Turtles as hopeful monsters

What the papers say
Turtles as hopeful monsters
Olivier Rieppel
Summary
A recently published study on the development of the
turtle shell (1) highlights the important role that development
plays in the origin of evolutionary novelties (1) . The
evolution of the highly derived adult anatomy of turtles is
a prime example of a macroevolutionary event triggered
by changes in early embryonic development. Early
ontogenetic deviation may cause patterns of morphological
change that are not compatible with scenarios
of gradualistic, stepwise transformation. BioEssays
23:987±991, 2001. ß 2001 John Wiley & Sons, Inc.
Introduction
Turtles are a very popular group of reptiles that have played an
important role in the folklore and mythology of many cultures
through the ages. Recently, turtles have gained new prominence
in the study of amniote relationships and evolution.
What makes turtles so special is their peculiar appearance.
Their body is ``boxed in'' between the carapace that covers
the back of the trunk, and the plastron, which covers the belly.
The turtle shell is also referred to as dermal armor, in reference
to the fact that the bones that make up the turtle shell ossify in
the dermis, the deep layer of the skin.
Department of Geology, The Field Museum, 1400 S Lake Shore Drive,
Chicago, IL 60605-2496.
Correspondence to: Olivier Rieppel, Department of Geology, The Field
Museum, 1400 S Lake Shore Drive, Chicago, IL 60605-2496.
E-mail: rieppel@fieldmuseum.org
The turtle shell
The capacity to form bone and other hard tissues such as
dentine and enamel in the skin, at the interface between the
dermis and the superficial epidermis, was realized in the
earliest vertebrates, jawless fishes whose fossil record dates
back some 450 Mio years. Well-known examples of dermal
ossifications among modern animals are the bony scales of
fishes, or the bony plates embedded in the skin of crocodiles
and many lizards. Such bony plates, developing in the dermis,
are referred to as osteoderms. The scales of fishes or the
osteoderms of crocodiles and lizards, are always arranged in a
single layer. In most turtles, the carapace also forms a single
layer of bone, but there are some species in which the
carapace consists of two superimposed sets of ossifications.
This then requires the distinction of deeper thecal and
superficial epithecal ossifications. (2±7) The thecal ossifications
are always arranged in a very regular, strictly defined geometry
that is closely comparable throughout turtles (Fig. 1A).
A central longitudinal row of neural plates cap the neural
spines of the dorsal vertebrae; a lateral row of costal plates is
closely associated with the dorsal ribs (Fig. 2); a marginal row
of marginal plates, an anterior nuchal plate, and one or two
posterior pygal plate(s) complete the carapace. Epithecal
ossifications are osteoderms that develop superficial to the
thecal ossifications (Fig. 1B). Their appearance and arrangement
varies among turtles.
Early theories (2,8) attempted to explain the evolution of the
turtle shell in the context of a step-wise, hence gradual process
of transformation. The distant ancestor of turtles was
hypothesized to have had a body loosely covered by
osteoderms. Within the evolutionary lineage leading to turtles,
the number of osteoderms would have gradually increased,
until the bony plates would eventually have provided a
complete covering of the trunk, thus forming an epitheca.
Thecal ossifications would have developed below the epitheca
at later stages in the evolution of the turtle body
plan, while epithecal ossifications would subsequently be
lost at even more advanced stages of turtle evolution.
This theory met with various difficulties, however, such as
the fact that the earliest fossil turtle (Proganochelys) (9) from
the Upper Triassic of Europe (215 Mio years) has a complete
theca. Furthermore, epithecal ossifications appear later than
ossifications of the theca in development and, in modern
turtles, epithecal ossifications tend to form in evolutionarily
relatively advanced forms only.
The capacity to form dermal bone is certainly not a specialty
of turtles, but their arrangement in a carapace and plastron
is. Turtles are unique among tetrapods, however, in that
the shoulder blade (scapula) lies inside the rib cage (Fig. 3).
The reason for this inverse relationship of the scapula is
the close association of the ribs with the costal plates of
the theca.
The development and evolution
of the turtle shell
In support of a gradual evolution of the turtle body plan, some
authors tried to explain this position of the scapula by a
backwards shift of the pectoral girdle in turtles. A recent theory
proposed the evolution of turtles from Paleozoic Pareiasaurs
by a process of ``correlated progression''. (10) Correlated key
elements of this progressive transformation are an increase in
the number of osteoderms until they form a closed dorsal
shield (carapace), the broadening of the ribs below this dorsal
BioEssays 23:987±991, ß 2001 John Wiley & Sons, Inc. BioEssays 23.11 987

What the papers say
Figure 1. Schematic representation of the turtle carapace in
dorsal view (A) and in a transverse section (B) (A: after A.F.
Carr, Handbook of Turtles, Cornell University Press, 1952,
Fig. 2c).
shield, the shortening of the trunk, the immobilization of the
dorsal vertebral column and a backwards shift of the pectoral
girdle.
This evolutionary scenario (10) not only requires that the
turtle carapace be derived from a fusion of ancestral osteoderms,
but also encounters the problem that there is no, or only
a very minor, backward shift of the pectoral girdle in turtles. (11)
Instead, the scapula of turtles comes to lie inside the rib cage
because of a deflection of rib growth to a more superficial
position. (12) Recent developmental work (1,13,14) has identified
inductive interaction generated by the carapacial ridge as
probable cause of this deflection of rib growth. Here is how it
works.
The vertebrate embryo is organized in three germ layers,
the outer ectoderm, the inner entoderm, and the mesoderm
between the two. During development, that part of the
mesoderm lying alongside the notochord (paraxial mesoderm)
becomes compartmentalized into segments, discrete blocks
of cells or somites. From the somites originate, by embryonic
cell proliferation and migration, the endoskeleton, the voluntary
body musculature, and the dermis, the deep layer of
the skin. The somites define the primary segmentation of the
vertebrate body, which in the adult is still reflected in the
segmented body musculature of fishes or salamanders, for
example. To bend the body into a lateral curve during
locomotion, the segmented body muscles must bridge the
intervertebral joint. This means that the joint between
vertebrae must lie within the primary body segments, which
also allows the spinal nerves to exit between two vertebrae on
their way to the body muscles. By that arrangement, the
vertebral body itself, and the rib attached to it, come to lie in
between the primary body segments. This repositioning of the
vertebrae relative to the primary body segments is achieved by
resegmentation of the somites. Each somite splits in half, and
the posterior part of one somite recombines with the anterior
part of the succeeding somite to form a vertebra.
As a turtle embryo grows and develops, the contours of the
future carapace are soon mapped out by an accelerated
growth and a thickening of the skin on its back. The carapacial
disk is formed. The margin of the carapacial disk shares
histological and chemical properties with the marginal apical
ridge of the early limb bud, a region that was long known to be
an important site for inductive tissue interaction in the
formation of the limb skeleton. Somite extirpation experiments
confirmed a somitic origin for those cells that form the ribs as
well as the thickened dermis of the carapacial disk in the
snapping turtle. (14) Extirpation experiments performed on
the margin of the carapacial disk, referred to as carapacial
ridge, confirmed its role in the deflection of rib growth to a more
superficial position in turtles. (13)
Exoskeleton versus endoskeleton
By definition, the endoskeleton of vertebrates develops deep
to the dermis of the skin, mostly preformed in cartilage that is
later replaced by bone in a process called ossification. For
most elements, ossification first forms a bony layer around the
cartilage (perichondral bone), before bone replaces the
degenerating cartilage inside the skeletal element (endochondral
bone). All this is also true for most of the endoskeleton of
turtles, but the ribs of turtles are unique among vertebrates in
that they chondrify within the deep layers of the thickened
dermis of the carapacial disk. Perichondral ossification starts
at the point of entry of the rib into the dermis, and from there
spreads laterally. Once the whole cartilage of the embryonic
rib is surrounded by perichondral bone, trabecular bone starts
to spread from the rib through the dermis of the carapacial disk,
and the costal plate is formed. In a similar pattern, the
cartilaginous tips of the neural spines of the dorsal vertebrae
pierce the deep layers of the dermis of the carapacial disk.
Following their perichondral ossification, trabecular bone
988 BioEssays 23.11

What the papers say
Figure 2. The carapace of a marine turtle (Caretta
caretta) in ventral view, showing the close association of
the ribs with the costal plates.
spreads from them through the dermis to form the neural
plates.
The unique pattern of development of the ribs at a level
outside the shoulder blade is not the end of the story, however.
The amniote vertebra is composed of two parts that originally
develop independently, but most of the time fuse in the adult
(Fig. 1B). One part is the centrum, which replaces the embryonic
notochord. Above the centrum develops the neural arch,
which together with the centrum forms the neural canal for the
spinal cord. In turtles, the neural arches of the dorsal vertebrae
shift forward by half a segment, carrying the ribs with them,
again a unique condition in amniotes. (15) The neural arches of
successive vertebrae consequently meet each other above
the midpoint of the centrum, whereas the ribs come to lie lateral
to the no longer functional intervertebral joints. The myomeric
and neuromeric segmentation is thus secondarily established
in the dorsal region of turtles, (16) as was already the case in
Proganochelys. (9) The functional reason for this anterior shift
of the neural arches is not clear, other than that it may
contribute to the mechanical strength of the carapace, as the
neural plates come to alternate with the costal plates (Fig. 1A).
The ossification of endoskeletal structures preformed in
cartilage, such as the ribs, within the dermis poses special
problems for the comparative anatomist. Three types of bone
have classically been recognized: dermal bone, which develops
in the dermis and typically characterizes the exoskeleton,
chondral bone, which replaces cartilage during ossification of
the endoskeleton, and membrane bone, which ossifies in deep
laying membranes as part of the endoskeleton without
cartilaginous preformation. The costal and neural plates of
the turtle carapace start to develop as perichondral bone, as is
typical for the initial phase of ossification of endoskeletal
elements. But this perichondral bone develops in the dermis,
and from it trabecular bone spreads through the dermis during
BioEssays 23.11 989

What the papers say
Figure 3. The relationship of the shoulder blade to the carapace in turtles (Graptemys geographica).
later developmental stages. In the turtle carapace, therefore,
the distinction of endoskeleton versus exoskeleton becomes
problematic. It had been recognized earlier that the endoskeleton
versus exoskeleton cannot be distinguished on the basis
of histogenesis, but must be defined with reference to a
phylogenetic framework. (17,18) Exoskeletal elements are
homologous to structures that in the ancestral condition
combine bone, dentine and enamel, i.e., develop at the
ectoderm±mesoderm interface. Thus, the bony scales of a
trout, or the osteoderms of a crocodile, are exoskeletal,
because these are structures that ultimately can be traced
back (are homologous) to the heavy scales of early fishes that
combine bone, dentine and enamel in a three-layered scale.
By contrast, endoskeletal elements are elements that in the
ancestral condition are preformed in cartilage, while the
cartilaginous stage may be deleted in the descendant
(membrane bone). In the turtle carapace, the neural and
costal plates ossify from, and in continuity with, the periost of
their endoskeletal component. This pattern of ossification
corresponds to the definition of Zuwachsknochen, (18) i.e.,
bone that complements an endoskeletal element that is itself
preformed in cartilage. As such, neural and costal plates are
endoskeletal components of the turtle carapace, and cannot
be derived from a hypothetical ancestral condition by fusion of
exoskeletal osteoderms. All other parts of the turtle carapace
are exoskeletal, however.
The origin of a new body plan
The turtle body plan is evidently highly derived, indeed unique
among tetrapods. The problem for an evolutionary biologist is
to explain these transformations in the context of a gradualistic
process. Given the recently obtained developmental evidence,
(1,11,13) the theory of ``correlated progression'' (10) presents
an incomplete explanation of the turtle body plan:
formation of the carapace is not simply the consequence of a
fusion of osteoderms, and the location of the scapula inside the
rib cage is not the result of a backward migration of the pectoral
girdle.
Early in the 19th century, EÂ tienne Geoffroy Saint-Hilaire
raised the question of how the lung of a reptile could be
transformed into the lung of a bird? He found it impossible to
postulate that the lung of a reptile, specialized in its own way,
could transform into an even more specialized lung of a bird.
But he recognized that both reptiles and birds share similar
early embryonic rudiments of the lung, and he hypothesized
that these could develop along different trajectories in the
two groups due to a minor change in early ontogenetic
development: ``It only required an `accident' [a change] that
990 BioEssays 23.11

What the papers say
is viable and of small magnitude in its origin, but which
could be of unpredictable importance with respect to its
effects''. (19)
The same can be said with respect to the axial skeleton of
turtles. The initial segmentation of the paraxial mesoderm is
the same in turtles as in all other tetrapods, but further
development of structures derived from the somites (dermis,
vertebrae and ribs) proceeds along a different trajectory in
turtles compared to all other tetrapods. Ribs can only be
located either deep to, or superficial to, the scapula. There are
no intermediates, and there is only one way to get from one
condition to the other, which is the redirection of the migration,
through the embryonic body, of the precursor cells that will
form the ribs. The study of such complex problems as the
evolution of the turtle body plan highlights the important role
that development can play in the origin of evolutionary
novelties.
References
1. Gilbert SF, Loredo GA, Brukman A, Burke AC. Morphogenesis of the
turtle shell: the development of a novel structure in tetrapod evolution.
Evol Dev 2001;3:47±58.
2. Hay OP. On Protostega, the systematic position of Dermochelys, and the
morphogeny of the chelonian carapace and plastron. Amer Nat
1898;32:929±948.
3. Kaelin J. Zur Morphogenese des Panzers bei den SchildkroÈten. Acta
Anat 1945;1:144±176.
4. ValleÂn E. BeitraÈge zur Kenntnis der Ontogenie und der vergleichenden
Anatomie des SchildkroÈtenpanzers. Acta Zool Stockholm 1942;23:1±
127.
5. VoÈlker H. UÈ ber das Stamm- Gliedmassen-, und Hautskelett von
Dermochelys coriacea L. Zool Jb Anat 1913;33:431±552.
6. Zangerl R. The homology of the shell elements in turtles. J Morphol
1939;65:383±406.
7. Zangerl R. The turtle shell. In: Gans C, Bellairs Ad'A, Parsons TS, editors.
Biology of the Reptilia Vol. 1, London: Academic Press; 1969.p. 311±
339.
8. Versluys J. UÈ ber die Phylogenie des Panzers der SchildkroÈten und uÈber
die Verwandtschaft der LederschildkroÈte (Dermochelys coriacea).
PalaÈont. Z 1914;1:321±347.
9. Gaffney ES. The comparative osteology of the Triassic turtle Proganochelys.
Bull. Amer Mus Nat Hist 1990;194:1±263.
10. Lee MSY. Correlated progression and the origin of turtles. Nature
1996;379:811±815.
11. Burke AC. The development and evolution of the turtle body plan:
inferring intrinsic aspects of the evolutionary process from experimental
embryology. Amer Zool 1991;31:616±627.
12. Ruckes H. Studies in chelonian osteology. Part II. The morphological
relationships between girdles, ribs and carapace. Ann NY Acad Sci
1929;31:81±120.
13. Burke AC. Development of the turtle carapace: implications for the
evolution of a novel bauplan. J. Morphol 1989;199:363±378.
14. Yntema CL. Extirpation experiments on the embryonic rudiments of the
carapace of Chelydra serpentina. J Morphol 1970;132:235±244.
15. Goette A. UÈ ber die Entwicklung des knoÈchernen RuÈckenschildes
(Carapax) der SchildkroÈten. Z wiss Zool 1899;66:407±434.
16. Hoffstetter R, Rage J-C. Vertebrae and ribs of modern reptiles. In: Gans
C, Bellairs Ad'A, Parsons TS, editors. Biology of the Reptilia Vol. 1
London: Academic Press; 1969. p. 201±310.
17. Patterson C. Cartilage bones, dermal bones and membrane bones, or
the exoskeleton versus the endoskeleton. In: Andrews SM, Miles RS,
Walker AD, editors. Problems in Vertebrate Evolution. London: Academic
Press; 1977. p. 77±121.
18. Starck D. Vergleichende Anatomie der Wirbeltiere auf evolutionsbiologischer
Grundlage, Vol 2 Berlin: Springer Verlag; 1979.
19. Geoffroy Saint-Hilaire E. Le deÂgre d'influence du monde ambiant pour
modiifier les formes animales; question inteÂressant l'origine des espeÂces
teÂleÂosauriennes et successivement celle des animaux de l'eÂpoque
actuelle. MeÂm Acad R Sci Inst. France 1833;12:63±92. (The quote is
from p. 80).
BioEssays 23.11 991