. des Steinheimer Beckens. Palaeonto- Schweiz. Palaont. Abhandl., vol. 83, pp. graphica, vol. 8, pp. 1-35. 1-190. Newman, H. H. Burke, A., M. Anderson, A. Weld, and E. S. Gaff- 1906. The significance of scute and plate "ab- ney normalities" in Chelonia. Biol. Bull., vol. 1983. The reconstruction and casting ofa large 10, pp. 68-114. extinct turtle, Meiolania. Curator, vol. Ogushi, K. 26, no. 1, pp. 5-26. 191 1. Anatomische Studien an der japanisch- 1985 GAFFNEY: MEIOLANIA PL,ATYCEPS 29

en dreikralligen Lippenschildkr6te (Tri- Hautskelet von Dermochelys coriacea onyxjapanicus). Part 1. Morph. Jahrb., L. Zool. Jahrb., vol. 33, pp. 431-542. vol. 43, pp. 1-106. Walther, W. G. Owen, R. 1922. Die Neu-Guinea Schildkrote Caret- 1882. Description of some remains of the gi- tochelys insculpta Ramsay. Nova gantic land-lizard (Megalania prisca Guinea, vol. 13, pp. 607-702. Owen), from Australia. Part III. Phil. Williams, E. E. Trans. Roy. Soc. London (1881), vol. 1950. Variation and selection in the cervical 172, pp. 547-556. central articulations of living turtles. 1888. On parts of the skeleton of Meiolania Bull. Amer. Mus. Nat. Hist., vol. 94, platyceps (Owen). Phil. Trans. Roy. Soc. pp. 505-562. London, ser. B, vol. 179, pp. 181-191. 1959. Cervical ribs in turtles. Breviora, no. Russell, L. S. 101, pp. 1-12. 1934. Fossil turtles from Saskatchewan and Woodward, A. S. Alberta. Trans. Roy. Soc. Canada, vol. 1901. On some extinct reptiles from Patago- 28, pp. 101-110. nia, of the genera Miolania, Dinilysia, V6lker, H. and Genyodectes. Proc. Zool. Soc. Lon- 1913. Ueber das Stamm-, Gliedmassen- und don, 1901, pp. 169-184.