LETTER doi:10.

1038/nature15397

Deep-time evolution of regeneration and preaxial

polarity in tetrapod limb development
Nadia B. Fröbisch1, Constanze Bickelmann1, Jennifer C. Olori2 & Florian Witzmann1,3

Among extant tetrapods, salamanders are unique in showing a
reversed preaxial polarity in patterning of the skeletal elements
of the limbs, and in displaying the highest capacity for regenera-
tion, including full limb and tail regeneration. These features are
particularly striking as tetrapod limb development has otherwise
been shown to be a highly conserved process1,2. It remains elusive
whether the capacity to regenerate limbs in salamanders is mechan-
istically and evolutionarily linked to the aberrant pattern of limb
development; both are features classically regarded as unique to
urodeles3. New molecular data suggest that salamander-specific
orphan genes play a central role in limb regeneration and may also
be involved in the preaxial patterning during limb development4,5.
Here we show that preaxial polarity in limb development was pre-
sent in various groups of temnospondyl amphibians of the
Carboniferous and Permian periods, including the dissorophoids
Apateon and Micromelerpeton, as well as the stereospondylo-
morph Sclerocephalus. Limb regeneration has also been reported
in Micromelerpeton6, demonstrating that both features were
already present together in antecedents of modern salamanders
290 million years ago. Furthermore, data from lepospondyl ‘micro-
saurs’ on the amniote stem indicate that these taxa may have shown
some capacity for limb regeneration and were capable of tail regen-
group of Palaeozoic and Mesozoic anamniotes that includes the clade
Dissorophoidea (Fig. 1, node 4), which is supported by most recent
studies as having given rise to all modern amphibians or at least
batrachians (frogs and salamanders)12,13; second, the lineage compris-
ing Lepospondyli (Fig. 1, node 6), a clade of mostly small-bodied
Palaeozoic anamniotes, and Amniota13. Some lepospondyl taxa also
are proposed to be lissamphibian predecessors in an alternative hypo-
thesis of amphibian origins14. A number of temnospondyl taxa have an
excellent fossil record that has provided detailed insights into their
ontogenetic development and life history patterns15. Particularly, the
fossil record of the Early Permian Branchiosauridae is distinguished by
a well resolved ontogenetic series, which allowed for the reconstruc-
tion of the ossification sequence in the limbs16. Ossification patterns in
Dipnoi Reg.
Tiktaalik
Acanthostega
Dvinosauria
1 II
Micromelerpeton Reg.
Reg.? Preaxial
4 Apateon
II

eration7, including re-patterning of the caudal vertebral column Preaxial

Gerobatrachus
that is otherwise only seen in salamander tail regeneration. The II

?
data from fossils suggest that salamander-like regeneration is an Preaxial
Caecilians
ancient feature of tetrapods that was subsequently lost at least once 3 IV
5 Frogs Reg.
in the lineage leading to amniotes. Salamanders are the only mod- Postaxial II

Salamanders Reg.
ern tetrapods that retained regenerative capacities as well as pre- Preaxial

axial polarity in limb development. Eryops

Research on tetrapod limb development is a prime example of how
II

II
IV Sclerocephalus
the integration of palaeontological and developmental data can pro- ? Or ? 2 Preaxial
Preaxial Stereospondyli
vide novel insights into the evolution of a crucial organ system1. In Postaxial

recent decades, research has revealed many aspects of the molecular Anthracosauria
mechanisms involved in limb development, and has shown that tet-
Hyloplesion IV

rapod limb development is a remarkably conserved process1,2. It is 7 Reg.

therefore even more surprising that salamanders are an exception to Microbrachis Postaxial

6
this overall conserved pattern of limb development, characterized by Nectridea
preaxial rather than postaxial or central polarity in the development of
Aistopoda
the zeugopodial elements and digits3,8,9. Moreover, salamanders show IV
IV

the highest regenerative capacity among tetrapods, including full limb ? Amniota No Reg.
Postaxial
and tail regeneration throughout the lifetime of an individual10. Both Postaxial

features have classically been considered highly derived for salaman- Figure 1 | Phylogenetic tree depicting major groups of tetrapods.
ders9,11, yet surprisingly the question of whether the remarkable regen- (1) Sarcopterygii; (2) crown-group tetrapods; (3) Temnospondyli;
erative capacities of salamander limbs and the aberrant pattern of limb (4) Dissorophoidea; (5) Lissamphibia; (6) Lepospondyli; and (7) Microsauria.
development are evolutionarily and/or mechanistically linked, has Stylized right hands indicate taxa for which pre- or postaxial polarity in limb
development has been reported. Neither pre- nor postaxial polarity can be
never been investigated, to our knowledge3,4. Here we present new
reconstructed as the plesiomorphic state for tetrapods (node 2). ‘Reg.’ indicates
evidence from the fossil record that provides a novel deep-time reported salamander-like regeneration; Reg. is depicted as faded for frogs,
perspective on the evolution of preaxial polarity in limb development because they have regenerative capacities in limbs and tails only in tadpole
and salamander-like regenerative capacities among early tetrapods. stages until metamorphic climax is reached; ‘No Reg.’ indicates that
The phylogeny of Tetrapoda comprises basal stem taxa and two regenerative capacity is absent; for taxa without a square, no data on
major lineages: first the Temnospondyli (Fig. 1, node 3), a diverse regenerative capacity are currently available.
1
Museum für Naturkunde, Leibniz Institute for Evolution and Biodiversity Science, Invalidenstrasse 43, 10115 Berlin, Germany. 2Department of Biological Sciences, State University of New York at Oswego,
30 Centennial Drive, Oswego, New York 13126, USA. 3Department of Ecology and Evolutionary Biology, Brown University, Providence, Rhode Island 02912, USA.

0 0 M O N T H 2 0 1 5 | VO L 0 0 0 | N AT U R E | 1
G2015 Macmillan Publishers Limited. All rights reserved
 RESEARCH LETTER

the limbs mirror general patterns of polarity of earlier events in ske- zeugopodial element, as well as a clear gradation in the degree of
letogenesis in extant and extinct tetrapods17 and the ossification ossification and differentiation of the metapodial and phalangeal ele-
sequence in the branchiosaurid Apateon revealed that it indeed had ments from pre- to postaxial (Fig. 2a–o). This demonstrates that pre-
preaxial polarity in limb development, previously exclusively attribu- axial polarity in limb development was not only a feature of derived
ted to salamanders16. However, it has remained uncertain whether this dissorophoids, but instead was already established at the base of
bears a phylogenetic signal or represents a homoplastic feature shared Dissorophoidea. More surprisingly, new data in support of preaxial
by branchiosaurids and salamanders owing to a similar larval ecology, polarity in limb development were also found in Sclerocephalus, a
as it was suggested that the early formation of the first and second stereospondylomorph temnospondyl and a much more distant rela-
digits of salamanders may be a larval adaptation9,18. The two temnos- tive of modern amphibians (Fig. 2p–w). While the preserved ontogen-
pondyl taxa Micromelerpeton and Sclerocephalus derive from the same etic trajectory of Sclerocephalus is considerably less comprehensive
fossil lakes (Lagerstätten) in southwestern Germany as the branchio- than for Apateon and Micromelerpeton, a few very well-preserved
saurid Apateon, and although the preserved ontogenetic sequences of larvae provide insights into patterns of limb ossification and indicate
these taxa are not as extensive as for Apateon, they provide detailed an advanced differentiation of preaxial zeugopodial and autopodial
insights into the developmental trajectory of these species19,20 (see elements. Together, these new data on temnospondyl limb ossification
Supplementary Information). Micromelerpeton represents the most patterns suggest an early evolution of preaxial polarity in limb
basal member of the Dissorophoidea. Here we show for the first time development at the base of Temnospondyli or even at the base of
that, as in branchiosaurids, fossils of successive ontogenetic larval Tetrapoda, as has been tentatively suggested previously3,16 (Fig. 1).
stages of Micromelerpeton reveal a delayed ossification of the postaxial Notably, while it remains unknown whether Sclerocephalus and

a b c d Figure 2 | Ossification patterns in the limbs

H H of Micromelerpeton and Sclerocephalus.
Micromelerpeton forelimbs: a, b, MB.Am.1207
right forelimb; c, d, MB.Am.1124 left forelimb,
ra
note the advanced differentiation of the radius
ul
*ra compared to the ulna. Sclerocephalus forelimb:
e, f, MB.Am.1314 right forelimb. Micromelerpeton
* ul hind limbs: g, h, MB.Am.1105 left hind limb;
mc
i, j, MB.Am.1341 right hind limb; k, l, MB.Am.1159
left hind limb; m, n, MB.Am.1207 right hind
e f g h i j limb; o, p, MB.Am.1124 right hind limb.
Sclerocephalus hind limbs: q, r, MB.Am.1305
H F hind limbs (left leg is on the left, right leg is on
ti F the right); s, t, MB.Am.1314 right hind limb;
ra fi
ul
fi u, v, MB.Am.1313 right hind limb. Elements in
ti black are fully ossified; stippling indicates the
* mt degree of ossification; asterisks indicate preaxial
mc
mt side of limb. w, Left hind leg of an extant
ph * ph
* salamander (Cryptobranchus, FMNH 814154)
ph cleared and stained for bone (red) and cartilage
(blue) showing ossification proceeding from pre-
k l m n to postaxial. x, Bilateral asymmetry in the forelimbs
F
of Microbrachis, cast of specimen MB.Am.822;
F the upper radius is crushed. Note the well-
ti ti developed scapula and differentiated state of upper
fi fi zeugopodial elements. Zeugopods in the bottom
part of the figure lag behind in development, are
*
undifferentiated and lack an olecranon process.
mt * F, femur; fi, fibula; H, humerus; mc, metacarpals;
mt
ph mt, metatarsals; ph, phalanges; ra, radius; sc,
scapula; ti, tibia; ul, ulna. Scale bars, 1 mm.
o p q r s t
F
F F

F fi ti

ti? ti
fi fi mt
ti
ph
*
mt
* *
* mt
u v w x
fi H
ti F

mt ul
ti fi
ph ra sc
* *
mt
H
ph
ra +u l

2 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 5
G2015 Macmillan Publishers Limited. All rights reserved

branchiosaurids were capable of regenerating their limbs, Micro-
melerpeton were shown to possess this capacity on the basis of a unique
pattern and combination of abnormalities associated with limb regen-
eration in modern salamanders6. Interestingly, new data about the
molecular control of limb regeneration show that salamander orphan
genes are involved in regeneration and probably also plays a role in the
preaxial patterning during limb development, indicating a possible
link between these two features in modern salamanders4,5. The data
from the fossil record show that both preaxial polarity and the capacity
to regenerate limbs were definitely present at the base of dissorophoid
temnospondyls 290 million years ago, but may even have evolved
earlier at the base of Tetrapoda (Fig. 1).
Furthermore, novel ontogenetic data from the lepospondyl lineage
of Tetrapoda provide important data points for an even deeper picture
of the evolution of limb development and regenerative capacities in
tetrapods. The available ontogenetic data of two microsaurian taxa,
Microbrachis and Hyloplesion, from the Upper Carboniferous coal
deposits of Nýrany, Czech Republic, suggest that, contrary to the tem-
nospondyl taxa, these ‘microsaurs’ probably displayed postaxial polar-
ity in limb development as indicated by a delayed differentiation of the
preaxial zeugopodial and metapodial elements21. Notably, a number of
Microbrachis specimens show a clear asymmetry in ossification
between the left and right limbs21. The degree of ossification and
morphological differentiation of limb elements on one side is well
progressed and in accordance with the overall developmental stage
of the specimen. However, on the other side, the proximal limb ele-
ments (scapula and humerus) are equally well developed, while more
distal limb elements show a much less mature state of differentiation
(Fig. 2x). This suggests that the distal limbs of these microsaur speci-
mens may be in the process of regenerating their lower limbs, leading
to the obvious disjunction of differentiation between proximal and
distal elements.
Even stronger support for salamander-like regenerative capacities in
lepospondyl microsaurs comes from clear evidence of tail regeneration
in Microbrachis and Hyloplesion. While tail regeneration in these
taxa has been previously mentioned in brief7,21, here we show that
the regeneration of vertebral elements during tail regeneration in
a Figure 3 | Tail regeneration in the microsaur
b bars, 1 mm (a and c) and 2 mm (b).
G2015 Macmillan Publishers Limited. All rights reserved
LETTER RESEARCH
microsaurs proceeded in a salamander-like fashion (Fig. 3a–c). In
extant vertebrates, tail regeneration is known to occur in lungfish22,
salamanders23,24, frog tadpoles25 and lizards26. Tail regeneration in
tadpoles is only possible until metamorphic climax and differs from
salamander tail regeneration in many respects, including tissue organ-
ization and the molecular markers involved25. Moreover, tadpoles lack
ossified vertebral structures in the tail, which is resorbed entirely dur-
ing metamorphosis. Unfortunately, very little is known about lungfish
tail and fin regeneration22. Among amniotes, many lizards are known
to autotomize their tail as a defence mechanism when threatened by a
predator26,27. While the subsequently regenerated tail is functional as a
balancing organ, articulated vertebrae are replaced by an unsegmented
hyaline cartilage tube26,27. In contrast, salamanders form a blastema at
the tail amputation site, similar to the blastema initiating limb regen-
eration23,28, and fully regenerate vertebral elements as well as the assoc-
iated musculature and innervation23,24. Initially, a long rod of cartilage
forms, which is already subdivided on the cellular level into regions of
hyaline cartilage cells with abundant interstitial matrix and interpo-
lated regions of fusiform cartilage cells lacking matrix24. These seg-
mented cartilages subsequently differentiate further and represent the
precocious centra of the regenerated tail23. The sequence in which
vertebral elements regenerate is reversed compared to original axial
development in that centra form first, followed by the musculature and
finally neural and haemal arches23,24.
In the microsaurs Microbrachis and Hyloplesion, the precocious cen-
tra of the regenerating tail vertebrae are clearly visible (Fig. 3a, c),
showing further differentiation from an unsegmented rod into segmen-
ted precocious centra. Microbrachis specimen MB.Am.815.6 (Fig. 3c)
shows a line of small, segmented elements, which attach to differen-
tiated caudal vertebrae at the former amputation site. In Hyloplesion
(specimen St.209, Fig. 3a), a clear differentiation between dense pre-
cocious centra and interspersed hyaline zones is visible in a pattern
notably similar to regenerating tails of modern salamanders23. The
degree of differentiation suggests a more advanced stage of tail regen-
eration in Hyloplesion specimen St.209 than in Microbrachis specimen
MB.Am.815.6, although taphonomic and taxonomic variation may
play a role as well. A further Hyloplesion specimen (RSM.1899.32.3.)
taxa Microbrachis and Hyloplesion in
comparison with an extant salamander.
a, Hyloplesion specimen St 209 showing tail
regeneration; inset represents enlarged region of
the regenerating tail of St 209, showing
segmentation and differentiation into primordial
centra (black arrowheads). b, Enlarged view of
regenerating tail of a cleared and stained
salamander (Ambystoma tigrinum) specimen
(FMNH 281665). c, Microbrachis specimen
MB.Am.815 showing tail regeneration. Inset shows
close-up of regenerating section of the tail. Scale
0 0 M O N T H 2 0 1 5 | VO L 0 0 0 | N AT U R E | 3
 RESEARCH LETTER
also shows tail regeneration, but owing to poorer preservation a more
detailed assessment of the stage of development of the central prim-
ordia is not possible. As in salamanders, haemal and neural arches are
still lacking at the available stages of tail regeneration in the two lepos-
pondyl taxa. This form of full tail regeneration is otherwise exclusively
known from modern salamanders and, together with the indication
for possible limb regeneration in ‘microsaurs’, provides evidence for
salamander-like regenerative capacities previously unrecognized for
lepospondyls.
Among modern tetrapods, only salamanders still conserve the capa-
city to regenerate their limbs and many other organs fully, including
fully functional tails, throughout their entire lifespan. New molecular
evidence indicates that salamander-specific genes are involved in the
capacity to regenerate limbs in modern salamanders and are implicated
in preaxial polarity in limb development4,5. The novel evidence from the
fossil record suggests that regenerative capacities are an ancient feature
of Tetrapoda or even Sarcopterygii (tetrapods and lobe-finned fish),
and were lost at least once in the evolutionary history of tetrapods along
the lineage leading to amniotes (Fig. 1). Tail regeneration and the
potential ability of lepospondyls to regenerate limbs show that sala-
mander-like regeneration was also possible in fossil tetrapod taxa that
probably had postaxial polarity in limb development.
The definitive co-occurrence of preaxial polarity in limb develop-
ment and the capacity to regenerate limbs at the base of the dissor-
ophoid clade within Temnospondyli demonstrates that both features
were already present together in the lineage leading to modern sala-
manders, at least 80 million years before the first occurrence of stem
salamanders in the fossil record. Moreover, preaxial polarity in limb
development and salamander-like regenerative capacities may be
ancient features of tetrapod vertebrates that are conserved only in
salamanders (Fig. 1).
Received 6 May; accepted 13 August 2015.
Published online 26 October 2015.
1. Shubin, N., Tabin, C. & Carroll, S. B. Fossils, genes and the evolution of animal
limbs. Nature 388, 639–648 (1997).
2. Zeller, R., Lopez-Rios, J. & Zuniga, A. Vertebrate limb bud development: moving
towards integrative analysis of organogenesis. Nature Rev. Genet. 10, 845–858
(2009).
3. Fröbisch, N. B. & Shubin, N. H. Salamander limb development: integrating genes,
morphology, and fossils. Dev. Dyn. 240, 1087–1099 (2011).
4. Brockes, J. P. & Gates, P. Mechanisms underlying vertebrate limb regeneration:
lessons from the salamander. Biochem. Soc. Trans. 42, 625–630 (2014).
5. Brockes, J. in Salamanders in Regeneration Research: Methods and Protocols (eds
Kumar, A. & András, S.) Ch. 1 (Springer, 2015).
6. Fröbisch, N. B., Bickelmann, C. & Witzmann, F. Early evolution of limb regeneration
in tetrapods: evidence from a 300-million-year-old amphibian. Proc. Biol. Sci.,
(2014).
7. Carroll, R. L. & Gaskill, P. The Order Microsauria. 122 (American Philosophical
Society, 1978).
8. Shubin, N. H. & Alberch, P. A morphogenetic approach to the origin and basic
organisation of the tetrapod limb. Evol. Biol. 20, 319–387 (1986).
9. Shubin, N. H. & Wake, D. B. in Amphibian Biology Vol. 5 (eds Heatwole, H. & Davies,
M.) 1782–1808 (Surrey Beatty & Sons PTY limited, 2003).
4 | N AT U R E | VO L 0 0 0 | 0 0 M O N T H 2 0 1 5
G2015 Macmillan Publishers Limited. All rights reserved
10. Gardiner, D. M. & Bryant, S. V. in Fins into Limbs (ed. Hall, B. K.) 163–182 (Univ.
Chicago Press, 2007).
11. Simon, A. & Tanaka, E. M. Limb regeneration. Wiley Interdiscip. Rev. Dev. Biol. 2,
291–300 (2013).
12. Anderson, J. S., Reisz, R. R., Fröbisch, N. B., Scott, D. & Sumida, S. S. A stem
batrachian from the Early Permian of Texas and the origin of frogs and
salamanders. Nature 453, 515–518 (2008).
13. Ruta, M. & Coates, M. I. Dates, nodes and character conflict: addressing the
lissamphibian origin problem. J. Syst. Palaeontology 5, 69–122 (2007).
14. Marjanović, D. & Laurin, M. The origin(s) of extant amphibians: a review with
emphasis on the “lepospondyl hypothesis”. Geodiversitas 35, 207–272 (2013).
15. Schoch, R. R. Life cycles, plasticity and palaeoecology in temnospondyl
amphibians. Palaeontology 57, 517–529 (2014).
16. Fröbisch, N. B., Carroll, R. L. & Schoch, R. R. Limb ossification in the Paleozoic
branchiosaurid Apateon (Temnospondyli) and the early evolution of preaxial
dominance in tetrapod limb development. Evol. Dev. 9, 69–75 (2007).
17. Fröbisch, N. B. Ossification patterns in the tetrapod limb – conservation and
divergence from morphogenetic events. Biol. Rev. Camb. Philos. Soc. 83, 571–600
(2008).
18. Schmalhausen, J. J. Die Entwicklung des Extremitätenskelettes von Salamandrella
kayserlingii. Anat. Anz. 37, 431–446 (1910).
19. Schoch, R. R. Early larval ontogeny of the Permo-Carboniferous temnospondyl
Sclerocephalus. Paleontology 46, 1055–1072 (2003).
20. Witzmann, F. & Pfretzschner, H.-U. Larval ontogeny of Micromelerpeton credneri
(Temnospondyli, Dissorophoidea). J. Vert. Paleontol. 23, 750–768 (2003).
21. Olori, J. Skeletal morphogenesis of Microbrachis and Hyloplesion (Tetrapoda:
Lepospondyli), and implications for the developmental patterns of extinct, early
tetrapods. PLoS ONE 10, e0128333 (2015).
22. Conant, E. B. Regeneration in the African lungfish, Protopterus. I. Gross aspects.
J. Exp. Zool. 174, 15–31 (1970).
23. Holtzer, H., Holtzer, S. & Avery, G. An experimental analysis of the development
of the spinal column IV. Morphogenesis of tail vertebrae during regeneration.
J. Morphol. 96, 145–171 (1955).
24. Vaglia, J. L., Babcock, S. K. & Harris, R. N. Tail development and regeneration
throughout the life cycle of the four-toed salamander Hemidactylium scutatum.
J. Morphol. 233, 15–29 (1997).
25. Beck, C. W., Izpisúa Belmonte, J. C. & Christen, B. Beyond early development:
Xenopus as an emerging model for the study of regenerative mechanisms. Dev.
Dyn. 238, 1226–1248 (2009).
26. Fisher, R. E. et al. A histological comparison of the original and regenerated tail in
the green anole, Anolis carolinensis. Anat. Rec. 295, 1609–1619 (2012).
27. Alibardi, L. Morphological and Cellular Aspects of Tail and Limb Regeneration in
Lizards. (Springer, 2010).
28. Holtzer, S. W. The inductive activity of the spinal cord in urodele tail regeneration.
J. Morphol. 99, 1–39 (1956).
Supplementary Information is available in the online version of the paper.
Acknowledgements We thank B. Ekrt, J. Müller, U. Göhlich, S. Walsh, C. Mehling,
P. Barrett and R. Schoch for access to the collections under their care. H. J. Götz took the
photograph in Fig. 3c; S. Lokatis took the photograph in Fig. 3b. This research was
funded by an DFG Emmy Noether Grant (FR 2647/5-1) to N.B.F.; the Jackson School of
Geosciences, the Banks Fellowship, Society of Vertebrate Paleontology Estes Memorial
Grant and Paleontological Society Lane Student Award to J.C.O.; and the Feodor-Lynen
Fellowship of the Alexander von Humboldt Foundation to F.W.
Author Contributions N.B.F., C.B. and F.W. designed the research; N.B.F., C.B., J.C.O.
and F.W. performed the research; N.B.F. wrote the manuscript; C.B., J.C.O. and F.W.
contributed to the manuscript.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Readers are welcome to comment on the online version of the paper. Correspondence
and requests for materials should be addressed to N.B.F.
(nadia.froebisch@mfn-berlin.de).