/. Embryol. exp. Morph. Vol. 71, pp. 109-120, 1982 109Printed in Great Britain © Company of Biologists Limited 1982

The effect of tail skin on themorphology and morphogenesis of limb

regenerates in the red-backed salamander,Plethodon cinereus

By CHARLES E. DINSMORE1

From the Department of Anatomy, Rush Medical College, Chicago andThe Mount Desert Island Biological Laboratory,

Salsbury Cove, Maine

SUMMARY

Tail skin cuffs have been grafted to the upper forelimb of red-backed salamanders ineither normal or 180°-rotated dorsoventral orientation. Subsequent amputation through thegraft region resulted in arrested regeneration, distally deficient or typical four-digit regenerates.Distribution was not substantially influenced by graft orientation nor were there any super-numerary limbs induced by the axially dislocated tail skin on the limb stumps. Furthermore,regenerates bore no indication of tail-like structures other than large granular skin glandsproximally. These data indicate that, unlike limb skin, tail skin is not morphogeneticallyactive in the epimorphic process of limb regeneration. In addition, the species used in thisstudy is a fin-less, round-tailed salamander. It is therefore suggested that the previouslyreported morphogenetic effects of tail skin on limb regeneration may be related to thepresence of tail fins on the species studied.

INTRODUCTION

The location of pattern information within various developmental fields andthe mechanisms by which morphogenesis is accomplished are only beginningto be discovered. One of the means by which this area is being explored isthrough the identification of obligatory morphogenetic tissue interactions atvarious stages of developmental processes. By selectively dissociating the con-stituents or altering the spatial relationships of specific tissue components in adeveloping system, whether early embryo or epimorphic field, one can deter-mine which manipulations affect morphogenesis, thereby identifying both thetiming of morphogenetic interactions and the tissues involved in primary orsecondary patterning.

The regenerating urodele limb is a system which has proven amenable to suchexperimental dissection. Building on the accumulated observations of manyearlier investigations, Carlson (1974, 1975) designed and performed a critical

1 Author's address: Department of Anatomy, Rush Medical College, 600 S. Paulina Street,Chicago, Illinois 60612, U.S.A.

110 C. E. DINSMORE

series of experiments from which the data and lucid analysis have providedinsight into many previously described but poorly understood regenerativephenomena. By surgically isolating and positionally dislocating, in turn, theindividual components of the limb (i.e. epidermis, dermis, skeletal muscle orbone), he demonstrated that only the dermis and skeletal muscle are morpho-genetically active. When a 'positional disharmony' (Lheureux, 1972, 1975) iscreated between either of these tissues and the rest of the limb stump, regeneratesare often abnormal and frequently multiple.

Recent efforts to confirm the morphogenetic activity of skin (Dinsmore,1981 a, b) and skeletal muscle (Dinsmore, 1979, 1981 b) in another epimorphicfield, the urodele tail, have failed. Neither positional dislocation nor unilateralablation of these caudal tissues disrupts the regenerative process, in markedcontrast to the effects such procedures have on the limb field. There are severalpossible explanations for this apparent difference in the modes by which theseepimorphic fields are governed. Morphogenetic information may not be intrinsicto these tissues but rather dependent upon their field of origin; present in thelimb territory, absent in the tail. Another explanation previously considered(Dinsmore, 1981 a) is that there are hierarchies among the tissues which com-prise a particular field. Within the tail, the regenerating spinal cord (sine quanon of this field) may present a dominant primary morphogenetic signal sourcewith other constituent tissues subordinate to its regulatory influence. This neednot assume a lack of positional information in other tissues. Rather, it maysimply be a field specific masking of the tissues' potentials. The expression ofa tissue's hidden potential might be affected by grafting it to a heterotypic fieldin which its counterpart is normally active in regulating regenerate morpho-genesis. It is also possible that a tissue such as skin bears information relativeto field morphology (e.g. tailness), as has been described by Glade (1957, 1963,1978), yet lacks axial stability relying on underlying tissues for appropriatedifferentiation at that level.

The present experiments address some of these possibilities by re-examiningthe effect that grafted tail skin may have on urodele limb regeneration. Earlierstudies of this nature (Glade, 1957, 1963, 1978) employed urodeles with distinctdorsal and ventral tail fins. This raised questions about the morphogeneticinfluence of fin mesenchyme and how it might alter pattern regulation outsideof the tail field. In order to avoid such potential problems, the round-tailedsalamander, Plethodon cinereus, has been used exclusively in this study assuringthat grafts consist of whole tail skin uncontaminated with fin mesenchyme.This terrestrial salamander regenerates its limbs in a typical urodele mannerand may produce supernumerary structures in response to limb skin rotationand subsequent amputation (Fig. 1). It is thus comparable to those urodelesused in the earlier studies by Lheureux (1972, 1975) and Carlson (1974, 1975)relative to morphogenetic regulation during limb regeneration as determined bythat technique.

Effect of tail skin on limb regeneration 111

0-2 mm

Fig. 1. A supernumerary regenerate from an upper forelimb whose skin wasexchanged with that of the contralateral limb. Only the anteroposterior axis of thegraft was reversed.

Fig. 2. Frontal section through a 61 -day normal control regenerate. Four metacarpalsare clearly visible while the carpals and phalanges are out of the plane of section due tonormal wrist extension and digit flexion. Amputation level is indicated by the arrows.

Two specific questions addressed by the ensuing experiments are: (1) willgrafts of tail skin to the limb influence the morphology of limb regenerates byintroducing a degree of'tailness' into the internal organization of the regenerate,and (2) does tail skin bear accessible positional information with an axial biassuch that a positionally dislocated graft of tail skin will alter limb regeneratemorphogenesis ?

MATERIALS AND METHODS

The experiments described below were performed on adult Eastern red-backed salamanders, Plethodon cinereus, ranging from about 8 to 10 cm intotal body length. Specimens collected in the wooded areas around the MountDesert Island Biological Laboratory, Salsbury Cove, Maine, were maintainedas in earlier studies (Dinsmore, 1979, 1981a).

Two series of experimental manipulations were employed. The basic procedureconsisted of anaesthetizing animals by immersion in 1-0% MS 222 (Eastman)and removing them to the stage of a dissecting microscope. A circumferentialcuff of tail skin was then carefully dissected free by making a longitudinal mid-ventral incision approximately 1-5 cm distal to the vent, elevating the skingently while cutting intervening myoseptal attachments and circumferential lyincising the skin at the proximal and distal extents (5-6 mm) of the longitudinalincision. The freed cuff was placed in a pool of Holtfreter's solution whereadhering myofibres and fat deposits were removed. The flattened piece of skinwas then trimmed to a square which would subsequently be fitted to a graft bedon the limb. The limb was prepared for the graft of tail skin by completelyremoving the skin from the upper forelimb from elbow to shoulder. Except fora strong attachment by an anterior intercompartmental connective tissue septum,the limb skin is only tenuously attached to the underlying musculature and iseasily removed. The graft tissue was then introduced onto the graft bed and a

112 C. E. DINSMORE

final trim assured close fitting of the tail skin onto the limb. At this point,grafts for specimens in series I (16 animals) were positioned so that their dorsalsurfaces corresponded with the dorsal aspect of the limb (normal orientation)and were secured by two or three interrupted 7-0 sutures (Ethicon). Designationof axes in this study is consistent with that applied to the whole body. Thecraniocaudal orientation of the tail skin grafts relative to the proximodistal axisof the limb was not recorded.

The animals in series II (24 animals) were prepared in the same manner asthose in series I except for the orientation of the graft. The squares of tail skinwere positioned with their dorsal surfaces covering the ventral side of the upperforelimbs and, as above, secured in that orientation with two or three sutures.The skin grafts were thus rotated 180° about the long axis cf the limbs producingmaximal dislocation of this tissue relative to the dorsoventral axis of the limb.

Subsequent treatment was the same for both groups of animals. The tailswere removed at the proximal edge of the denuded areas by pinching throughan autotomy plane. This reduced the risk of infection and death which was aproblem in early trials. The animals were then placed in petri dishes containingHoltfreter's solution for recovery, after which they were returned to theircontainers for a 10-day post-operative period to allow graft stabilization andhealing. At the end of this period, animals were again anaesthetized and underlow power magnification the limbs bearing tail skin were amputated throughthe middle of the graft (mid-humerus) with the projecting humerus trimmedeven with the retracted soft tissues. Distal segments were usually fixed in Bouin'ssolution for subsequent histological analysis of graft morphology and stabilityat the time of amputation. The contralateral limbs were amputated at the sametime to serve as normal controls on each specimen.

Animals were again anaesthetized at intervals of from 50 to 98 days post-amputation at which time both the experimental and contralateral control limbswere dissected free at the shoulder. The limbs were examined closely under thedissecting microscope for superficial morphology, noting also the thickness andpigmentation of the stump and regenerate on the experimental limb relative tothat of the control. Due to the larger skin glands and heavier pigmentation ofthe grafted tail skin, the experimental limbs were thicker and darker, at leastat the stump level, preliminarily verifying graft retention. The limbs were thenfixed in Bouin's solution, decalcified and prepared for routine paraffin sectioning.Blocks were sectioned serially at 8 /*m and stained with haematoxylin and eosinor by Mallory's trichrome method.

RESULTS

The control limbs which were simply amputated at the same time as the con-tralateral experimental limbs produced normal four-digit regenerates (Fig. 2)with one exception. The single abnormal regenerate (1 out of 40) consisted of

Effect of tail skin on limb regeneration 113

Table 1. Summary of results comparing regenerative success versustail skin graft orientation

Graft orientation

NormalRotated 180°Pooled dateContralateral

limbs (no graft)

Number oflimbs

16244040

No epimorphicregeneration

6(37-5%)5 (20-8%)

11 (27-5%)0(0%)

Deficientregenerate

6 (37-5%)11 (45-8%)17(42-5%)1(2-5%)

4-digitregenerate

4 (25%)8(33-3%)

12(30%)30(97-5%)

a normal elbow joint with the proximal segments of radius and ulna articulatingappropriately. There was no development beyond mid-zeugopodium. However,a low percentage of defective regenerates following simple amputation is notunexpected (Dearlove & Dresden, 1976).

Regenerates from limbs bearing grafts of tail skin in either normal or 180°-rotated orientation have been grouped into three general categories: no epi-morphic regeneration, deficient regenerate, and four-digit regenerate (Table 1).Stumps which were grossly truncated and upon histological examination showedno sign of forming an elbow joint have been arbitrarily classified as showing noepimorphic regeneration. The covering of the humeral stump in specimens fromthis category ranged from a simple fibrous cap (Fig. 3) to an elongate cartila-ginous rod (Fig. 4). These stumps were enclosed in a jacket of skin endowedwith thick and obviously tail-like glands (compare with Fig. 5).

Limb regenerates which were deficient bore at least an elbow joint but fewerthan four digits on the distal segment. The range of skeletal expression includedsimple spikes (Fig. 6) to nearly complete, three-digit regenerates. Figure 7 showsone plane in a serially sectioned, three-digit regenerate demonstrating the levelof amputation, the obvious retention of the tail skin graft and its influence onregenerate skin gland morphology. Serial sectioning provided a more completemeans of evaluating the internal organization of regenerates. Distal fusion ofthe radius and ulna as well as fusion of the carpal and metacarpal elements wasalso observed in this group. Again, skin proximal to the level of amputationwas obviously of tail type. That covering the regenerate was more variable,however, usually bearing progressively smaller, more limb-like glands in aproximodistal gradient (compare with Figs. 2 and 3).

Although they were a minority, 30% of the amputated stumps bearing cuffsof tail skin produced four-digit regenerates (Table 1). Examination of serialsections showed that regenerates in this category bore elbow joints with synovialcavities, a radius and ulna articulating appropriately, carpals, metacarpals, andfour phalanges. The absolute numbers of skeletal elements in the autopodia werenot counted and may be reduced in some cases. This also occurs in normal

114 C. E. DINSMORE

Ei

LO

Effect of tail skin on limb regeneration 115regenerates and was therefore not important to the fundamental observations ofthis study. Skin gland morphology was again the criterion for ascertaining graftretention at the level of the stump. Except at the proximodistal line along whichthe skin cuff was sutured closed around the limb, typical tail skin was presenton the stump and proximally on most of the regenerates in this category (Fig. 8).

These data demonstrate that grafts of tail skin to the limb with subsequentregeneration induced in the limb-tail chimaera do not, in this species, alter thegross morphology of the limbs which regenerate. Although distal reduction wasthe most common outcome in this procedure (70% producing either no re-generation or distally incomplete outgrowths), tail structures other than largegranular glands were not observed in any of the regenerates.

Table 1 shows a basic similarity of results irrespective of the graft orientationindicating that, unlike rotated limb skin, tail skin is morphogenetically inertrelative to axial positional information. Regenerates developed grossly normalaxiation as judged by the elbow flexure and, when present, orientation of thedigits. No supernumerary limbs were induced by apposition of dorsal skin withthe ventral limb surface. A small digitiform outgrowth was observed, however,on the palm of one four-digit regenerate. From the curious location of this super-numerary projection, it is unclear how its development may have been stimulated.

In summary, the data indicate that while morphogenesis of the limb re-generate may be arrested or stunted by a cuff of tail skin at the level of amputa-tion, tailness has not been introduced into the system beyond the level of theintegument. Furthermore, a rotated cuff of tail skin which conceptually presentspositionally dislocated axial information to the limb stump, and thereforecreates potential morphogenetic disharmony, produces results which are notsubstantially different from the normally oriented skin series. Under the con-straints of the procedures used in this study, one may conclude that tail skinis not a morphogenetically active tissue though it can support normal limbregeneration in this species.

DISCUSSION

Regeneration of urodele limbs and tails requires not only a source of cellswith which to reconstruct the lost appendage but also a pattern by which theappropriate appendage will be constructed and in proper alignment. There are,

Fig. 3. Limb stump bearing tail skin graft in inverted orientation 62 days post-amputation. Regeneration has been totally arrested and an obviously tail-typeintegument completely envelops the stump.Fig. 4. Another example of arrested regeneration in a limb bearing an invertedcuff of tail skin. Although the humeral stump produced an elongate cartilaginouscap, no elbow was formed, hence its being categorized as showing no epimorphicregeneration. Specimen was fixed 63 days postamputation.Fig. 5. Cross section of a normal tail regenerate, the skin morphology of which maybe compared with that of both control and experimental limbs in this study. Dorsalside at top; spinal cord in centre above vertebral centrum.

116 C. E. DINSMORE

8

Effect of tail skin on limb regeneration 117however, apparent differences between limbs and tails in the ways in whichthey regulate their respective patterns. While skeletal muscle and skin are thepredominant sites of morphogenetic signalling in the limb, the spinal corddominates tail regeneration with tail skin and muscle being relatively inert(discussed in Dinsmore, 1979, 1981 b). This conclusion is based on several typesof experiments which were characterized and discussed early on by Goss (1961).Three types are drawn on in this discussion as bearing directly upon the problemof identifying morphogenetic activity in a tissue: deletion of a specific tissue,rearrangement of that tissue relative to adjacent tissues and substitution with acomparable tissue from another locus.

The importance of limb skin in pattern regulation during urodele limb re-generation has been demonstrated in several laboratories. A complete circum-ference of limb skin supports normal regeneration from a longitudinally splithalf stump, while inclusion of the skin when the limb is split will most oftenresult in a half regenerate (Goss, 1957; Dinsmore, 1982). This example ofdeletion indicates that a complete circumference of skin is sufficient for normalpattern regulation in the absence of other stump constituents. Nevertheless, thepattern itself is a mosaic in that a half stump produces an appropriately halvedregenerate, at least in the lower forelimb.

Positional dislocation of the limb skin by various degrees of cuff rotationalso produces distinct morphogenetic effects on limb regenerates. Supernumerarylimbs may develop on stumps bearing skin cuffs rotated 90° about the dorso-ventral axis of the limb thereby presenting a complete circumference of dorsalpositional information (Settles, 1967, 1978). Several studies have used rotationof limb skin about the longitudinal axis of limbs to demonstrate further themorphogenetic activity of this tissue (Lheureux, 1972,1975; Carlson, 1974,1975).Whether one axis (e.g. dorsoventral) or two axes (dorsoventral and antero-posterior) of the limb skin cuff are reversed relative to the internal limb tissues,multiple or supernumerary regenerates are often produced (Lheureux, 1972,1976), although this may vary depending upon the species employed (Carlson,1974).

Fig. 6. An experimental limb regenerate from a stump with normally oriented tailskin. Sacrificed 78 days postamputation, this regenerate differentiated an elbowjoint whose distal skeletal element was a single cartilaginous spike. The bone-cartilage transition in the humerus marks the approximate level of amputation.Fig. 7. Another example of distally deficient regenerate, this specimen had threedigits, the proximal segments of the metacarpals evident in this section. The tail-type skin is seen not only at the stump level but also over the proximal regenerate.Compare with Fig. 1. The arrow again indicates the level of amputation.Fig. 8. This frontal section through a four-digit regenerate bearing inverted tailskin shows the bases of four metacarpals and obvious tail skin on the proximal partof the regenerate. The arrow marks the approximate level of amputation.Bar = 01 mm.

118 C. E. DINSMORE

Focusing on the potential of the skin.and recalling Goss's (1961) classificationof experiments, the protocol of the present study incorporates both' substitution'of a tissue in the limb field by its counterpart from the tail and 'rearrangement'of that tissue in the host field by axial rotation. By grafting tail skin to the limbin the normal dorsoventral orientation, the question of whether or not general,field-specific information is borne by this tissue was answered in the negative.Although the skin in the area of the stump-regenerate interface has largegranular glands characteristic of the axial skin of this species, the morphologyof the limbs which regenerated (including the distal skin of the regenerate) wasappropriate for that appendage. This appears to conflict with earlier reports ofchimaeric regenerates from limbs bearing grafts of tail skin (e.g. Glade, 1957,1963, 1978). A comparison of details actually provides insight into diversity ofmorphogenetic control mechanisms among the tails of different urodele species.The species in the present work, Plethodon cinereus, lacks a tail fin and itsassociated fin mesenchyme. Both species employed by Glade, Notophthalamusviridescens and Ambystoma mexicanum, have distinct caudal fins along thedorsal and ventral margins of their tails. Without necessarily being 'contami-nated' with fin mesenchyme, dermis from finned tails may carry a bias towardorganizing adjacent undifferentiated connective tissue or blastemal mesenchymeinto a fin-like structure. The physically reorganized materials could then have a'second morphogenetic effect' (Glade, 1978) by mechanically disrupting thehost epimorphic field. Another way of interpreting the disrupted morphologyis to consider that competitive recruitment of cells into either a fin or a normallimb regenerate by the two field influences causes the formation of unresolvedregions where anomalous structures result. Whatever the eventual explanationfor the latter phenomena, they do not arise when the experimental animal lacksa caudal fin, at least not with the present experimental method.

Rearrangement or axial rotation of the substituted tissue is the other com-ponent of the foregoing protocol. The procedure asked what, if any, axialinformation was present or accessible in tail skin when grafted to the limb field.Again, the species employed showed no morphogenetic effect even though theskin graft produced apposition of dorsal position (skin) with ventral (internallimb tissues). Such maximal positional dislocation of limb skin results in amajority of regenerates being multiple (Lheureux, 1972; Carlson, 1974, 1975).In these cases, however, the anteroposterior as well as dorsoventral axis isrotated 180°. Since tail skin lacks an anteroposterior axis, a more appropriatecomparison of the tail-skin-to-limb data may be found in the single-axis-rotation experiments where cuffs of limb skin were grafted to the contralaterallimb with only one axis maximally dislocated. When only the dorsoventral axisof the limb skin graft was dislocated 180°, Lheureux (1972) found that 46%of the subsequent limb regenerates bore supernumerary structures. Maden &Goodwin (1980) have also shown that when the tips of axolotl limb buds areexchanged with the contralateral bud and rotated 180° such that only the dorso-

Effect of tail skin on limb regeneration 119ventral axis is inverted, 40 % of the limbs develop supernumeraries. However,Carlson (1974) repeated the skin graft experiment on ten axolotls and foundno supernumerary outgrowths. This may again indicate species differences ordifferences between larval and adult forms in the way morphogenetic informationis maintained, transmitted, received or any combination of these and otherfactors involved in regulating pattern formation. Supernumerary limb regene-rates have been produced on Plethodon cinereus in this laboratory by reversingeither the dorsoventral or anteroposterior asis of the limb skin or both simul-taneously (Fig. 1). However, the frequency and complexity of supernumerariesin this species for any of the axial skin dislocations is not as great as those foundby Lheureux (1972, 1975) using Pleurodeles wait Hi.

One final analogy for the tail-skin-to-limb procedure takes into account thelarger circumference of the tail relative to the limb. Care was taken to assurethat the dorsal aspect of the tail skin was always included in the graft, as itusually bore the red pigment which was a convenient marker for graft orienta-tion. Trimming the tail skin to fit the graft bed reduced the ventral skin contri-bution but invariably produced a cuff representing, at the least, dorsal andlateral caudal position.

Analogous with this reduced representation of normal axial position are otherexperiments performed by Lheureux (1975). He removed cuffs of whole skinor longitudinal strips of skin from the upper arm and repositioned them withthe proximodistal axis of the graft wrapped around the circumference of thelimb. By selecting either the level of amputation through the repositioned cuffor the origin of the graft strip, he produced limb stumps whose skin representeda single axial quality. When the stump bore skin of dorsal origin around itsperimeter, 28 % of the resulting regenerates had supernumerary structures butonly 34% approximated normal regeneration with three or four digits. Ampu-tation through a cuff of posterior or postaxial skin had an even greater disruptiveinfluence producing 49% supernumerary regenerates. It is thus unlikely thatthe reduced ventral representation in the tail skin grafts in the present experi-ment had any significant effect on the overall results. Morphogenetic influenceis thus considered to be a product not of a particular tissue in an epimorphicfield, but rather the result of its development in a particular field. Urodele tailskin thus appears to be morphogenetically inert, at least in this species.

This work was supported in part by BRSG Grant S07RR 05477 from the NIH and aGrant to the MDIBL from NSF.

I wish to thank Drs R. J. Goss, R. W. Glade and D. L. Stocum for their helpful commentson and criticisms of the initial manuscript. Their thoughtful suggestions were greatlyappreciated.

120 C. E. DINSMORE

REFERENCES

CARLSON, B. M. (1974). Morphogenetic interactions between rotated skin cuffs and underlyingstump tissues in regenerating axolotl forelimbs. Devi Biol. 39, 263-285.

CARLSON, B. M. (1975). The effects of rotation and positional change of stump tissues uponmorphogenesis of the regenerating axolotl limb. Devi Biol. 47, 269-291.

DEARLOVE, G. E. & DRESDEN, M. H. (1976). Regenerative abnormalities in Notophthalmusviridescens induced by repeated amputations. J. exp. Zool. 196, 251-262.

DINSMORE, C. E. (1979). Morphogenetic control during tail regeneration in Plethodon cinereus:The role of skeletal muscle. Devi Biol. 72, 244-253.

DINSMORE, C. E. (1981a). Regulative ability of the regenerating urodele tail: The effect ofunilateral soft tissue ablation. Devi Biol. 82, 186-191.

DINSMORE, C. E. (19816). Morphogenetic effect of rotated skin cuffs on tail regeneration inPlethodon cinereus. J. exp. Zool. 215, 151-161.

DINSMORE, C. E. (1982). Regeneration from split forelimbs in two species of urodele amphi-bian: A revised interpretation. Anat. Rec. 202, 46A.

GLADE, R. (1957). The effects of tail tissue on limb regeneration in Triturus viridescens. J.Morph. 101, 477-522.

GLADE, R. (1963). Effects of tail skin epidermis and dermis on limb regeneration in Triturusviridescens and Siredon mexicanum. J. exp. Zool. 152, 169-193.

GLADE, R. (1978). Morphogenetic influence of tail skin on limb regeneration during stagesin development of the skin in Ambystoma mexicanum. Growth 42, 253-262.

Goss, R. J. (1957). The relation of skin to defect regulation in regenerating half limbs. / .Morph. 100, 547-564.

Goss, R. J. (1961). Regeneration of vertebrate appendages. In Advances in Morphogenesis.(ed. M. Abercrombie & J. Brachet), vol. 1, pp. 103-152. New York: Academic Press.

LHEUREUX, E. (1972). Contribution a l'etude du role de la peau et des tissus axiaux dumembre dans le declenchement de morphogeneses regeneratrices anormales chez le triton,Pleurodeles waltlii Michah. Ann. Embryol. Morph. 5, 165-178.

LHEUREUX, E. (1975). Nouvelles donnees sur les roles de la peau et des tissus internes dansla regeneration du membre du triton, Pleurodeles waltlii Michah (Amphibien, Urodele).Wilhelm Roux' Arch, devl Biol. 176, 285-301.

LHEUREUX, E. (1976). Roles du derme et de l'epiderme dans la regeneration du membre dutriton, Pleurodeles waltlii. Bull. Soc. Zool. Fr. 101 Suppl. No. 3, 109-118.

MADEN, M. & GOODWIN, B. C. (1980). Experiments on developing limb buds of the axolotlAmbystoma mexicanum. J. Embryol. exp. Morph. 57, 177-187.

SETTLES, H. E. (1967). Supernumerary regeneration caused by ninety degree skin rotation inthe adult newt, Triturus viridescens. Diss. Abst. 28, 3567.

SETTLES, H. E. (1978). Supernumerary regeneration caused by ninety degree turning of limbskin in adult Notophthalmus. Growth 42, 297-307.

(Received 30 November 1981, revised 24 May 1982)