Amniote ankle

7/30/2019 Amniote ankle

1/11

Evolution and Homology of the Astragalus in EarlyAmniotes: New Fossils, New Perspectives

F. Robin OKeefe,1* Christian A. Sidor,1 Hans C.E. Larsson,2 Abdoudaye Maga,3 andOumarou Ide3

1Department of Anatomy, New York College of Osteopathic Medicine, Old Westbury, New York 115682Redpath Museum, McGill University, Montreal, Quebec H3A 2K6, Canada3Institut de Recherches en Sciences Humaines, Niamey, Niger Republic

ABSTRACT The reorganization of the ankle in basalamniotes has long been considered a key innovation al-lowing the evolution of more terrestrial and cursorial be-havior. Understanding how this key innovation arose is acomplex problem that largely concerns the homologizingof the amniote astragalus with the various ossifications in

the anamniote tarsus. Over the last century, several hy-potheses have been advanced homologizing the amnioteastragalus with the many ossifications in the ankle ofamphibian-grade tetrapods. There is an emerging consen-sus that the amniote astragalus is a complex structureemerging via the co-ossification of several originally sep-arate elements, but the identities of these elements re-main unclear. Here we present new fossil evidence bear-ing on this contentious question. A poorly ossified, juvenileastragalus of the large captorhinid Moradisaurus grandisshows clear evidence of four ossification centers, ratherthan of three centers or one center as posited in previousmodels of astragalus homology. Comparative material ofthe captorhinid Captorhinikos chozaensis is also inter-pretable as demonstrating four ossification centers. A

new, four-center model for the homology of the amnioteastragalus is advanced, and is discussed in the context ofthe phylogeny of the Captorhinidae in an attempt to iden-tify the developmental transitions responsible for the ob-served pattern of ossification within this clade. Lastly, thebroader implications for amniote phylogeny are discussed,concluding that the neomorphic pattern of astragalus os-sification seen in all extant reptiles (including turtles)arose within the clade Diapsida. J. Morphol. 267:415425,2006. 2006 Wiley-Liss, Inc.

KEY WORDS: amniote; ankle; astragalus; captorhinid;homology

The tarsus in amniotes differs radically from thatof anamniote tetrapods. Relative to amphibian-grade taxa, amniotes have fewer bony elements inthe ankle, and the elements play an increased, moreactive role in locomotion by stiffening and stabiliz-ing the pes (Sumida, 1997). This repatterning of thetarsus has long been considered a key innovation inthe transition between amphibian-grade andreptilian-grade tetrapods (Romer, 1956; Gauthier etal., 1988), allowing increased terrestriality and loco-motor efficiency. Unfortunately, this transition isnot well understood; the homologies of the various

tarsal elements are unclear, rendering the study ofspecific joint articulations difficult (Sumida, 1997, p.387). New data and new insights on the developmen-tal and evolutionary repatterning of the amnioteankle are therefore of prime importance, for they

bear on one of the key transitions in tetrapod his-tory: the attainment of full terrestriality after thelong transition from fin to limb (Clack, 2002).

Perhaps the most important novelty within theamniote tarsus is the astragalus, a large, complexbone comprising the primary area of articulation forthe distal tibia. The definitive astragalus first ap-pears at the origin of the clade Amniota (Fig. 1A);amphibian-grade tetrapod taxa below this node,such as seymouriamorphs and anthracosaurs (re-gardless of the current debate over their relation-ships to each other and to Amniota; Clack, 2002),generally maintain the pattern of tarsal bones char-

acteristic of most anamniote tetrapods, while synap-sids and reptiles possess a true astragalus (relation-ships from Coates, 1996; Sumida, 1997). This obviouscorrelation between amniote origins and the appear-ance of the astragalus has lead to a long discussionof the homologies of this bone (Gegenbauer, 1864;Zittel, 1932; Romer, 1956; Rieppel, 1993; Kissel etal., 2002; Berman and Henrici, 2003). However, alack of definitive evidence has precluded a satisfac-tory resolution. Here we present new fossil evidencethat may help to resolve this debate.

Contract grant sponsor: National Geographic Society; Contractgrant number: 7258-02 (to C.A.S.); Contract grant sponsor: New YorkCollege of Osteopathic Medicine (research funds to C.A.S. and F.R.O.);Contract grant sponsor: National Sciences and Engineering Research

Council, Canada Research Chairs; Contract grant sponsor: FondsQuebecois de la Recherche sur la Nature et les Technologies (toH.C.E.L.).

*Correspondence to: F.R. OKeefe, Assistant Professor, Dept. ofAnatomy, New York College of Osteopathic Medicine, NYCOM II Rm.

326, Old Westbury, NY 11568-8000. E-mail: [email protected]

Published online 18 January 2006 inWiley InterScience (www.interscience.wiley.com)DOI: 10.1002/jmor.10413

JOURNAL OF MORPHOLOGY 267:415425 (2006)

2006 WILEY-LISS, INC.


2/11

Captorhinid Reptiles and Comparative Taxa

Reptiles of the clade Captorhinidae (Fig. 1B) offer

a unique opportunity to study the ossification of thetarsus in early amniotes. Captorhinids are a groupof basal, generalized early reptiles with an excellentfossil record beginning in the Pennsylvanian (Wide-man and Sumida, 2004) and spanning the Permian(300 million to 250 million years ago; Modesto andSmith, 2001). Because of their great age, excellentfossil record, and generalized anatomy, captorhinidshave long been used as exemplars in discussions ofreptile evolution (Romer, 1956). From a modern phy-logenetic perspective, captorhinids are appealing be-cause they are the most primitive eureptiles(Modesto and Anderson, 2004; Fig. 1), and as such

are archaic members of the clade that includes allmodern reptiles. The ankles of extant diapsids arequite derived in terms of both morphology and de-

velopment relative to more basal amniotes (Rieppeland Riesz, 1999), whereas the ankle morphology anddevelopment of captorhinids is quite primitive(Kissel et al., 2002; Holmes, 2003). Recent functional

work on the ankle of Captorhinus has demonstratedthat it had a mesotarsal joint, an important innova-tion allowing more efficient terrestrial locomotion(Holmes, 2003). Captorhinids are therefore a logicalgroup in which to study ankle evolution, becausethey form an anatomical and functional link be-tween modern, derived diapsids on the one hand andmore basal amniotes and anamniote tetrapods onthe other. Lastly, the excellent fossil record of thegroup, consisting of literally thousands of speci-mens, affords ample opportunity for the study ofboth juvenile and adult animals within a phyloge-netic framework.

Phylogenetic uncertainty has hindered a clearunderstanding of the morphological configurationpresent in the ankles of amniote precursors(Clack, 2002, pp. 267277). The one clade that allagree is very close to the origin of amniotesDiadectomorphahas autapomorphic ankle mor-phology (Sumida, 1997; Berman and Henrici,2003). Fortunately, the morphology of the tarsusin most basal tetrapods is generalized and fairlyconservative. Permo-Carboniferous taxa as variedas the colosteid Greererpeton, the temnospondyl

Acheloma, and the highly terrestrial embolomereProterogyrinus share a canonical pattern of seventarsal ossifications comprising intermedium, tibi-

ale, fibulare, and four centralia (Holmes, 1984;Godfrey, 1989). In this article we use the temno-spondyl Acheloma (previously known as Trema-tops; this taxon was declared a junior synonym of

Acheloma by Dilkes and Reisz, 1988) as an exem-plar taxon for the primitive condition outside of

Amniota. There are several justifications for doingthis, the two most germane being the highly con-servative nature of the tarsus in basal tetrapodsand the fact that Acheloma is a large animal witha well-ossified tarsus. Many basal tetrapods, espe-cially of small body size, have poorly ossified tarsiand therefore lack detailed information on the

articulations of some elements and on the locationof soft tissue structures such as the perforatingartery. A last justification for using Acheloma asan exemplar taxon is simply historical: authorsfrom Peabody (1951) to Romer (1956) to Shubin(2002) all accept Acheloma as an accurate repre-sentation of the generalized condition from whichboth amniotes and lissamphibians evolved.

Previous Models of Astragalus Homology

Before the work of Peabody (1951), Romer andearlier workers had entertained various hypotheses

Fig. 1. A: Tetrapod phylogeny showing relationships of taxadiscussed in this article. Topology follows Modesto and Anderson(2004), Hill (2005). B: Skeleton of the representative captorhinidreptile Captorhinus laticeps, after Heaton and Reisz (1980).

416 F.R. OKEEFE ET AL.

Journal of Morphology DOI 10.1002/jmor


3/11

for the homology of the amniote astragalus (re-viewed in Romer, 1956, pp. 392393); however, inOsteology of the Reptiles Romer accepted Peabodyscontention that the amniote astragalus is formedfrom the fusion of three elements present inamphibian-grade tetrapods: the tibiale, the interme-dium, and the proximal centrale (Fig. 2). This

three-center model was the conventional wisdomuntil challenged by Rieppel (1993), who advanced anew model based on his study of tarsal ossificationin extant reptiles. Rieppel disputed Peabodys inter-pretation of juvenile Captorhinus aguti fossils, andconcluded that the ossification pattern seen in allmodern reptiles (i.e., ossification from a single cen-ter, or the one-center model; Fig. 2) was phyloge-netically ancient, because it was also present inmesosaurs, a group of aquatic reptiles from theEarly Permian. Therefore, the developmental reor-ganization necessary to replace the various stemtetrapod tarsal ossifications with the novel amniote

astragalus would be a neomorph of all Amniota. Inthis view the amniote astragalus lacks a direct ho-mologous precursor among amphibian-grade tetra-pods, and furthermore does not bear on the ongoingdebate concerning parareptile versus diapsid affini-ties of turtles (Rieppel and Reisz, 1999; Hill, 2005,and references therein).

Recent work has seemingly overturned Rieppelsone-center model. Kissel et al. (2002) presented ju-

venile, partially ossified astragali from the capto-rhinid taxa Labidosaurus hamatus and Captorhinusmajor that clearly show three ossification centers.

Additionally, Berman and Henrici (2003) docu-mented several well-preserved, partially ossified

diadectid astragali that similarly show at least threeossification centers. All of these authors haveadopted Peabodys original three-center model, i.e.,that the astragalus in basal amniotes is formed fromthe fusion of three ossification centers directly ho-mologous to the amphibian intermedium, tibiale,and proximal centrale. Phylogenetically, this find-ing implies that the ossification pattern seen in allextant reptiles, in which the astragalus ossifies froma single center (reviewed in Rieppel, 1993), mustcharacterize a clade crownward of Captorhinidae(Fig. 1). In this view the one-center model is notphylogenetically ancient, and diagnoses a clade of

derived diapsids only.One uncertainty persists concerning the three-center modelthe identity of the proximal cen-trale. Peabody (1951) is clear at the beginning of hisarticle that the amphibian tarsus (exemplified by

Acheloma) includes four centralia (c1c4); the samecondition occurs in Eryops (pers. examin., see below)and is the generally accepted ancestral conditionfrom which recent anamniote and amniote tetrapodspresumably evolved (Romer, 1956; Shubin, 2002).Distal centralia c1 and c2 (Fig. 2) are thought toform the centrale proper of amniotes (Peabody,1951; Romer, 1956), and are in fact separate in some

basal synapsids (Romer and Price, 1940). However,Peabody is vague about the fate of c3, one of the twomore proximal centralia. He maintains that the sec-ond, larger proximal centrale (c4) contributes to theamniote astragalus, and terms c4 the proximal cen-trale. Yet c3 and c4 are both proximal relative tothe positions of c1 and c2. Peabody is clearly refer-

ring to c4 as the proximal centrale; he makes onlyone brief comment on the fate of c3 (Peabody, 1951,p. 343), hypothesizing that it becomes incorporatedin distal tarsal 4 (dt4). Peabody presents no evidencefor this assertion. In the rest of the text he refers toc4 as the proximal centrale, and in all later liter-ature workers have adopted this terminology. Thereis no further discussion, however, of the fate of c3,although new fossil evidence indicates that this el-ement may in fact play a role in the formation of theamniote astragalus.

MATERIALS AND DESCRIPTION

New Fossil Evidence

Recent field work in Niger (Sidor et al., 2005) hasresulted in the collection of significant new materialof Moradisaurus grandis Taquet 1969, the largest,most derived, and youngest occurring captorhinid.The discovery of a complete, articulated, and verypoorly ossified pes (MNN MOR79, Fig. 3), in con-

junction with well-ossified subadult pedal material(MNN MOR78), affords an unprecedented opportu-nity to study the ossification history of the capto-rhinid tarsus in a large-bodied form. In the juvenilespecimen, none of the individual ossification centerscontributing to the astragalus or centrale are co-

ossified (Fig. 4A).The fossil shown in Figure 4A comprises six ossi-

fied masses, and comprises the proximo-medial ele-ments of the complete juvenile pes (OKeefe et al.,2005; Fig. 3). All elements are poorly ossified, and alllack cortical bone surface indicative of perichondralossification except for the central element (c4, ter-minology from Fig. 2), which possesses small islandsof cortical ossification on both its ventral and dorsalsurfaces. The gross homologies of this complex areeasily established; the fossil was discovered in artic-ulation (fig. 6 in OKeefe et al., 2005), just medial toa poorly ossified but identifiable calcaneum, and

proximal to a complete distal tarsal row. The centralelement also possesses a shallow groove for the per-forating artery on its ventral surface, establishingbeyond doubt that this complex contains the astra-galus. Just proximal to the central element, onelarge element is identifiable as the tibiale based onits location and its possession of a broad surface forarticulation with the tibia. Distal to this element aretwo masses that, based on position, are probablyhomologous to the distal centralia c1 and c2 of Pea-body (1951); the lack of fusion between these ele-ments demonstrates that the captorhinid centralearises via the fusion of these two centralia, a condi-

417AMNIOTE ASTRAGALUS HOMOLOGY



4/11

Fig. 2. Morphology and models of homology between the tarsi of anamniote and amniote tetrapods, based in part on Peabody(1951). Acheloma is a temnospondyl, an amphibian-grade tetrapod, and its tarsus is thought to represent the primitive condition fromwhich the amniote tarsus evolved. Captorhinus is a captorhinid eureptile. The arrows with large numbers represent models ofastragalus formation; the one-center model of Rieppel (1993), the three-center model of Peabody (1951) and later authors, and thefour-center model (this study). Four ossifications in Acheloma are color-coded to the parts of the reptilian astragalus each is thoughtto represent under each model: intermedium, blue; tibiale, yellow; c4, green; c3, orange. A, astragalus; c1c4, centralia 14; cn, fe,calcaneum, fibulare; dt15, distal tarsal 15; f, fibula; fe, fibulare; in, intermedium; t, tibia; te, tibiale.



5/11

tion previously supposed but never demonstrated.In general, the morphology of this complex is verysimilar to other juvenile captorhinid astragali illus-trated by Kissel et al. (2002), except that the presentspecimen is relatively poorly ossified.

We also studied comparative material of anotherderived captorhinid, Captorhinikos chozaensis Ol-son 1954, previously figured and discussed briefly byOlson (1962). This juvenile astragalus (USNM21275; Fig. 4B) shows three ossification centers,which Olson homologized to the amphibian interme-dium, tibiale, and proximal centrale following Pea-

bodys three-center model. However, our reexamina-tion of this specimen reveals that the astragalus isincomplete; the tibale ossification had not fused withthe rest of the astragalus at the time of death, and isabsent in the fossil. This interpretation is confirmedby the fact that the mass identified as the tibiale byOlson possesses a notch for the perforating artery, afeature always carried by the intermedium, neverby the tibiale (Romer, 1956). The juvenile C. choza-

ensis astragalus therefore replicates the conditionseen in the juvenile Moradisaurus grandis astraga-lus, and apparently ossified from four centers. TheC. chozaensis astragalus is more completely ossified,

however, and the line of fusion between the twomasses comprising the intermedium is a distinctline running completely around the element. Thelocation of this line of fusion is in an identical loca-tion to the hiatus between the two masses compris-ing the intermedium in the M. grandis astragalus.The two fossil astragali figured here thereforepresent strong evidence that the astragalus of largecaptorhinids arises through the co-ossification offour elements, not three as hypothesized by Peabody(1951).

RESULTS

Homologies of Observed Elements

According the Peabodys three-center model, theastragalus comprises ossifications homologous tothe amphibian tibiale, intermedium, and proximalcentrale (c4). However, the juvenile astragalus of

Moradisaurus grandis clearly contains four masses,not three. As stated above, the identification of thetibiale seems certain. However, the identities of theother three elements of the complex are more diffi-cult to establish.

Fig. 3. Juvenile left pes of the capto-rhinid reptileMoradisaurus grandis, MNNMOR 79, from the Upper Permian MoradiFormation of northern Niger. Fossil is in

ventral view; length 15 cm. Note lack offusion among ossification centers contrib-uting to the astragalus, bottom.




6/11

As currently prepared, the Moradisaurus grandisfossil is in two pieces, with the most proximal massseparated from the rest of the astragalus. This ele-ment is very poorly ossified and lacks any morpho-logical clues that might help establish its identity.Based on its proximal position, however, the mass isprobably the intermedium or a part of it. More distalto this mass is a second, better-ossified mass. This

mass is also securely identifiable as part of theintermediumas defined by the three-centermodelbased on its location, and by the presence ofa groove for the perforating artery on its ventralsurface. Therefore, two masses seem assignable tothe intermedium. There are at least two possibleexplanations for this: the intermedium portion of theastragalus is simply broken, or the intermediumportion ossified from two centers. There is not suffi-cient evidence in the fossil alone to differentiatebetween these two possibilities. Suggestive, how-ever, is the globular state of the proximal fragment,and the lack of a clear fit or plane of fracture be-

tween the two pieces. We can therefore constructtwo hypotheses for this fossil.

Three-center hypothesis. The juvenile Moradi-saurus grandis astragalus is comprised of the am-phibian c4, intermedium, and tibiale. The amphib-ian c3 is absent, either lost or incorporated intoanother element. The intermedium is simply brokenin half, and the lack of a clear fracture between the

pieces of the intermedium is due to the poor ossifi-cation of the element. The element c3 is either lostentirely or incorporated in dt4, although no evidenceexists for either possibility.

Four-center hypothesis. The amniote astraga-lus is comprised of both the amphibian c3 and c4, aswell as the intermedium and tibiale. Because Mora-disaurus grandis is very large relative to other cap-torhinids, ossification is delayed, and the specimenis quite young. It therefore records an early devel-opmental stage before the fusion of the c4 ossifica-tion center with the intermedium ossification center.The fusion of these two centers would occur before

Fig. 4. Fossilized juvenile astragali described in this article. A: MNN MOR79, Moradisaurusgrandis Taquet 1969, left astragalus in dorsal view, reversed. B: USNM 21275, Captorhinikoschozaensis Olson 1954, right astragalus in dorsal view; dotted line represents location of thetibiale. Abbreviations as in Figure 1. Scale bars 2 cm in A; 1 cm in B.




7/11

the fusion of the composite intermedium, tibiale,and c3 observed in smaller, more primitive capto-rhinid genera.

DISCUSSION

The realization that the astragalus of large capto-rhinids seemingly ossified from four centers sug-gests the question of the homology of these ossifica-tion centers. The tarsus of anamniote tetrapods,exemplified by Acheloma, contains four separate os-sifications in this region: the intermedium, the tibi-ale, and two proximal centralia (c3 and c4; Figs. 2,5). The identity of the most medial of these ele-ments, the tibiale, seems secure. This element is theprimary location for the articulation of the tibia, andits chondrogenetic precursor segments from that of

the tibia proper early in development (Shubin andAlberch, 1986). The tibiale also articulates distallywith the distal centralia and laterally with the cen-tral series of ossifications (Figs. 2, 5). These relation-ships of the tibiale are unchanged within amniotes(Romer, 1956), and identification of the tibiale in the

Moradisaurus grandis tarsus is based on these po-sitional relationships, as well as the presence of abroad surface for articulation with the tibia. Thissurface is well-developed even though the bone ispoorly ossified, and is quite similar to the tibialsurface found in captorhinids generally (Kissel etal., 2002).

The homologies that are most difficult to deter-mine are those of the central series, namely, theintermedium and the two centralia that lie betweenthe tibiale medially and the fibulare laterally. One

great conceptual aid in establishing the homologiesof these elements is the metapterygial axis of Shu-bin and Alberch (1986; gray elements, Fig. 5). Thispattern of limb topology is highly conservedthroughout tetrapods, and in fact is one of the fun-damental components of the pentadactyl tetrapodbauplan as defined by Hinchliffe (2002, p. 843). Theelements of this metapterygial axis (the fibulare anddistal tarsals 4 through 1) form a stable frame ofreference within the tetrapod tarsus, a frame thatdoes not change throughout early amniote phylog-eny. This frame can be used to establish the homol-ogies of other structures surrounding it. We there-fore take the calcaneumthe adult derivative of thefibulareas our landmark when interpreting thehomologies of the more medial tarsal ossifications.

Given that there are apparently four ossificationscontributing to the astragalus in the fossils de-scribed above, and four bones in this area inamphibian-grade tetrapods, it is tempting to simplyassign the identities of the amphibian bones to thereptilian ossification centers. However, there aretwo reasons to be cautious in doing this. The first isthat the tarsus in reptiles is reorganized, with co-ossifications and novel contacts between bones thatmake the establishment of homology difficult based

Fig. 5. Block models of tarsus ossificationhomologies, following the convention of Shubinand Wake (1996). Color coding matches that inFigure 1; additionally, the elements of the

metapterygial axis are colored gray. The ele-ments contributing to the amniote astragalusare outlined in bold and fused elements are in-dicated by dotted lines. The red circle indicatesthe location of the passage of the perforatingartery. The three-center model is advocated byPeabody and more recent authors; the four-center model is proposed in this article.




8/11

on positional information alone. The second, andperhaps more compelling, reason to be cautious isthe position of a soft-tissue structure, the perforat-ing artery. This vessel passes from the ventral to thedorsal surface of the autopod in both amphibian-grade and reptilian-grade tetrapods, traveling be-tween the tarsal ossifications, and leaving evidence

of its passage in the form of prominent grooves inlarger taxa (Romer, 1956). In temnospondyls, theperforating artery passes between the intermediumand calcaneum (Fig. 5); we verified this condition byexamining material of the large temnospondyl

Eryops (MCZ 7551, 7552). In captorhinids, the per-forating artery also passes between the calcaneumand the central series of ossifications, and the pres-ence of a canal for this artery formed the basis forPeabodys (1951) assertion that the large proximalmass of the astragalus was homologous to the am-phibian intermedium only.

However, in Eryops and in temnospondyls gener-

ally, the perforating artery passes between calca-neum and intermedium in a very proximal location,at the level of the proximal third of the calcaneum.In captorhinids and other amniotes, the perforatingartery skirts the calcaneum near its distal end.Therefore, relative to the calcaneum, the location ofthe perforating artery has moved distally in am-niotes, and the homology statement presence ofperforating artery canal intermedium must becalled into question. In fact, the location of the per-forating artery in captorhinids is topologically iden-tical to the position of the junction of the fibulare, c4,and dt4 in amphibians. If the hypothesis that theperforating artery has moved distally is correct, it

implies strongly that the intermedium of thethree-center model contains the ossification centersof both the amphibian intermedium and c4.

The condition of the astragalus in pelycosaur-grade synapsids would be of great interest giventhat this clade is the sister taxon to the rest of

Amniota; however, we have been unable to find anydescribed juvenile synapsid material that mightshed light on the ossification history in this clade. Itis worth noting, however, that in some synapsids theperforating artery is more proximal than in otherbasal amniotes, although it is still quite distal rela-tive to amphibian-grade tetrapods (Romer and

Price, 1940). The significance of this observation isunknown, and further research in this clade isneeded.

There is precedent for independent movement ofthe location of the perforating artery: in extant lepi-dosaurs and turtles the artery has moved to a veryproximal location between the distal ends of thetibia and fibula (Rieppel, 1993). Rieppel linked thismovement to the novel formation of an active jointbetween the astragalus and calcaneum in lepido-saurs. Captorhinids are also interpreted as possess-ing an active joint between calcaneum and astraga-lus (Holmes, 2003; OKeefe et al., 2005), and thus

might also have required a relocation of the artery.The fact that the artery moves distally in captorhin-ids, but proximally in lepidosaurs, lends credence toHolmes belief that the mesotarsal joint in capto-rhinids and other basal amniotes is not homologousto that seen in lepidosaurs.

The hypothesis that the intermedium of the

three-center model incorporates the amphibian c4 isalso supported by positional information (Fig. 5).The three-center proximal centrale is much closerin size and positional relations to the amphibian c3rather than c4. The amniote proximal centraledoes not come anywhere near the tibia, while thereis a contact between the tibia and the amphibian c4.

Also, the amphibian c4 has a large contact with thefibulare, whereas c3 does not. In order to create thecondition of the three-center model, the interme-dium must be greatly expanded to fill the position ofc4, c4 must shrink and move distally to displace c3,and c3 must either disappear or move distally for

incorporation in dt4. In the four-center model noneof these changes are necessary, and c3 and c4 retaintheir original sizes and positional relationships.Based on positional information, therefore, theproximal centrale is mostly likely identifiable withthe amphibian c3, not c4.

Developmental Transitions in Captorhinids

In the four-center model proposed here the ossifi-cation centers of the amphibian intermedium and c4retain their original sizes and locations, and bothcontribute to the captorhinid astragalus. The fact

that these two centers have not been recognized asseparate ossifications in smaller captorhinids hasseveral possible explanations. We may simply nothave fossils from young enough animals to preservethis stage of development; early proximal ossifica-tion is a common feature of tetrapod limb ossifica-tion (Hinchliffe, 2002), and the intermedium and c4might therefore be expected to co-ossify before theother elements contributing to the astragalus. How-ever, thousands of specimens of the small capto-rhinid Captorhinus aguti are known from the FortSill fissure fill deposits. Peabody (1951) believedthat he saw evidence of a tripartite astragalus in

some of this material, but his evidence was ques-tioned by Rieppel (1993) and is not definitive. Onlyin larger captorhinids (e.g., Kissel et al., 2002) isthere strong evidence for a tripartite or tetrapartiteastragalus, and the recognition of this pattern isremarkable given the relative paucity of specimensknown from these animals. The observed pattern ofprogressively delayed fusion of ossification centersseen within the Captorhinidae (Fig. 6) may there-fore be real. It is possible that a change in develop-mental timing occurred with increased body size, sothat the fusion of originally separate elements hap-pened later relative to their ossification history in




9/11

larger animals. This hypothesized change in timingwould afford progressively earlier views of astraga-lus ossification history in animals of progressivelylarge body size. The possible mechanics of this tran-

sition are a topic of current research.

Conclusion and Phylogenetic Implications

A four-center model for the homology of the am-niote astragalus is supported by several lines ofevidence. The first is new fossil data from Moradis-aurus grandis and Captorhinikos chozaensis. Well-preserved juvenile astragali from both taxa clearlydisplay four ossification centers, not one or threecenters as proposed in previous models of astragalushomology. We propose that these four centers arehomologs of the intermedium, tibiale, c3, and c4 of

anamniote tetrapods. The astragali of larger, de-rived captorhinids may show all four contributingmasses due to a delay in the onset of fusion amongtarsal elements relative to smaller, more primitive

animals; captorhinid taxa of decreasing body sizeshow a corresponding decrease in the number ofidentifiable ossifications contributing to the astraga-lus. The early state of ossification of the M. grandis

juvenile astragalus may therefore arise from a het-erochronic delay in ossification related to the attain-ment of large body size, as hypothesized for otherrelatively large captorhinids by Kissel et al. (2002).However, pelycosaur-grade synapsids reach compa-rable body sizes, apparently without this type ofheterochrony. This effect may therefore be a featureof captorhinids only rather than a generalized prop-erty of basal amniotes.

Fig. 6. Captorhinid phylogeny, body size evolution, and degree of ossification. Cladogram topology based on that of Modesto andSmith (2001) and de Ricqles (1984). Pattern of astragalus ossification is schematic, with either one, three, or four ossifications shown(see text for discussion). Thickened lines denote decreased consolidation of the braincase. Skull outlines depict relative size, based onskull measurements given in the following sources: Protothyris (Muller, pers. commun.); Romeria prima (Muller, pers. commun.);

Romeria texana (Muller, pers. commun.); Protocaptorhinus (Muller, pers. commun.); Rhiodenticulatus (Berman and Reisz, 1986);Saurorictus (Modesto and Smith, 2001); Captorhinus laticeps (Heaton and Reisz, 1980); Captorhinus aguti and Captorhinus magnus(Kissel et al., 2002); Labidosaurus (Modesto, pers. commun.); Captorhinikos (Olson, 1962); Labidosaurikos (Dodick and Modesto,1995); Moradisaurus (de Ricqles and Taquet, 1982).




10/11

The location of the perforating artery has moveddistally relative to the calcaneum in captorhinidsand other basal amniotes, and so is a poor landmarkfor use in establishing the homology of the interme-dium. Additionally, the positional relationships of c3and c4 are much more conservative under the four-center model, requiring less reorganization of the

tarsus than does the three-center model. We there-fore conclude that the captorhinid astragalus arosefrom the co-ossification of four elements present inanamniote tetrapods: intermedium, tibiale, c3, andc4. More comparative research is clearly needed onthe detailed evolution of the astragalus across theanamnioteamniote boundary, particularly in dia-dectids, where partially ossified material is known.The condition in pelycosaur-grade synapsids is alsocritical and in need of further examination.

The recognition that the astragalus of early am-niotes ossified from multiple centers is importantfrom a phylogenetic perspective because it implies

that the ossification pattern seen in all Recentreptilesossification from a single centeris a phy-logenetically recent innovation, having occurrednear the origin of Diapsida. The primitive diapsid

Hovasaurus, for which good growth series areknown, demonstrates ossification of the astragalusfrom a single center (Carroll, 1997, p. 248). Thephylogenetic position ofHovasaurus is currently un-resolved, however, and the taxon is also aquatic, atrait known to influence timing of ossification. Theearly araeoscelid diapsid Petrolacosaurus, on theother hand, shows evidence of ossification from mul-tiple centers (Peabody, 1951; Carroll, 1964). Thechange in development to a single ossification center

therefore seems localizable to somewhere within thebase of Diapsida. Furthermore, it seems clear thatthe marked similarities in the ossification and struc-ture of the tarsus between lepidosaurs and turtlesnoted by Rieppel and Reisz (1999) are shared de-rived characteristics uniting these taxa within Di-apsida. Importantly, the presence of a primitive,four-center ossification pattern in captorhinids im-plies that all nondiapsid reptiles save Parareptiliahave this primitive ossification pattern, and thatthis is strong evidence precluding an anapsid ances-try for turtles. The sole exception is Parareptilia,where the presence of the aquatic Mesosauridae as

an outgroup renders the character reconstructionequivocal. Establishing the ossification history ofthe astragalus in parareptilian taxa is therefore aprime subject for future research.

ACKNOWLEDGMENTS

We thank B. Gado, T. Lyman, S. Steyer, R. Smith,D. Sindy, and A. Dindine for assistance in the field.We thank V. Heisey, whose skilled preparation of adifficult fossil made this article possible. C. Schaffand F. Jenkins facilitated access to material in the

MCZ. J. Muller, R. Reisz, M. Coates, and K. Searssupplied stimulating discussion and relevant refer-ences. S. Sumida and an anonymous reviewer pro-

vided helpful and timely reviews.

LITERATURE CITED

Berman DS, Henrici AC. 2003. Homology of the astragalus andstructure and function of the tarsus of Diadectidae. J Paleont77:172188.

Berman DS, Reisz RR. 1986. Captorhinid reptiles from the EarlyPermian of New Mexico, with description of a new genus andspecies. Ann Carnegie Mus 55:128.

Carroll RL. 1964. The earliest reptiles. Zool J Linn Soc 45:6183.

Carroll RL. 1997. Patterns and processes of vertebrate evolution.Cambridge, UK: Cambridge Paleobiology Series 2, CambridgeUniversity Press.

Clack JA. 2002. Gaining ground. The origin and evolution oftetrapods. Bloomington: Indiana University Press.

Coates MI. 1996. The Devonian tetrapod Acanthostega gunnariJarvik: postcranial anatomy, basal tetrapod relationships andpatterns of skeletal evolution. Trans R Soc Edinb Earth Sci87:363421.

de Ricqles A. 1984. Remarques systematiques et method-olog iques pour s ervir a letude d e la fam ille des cap torhi nides(Reptilia, Cotylo sauria, Captorhinomo rpha). Ann Paleo nt 70:139.

de Ricqles A, Taquet P. 1982. La fau ne de vertebres du Per-mien Superieur du Niger. I. Le captorhinomorphe Moradis-aurus grandis (Reptilia, C otylosauria). Ann Paleont 68:33106.

Dilkes DW, Reisz RR. 1988. Trematops milleri Williston, 1909identified as a junior synonym of Acheloma cumminsi Case,1882. Am Mus Novit 2902:112.

Dodick JT, Modesto SP. 1995. The cranial anatomy of the capto-rhinid reptileLabidosaurus meachami from the Lower Permianof Oklahoma. Palaeontology 38:687711.

Gauthier JA, Kluge AJ, Rowe T. 1988. Amniote phylogeny and

the importance of fossils. Cladistics 4:105209.Gegenbauer C. 1864. Untersuchungen zur vergleichenden Anato-

mie der Wirbeltiere, I. Carpus und Tarsus. Leipzig: WilhelmEngelmann.

Godfrey SJ. 1989. The postcranial skeletal anatomy of the Car-boniferous tetrapod Greererpeton burkemorani Romer, 1969.Philos Trans R Soc Lond B 323:75133.

Heaton MJ, Reisz RR. 1980. A skeleton reconstruction of theEarly Permian captorhinid reptile Eocaptorhinus laticeps (Wil-liston). J Paleont 54:136143.

Hill RV. 2005. Integration of morphological data sets for phylo-genetic analysis of Amniota: the importance of integumentarycharacters and increased taxonomic sampling. Syst Biol 54:530547.

Hinchliffe JR. 2002. Developmental basis of limb evolution. Int JDev Biol 46:835845.

Holmes RB. 1984. The Carboniferous amphibian Proterogyrinusscheelei Romer, and the early evolution of tetrapods. PhilosTrans R Soc Lond B 306:431524.

Holmes RB. 2003. The hind limb of Captorhinus aguti and thestep cycle of basal amniotes. Can J Earth Sci 40:515526.

Kissel RA, Dilkes DW, Reisz RR. 2002. Captorhinus magnus, anew captorhinid (Amniota: Eureptilia) from the Lower Permianof Oklahoma, with new evidence on the homology of the astra-galus. Can J Earth Sci 39:13631372.

Modesto SP, Anderson JS. 2004. The phylogenetic definition ofReptilia. Syst Biol 53:815821.

Modesto SP, Smith RHM. 2001. A new Late Permian captorhinidreptile: a first record from the South African Karoo. J VertPaleont 21:405409.

OKeefe FR, Sidor CA, Larsson HCE, Maga A, Ide O. 2005. Thevertebrate fauna of the Upper Permian of Niger. III. Morphol-




11/11

ogy and ontogeny of the hindlimb of Moradisaurus grandis(Reptilia: Captorhinidae). J Vert Paleontol 25:309 319.

Olson EC. 1962. Permian vertebrates from Oklahoma and Texas.Part 2. The osteology of Captorhinikos chozaensis Olson. OkGeol Sur Circular 59:4968.

Peabody FE. 1951. The origin of the astragalus of reptiles. Evo-lution 5:339344.

Rieppel O. 1993. Studies on skeleton formation in reptiles. IV.

The homology of the reptilian (Amniote) astragalus revisited. JVert Paleontol 13:31 47.Rieppel O, Reisz RR. 1999. The origin and early evolution of

turtles. Annu Rev Ecol Syst 30:122.Romer AS. 1956. Osteology of the reptiles. Chicago: University of

Chicago Press.Romer AS, Price LI. 1940. Review of the Pelycosauria. New York:

Geol Soc Am Special Paper 28.Shubin NH. 2002. Origin of evolutionary novelty: examples from

limbs. J Morphol 252:1528.

Shubin NH, Alberch P. 1986. A morphogenetic approach to the

origin and basic organization of the tetrapod limb. Evol Biol20:318390.

Shubin NH, Wake DB. 1996. Phylogeny, variation, and morpho-logical integration. Am Zool 36:5160.

Sidor CA, OKeefe FR, Damiani R, Steyer JS, Smith RMH, Lars-son HCE, Sereno PC, Ide O, Maga A. 2005. Permian tetrapodsfrom the Sahara show climate-controlled endemism in Pangea.

Nature 434:886889.Sumida SS. 1997. Locomotor features of taxa spanning the origin

of amniotes. In: Sumida SS, Martin KLM, editors. Amnioteorigins. San Diego: Academic Press. p 353398.

Wideman NK, Sumida SS. 2004. Taxonomic status of the tetra-pod Limnostygis relictus and its bearing on the temporal dis-tribution of basal amniotes. J Vert Paleontol 24(Suppl):129A.

Zittel von K. 1932. Textbook of paleontology, 2nd English ed.

London: MacMillan.



Documents

Amniote ankle