the development and homology of the chelonian carpus and tarsus

JOURNAL OF MORPHOLOGY 186:119-131(1985)

The Development and Homology of the Chelonian Carpus and Tarsus

ANN C. BURKE AND PERE ALBERCH Museum of Comparatiue Zoology, Haruard Uniuersity, Cambridge, Massachusetts 01238

ABSTRACT The long-standing controversies involving the number and homologies of the elements of the carpus and tarsus of turtles are reviewed from a developmental perspective. The analysis is based on a detailed description of the chondrogenesis of the carpus and tarsus in the species Chelydra serpentina and Chrysemys picta. The first stage described is the differentiation of a Y-shaped chondrogenetic condensation involving the humerus (femur)-ra- diusl ulna (tibidfibula). This stage is followed by the early formation of a series of connected condensations off the distal end of the postaxial element (ulna or fibula). This linear array, which we refer to as the primary axis, comprises the ulnare-distal carpal 4-metacarpal 4 in the carpus and the fibulare-distal tarsal 4-metatarsal 4 in the tarsus. There are two precondensations that branch off the primary axis. The proximal one will soon form the intermedium while the distal one will generate a digital arch that will give rise sequentially to digits 3-2-1, in this order. Digit 5 is not part of the digital arch and forms as an independent condensation.

We emphasize that chondrogenetic foci often form as “branches” off existing condensations. This well-defined pattern of connectivity is used to establish unambiguous homologies and allows comparisons with other vertebrates. We conclude that preaxial elements such as the radiale and tibiale are absent in the turtles examined and probably in all turtles. The observed proximal elements that form in the anterior region of the limb and that are often homologized as radiale or tibiale have clear connections to the postaxial elements. For this reason we argue that these elements should be homologized as centralia. Therefore, we find two chondrogenetic condensations in the tarsus and three in the carpus, which we consider centralia. They subsequently fuse with neighboring elements in a complex pattern. We also describe the variable presence of a late-developing element in the tarsus of Chelydra, which, to our knowledge, has never been described. We propose this element to be an atavistic pretarsale.

Comparison of the chondrogenetic pattern in turtles with those described in the literature for other vertebrates indicates that there are two invariant patterns in all tetrapods with the exception of the urodeles. These are 1) the primary axis and 2) the digital arch.

The principle of ontogenetic recapitulation has had a pervasive influence in evolutionary morphology. Comparative and developmental anatomists, implicitly or explicitly, expected that during the ontogeny of higher forms primitive patterns should be observed as intermediate stages. Nowhere is this phe- nomenon more apparent than in the em-

bryological analysis of the carpus and tarsus. These are very complex systems characterized by a dynamic pattern of cartilage condensations, often poorly defined, that interact and fuse with each other. As Hinchliffe and Griffiths (‘83) have reviewed recently, the ex- pectation of recapitulation has led to much misleading information in the literature.

0 1985 ALAN R. LISS, INC.

120 A.C. BURKE AND P. ALBERCH

Since paleontological and comparative evidence indicates that the evolution of the carpus and tarsus has proceeded predominantly by a progressive reduction of elements (e.g., Schaeffer, '41; Romer, '56), many classical authors have tended to overestimate the number of cartilage condensations present during the early development of these struc- tures (e.g., Steiner, '34; Holmgren, '33). Hinchliffe and Griffiths ('83) have reviewed the evidence for the development of the carpus and tarsus in amphibians, birds, and mammals. Their work illustrates the neces- sity of reexamining the development of the carpus and tarsus in the various vertebrate groups without an initial bias expecting recapitulation.

Here, we present a detailed description of the development of the turtle carpus and tarsus as exemplified by Chelydra serpentina serpentina and Chrysemys picta picta. Tur- tles are a key group in evolutionary studies, because they may be the most primitive liv- ing amniotes (Gaffney, '80). In addition, Che- lydra is also considered phylogenetically the most primitive cryptodire based on the per- sistence of opisthocelous caudal vertebrae (Gaffney, '84). The primitive position of Che lydra is reflected in the carpus and tarsus, since Chelydra is characterized by the presence of three proximal elements in the tarsus and has as many cartilaginous foci in both hand and foot as any other turtle. Chryse mys, although not sharing the primitive phylogenetic position of Chelydra, has retained a fairly generalized tarsus and carpus.

The adult chelonian carpus and tarsus exhibit various modifications and variability ranging from the flippers of marine turtles to the shortened limbs of tortoises. This variation has been reviewed by Walker ('73); the tarsus has been reviewed by Zug ('71); and Auffenberg ('66) reviewed the derived morphology of the carpus in tortoises (but see Bramble [SZ] and Crumly [84] for a reinter- pretation of some of the skeletal homologies). Most of the differences are those of propor- tion, but there is some variation in the number of elements. Commonly cited fusions in the carpus are 1) between the radiale and centrale (kinosternids, testudinids, and pleu- rodires; Walker, '73); 2) between distal carpals 1 + 2 in some testudinids (Bramble, '82; Crumly, '84); and 3) between distal carpals 4 + 5 in most groups. In the tarsus there are always only four distal tarsals, the fourth generally articulating with metatarsals 4

and 5. Most of the variability is found in the proximal region. Primitive turtles (i.e., pleu- rodires and chelydrids) often have three distinct elements in the adult, referred to as the astragalus, calcaneum, and centrale. There is some controversy about the number of elements involved in the formation of the astragalus, which will be addressed later. Most other groups have the calcaneum fused with the astragalus without any indication of a centrale. (Zug, '71).

There is considerable debate regarding the homologies of elements as well as composi- tion of the adult bones. As pointed out by Romer ('56), most authors have considered embryonic fusion to be the main mode of reduction in number of skeletal elements in the autopodium. Holmgren ('33) (Fig. 1) con- tended that the entire primitive reptile series of ll elements appears in the turtle embryo, and this hypothesis has been widely accepted (e.g., Romer, '56, p. 381). To test this hypothesis, detailed embryological information is required. However, a review of the literature indicates that few complete studies on chelonian carpal and tarsal development have been carried out. The work of Sewertzoff ('08) on Emys lutaria (= Emys orbicularis) and Holmgren ('33) on Chryse mys marginata (C. picta marginata) are the most complete studies. Additional ontogenetic observations in Emys have been re- ported by Rosenberg (18921, Mehnert (1897), Hoffmann (18901, and Rabl ('03). However, these studies are based on limited series and are unreliable in some aspects. As empha- sized by Hinchliffe and Griffiths ('83) and Mathur and Goel ('76), these studies have often been saddled by expectations of recapitulation. For example, most studies on the evolution of the tarsus of extant reptiles use the pattern found in rhachitomous amphibians such as Eryops as the primitive condition (e.g., Auffenberg, '66). Since Eryops, as well as other labyrinthodonts, appears to have four centralia, many authors (most impor- tantly Holmgren in his influential 1933 review of the development of the carpus and tarsus in vertebrates) have postulated that the embryonic reptilian carpus and tarsus should have four centralia and that the re- duced number of such elements found in adults results from fusion during development.

Figure 1 illustrates Holmgren's schematic reconstruction of the embryonic carpusltar- sus of Chrysemys. He identified four cen-

CARPUS AND TARSUS DEVELOPMENT 121

Abbreviations a, Astragalus Ca, Calcaneum C1-C4, Centralia 1-5. Distal carpalsitarsals

Digits 1 through 5 Fibula Fibulare Femur Intermedium

Metacarpals/tarsals Pisiform Prepollex Pretarsale Radius Tibia Ulna Ulnare

Fig. 1. Holmgren’s reconstruction of a generalized embryonic turtle carpus based on his observations of Chrysemys. Synonymy of Holmgren’s terminology for the centralia and that used by other authors is as fol- lows: centrale 1 = proximal centrale; centrale 3 = medial centrale = centrale radiale; centrale 4 = lateral centrale = centrale ulnare. The presence of a prepollex in this figure is most likely an error, as Holmgren states in the text that this element never develops (‘33, p. 252).

tralia. One is a well-defined cartilaginous condensation proximal to carpalltarsal 4 and distal to the intermedium, which Holmgren refers to as centrale 4 . He recognized three additional centralia: centrale 1 was a weakly defined condensation which posteriorly merged with the intermedium. Dorsal and anterior to centrale 1, two vaguely defined condensations were homologized as centralia 2 and 3. These three centralia never separated and soon fused with the intermedium in the tarsus, or centrale 4 in the carpus, to form single elements. According to Holm- gren (‘331, the radiale and the tibiale are absent in the species studied. There is contro-

versy regarding the validity and generality of this conclusion. Many authors have identified proximal, preaxial elements as radi- ales in several species of turtles (e.g., Romer, ’56; Bramble ’82; Walker, ’73). Sewertzoff (‘08) identified a cartilaginous condensation as a radiale in his embryological study, which would correspond to Holmgren’s centrale 1 or centrale 2. In this paper, we review these controversies after describing the development of the carpus and tarsus in Chelydra serpentina and Chrysemys picta.

MATERIALS AND METHODS

Newly laid eggs of the eastern painted turtle, Chrysemys p. picta, and the common snapper, Chelydra s. serpentina, were collect- ed in June 1984 in Merrimack County, New Hampshire. The eggs were kept in their orig- inal dirt for 4-10 days until they were placed in glass bowls with damp towels in an envi- ronmental control chamber at 22°C. Eggs were periodically opened and the embryos fixed in either 10% buffered formalin, Bouin’s fixative, or half-strength Karnovsky’s for 24 hours followed by storage in cacodylic buffer.

Staging of Chelydra embryos was done based on external morphology according to Yntema (‘68). No formal staging of Chryse- mys was attempted, but stages within the developing limb buds were easily equated with Chelydra.

Eighteen specimens of Chelydra serpentina, ranging in age from day 37 through hatching (day 106 +), corresponding to Ynte- ma’s stages 14-25 (‘681, and 13 specimens of Chrysemys picta, of ages day 39 through hatching at day 94+, were cleared and stained in toto with Alcian blue 8 GX (Sigma Co, St. Louis, Mo.) for cartilagenous mucopoly- saccharides (method slightly modified after Wassersug, ’76).

Limb buds were removed from whole mount specimens, placed on glass slides, and photographed with light field and phase contrast microscopy.

In addition, fore and hind limbs of seven specimens of Chelydra and five specimens of Chrysemys were embedded in paraplast, sec- tioned at 5 and 10 pm, and stained with to- luidine blue 0 (e.g., Humason ’79).

RESULTS Adult morphology

The adult morphology of the carpus and tarsus of Chelydra serpentina serpentina and Chrysemys picta picta is illustrated in Figure


2. The adult carpus of Chelydra (Fig. 2a) usually consists of five distal carpals, a central element, and a proximal row consisting of ulnare and intermedium. The homology of the intermedium is unclear and will be discussed later. In adult specimens the central element is a long bar which extends from the medial border of the ulnare, across the intermedium, and articulates with the radius. There is no distinct radiale and the preaxial edge of the radius, not covered by the centrale, contacts distal carpal one. In some juvenile specimens, the central element is made up of two discrete elements; the postaxial one is in the position of Holmgrens’ centrale 4, and the larger, preaxial one occupies the position of the other three (Fig. 1). Distal carpals 4 + 5 generally fuse in older individuals; the fifth is extremely small. A pisiform lies just lateral to the ulnare. It is quite small and is often lost in preparation of dry skel- etons.

2a

P

The tarsus of Chelydra (Fig. 2b) undergoes a good deal of fusion during ontogeny, though the timing of this is extremely variable. In large individuals there is a single proximal element that articulates with the four distal tarsals; distal tarsal 5 is assumed to be absent. There is generaly an obvious suture marking the union of the fibulare (calcaneum) to the intermedium (astragalus). In young specimens the intermedium, fibulare, and distal centrale are often free. This condition has generally been attributed to the primitive phyletic position of this animal. The fusion may also occur in the embryo, resulting in hatchlings with a single proximal element.

The carpus of adult Chrysemys (Fig. 2c) differs from that of Chelydra in the extent of fusion and the presence of a separate “radiale.” Distal carpals 4 + 5 are fused and the centrale is always a single element in hatchlings. It is considered the product of fusion

c2 -

2c

I

4 + 5

2

Fig. 2. Adult morphology of manus and pes in Chelydru and Chrysemys. Note that this specimen of Chelydru exhibits incomplete fusion of the centralia and proximal tarsal elements. a) Chelydru right manus; b) Chelydra right pes; c) Chrysemys right manus; d) Chrysemys right pes. Bar = 2.5mm.


between a medial and lateral centrale (Bram- ble, ’82). Its position is similar to that of Chelydra. Distally it articulates with carpals 1,2, and 3, proximally with the ulnare, intermedium, and the radius. A “radiale” is lo- cated at the preaxial border of the centrale in a more ventral plane. It separates the radius from the first distal carpal. A pisiform is present as in Chelydra.

The adult tarsus of Chrysemys (Fig. 2d) shows a higher and more consistent degree of fusion than Chelydra. There is only one proximal element, the fusion of elements oc- curring during embryonic development, and there are generally no sutures visible. This proximal element articulates with four distal tarsals.

Development Y-condensation stage

The earlier stages of chondrogenesis observed in our cleared and in toto stained preparations occur as the limb bud begins to broaden distally to form a paddle (for Chely-

dra, stage 14 of Yntema ’68). At this stage the three proximal elements of the limb are visible as a continuous Y-shaped condensation (Fig. 3). The femurhumerus stains darker, which is suggestive of an earlier differentiation and onset of matrix secretion. The radius and ulna (tibia and fibula) appear as divergent cartilaginous rods terminating in loose mesenchymal condensations at the level of the base of the paddle. At this stage the joint region has not yet differentiated. This initial formation of cartilaginous elements as a continuous structure that subsequently breaks up into its final components, the joints forming later in the intermediate region, is a common feature not only in turtle limb development but also in other vertebrates (Shubin and Alberch, ’86).

Primary axis stage and derivatives (intermedium and digital arch)

This stage is very characteristic in the development of both carpus and tarsus in the two species studied. At Yntema’s stage 15 for Che-

Fig. 3. Y-condensation stage of Chelydra serpentina, stage 14, day 37, right hind limb. Note the continuous chondrification of the femur branching into the tibia and fibula. Bar = 266 pm.

Fig. 4. The primary axis stage of Chelydra, stage 15, day 44, right fore limb. Note connectivity between the ulna and ulnare. Carpal 4 has “budded off’ the ulnare and the primordia of the metatarsal is connected to it. Asterisk indicates the early precondensation of the digital arch. Faint condensations of the prospective intermedium and centrale 4 can also be observed (see Fig. 5 for a diagram). Bar = 154 pm.


lydra, and day 40 for Chrysemys, a proximo- distal series of connected condensations appear at the distal end of the ulndfibula. These condensations are the ulnarelfibulare and distal carpalltarsal 4 (this element, de- spite its rounded morphology in the adult, initially forms as an elongated condensation). We refer to this pattern, illustrated in Figure 4, as the primary axis. The axis is continued by the distal differentiation of metacarpal (-tarsal) 4. The elements of the primary axis are clearly connected. They appear as darkly stained cartilaginous condensations connected by loosely condensed, lightly stained areas. The metachromatic stain indicates the presence of cartilage-specific glycosamino- glycans.

Two branches originate from the primary axis. They appear as faintly stained regions of elongated shape that posteriorly connect to the primary axis and extend anteriorly in a 45” angle (Fig. 5). These are the precondensations of the intermedium and of the digital arch. The intermedium, at its proximal end, branches off of the distal end of the ulna (fibula), merging medially with the distal tip

of the radius (tibia). Its proximal border is clearly defined, being continuous with the primary axis. There is no clear distal or anterior border at that stage. Later on, the intermedium splits from the primary axis and forms a well-defined cartilage condensation between the radius and ulna (tibia and fibula). The second branch appears at the level of the ulnare (fibulare) extending distally and forming a “V” with the primary axis. At the level of carpal (tarsal) 4, this branch turns anteriorly at a right angle to the primary axis. This is the anlage of the digital arch (Fig. 6), which will give rise sequentially to distal carpals (tarsals) 3, 2, and 1. Digit 4 forms as a continuation of the primary axis. Digits 3, 2, and 1 form by the sequential condensation of distal carpalsltarsals within the loose precondensation of the arch. The metacarpals (-tarsals) form as a continuous branch of the corresponding distal carpal (tarsal).

Digit 5 appears concurrently with digit 3 and has no obvious connection with either the primary axis or the digital arch. Its formation is initiated, in the manus, with the

Fig. 5 . Illustration to show proximal and distal branches off of the primary axis. The former will be the intermedium and the latter will give rise to the digital arch.


Fig. 6 . Digital arch, Chelydru, stage 15+, day 44. a) Right fore limb; b) right hind limb. Asterisks show the anterior edge of the digital arch at these stages. Centrale 4 is already a well-defined chondrogenetic foci at this stage in the manus. Also note digit 5 forming as a “ray” independent of the digital arch that develops anteriorly.

condensation of carpal 5, which subsequently extends distally and gives rise to the metacarpal and phalanges. In contrast, the metatarsal is the first element to differentiate in the fifth digit of the pes, our data indicate that distal tarsal 5 does not form at all in the two species examined (Fig. 6). Thus, the sequence of digital differentiation is 443, 51-2-1.

Differentiation of the central region of the carpus

During the differentiation of the derivatives of the digital arch, the central regions of the carpus and tarsus remain undifferen- tiated with the exception of the previously mentioned condensation of the intermedium. As shown in Figure 7, there is a region of precondensation that stains lightly with Al- cian blue, extending from the distal end of the ulna (fibula) to the first distal carpal (tarsal). Within this region the intermedium appears as a well-defined condensation. Cen- trale 4 is the next carpal element to differentiate. It is surrounded by the ulnare and intermedium proximally, and distal carpals 3 and 4 distally (Fig. 7).

Figure 8 illustrates a slightly later stage in the development of the carpus. The elements previously discussed are clearly defined and discrete condensations. Further- more, it underscores the fact that most of the carpal development occurs in the postaxial region. There is no evidence of anterior condensations in the region immediately distal to the radius. At this stage (Chelydra, stage

161, the initial condensation of another centrale is visible as a faintly stained region medial to the centrale 4.

Figure 9 shows this centrale already well differentiated as an elongated structure that appears to be growing in an anterior direction. In later stages, an element appears in the position of the radiale. It lies at the preaxial boundary of the elongate central element and slightly ventral to it. It appears to be connected to the centrale, and it is not clear from our series whether it arises independently or as a continuation of the centrale, separated only by a constriction of the condensation. It is apparent in the embryos of both species studied, but ossifies as an independent element in Chrysemys only. This element and the pisiform are the last carpal elements to differentiate. According to Romer (’56), the pisiform is probably a sesamoid cartilage, a neomorph in reptiles associated with a tendinous insertion.

Differentiation of the central region of the tarsus

Shortly after, or simultaneously with, the differentiation of the distal tarsals and metatarsals along the digital arch, we observe an amorphous rectangular precondensation that occupies most of the proximal tarsal region (Fig. 10). When this region is examined under a phase-contrast microscope, we can see three proximal condensations. One is the already well-differentiated fibulare (calcaneum). The medial condensation is also well delimited and corresponds to the intermedium. The third is a faint condensation an-


Fig. 7. a) Chrysemys, day 45, right fore limb showing relationship of the primary axis, intermedium, and the centrale 4. The digital arch is complete. Note that the radius does not contribute to the central blastema. Bar = 227 pm. b) Increased magnification of boxed area showing the condensation that extends off the primary axis to base of digit I encompassing the intermedium and centrale 4. Bar = 75 pm.

Fig. 8. Chelydru, stage 16, day 49, left fore limb. Note predominance of postaxial elements. Bar = 290 pm.

Fig. 9. Chelydru, stage 19, day 56, left fore limb. Condensation of centrale 3 and of the pisiform. Bar = 210 pm

terior to this element. The postaxial border of the intermedium is clearly separated from the fibulare but the medial edge is continuous with the third area of condensation, which is particularly obvious in Chrysernys. This condensation does not seem to be closely associated with the tibia but rather has the position of Holmgren’s centrale 1, or the proximal centrale. By the stage when the first phalanges are appearing, the anterior

edge of the intermedium has expanded and apparently incorporated the preaxial condensation, resulting in a single element (Fig. 11). An alternative explanation to this new pattern would be that the preaxial condensation disappears and the intermedium, instead of fusing, is invading the preaxial region. How- ever, the morphology of the condensation shown in Figure 11 seems to support the fusion hypothesis.


In Cheldyra, one centrale (Holmgren’s centrale 4) appears as an independent element immediately proximal to tarsal 2 (Fig. 12). This element is later incorporated into the growing mass of the intermedium. There- fore, we conclude that in Chelydra, at least, the astragalus is composed of three distinct elements: the intermedium proximomedi- ally, the centrale 4 distally, and the centrale 1 anteriorly.

We have observed the following variation regarding the final tarsal morphology of the two species studies. In both Chrysemys and Chelydra the fibulare (calcaneum) fuses with

the astragalus (Fig. 2). Chelydra has also been observed to have a free distal centrale, fibulare, and astragalus. The presence of this variation should not be surprising, since it reflects patterns of incomplete fusion within the normal ontogeny.

Of special note is the variable presence in Chelydra of an additional element in the tarsus (Fig. 13). This element can either be a neomorph or, based on its position, it may be a pretarsale like that found in the rachitome Trematops, presumably reflecting the presence of a prehallux in the chelonian ancestor (Romer, ’56). It has been observed in late

Fig. 10. Chrysemys, day 45, left hind limb. Note the condensation anterior to the intermedium, which we refer to as the proximal centrale (centrale l), is not connected to the tibia and is part of the post axial blastema. Bar = 225 pm.

Fig. 11. Chrysemys, day 49, right hind limb. Arrow indicates region of “fusion” between the intermedium and the proximal centrale. Bar = 200 pm.

Fig. 12. Chelydra, stage 19, day 56. Left hind limb showing condensation of the distal centrale (centrale 4). Bar = 190 pm.

Fig. 13. Chelydra, stage 20, day 71. Ventral view of right hind limb showing an atavistic pretarsale. Note that the distal centrale has fused to the proximal element to form the astragalus.


embryos and hatchlings and is the size and shape of the pisiform in the hand. It is the last element to form.

DISCUSSION Primary axis

The earliest event in the development of the carpus and tarsus is the formation of a linear array of chondrogenetic foci linked to each other by faintly staining cartilaginous matrix and proximally connected to the ulna (fibula). These cartilaginous primordia correspond to the ulnare (fibulare) carpale (tar- sale) 4 and metacarpal (metatarsal) 4. We refer to this pattern as the primary axis. The term is only meant to emphasize the early appearance of this arrangement and its po- tential organzing role in autopodial morpho- genesis. It is not necessarily equivalent to the metapterygial axis of earlier authors, such as Sewertzoff (‘08) and Holmgren (‘331, a matter which is not addressed in this paper.

Besides the turtles studied here, comparative data show that the early formation of a linear array of elements at the level of the fourth digit is also found in the fore and hind limbs of the lizard Calotes (Mathur and Goe1,’76), the lizard Agama and crocodiles (Holmgren, ’33), and in the chick and the frog Xenopus (Hinchliffe and GriEths, ’83). Of the species surveyed, it is absent only in the salamander Ambystoma (Shubin and Al- berch, ’86).

Two prechondrogenic blastemas “branch off’ the primary axis in an anterior direction (Fig. 5). The proximal one will give rise to the intermedium and the distal one will form the digital arch. This very characteristic pattern of “connectivity” among developmental anlagen not only suggest specific morphogenetic mechanisms but also can be used to define homologies (see below).

The species examined exhibit a well-defined arch of prechondrogenic condensations extending from the primary axis in an anterior direction (Fig. 6). This arch will give rise to carpals (tarsals) and metacarpals (metatarsals) in the following sequence: 4, 3 -+ 2 + 1. Digit 5 is not part of the digital arch but develops as an independent focus. The presence of a digital arch has been observed in most tetrapods studied (Holmgren, ’33); Milaire, ’78; Hinchliffe and GriEths, ’83). Again, the sole exceptions are the urodeles, where a digital arch is present but develops in an anterior to posterior sequence (i.e., 1 -+ 5) (unpublished data).

Our observations corroborate the earlier observations of Sewertzoff (‘08) and Holm- gren (‘331, who pointed out that most cartilaginous elements form in the postaxial region of the limb bud (Figs. 8,ll). The three preaxial digits form as a part of the digital arch which branches off the primary axis. Simi- larly, the intermedium and centralia originate in the postaxial region and seem to ex- pend anteriorly.

Homology of tarsal and carpal elements There is considerable confusion regarding

the homologies of the preaxial and central elements in the carpus and tarsus of turtles. Most decisions on homology have been based on anatomical position in the adult or on a recapitulationist idea of an archetypal pattern. Here we base our homologies on the embryonic origin of the elements. As pointed out by Holmgren and many other classical authors, embryonic tarsal and carpal elements differentiate in well-defined relationship with each other. Cartilaginous elements do not usually appear as independent foci, but very often are observed to derive as branches or “buds” off of existing chondro- genic or prechondrogenic blastemas. This pattern of connectivity appears to be highly invariant. For example, the intermedium can be defined as a postaxial element branching off the primary axis at the level of the distal end of the ulndfibula (Fig. 6). Similarly, the centralia are defined as postaxial elements emerging from the prechondrogenetic blastema that appears posteriorly connected to the primary axis (Fig. 7). In fact, the pattern of connectivity among the carpal and tarsal elements is almost as invariant as the presence of a single proximal element in the limb (humerus/femur) which branches distally into two components (radiushlna, tibidfibula).

Except in those specimens of Chelydra where two separate elements persist, there is a single bar-like centrale in the manus of adult Chrysemys and Chelydra (Fig. 2). There is, however, much controversy in the literature regarding the number of embryonic primordia present in the central region of the carpus. Holmgren (‘33) interpreted the embryonic carpus of Chrysemys as being close to the primitive condition observed in the rachitomous amphibian Eryops, i.e., he postulated the presence of four centralia in the embryo (Fig. 1). Steiner (’34) also identified four, whereas Sewertzoff (‘08) found only two. This discrepancy is easy to understand since


many of the elements were defined not on the basis of well-formed cartilaginous foci, but on amorphous precondensations (“Verk- norpelung begriffene Anlage,” Sewertzoff, ‘08, p. 157). Our data support the presence of three foci in the carpus; we cannot find evidence of a proximal centrale or of two foci in the intermedium. The two early condensations seen in Figure 9 correspond to Holm- gren’s centrale 3 and centrale 4. On the basis of their position, these elements can be homologized with Romer’s lateral and medial centralia (‘56) and Sewertzoff s centrale ulnare and centrale radiale (‘08). The third centrale forms very late and in close association with the preaxial edge of centrale 3. As discussed below this later element, often identified as a radiale, corresponds to Holmgren’s centrale 2 (Fig. 14a).

The central region of the tarsus shows the greatest contrast between the two species studied. Chrysemys has no remnant of a distal centrale, while in Chelydra it appears as a distinct condensation and often persists to ossify independently in hatchlings.

In contrast to the situation in the carpus, there is a tarsal precondenstion that corresponds to Holmgren’s proximal centrale (centrale 1). This element is soon incorporated into the expanding intermedium and is never more than an ill-defined condensation within the expansive postaxial blastema (Fig. 14b).

In amphibians, the preaxial radiale and tibiale form as condensations initially continuous with the distal end of the radius or tibia, much like the elements in the primary

axis of frogs and amniotes do from the ulna or fibula (Subin and Alberch, ’86). Using the criterion of embryonic origin (connectivity) to define homology, we can only interpret the anterior precondensation in the tarsus to be the remnant of the proximal centrale, which merges with the intermedium to form the astragalus.

This conclusion contrasts with a tradi- tional view that the astragalus is made up of the intermedium and the tibiale (Sewertzoff, ’08; Zittel, ’32; Hyman, ’42) or that it represents only the tibiale or only the intermedium (Romer, ’56). Peabody (‘51) has shown that in the primitive reptilian tarsus, as illustrated by Captorhinus, the astragalus is made up of the tibiale, the proximal centrale, and the intermedium. Schaeffer (‘41) has postulated that the tibiale has been lost from the tarsus of later reptiles. Since no condensation connected to the distal end of the tibia is observed, we conclude the tibiale to be absent in the species studied, in agreement with Holmgren and Schaeffer.

In the manus of many adult chelonians, the element having a preaxial position and articulating with the radius and distal carpal is usually identified as the radiale. In late stages of both Chelydra and Chrysemys, there is a small element at the preaxial end of the centralia. In Chrysemys this ossifies independently wig. 24, but in Chelydra it either never ossifies or becomes intimately fused with the centralia. This element does not seem to be derived from the radius but rather to arise as an extension of the centrale. Us-

:\ 0 /; , I

14 Fig. 14. Diagrammatic summary of our observations regarding the elements that differentiate in the Chelydru

and Chrysemys carpus fa) and tarsus (b). Dotted lines indicate the pattern of fusion. The astragalus is illustrated in b as the area enclosed by the broken lines and results from the fusion of the intermedium and centralia 1 and 4 (this latter element was not present in Chrysemys).


ing the criteria for the establishment of homology based on embryonic connectivity, we propose that, instead of the radiale, this element represents the centrale 2 identified by Holmgren (Fig. 1).

Thus, we conclude that the chelonian carpus and tarsus, a t least as exemplified by the two species studied, is characterized by the absence of truly preaxial elements in the carpus and tarsus. A possible exception may be the observed “atavistic” pretarsale observed in Chelydra (Fig. 13). However, its later appearance seems t o suggest that this is a sesamoid cartilage. Further work on the origin and identity of this element is necessary prior to further conclusions. In particular, an examination of the carpus and tarsus in a generalized member of the primitive group Pleu- rodira is necessary to test the generality of the pattern we have observed in cryptodires. The pleurodire Elseya is described as having a radiale (Elrambe, ’82). The origin and therefore homology of this element in this primitive group is a matter that deserves further investigation.

Using this argument of homology we can reassess nomenclatorial decisions made by some other authors. For example, Hinchliffe and GriEths (‘83) have identified a radiale in the chick wing bud. Their photographs and schematic illustrations (Figs. 4 and 5, p. 106) show this “radiale” to branch off the ulna and connect with distal carpal 3. This would correspond to the position and origin of the intermedium in turtles. Our interpretation of Hinchliffe and Griffiths’ Figures 4 and 5 suggests the presence of two precondensations, the medial one corresponding to an intermedium, and the preaxial to either radiale or a proximal centrale. The connectivity of the element cannot be determined from their figures, but following the homologies in the amphibian limb, it is a radiale only if it arises from the radius.

To cite another example, Zug (‘71) con- cluded that the singre proximal element in the tarsus of trionychids was composed only of astragalus. He rejects the contribution of a calcaneum (fibulare) because he found only one center of ossification in this element in juvenile specimens. This interpretation is very unlikely in light of the presence of the primary axis in a wide range of organisms. The fibulare, as we have observed in Chely- dra and Chrysemys, occupies an extremely strategic position in the tarsus. Its existence may be ephemeral in terms of subsequent

ossification, but it is doubtful that trionychids deviate from this general pattern. An examination of trionychid tarsal development offers an excellent opportunity to test the validity of the general pattern in the face of conflicting data from subsequent ontogeny.

In addition, knowledge of the sequence of development can be used to make some pre- dictions about what kind of morphological transformations are more likely to occur in phyIogeny. Empirically, is has been shown that elements forming late in ontogeny are more variable phylogenetically (e.g., Alberch, ’83). This is particularly true if hetero- chrony-the process of phylogenetic diver- sification by regulation of timing and rates of development (Gould, ’77; Alberch et al., ’79)-is the mechanism mediating morphological evolution. We can use this perspective to comment on the expected patterns of tarsal and carpal variation in turtles. For example, the centrale 2, the pisiform, and the pretarsale, which are the last elements to differentiate in the carpus and tarsus, respec- tively, are bound to be lost at high frequen- cies. There are high levels of interspecific variation regarding the presence or absence of the pisiform and the “radiale” (Romer, ’56; Crumly, ’84) and the pretarsale is probably an atavistic element, which is found to be variable at the intrapopulational level in Chelydra. Another expected source of variation involves the number of central and proximal elements. We would expect patterns of incomplete fusion within the normal ontogeny in this region. We have observed some adult Chelydra with two centralia. Finally, there has been a fair amount of controversy in the specialized literature regarding the loss of digits in four-digited tortoises (Crumly, ’84). For instance, in Homopus areolatus, Auffenberg (‘66) postulated that the fifth digit is the absent one, while Hewitt (’37) considered the first digit to be mising. As shown by Alberch and Gale (‘85) the patterns of digital loss in amphibians are highly invariant and congruent with developmental factors. Frogs, which have a digitial arch similar to turtles, lose the first digit, while salamanders, which have a “reversed” digital arch, lose the fifth. On the basis of this evidence we would expect turtles preferentially to lose the first digit. This should especially be the case if the transformation is the result of paedomorph- osis.

In conclusion, our observations agree with Holmgren’s interpretation of chelonian limb


development on several points. These are the absence of a tibiale and radiale, the presence of a proximal centrale in the tarsus and three distal centralia in the carpus. Our approach is, however, very different from classical views involving recapitulation. The general patterns that we illustrate, the primary axis and the digital arch, do not represent Hae- ckelian recapitulation. They do not correspond to the adult morphology of a distant ancestor but are transient embryonic stages that indicate a common mode of morphogen- esis. The ubiquitous nature of this simple pattern in the early stages of limb development in frogs and amniotes allows for unambiguous definitions of homology in the tarsal and carpal region (see Shubin and Al- berch [86] for further discussion on this topic). Homologies are here established not on comparisons between embryonic stages and an- cestral morphology but rather on conserved morphogenetic rules that generate organized patterns.

ACKNOWLEDGMENTS

David Carroll assisted in the collection of eggs and Annette Cote in histological preparation. C. Crumly critically read the manu- script. Laszlo Meszoly drew the illustrations. The typologists provided many criticisms and moral support. This work was funded by

’ National Science Foundation grant BSR- 8407437 to P.A. A.C.B. is an NIH trainee under GM-07598-07.

LITERATURE CITED

Alberch, P. (1983) Morphological variation in the neo- tropical salamander genus Bolitoglossa. Evolution 37:906-919.

Alberch, P., and E. Gale (1985) A develomental analysis of an evolutionary trend digital reduction in amphibians. Evolution 39:8-23.

Alberch, P., S.J. Gould, G.F. Oster, and D.B. Wake (1979) Size and shape in ontogeny and phylogeny. Paleobiol- ogy 5296-317.

Adenberg, W. (1966) The carpus of land tortoises (Tes- tudininae). Bull. Florida State Mus. 10:159-191.

Bramble, D.M. (1982) Schaptochelys: Generic revision and evolution of gopher tortoises. Copeia 4:852-867.

Crumly, C. (1984) The evolution of land tortoises (family Testudinidae). Unpublished Ph.D. Dissertation. New- a r k Rutgers University.

GaSney, E.S. (1980) Phylogenetic relationships of the major groups of amniotes. In A.L. Panchen (ed): The Terrestrial Environment and the Orign of Land Verte- brates. New York Academic Press, pp. 593-610.

Gaffney, E.S. (1984) Historical analysis of theories of chelonian relationship. Syst. Zool.33:283-301.

Gould, S.J. (1977) Ontogeny and Phylogeny. Cambridge: Harvard University Press.

Hewitt, J . (1937) A note on the relationships of the Cape genera of land tortoises. S. Afr. J. Sci. 33:788-796.

Hinchliffe, J.R., and P.J. Griffiths (1983) The prechondrogenic patterns in tetrapod limb development and their phylogenetic significance. In B.E. Goodwin, N. Holden, and C.G. Wylie (eds): Development and Evolution. England Cambridge University Press, pp. 99-121.

Hoffman, C.K. (1890) I. Schildkroten. In H.G. Bronn’s Klassen und Ordnung des Thier-Reichs. Leipzig: C.F. Wintersche Verlag.

Holmgren, N. (1933) On the origin of the tetrapod limb. Acta Zool. 14:184-295.

Humason, G.L. (1979) Animal Tissue Techniques, 4th ed. San Francisco: W.H. Freeman and Co.

Hyman, L.H. (1942) Comparative Vertebrate Anatomy. Chicago: University of Chicago Press.

Mathur, J.K., and S.C. Goel (1976) Patterns of chondrogenesis and calcification in the developing limb of the lizard Calotes versicolor. J. Morphol. 149:401-420.

Mehnert E. (1897) Kainogenesis eine gesetzmassige Abanderung der embryonalen Entfaltung in Folge von erblichen Uebertragung in der Phylogenese erworbe- ner Eigenthumlichkeiten. Jena G. Schwalbe, Morphol. Arbeiten 7;l-157.

Milaire, J . (1978) Etude morphologique, histochemique et autoradiographique du development des squelette des membres chez l’embryon de souris: I Membres An- tetieurs. Arch. Biol (Bruxelles) 89:169-216.

Peabody, F.E. (1951) The origin of the astragalus in reptiles. Evolution 5:339-344.

Rabl, C. (1903) Uber einige Probleme des Morphologie. Verh. Anat. Ges. 17: 154-190.

Romer, A.S. (1956) Osteology of the Reptiles. Chicago: University of Chicago Press.

Rosenberg, E. (1892) Uber einige Entwicklungsstadien des Handskelettes der Emys lutunia. Morphol. Jahrb. 18:l-54.

Schaeffer, B. (1941) The morphological and functional evolution of the tarsus in amphibians and reptiles. Bull. Am. Mus. Nat. Hist. 78t395-472.

Sewertzoff, A.N. (1908) Studien uber die Entwicklung der Muskein, Nerven und des Skelettes der Extremi- taten der niederen Tetrapoda. Beitrage zu einer Theo- rie des pentadactylen Extremitat der Wirbeltiere. Bull Soc. Nat. Moscou (N.S.) 21:l-432..

Shubin, N., and P. Alberch (1986) A morphogenetic approach to the origin and organization of the tetrapod limb. In M. Hecht, G. Prance and B. Wallace (eds): Evolutionary Biology 19B. New York Plenum Press.

Steiner, H. (1934) Ueber die embryonale Hand-und-Fuss- Skelett- Anlage Bei den Crocodiliern, sowie iiber ihre Beziehungen zur Vogel-Flugelanlage und zur ursprung- lichen Tetrapoden-Extremitat. Rev. Suisse Zool. 41: 383-396.

Walker, W.F. Jr. (1973) The locomotor apparatus of Tes- tudines. In C. Gans (ed): Biology of the Reptilia. Lon- don: Academic Press Vol. 4, pp. 1-100.

Wassersug, R. (1976) A procedure for differential staining of cartilage and bone in whole formalin fixed vertebrates. Stain Technol. 51:131-133.

Yntema, C.L. (1968) A series of stages in the embryonic development of Chelydra serpentina. J. Morphol. 125: 219-252.

Zittel, K. Von (1932) Textbook of Paleontology 2nd Eng- lish ed, Vol. 11. London: Macmillan and Company.

Zug, G.R. (1971) Buoyancy, locomotion and morphology of the pelvic girdle and hindlimb, and systematics of Cryptodiran turtles. Misc. Publ. Mus. Zool. Univ. Michi- gan No. 142, pp. 1-98.

the development and homology of the chelonian carpus and tarsus

Documents