Functional Morphology and Evolution of Tail Autotomy in Salamanders

DAVID B. WAKE AND IAN G. DRESNER Department of Anatomy, The University of Chicago, Chicago, lllinois

ABSTRACT Basal tail constriction occurs in about two-thirds of the species of plethodontid salamanders. The constriction, which marks the site of tail autotomy, is a result of a reduction in length and diameter of the first caudal segment. Gross and microscopic anatomical studies reveal that many structural specializations are associated with basal constriction, and these are considered in detail. Areas of weak- ness in the skin at the posterior end of the first caudal segment, at the attachment of the musculature to the intermyotomal septum at the anterior end of the same segment, and between the last caudosacral and first caudal vertebrae precisely define the route of tail breakage. During autotomy the entire tail is shed, and a cylinder of skin one segment long closes over the wound at the end of the body. It is suggested that specializations described in this paper have evolved jndepend- ently in three different groups of salamanders. Experiments and field observations reveal that, contrary to expectations, frequency of tail breakage is less in species with apparent provisions for tail autotomy than in less specialized species. The tail is a very important, highly functional organ in salamanders and it is suggested that selection has been for behavioral and structural adaptations for control of tail loss, rather than for tail loss per se.

Tail loss and subsequent regeneration is sent facts concerning anatomical details a well documented phenomenon among of the basal tail region, a functional inter- the lower vertebrates. Although much is pretation of the morphology of the region, known concerning regeneration, the litera- and a hypothesis concerning the evolution ture on tail breakage is scattered and of the observed morphological specializa- scanty. This is particularly true of sala- tions. manders, a group that contains a number of species reputed to practice tail autot- MATERIALS AND METHODS omy. The present study was undertaken Cleared and stained as well as skeleton- in an attempt to determine those anatom- ized specimens of adults of all genera and ical features that permit and facilitate tail most species of plethodontid salamanders breakage in salamanders. In addition, we have been available for the study (see wished to investigate morphological spe- Wake, '66, for complete list). In addition, cializations in those species reported to large numbers of x-rays have been exam- practice tail autotomy, and to attempt an ined to obtain osteological information. analysis of the evolution of such specializa- Histological investigations were made on tions. Our working hypothesis has been adults of the following species of the fam- that species which demonstrate specialized ily Plethodontidae : Desmognathus fuscus, behavioral patterns resulting in tail loss D. quadramaculatus, D. ochrophaeus, likely have evolved physiological and ana- Pseudotriton ruber, Gyrinophilus porphy- tomical specializations which facilitate the riticus, Eurycea bislineata, Hemidactylium phenomenon. These morphological fea- scutatum, Plethodon cinereus, P. jordani, tures might, in turn, provide insight into Aneides lugubris, Ensatina eschscholtzii, the processes leading to structural and Hydromantes brunus, Batrachoseps attenu- functional specialization. atus, B. pacificus, Bolitoglossa subpalmata, Our study has revealed that basal tail B. morio, B. rufescens, Chiropterotriton constriction in salamanders is an indica- chiropterus, C. abscondens, C. multidenta- tion of a series of interrelated morpholog- tus, Pseudoeunjcea leprosa, Oedipina uni- ical specializations. In this paper we pre- formis, Lineatriton lineola, and Thorius

J. MORPII., 122: 265-306. 265 266 DAVID B. WAKE AND IAN G. DRESNER dubitus. Emphasis was placed on three all of the tail is lost. Many authors have genera, Hemidactylium, Ensatina, and reported tail autotomy in salamanders, liz- Bolitoglossa. ards, snakes, and mammals. The phenom- Material utilized in the study was fixed enon is most prevalent and most thor- in neutral buffered formalin, Bouin's or oughly understood in lizards (Quattrini, Zenker's fluid, decalcified in dilute HC1 or '52a,b, '54; Etheridge, '67) where breakage buffered EDTA, embedded in paraffin or is intravertebral. The breakage plane fol- parlodion, and sectioned serially in frontal, lows an unossified area that extends trans- sagittal, or transverse planes at 5 to 10 P. versely through from one to many Stains were chosen to best demonstrate vertebrae, along the caudal side of the connective tissue relationships. These in- associated intermyotomal septum and cluded routine Ehrlich's hematoxylin and through the skin. An unsegmented cartila- eosin, orcein (Mallory, '38), aldehyde fuch- ginous rod and rather atypical musculature sin (Gomori, '50), Heidenhain's azan (Lil- and skin is regenerated (Moffat and Bel- lie, '54), Van Gieson's connective tissue lairs, '64). Breakage is between vertebrae stain (Lillie, '54), and Bodian's Protargol in the other groups, but only in salaman- method of silver staining (Bodian, '36). ders is there any apparent morphological Living examples of many species of sal- specialization and only salamanders regen- amanders have been studied in the field erate a virtually normal tail. and laboratory, including Hemidactylium Salamanders may be placed in three scutatum, Ensatina eschscholtzii, Bolito- very general groups based on the structure glossa subpalmata, B. yucatana, Pseudo- of the tail and the pattern of tail breakage eurycea leprosa, Chiropterotriton multiden- (Wake, '66). These groups are not phy- tatus, C. chiropterus, Thorius dubitus, Ietic, but rather represent grades of mor- Batrachoseps attentuatus, B. pacificus, and phological and functional specialization. Plethodon cinereus. Large numbers of pre- Because this classification is convenient served animals have been studied in the and facilitates communication we have Field Museum of Natural History and the adopted it in the present report. University of Michigan Museum of No obvious specializations are associ- Zoology. ated with the breakage of tails in the more The classification scheme adopted here primitive families of salamanders. Break- for the salamander family Plethodontidae age is the result of traumatic events, such is that proposed by Wake ('66). Because as attacks by predators, rock falls, entrap- this classification is as yet relatively un- ment, and similar accidents. These ani- familiar, it is briefly summarized below mals usually have tails that are not sharply (genera in parentheses) : differentiated from the body and hence are Subfamily Desmognathinae (Desmognathus, of large diameter basally. Breakage is usu- Leurognathus, Phaeognathus). ally limited to posterior parts of the tail. Subfamily Plethodontinae Tribe Hemidactyliini ( Gyrinophilus, Pseudo- These animals are said to have thich-based triton, Stereochilus, Typhlotriton, Eurycea, tails in this study. Species characteristic Typhlomolge, Haideotriton, of this group included in our study are Hemidactylium). Pseudotriton Tuber, Desmognathus fuscus, Tribe Plethodontini (Plethodon, Aneides, Ensatina). and Eurycea bislineata. Tribe Bolitoglossini Some members of the family Salaman- Supergenus Hydromantes (Hydromantes). dridae (e.g. Chwglossa lusitanica) and Supergenus Batrachoseps (Batrachoseps). many members of the family Plethodon- Supergenus Bolitoglossa (Bolitoglossa, Oedipina, Pseudoeulycea, Chiropterotriton, tidae (e.g. Batrachoseps pacificus, Pletho- Lineatriton, Parvimolge, Thorius). don cinereus) have what we call slender- based tails. In these species breakage may Patterns of tail morphology occur at any point in the tail, but the route The term tail autotomy has been used of breakage is specialized in that the skin freely in the literature and it is impossible break occurs one segment posterior to the to determine exactly what various authors muscle break. The result is that, following mean by it. Nevertheless, the end result of breakage, the skin covers the open wound tail autotomy is always the same - part or and facilitates healing. This pattern, first TAIL AUTOTOMY IN SALAMANDERS 267 mentioned by Piersol ('lo), is called the sines) species. The amount of constriction wound-healing specialization by us. and details of structure differ considerably A final group that includes almost two- among the species. Constriction is pro- thirds of the species of the family Pletho- nounced in most, but is barely evident in dontidae has constricted-based tails. A dis- Oedipina, Lineatriton, and some species of tinct tail constriction is found at the base other genera. A universal feature of the of the tail and breakage, when it occurs, group is the presence of a first caudal seg- is ordinarily at this point. Members of the ment that is short relative to the segments group have been reported to practice tail that immediately precede and follow it. autotomy by many authors. We have stud- This segment is immediately posterior to ied this group more intensively than the the cloaca1 slit and is the first to entirely others, especially the following species : surround the vertebral column. Shortening Hemidactylium scutatum, Ensatina esch- of the segment is most pronounced in those scholtzii, and Bolitoglossa subpalmata. species with obvious basal constriction, With the exception of some brief and gen- such as members of the genus Bolitoglossa. eralized comments, nothing has been writ- In Hemidactylium, both the first and sec- ten concerning the anatomy of the basal ond segments are short. tail region or the mechanisms of autotomy The basal region of species with con- and tail breakage in these species. stricted-based tails is of smaller diameter than some more posterior parts of the tail. The tail base region. Decrease in the thickness of the muscula- Thick-based tails are well adapted pro- ture of the first caudal segment is the pri- pulsive organs important for locomotion in mary factor, but the constriction is often aquatic to semiaquatic habitats. Such hab- accentuated by the abrupt increase in di- itats were probably occupied by ancestors ameter of the tail behind the shortened of the family Plethodontidae and are still segment. This increase is the result of characteristic of most species of the sub- hypertrophied dorsal skin glands (Hub- family Desmognathinae and tribe Hemi- bard, '03; Esterly, '04) in some species dactyliini (Wake, '66). These groups all (e.g. Ensatina eschscholtzii, Hemidactyl- have thick-based tails that are often later- ium scutatum). The tail is a site of fat ally compressed and may have dorsal keels. deposition in all species of plethodontid Within both of the above groups distinct salamanders and, in some of the tropical trends in the direction of terrestriality are species (e.g. Bolitoglossa yucatana), the found. As species become increasingly ter- tails are extraordinarily swollen. restrial in habits, for example, the series A series of caudal grooves encircle the Desmognathus quadramaculatus to D. fus- tail, marking the position of the intermyo- cus to D. ochrophaeus to D. wrighti, tails tomal septa. These grooves tend to be deep tend to become round in cross section and in the basal region of species with con- increasingly slender near the base. A sim- stricted tails and especially in the northern ilar trend is seen in the hemidactyliines, highland species of Bolitoglossa. The culminating in Eurycea lucifuga. Members groove at the posterior end of the first of the remaining plethodontid genera, caudal segment may be so deep that the grouped in the tribes Plethodontini and skin on either side of it comes into contact. Bolitoglossini, are all terrestrial (Wake, This gives the impression of a clean break '66). No species of these latter groups in the epidermis, although such a break is have thick-based tails. Slender-based tails not present. Such deep grooves also accen- are found in Plethodon and Aneides among tuate the basal constriction. the plethodonines and Hydromantes and The tail defined. In species with con- Batrachoseps among the bolitoglossines. stricted-based tails, breaks occur much Constricted-based tails are found only in more frequently in the basal region than semiterrestrial (Hemidactylium, a hemi- elsewhere. Such breaks extend through dactyliine) and terrestrial (Ensatina, a the skin at the posterior end of the first plethodonine; Bolitoglossa, Oedipina, Pseu- caudal segment, between the musculature doeurycea, Chiropterotriton, Lineatriton, of the first caudal segment and its anterior Parvimolge, and Thorius, all bolitoglos- attachments, and between the first caudal 268 DAVID B. WAKE AND IAN 6. DRESNER vertebra and the vertebra immediately pre- processes and other parts of the last caudo- ceding it. This pattern defines the tail, in sacral vertebra. a functional sense. A more precise defini- Vertebral column. A more or less grad- tion is possible if one includes all aspects ual reduction in vertebral length charac- of the morphology of the tail base and terizes the region from the sacrum to the immediate postsacral region of sala- tail tip in generalized salamanders, with manders. the most pronounced shortening near the A distinct graded morphological series tip. In those species with basal tail con- of vertebrae is found between the sacral striction, the first caudal vertebra has (identified by its ribs which articulate with many specializations, the most obvious be- the ilia) and the first caudal vertebrae in ing the decreased length relative to the all salamanders. These have been called immediately preceding and following ver- the caudosacral vertebrae (Wake, '63, '66) and are two or three in number in pletho- tebrae. The first caudal vertebra of these dontid salamanders. These numbers have species is always measurably shorter than taxonomic significance and three is the the last caudosacral and the second primitive number. Two are present in the through fifth or more caudal vertebrae, subfamily Desmognathinae, some individ- but in Hemidactylium, the first and second uals of several species of Batrachoseps, caudal vertebrae are almost equally short- and all species of the supergenus Bolito- ened. Vertebral shortening is a very con- glossa (including the genera Bolitoglossa, spicuous feature of the most highly spe- Oedipina, Pseudoeurycea, Chiropterotriton, cialized members of the group with Lineatriton, Parvimolge, and Thorius). constricted-based tails, and is especially Although caudosacral vertebrae are tran- marked in Hemidactylium, Ensatina, and sitional in morphology, they are function- Bolitoglossa. Vertebral shortening is much ally distinct from caudal vertebrae and less obvious in the elongated species of must be considered part of the trunk. The Lineatriton and Oedipina, and is somewhat cloaca1 region is in a post sacral position intermediate between the extremes in the in salamanders and lies directly below the other genera. caudosacral series. Some haemapophysial Within the family Plethodontidae, post- rudiments may be present more anteriorly, sacral ribs are never found. The caudo- but the first fully developed haemal arch sacral vertebrae bear long transverse is located on the last caudosacral vertebra. processes, and transverse processes are This haemal arch is unique in being swept typically present on most caudal vertebrae. posteriorly so that its distal tip is well In most of the genera, the transverse proc- behind the posterior end of the centrum. esses arise near or slightly anterior to the In addition, the arch is greatly broadened midpoint of the centrum. The processes distally and lacks the anteriorly directed, may extend almost directly laterally with median hypophysial spine that character- only a slight posterior bend at the distal izes caudal haemal arches. The cloaca and tip (Batrachoseps), or they may extend associated structures are firmly attached posterolaterally forming acute angles be- by massive amounts of connective tissue tween the posteriorly swept process and to the ventral side of the centrum of the the long axis of the vertebral column last caudosacral vertebra and the ex- (fig. 1). The angle varies considerably, panded, blade-like part of its haemal arch. from very large (e.g. Hydromantes), to In addition the caudalifemoralis and intermediate (most genera, e.g. Eurycea, ischiocaudalis muscles attach to the inter- Plethodon, Pseudotriton), to rather small myotomal septum that is a continuation of (Aneides). Typically the first caudosacral the haemal arch of the last caudosacral vertebra bears the longest transverse proc- vertebra, and the caudalipubo-ischiotibialis esses and the processes steadily diminish muscles originate on the margins of that in size on succeeding vertebrae until they arch. Clearly breakage cannot occur ante- disappear entirely. The point of disappear- rior to this point and the tail can be said ance of the transverse processes is subject to start at the intermyotomal septum that to ontogenetic as well as interspecific and connects to the haemal arch, transverse intergeneric variation. TAIL AUTOTOMY IN SALAMANDERS 269

A c s-3 c-1

B

Fig. 1 A dorsal view of the right side of three caudosacral (CS-3)vertebrae and two caudal (C-1) vertebrae of (A) Ensatina eschscholtzii and (B) Plethodon jordani. Orientation of the transverse processes (Tp) in these two species differs considerably.

The genera with constricted tail bases very stout structures, rather irregular in differ markedly from the less specialized shape and dorsoventrally expanded (Wake, plethodontid genera in the orientation of '63). The greatly expanded distal tips are the transverse processes in the tail base oriented almost vertically in Ensatina, region. In these genera, the orientation of while the tips of processes on succeeding the processes changes from posterolateral vertebrae are tilted anteriorly (fig. 1). to sharply anterolateral in a space of three Processes are better developed and less ir- to five vertebrae (fig. 1). Hemidactylium regular in shape in Ensatina than in Hemi- and Ensatina have three caudosacral ver- dactylium. tebrae, but other genera with constricted- The other genera with constricted-tail based tails have two. In all, the pattern of bases differ from Ensatina and Hemidac- the transverse processes on the first is tylium in that they do not have especially essentially similar to that in less special- stout processes, but they are similar in ized groups. In Hemidactylium and Ensa- having processes that are directed ante- tina, the processes of the second caudo- riorly rather than posteriorly. The proc- sacral vertebra form a larger angle with esses arise at the anterior ends of the cen- the long axis of the vertebral column than tra and have relatively broad bases. do those of the first. The last caudosacral Typically the processes are spinous with vertebra in all species with constricted-tail acuminate, anteriorly directed tips (fig. 2, bases bears transverse processes that are 3). The processes of the first caudal ver- almost perpendicular to the long axis. In tebrae are very elongate in most species of some species of Bolitoglossa, these proc- Chiropterotriton and Bolitoglossa and in esses may be directed slightly anteriorly. some species of Pseudoeurycea. In species The usual situation in other species is for of both Chiropterotriton (e.g. magnipes) the processes to be directed slightly poste- and Bolitoglossa (most species), the trans- riorly or to be almost at right angles with verse processes are sinuously curved ante- the axis (fig. 1 ) . All caudosacral processes riorly and approach the posteriorly di- arise near the vertebral midpoints. rected processes of the last caudosacral Transverse processes on the first caudal vertebra. Southern species of Bolitoglossa, vertebrae in genera with constricted tails especially altamazonica and chica, have differ markedly in position and shape from very Iong processes that are directed those of other plethodontids. In all these sharply anteriorly. The last caudosacral genera, the processes originate near the processes of these species are also directed anterior end of the vertebra and they are rather sharply anteriorly; there is thus no directed slightly to strongly anteriorly. In overlapping of processes. In many species Ensatina, and to a lesser extent, in Hemi- of Bolitoglossa, from Mexico and Central dactylium, the transverse processes are America, the processes of the last caudo- 270 DAVID B. WAKE AND IAN G. DRESNER

Fig. 2 A slightly oblique frontal section through the tail base of Bolitoglossa subpalmata. The first caudal segment (S-1) is much shorter than the second caudal segment (S-2). The transverse processes of the first caudal vertebra (C-1) are directed anteriorly. M, mucous gland; P, poison gland; T, last trunk segment; PM, pelvic musculature; V, cloaca. X 219. TAIL AUTOTOMY IN SALAMANDERS 271

A

Fig. 3 A slightly oblique, diagrammatic frontal section through a broken tail of BoEito- glossa subpalmata. The skin (S) is beginning to close over the broken portion of the body (A). The muscle (M) of the tail (B) is thrust laterally following separation from the body. E, empty space often containing blood clot; S-2, second caudal segment. x 219. 272 DAVID B. WAKE AND IAN G. DRESNER sacral and the first caudal vertebrae over- most vertebrae. All are located near the lap, with those of the first caudal being anterior ends of the centra and most are located somewhat ventral to the others. oriented sharply anteriorly, in contrast to Usually in these species, the transverse the situation in less specialized species. processes of the first caudal are bifurcated There is a gradual posterior movement of distally, with the amount of bifurcation the processes toward the end of the tail reaching a peak in such large species as (fig. 4). Characteristically, these proces- B. dunni and B. macrinii. As the bifurca- ses are oriented anteriorly in the most tion becomes increasingly pronounced, the specialized species, in contrast to the prim- area between the widely separated tips be- itive condition. comes filled with bone and the process The last caudosacral and all caudal ver- acquires massive proportions, with two tebrae but the most recently formed have spines and a horizontally oriented web of posteroventrally sloping haemal arches. bone (figs. 4, 5). The intermyotomal sep- Large median hypophysial keels are borne tum attaches to the more posterior of these on the anteroventral surfaces of all caudal spines. Overlapping processes and distal vertebrae of generalized species. Genera bifurcation are also encountered in some with basally constricted tails differ from populations of Pseudoeurycea leprosa and other plethodontid salamanders in the some related species in the southern moun- shape of the haemal arch of the last caudo- tains bordering the Mexican Plateau. sacral and first caudal vertebrae. The Processes of caudal vertebrae beyond the haemal arch of the last caudosacral verte- first, in all species with basal tail constric- bra is broad and flat distalIy, and the re- tion, become increasingly shortened. Gen- sulting lamellate structure is oriented al- erally, these processes are as sharply or most transversely in the cloaca1 region. more sharply (Ensatina) directed to the This structure is much larger, more flat- anterior than are the processes of the first tened, and more laterally extended than caudal vertebra. There is considerable the corresponding structure in more gen- variation in the size of the processes. In eralized forms. As in other salamanders, some forms, such as Ensatina and Hemi- this arch lacks a hypophysial keel. The dactylium, they diminish in size rapidly posterior margin of the arch is usually in and are absent from most posterior verte- very close proximity to the corresponding brae, but in others they are present on structure on the first caudal vertebra, a

Fig. 4 A left lateral view of the first four caudal vertebrae of a detached tail of Bolitoglossa rostrata. Drawn from cleared and stained specimen. The bifurcation of the transverse process of the first caudal vertebra (Tp-l) is shown. S-1,first caudal segment; H, haemal arch; K, hypophysial keel; SM, intermyotomal septum. X 15. TAIL AUTOTOMY IN SALAMANDERS 273 situation not encountered in more general- zygapophyses of the last caudosacral ver- ized forms. The haemal arch of the first tebra are oriented with the articulating caudal vertebra of species with constricted- facets in an almost horizontal plane, as in based tails is more vertically oriented than the preceding and following vertebrae. In in other species and there is a tendency Ensatina, in contrast, the entire process, for the median keel to be reduced or lost. and especially the lateral part, is tilted First caudal vertebrae of Ensatina, Hemi- dorsally, and the articulating facet faces dactylium, Bolitoglossa, and Pseudoeurycea in a somewhat ventrolateral direction. In either lack such keels, or have a low me- addition, a well marked zygapophysial dian ridge. Keels are very reduced and are ridge, only slightly indicated on other ver- little more than median ridges in Chirop- tebrae, extends from the posterior margins terotriton and Lineatritcm. Reduction is of the prezygapophyses to the postzyga- less extreme but very evident in Thorius, pophyses (fig. 1). The combined effect of Parvimolge, and Oedipina. The keels are the upturned postzygapophyses and the rather small on the second caudal verte- well developed zygapophysial ridge lends a brae of Ensatina, Bolitoglossa, Pseudoeur- depressed aspect to the posterior portion of ycea, and Chiropterotriton and increase the neural arch. The prezygapophyses of progressively in size posteriorly (fig. 4). the first caudal vertebra are twisted dor- Maximum size is attained at about the sally, matching the angle of the upturned fourth to sixth vertebrae. In other genera, postzygapophyses. The net result of these keel enlargement is less obvious. modifications is that the zygapophyses at Generalized salamanders have zygapoph- this particular joint are smaller than at the ysial joints in which the amount of contact joints immediately preceding and follow- between post- and prezygapophyses of ad- ing, and there is a great reduction in the jacent vertebrae is relatively great in the amount of actual articulation between anterior portion of the tail, with the these zygapophyses. amount of contact decreasing posteriorly. The arrangement seen in Ensatina is re- In those forms with constricted tail bases, peated in Bolitoglossa, but the zygapoph- this generalization does not hold true. Per- ysial joint between the last caudosacral haps the most extreme conditions are and first caudal vertebra is less specialized. found in the genera Bolitoglossa, Ensatina, The zygapophyses at this joint are reduced and Hemidactylium. Primitively, the post- in size and are upturned, and there is a distinct decrease in the area of zygapoph- ysial overlap. The situation in Chiroptero- triton and Pseudoeurycea is very similar to that in Bolitoglossa, but the zygapophyses are less tilted. Slightly upturned zygapoph- yses of reduced size are also found in Lineatriton and Oedipina, but the modifi- cations are slight compared with Bolito- glossa and Ensatina. The form of the zygapophyses in Hemi- dactylium differs radically from that seen in the genera discussed above. The neural arches and associated processes on the postsacral vertebrae are very high in Hemi- dactylium. In addition, there has been much dorsal and posterior growth of the posterior margin of the neural arch of the last caudosacral vertebra. Thus, the mar- gins of the neural arch of this vertebra and the associated hyperapophyses are mas- sive. The posterior growth of the neural Fig. 5 Dorsal view of the first caudal vertebra arch and associated structures of the last of Bolitoglossa morio. X 36. caudosacral vertebra is so great that the 274 DAVID B. WAKE AND IAN G. DRESNER region almost contacts the anterior margin volved in the reduction and subsequent of the arch of the first caudal vertebra. constriction: (1) shortening of the seg- As a result, the fibers of dorsal muscula- ment as a whole and thus of the individual ture binding these vertebrae are greatly muscle fibers, (2) reduction in the thick- reduced in length and numbers. The pos- ness of the musculature as a result of terior margins of the neural arch of the apparent decrease in the number of muscle last caudosacral vertebra extend to a point fibers in the segment. The result of these dorsal to the anterior limit of the centrum factors is a tail that is structurally weaker of the first caudal vertebra. The postzyga- and more flexible at its base than it is a pophyses of the last caudosacral vertebra short distance posteriorly. are reduced in size, somewhat tilted later- Shortening of the first caudal segment ally and elevated. However, the processes is related to the forward movement of the are located in a position rather simliar to transverse processes of the first caudal ver- that in generalized species and have not tebra. Such a position for these processes been carried posteriorly by the growth of is found only in those species with con- the neural arch. The effect of the posterior stricted-based tails. The anterior position of growth of the neural arch has been to the processes causes the intermyotomal increase the distance between the post- septum, which attaches to the distal tip of zygapophyses of the last caudosacral verte- the transverse processes, to be moved an- bra and the prezygapophyses of the first teriorly. In those groups, such as some caudal vertebra. As a result the latter are species of Bolitoglossa, in which the trans- very elongated. The articulation between verse processes are not only moved ante- the zygapophyses at this joint is more re- riorly but are also very elongated and duced than that at the preceding or fol- extend far anterolaterally, the effect is pro- lowing joints and the lateral margins of found and segment shortening is great. In the processes are tilted dorsally, but to a members of one species group of Bolito- lesser extent than in Bolitoglossa and En- glossa, in which the processes are over- satina. It is apparent that the zygapoph- lapped by those of the last caudosacral ysial joints of the species with constricted vertebra, muscle fiber shortening reaches tail bases are weaker than those in the an extreme. A further complexity is the more generalized species. terminal bifurcation of the transverse proc- MuscuZature. The first caudal segment esses of the first caudal vertebra in mem- starts at the intermyotomal septum at- bers of this species group. The bifurcation tached to the last caudosacral vertebra. occurs only in those forms in which antero- The musculature of this segment differs in lateral extension of the process is pro- configuration from that of the caudosacral nounced. The intermyotomal septum that region in that the septum at the anterior joins the first and second caudal segments end of the segment encircles the vertebra. attaches to the more posterior projection. In contrast, the musculature of the caudo- The anterior projection is elongated and sacral region is limited to the dorsal and extends into the first caudal segment. As lateral sides of the vertebrae and is re- a result, the muscle attached to the ante- placed ventrally by the pelvic musculature. rior projection is very short and the bone Musculature of all caudal segments encir- of the enlarged process occupies space nor- cles the vertebrae and individual fibers are mally occupied by muscle fibers. The bi- oriented primarily parallel to the body axis. furcation is also important because it al- In primitive salamanders, including gen- lows extreme segment shortening without eralized plethodontids, the length of the inordinate lengthening of the second seg- caudal segments gradually decreases from ment. As a result, the "normal" tail seg- the base to the tip of the tail. Basal seg- mentation pattern is restored in about ments are usually about equal in length three segments (fig. 4). and the first is at least as long as the next The circumference of the first caudal few segments and longer than the remain- segment is much less than that of the last der. In sharp contrast is the reduced first caudosacral and the second caudal seg- caudal segment found in species with con- ments, in those species with constricted- stricted-based tails. Two factors are in- base tails. The result of this diminution is TAIL AUTOTOMY IN SALAMANDERS 275 that there is less musculature to be sepa- many glands and much loose connective rated from the intermyotomal septum of tissue. the first caudal segment than in the seg- The dermal and subcutaneous layers of ment that immediately follows. Conse- slender-based tails are thinner than the quently, the junction of the first caudal corresponding layers in thick-based tails. with the last caudosacral segment is an The deep layer of the dermis is relatively area of potential weakness. thin and less dense, and the subcutaneous The genus HemidactyZium differs from region is greatly reduced. Dermal glands the other genera with constricted-based are abundant and the poison glands are tails in that the second caudal segment especially numerous and large. The epi- also is shortened, although to a lesser de- dermis is somewhat thicker than in thick- gree than the first. In addition, the second based tails. No distinction between inter- segment has a smaller circumference, and intramyotomal regions is noted in based on thickness of musculature and ig- either thick-based or slender-based tails, noring the skin, than the third. Thus in except for the fact that the attachment of Hemidactylium the normal segmental rela- the intermyotomal septum causes the en- tionships are restored somewhat more grad- tire skin to be pulled medially, toward the ually than in other genera. In Hemidac- body axis. The result is the conspicuous tylium there is a constricted region rather caudal groove seen in these species. Cau- than a constricted segment at the tail base. dal grooves are generally less marked in species with thick-based tails. Comparative Microscopic Anatomy Skin of the lateral and ventral surfaces Skin and subcutaneous connective tissue. of species with constricted-based tails re- Skin of the tail base region of species with sembles that in species with slender-based thick-based tails, such as Desmognathus tails. Mucous glands of the dorsal and quadramaculatus, consists of a relatively dorsolateral skin are also somewhat simi- thin epidermis and a very thick dermis. lar in the two groups but tend to be larger The dermis contains a few, small mucous in some genera with constricted-based glands and moderate numbers of much tails, especially Ensutina and Hemidac- larger poison glands which lie in a bed of tylium. In other respects, however, there relatively loose connective tissue (fig. 6). are striking differences. The dermal and The glandular region lies between the base- subcutaneous connective tissue forms a ment membrane (in the sense of Hay, '66) fibrous sheath surrounding the tail in all and a deep, very thick layer of collagenous salamanders. In species with thick- and connective tissue. Between this layer and slender-based tails, there is no regional the underlying musculature is a subcuta- differentiation in the sheath and it is about neous layer of loose connective tissue. This equally thick throughout the basal region. fairly extensive tissue is most fibrous in The same condition prevails, except at the the intersegmental regions, where the fi- tail base, in species with constricted-based bers of the intermyotomal septa join it. tails (fig. 7). In contrast, a marked differ- The fibers of the subcutaneous region are entiation in this sheath is found at the essentially in continuity with those of the level of the posterior end of the first caudal intermyotomal septa. segment in species with basal tail con- Some variation is encountered within striction. At this point, the dermal and the group with thick-based tails. Pseudo- subcutaneous connective tissue is reduced triton Tuber has a very thin epidermis and or absent. Details of this connective tis- an extraordinarily thick dermis. Dermal sue discontinuity in one highly specialized glands are small and widely dispersed, and group and comparisons with other groups the dense dermal layer is much thicker are presented below. than that in D. quadrumaculatus. In con- Adult individuals of several species of trast, some of the more terrestrial species Bolitoglossa have very well marked con- of Desmognathus (e.g. D. ochrophaeus) strictions at the bases of their tails. In have skin resembling that found in species addition to the shorter and smaller first with slender-based tails. Here the epider- caudal segment, there is a deep groove mis is relatively thick and the dermis has marking the junction of the first and sec- 2 76 DAVID B. WAKE AND IAN G. DRESNER ond caudal segments. This groove is so there is a decrease in the amount of der- marked in some (e.g. B. rostrata) that it mal and subcutaneous connective tissue. gives one the impression that the skin is Only a few cutaneous fibers are continu- interrupted at that point. The groove is ous between the first and second segments, somewhat less marked in B. subpalmata, and obviously the region is structurally but is still pronounced (figs. €49). Sev- weak. The skin of Ensatina does not be- eral factors are involved. The first is the come as thin in the area of constriction as marked reduction in thickness of the epi- in Bolitoglossa because the dermal glands dermis at that point. More important is tend to be somewhat larger and a moderate the reduction in size and number of the amount of loose connective tissue is pres- dermal glands in the groove region. The ent (figs. 10, 11). The greatest reduction result is that the epidermis rests directly occurs in the deep layer that ordinarily on the connective tissue sheath, a condi- consists of densely packed collagenous tis- tion not encountered elsewhere in the sue. At the end of the first caudal segment tail (fig. 9). Finally, the connective tis- in Ensatina, this layer cannot be distin- sue sheath of the tail is greatly reduced guished. in thickness immediately anterior to the A discontinuity in the connective tissue groove. As one nears the intermyotomal sheath is present at the end of both the septum at the end of the first caudal seg- first and second caudal segments in Hemi- ment, the connective tissue decreases in dactylium (figs. 12, 13, 14). The dermal thickness and density. There is essentially and subcutaneous connective tissue at the an interruption in the sheath immediately anterior ends of these segments is about anterior to the point where the intermy- as thick as that in other tail segments. At otomal septum joins the skin (see below), about the midpoint of the segment the con- and only a few fibers join the dermis of nective tissue layer begins to diminish in the first caudal segment with that of the thickness and, near the end of the seg- second. Certainly this area is one of struc- ment, very little remains. Glands are ab- tural weakness. sent or greatly reduced in size in the skin The intermyotomal septum between the near the ends of these segments and the first and second caudal segments in Bolito- epidermis rests almost directly on the un- glossa subpalmata is oriented posterolat- derlying musculature, close to the postero- erally from its origin on the first caudal laterally oriented intermyotomal septa. Our vertebra. As the thick, dense collagenous observations on cleared and stained, as sheet nears the skin it curves posteriorly well as on museum preserved specimens, and joins smoothly with the subcutaneous indicate that breakage occurs at the end connective tissue (figs. 8, 9). As a result, of the first segment, but, in view of the the deep dermal and subcutaneous fibrous weakened nature of the skin at the end of layers of the second caudal segment pass the second segment, breakage could also without interruption into the intermyoto- be expected at that point. ma1 septum at the anterior end of the In all species of Pseudoeurycea, Chirop- segment. This continuous fibrous layer ap- terotriton, Lineatritm, and Thorius exam- proaches a straight line orientation in the ined, the connective tissue defect is less region where the skin of the first segment pronounced than in Bolitoglossa, Ensatina, joins that of the second (fig. 8). It is at this or Hemidactylium. Grooves may be pres- junction that the connective tissue discon- ent, but they are usually not as well de- tinuity is most apparent and it is here that veloped as in Bolitoglossa. The connective the skin separates during tail breakage. tissue defect at the posterior end of the The cutaneous groove is ordinarily not first caudal segment is present, but it is so obvious in other genera as in Bolito- much less obvious than in the above gen- glossa. Nevertheless, although there are era. In Thorius, the connective tissue sim- slight differences in general morphology, ply appears to be diminished in thickness the connective tissue specialization in and the dermal glands reduced in size and other genera approaches the condition in density in the area of the groove. Bolitoglossa. In Ensatina, a groove is usu- Diminution in the amount of fibrous ally evident and, in the area of the groove, connective tissue at the end of the first TAIL AUTOTOMY IN SALAMANDERS 277 caudal segment is the major skin speciali- at any point in the tail from the septum zation related to weakness or fragility in (figs. 11, 18). Ensatina and Hemidactylium. This feature In contrast, some of the species with is of considerable importance in Bolito- constricted-based tails have a unique pat- glossa, Chiropterotriton, and Pseudoeury- tern of muscle fiber attachment in the first cea as well, but in these genera, and pos- caudal segment. The attachment at the sibly in Ensatina, another factor is also posterior end of the segment is normal, important. In the most specialized condi- but that at the anterior end is apparently tion, such as is seen in Bolitoglossa sub- weaker than at any other point in the tail. palmata and Pseudoeurycea leprosa, the Among the species studied, the most ex- subcutaneous region of only the first cau- treme condition is encountered in Bolito- dal segment contains a number of small, glossa subpalmata. The anterior side of the highly basophilic cells (fig. 15). These intennyotomal septum at the anterior end cells form a distinct layer below the dense of the first caudal segment is identical to connective tissue layer at the base of the muscle septum relationships elsewhere in dermis. Posteriorly this cellular layer con- the body and tail. On the posterior side of tacts the posterolaterally oriented intermy- the same septum, however, the muscle otomal septum at the end of the segment. fibers of the first caudal segment are deli- The cells follow the septum to the point cately attached to the main mass of col- at which it becomes continuous with the lagenous fibers in the septum by slender fibrous layer of the second caudal segment. bundles of collagenous fibers (fig. 19). Here the cell layer turns rather sharply The fibers in this bundle are relatively few toward the surface and extends almost to in number and small in diameter. In addi- the base of the epidermis (figs. 16, 17). tion, the bundle contains small cells with The cell layer thus forms a partition be- large nuclei, which apparently result in a tween the dermis of the first and second further weakening of the attachment by caudal segments. One can also find inward displacing fibrous elements. extensions of the epidermis at the point Breakage occurs at this point, and the where the cell line is located. The orienta- connective tissue of the delicate bundle tion of these cells and of those in the epi- either separates from its attachment to the dermal invagination forms a separation fiber or breaks, and the sarcolemma of the plane that orients and facilitates breakage muscle fibers normally ruptures. This situ- of the skin at this point (fig. 17). ation is somewhat analogous to the rupture Weakening of the skin at the end of the that occurs at the myotendinous junction constricted region results from a simple in most vertebrates (Barrnett, '66). Rup- decrease in the amount of connective tis- ture of the sarcolemma results in an ex- sue in some species. In the most highly pansion of the released end of the muscle specialized groups, the insertion of a par- fiber, producing a mushroom-like cap (figs. tition of cells of unknown origin and func- 21, 24). tion into the dermis results in a separation Cells that are found in the delicate at- in the dermal and subcutaneous connec- tachment bundle may contribute to the tive tissue sheath that further weakens the blastema of the regenerating tail. Their skin at this region. Currently, investiga- functional versatility was apparent in one tions are in progress as to the origin and individual of Pseudoeurycea leprosa in biological properties of these cells. which the muscle had become separated Intermyotoma1 septa -muscle fiber re- from the septum on one side of the tail. lationships. The attachments of the indi- The cells of the attachment bundles in- vidual muscle fibers to the intermyotomal creased greatly in number, and some septa in the tails of all salamanders are moved posteriorly (fig. 20). These latter identical, with the exception of the first cells were apparently transformed into caudal segment of species with constricted- myoblasts, and all intermediate stages are based tails. Typically each muscle fiber is encountered. Detailed studies of the at- firmly attached to the septum proper by a tachment of the muscle fibers to the sep- rather dense bed of collagenous fibers, and tum and of the regenerative capacities of there is no obvious separation of the fibers these peculiar cells are currently under 278 DAVID B. WAKE AND IAN G. DRESNER investigation. Whatever the significance of tive tissue bind all vertebrae together and the anatomical detail, the delicate attach- there is no localized differentiation in the ing bundle of connective tissue is much column. weaker than the attachment at any other The amount of intervertebral connective point in the tail. tissue is reduced in quantity, thickness, The pattern described in B. subpalmata and density in those species with slender- is not nearly as apparent in the other spe- based tails in which tail breakage may oc- cies examined. However, in all available cur basally or at any place within the tail. specimens of Bolitoglossa, Thorius, Linea- No differentiation of one intervertebral triton, Pseudoeurycea, and Chiropterotri- joint relative to any others is apparent. ton, the attachment of the muscle fibers of A remarkable differentiation is found the first caudal segment to the septum in forms that have marked tail constric- joining the last caudosacral and first cau- tion. In all specimens of Bolitoglossa, Chi- dal segments exhibits relative weakness. ropterotriton, Pseudoeurycea, Thorius, En- Generally the fibers are pulled slightly satina, and Hemidactglium that we have away from this septum and a delicate examined, the amount of connective tissue bundle of connective tissue, occasionally binding the first caudal to. the last caudo- containing cell bodies, joins the muscle to sacral vertebra is dramatically reduced. the septum. Most of our material was in- The connective tissue between neural arch filtrated with paraffin and doubtless the pedicels is restricted to a few randomly condition seen in our slides is in part an distributed strands of collagen and there artifact of preparation. However, since the is essentially no connective tissue between condition is observed only at this position the centra (fig. 22). The amount of con- it is obvious that the attachment of the nective tissue in the zygapophysial capsule muscle to this particular septum in these is reduced relative to the joints immedi- species is much weaker than in neighbor- ately in front of and behind this joint. The ing segments. region is certainly one of weakness rela- The picture is even less clear in the tive to other parts of the tail. For example, other species with constricted-based tails. the joint between the first and second In some specimens, a few of the connec- caudal vertebrae might be considered “nor- tive tissue bundles at the anterior end of mal,’’ that is, large amounts of connective the first caudal segment are visible, but the tissue bind all parts of the vertebrae to- condition is certainly not typical. The mus- gether (fig. 23). cle usually appears to be less tightly at- Thorius is a diminutive species that does tached than elsewhere in the tail, and this not have an obviously constricted tail. Yet suggests that the region is one of relative it seems fully as specialized as those spe- weakness. We have determined experi- cies with marked constriction, in terms of mentally that the breakage plane extends lack of connective tissue between the last across these attachments. caudosacral and first caudal vertebrae. In Tail breakage occurs with apparent ease addition to its absence in sectioned ma- in species of genera with slender-based terial of unbroken tails, no connective tis- tails, for example, Batrachoseps and Pleth- sue is evident in the articular region in odon. We have been unable to detect any one specimen of Thorius dubitus which differences between the muscle attach- was induced to break its tail. The caudal ments on the anterior and posterior sides part of the broken tail was sectioned and of the intermyotomal septa of these forms. stained, and as is apparent from figure 24, Intervertebral attachnents. In general- the margins of the centrum are entirely ized salamanders, successive vertebrae are devoid of connective tissue. bound very tightly together by relatively A summary of characters associated large amounts of collagenous connective with tail base morphology is presented in tissue. The connective tissue is particularly table 1. well developed between the pedicel walls, around the zygapophysial capsules, and DISCUSSION around the margins of the centra. The Within the Order Caudata, specializa- thick, relatively dense bands of connec- tions related to tail breakage have evolved TAIL AUTOTOMY IN SALAMANDERS 279

TABLE 1 Thick-based tails Slender-based tails Constricted-based tails Break pattern Skin and muscle Skin break at posterior Skin break at posterior and break at same point and muscle break at muscle break at anterior anterior end of seg- end of first caudal seg- ment (wound- ment (wound-healing healing specializa- specialization ) tion) Length of first Equal to or longer Equal to or longer Shorter than adjacent pos- caudal segment than adjacent than adjacent pos- terior segment posterior segment terior segment Length of first Equal to or longer Equal to or longer Shorter than adjacent pos- caudal vertebra than adjacent than adjacent pos- terior vertebra posterior vertebra terior vertebra Orientation of trans- Sharply posterolateral Moderately to sharply Sharply anterolateral verse processes on posterolateral anterior caudal vertebrae Epidermis Thin Thick Thick Dermal glands Relatively small Relatively large and Very large and numerous; and scattered numerous largest and most concen- trated on dorsal part of tail Deep fibrous layer Thick; continuous Thin; continuous be- Thin; discontinuous be- of dermis between first and tween first and tween first and second second caudal second caudal caudal segments segments segments Attachment of muscle Strong; identical to Strong; identical to Weaker than in other fibers at anterior that in other that in other segments end of first caudal segments segments segment Connective tissue Numerous and dense Numerous and dense Few and scattered; absent fibers between last between centra. caudosacral and first caudal segments in at least two families, the Salamandridae that the skin of the segment in which the and the Plethodontidae. The lowest grade break occurs is pulled toward the exposed of specialization is the wound-healing vertebra. The folded skin collapses over mechanism, a breakage pattern in which the wound, facilitating both immediate the cutaneous break occurs at the pos- blood clotting and subsequent healing and terior end and the muscle break at the regeneration. Regeneration characteristi- anterior end of a given caudal segment. cally proceeds rapidly with a tail blastema Although this specialization basically re- forming that regenerates all tissue except sults from no more than a slight topo- the notochord (see also, Holtzer, Holtzer graphic reorientation of the intermyotomal and Avery, '55). The earliest reference to septa, it is highly significant to the orga- nism. When the tail breaks, all of the this pattern of breakage is a brief discus- muscle of the segment in which the break sion by Piersol ('10). He discussed the pro- occurs is lost. The end of the body then tective function of the collapsed skin and consists of the exposed vertebra and the the effect exposure of the muscle would intermyotomal septum that extends from have on increasing the rate of the contor- the vertebra in all directions out to the tions of the severed tail. Since that time, segment-long cylinder of skin. Orientation nothing of significance has been added of the septum relative to the skin is such on the subject, but his observations have 280 DAVID B. WAKE AND IAN G. DRESNER been repeated by several authors on sev- about the midpoint of the vertebrae and eral species. this feature tends to accentuate the pos- We have not widely sampled families terior shift of the segment attaching to the other than the Plethodontidae, but wound- caudal half of each vertebra. As a result of healing specializations are present in other this posterior shift of the segments and groups and the feature has apparently the lateral compression of the tail, the evolved several different times. Chioglossa tissue that extends beyond the point of lusitanica is an exceptionally long-tailed skeletal separation collapses medially and salamandrid that has been available for a vertically oriented wound closure ensues. study. It clearly has the same kinds of This is quite a different and less effective wound-healing specializations that occur kind of wound-healing specialization than in plethodontids. We have not found simi- is found in other plethodontids and seems lar specializations in other families. to be the result of evolutionary conver- Within the family Plethodontidae, gence. The genus Hemidactylium is also a wound-healing specializations are rather member of this tribe. widespread. Wake ('66) has recently placed The Tribe Plethodontini includes the plethodontid salamanders in four major genera Plethodon, Aneides and Ensatina. groups. The subfamily Desmognathinae Wound-healing specializations are espe- contains three genera, Desmognathus, cially well developed in Plethodm, partic- Leurognathus, and Phaeognathus. In all ularly in those species grouped as Eastern three, tail breakage is apparently common Small Plethodons by Highton ('62). Long, but the wound-healing specialization is not slender tails characterize members of this present, or at least, not well developed. group and tail breakage is common. The When an animal breaks its tail, always by specialization is equally well developed in trauma or by mechanically twisting the the larger species of the genus and we body away from the restrained tail, the have been able to demonstrate it in east- skin and muscle breaks occur at approxi- ern large species such as Plethodon jordani mately the same spot. Occasionally, how- and western species such as Plethodon ever, the skin break may occur relatively vehiculum. Members of the genus Aneides far posteriorly and may partially cover the are close relatives of Plethodon and the exposed wound. The skin break usually ex- wound healing specializations also occur tends in a rather jagged-edged ring around commonly in the more terrestrial species, the tail. A. flavipunctatus and A. hardii. Tail break- The Tribe Hemidactyliini includes a age is rare in A. lugubris, A. ferreus, and rather diverse array of genera and species. A. aeneus, climbing species which use The more primitive genera (Stereochilus, their tails in locomotion. Gyrinophilus, Pseudotriton) include spe- The final group of lungless salamanders, cies that have thick-based tails and no the Tribe Bolitoglossini, includes a variety specializations for tail breakage. Members of forms. The genus Hydromantes uses its of the genus Eurycea and some close rela- tail as a propulsive and supportive organ tives (Typhlomolge, Typhlotritm, Haideo- during terrestrial locomotion (Stebbins, triton) vary from thick-based (e.g. E. '47) and we know nothing concerning aquatica) to an almost slender-based (e.g. break patterns. Tail breakage is rare in E. lucifuga) condition. Tails of these spe- the genus. Batrachoseps demonstrates very cies are laterally compressed and, when well developed wound-healing specializa- breakage occurs, a partial closure of the tions. This genus includes very long-tailed wound follows. Analysis of sectioned ma- species which have tails that may be twice terial reveals that the skin and muscle the body length. Breaks occur at any point breaks are identical in topography at the along the slender tail. Although basal tail caudal side of an intermyotomal septum. breaks are uncommon, the genus must be Wound closure is the result of the sharp considered to be somewhat transitional in posterolateral orientation of the intermy- tail morphology, between the slender-based otomal septa and transverse processes and constricted-based groups. Breaks occur from the centra of the caudal vertebrae. with ease in Batrachoseps and tend to be The transverse processes are located at most common in the distal parts of the TAIL AUTOTOMY IN SALAMANDERS 281 tail. In large individuals with extensive fat eschscholtzii, ranges from extreme south- deposits in the tail, a region at the base of ern California into Canada, west of the the tail has a distinctly smaller diameter Sierra Nevada-Cascade Mountains crests. than adjacent posterior regions. This re- The strongly constricted tail has a similar gion is at least as constricted externally shape to that of Hemidactylium. The as is the corresponding region in Oedipinn, genus Bolitoglossa is a diverse assemblage a genus we place in the constricted- of about 50 species of neotropical sala- based group. The first caudal segment of manders. Members of the genus range Batrachoseps appears to be slightly shorter from northern Mexico into South America. than the second in our sections. However, All of the species are terrestrial. The con- within Batrachoseps, no sharp demarca- striction is usually strongly marked, espe- tion is evident between the caudosacral cially in the highland species such as B. and caudal vertebrae, such as occurs in rostrata and B. subpalmata. Tails tend to other plethodontids (Wake, '66). It is the be almost round and dorsal concentrations only plethodontid genus in which caudo- of skin glands are not so obvious as in the sacral vertebrae vary in number and there above genera. The constriction in many is no vertebral shortening in the tail base species of Bolitoglossa is emphasized by region. Wound-healing specializations are thinning of the skin at the posterior mar- very well developed, but there is no evi- gin of the first caudal vertebra. In healthy dence of a connective tissue defect or other individuals with fat tails, this region tends specializations associated with basal tail to be pulled in, toward the vertebral col- constriction. Other members of this tribe umn, by the intermyotomal septum. A have basally constricted tails. very conspicuous, deep groove results. Tail constriction. Basal tail constriction All remaining genera that have con- is found only among members of the fam- stricted tail bases are also members of the ily Plethodontidae. However, this family is tribe Bolitoglossini. Although species of by far the largest family of salamanders, Bolitoglossa have the most obviously con- and the more than 115 species that dem- stricted tails, constriction is also very evi- onstrate basal tail constriction constitute dent in members of the genera Chiroptero- about 40% of all living species of sala- triton and Pseudoeurycea. In the genera manders. Parvimolge, Lineatriton and Thorius tail In its most extreme expression, such as constriction is much less obvious. The last in a fat-tailed old adult of Bolitoglossa three are all diminutive forms with few rostrata, the constriction is very conspicu- fat stores in their tails and no great hyper- ous, but, in some members of every spe- trophy of the dorsal tail glands. Lineatriton cies in which constriction occurs, individ- has a secondarily elongated tail in which uals are found which do not demonstrate the length of vertebral centra as well as obvious external tail constriction. Such numbers of tail vertebrae are increased individuals are usually the most emaciated relative to more primitive groups. These and slender to be found. In this study we factors all tend to deemphasize the con- have emphasized three genera with very stricted area. Finally, the highly special- conspicuously constricted tail bases. Hemi- ized, extremely long-tailed species of the dactylium scutatum is the most terrestrial genus Oedipina have slight to no external member of the Tribe Hemidactyliini and tail constriction. Parvimolge, Lineatriton, lives in woodlands bordering bogs and Thorius, and Oedipina must be considered low-lying areas primarily in the northern members of the functional category that sections of eastern North America. It has includes forms with constricted tail bases. a tail with a large, rounded dorsal surface All have first caudal vertebrae that are and an almost knife-edged ventral surface. shorter than either the last caudosacral or The conspicuous tail constriction is located the second caudal vertebrae. This is a at the base of the tail, immediately behind character found elsewhere only among the vent and in front of the area of dorsal species with obviously constricted tail skin gland enlargement. Ensatina is per- bases. Confirmation has been obtained haps the most terrestrial member of the from serial sections of species of Linea- tribe Plethodontini. The single species, E. triton, Thorius, and Oedipina which reveal 282 DAVID B. WAKE AND IAN G. DRESNER that the muscle fibers of the first caudal effectively shortens the first caudal seg- segment are distinctly shorter than are ment. The combination of perpendicular those of the segments that immediately to somewhat posterior orientation of the follow. transverse processes of the last caudo- The pattern and mechanism of tail sacral vertebra and acute anterior orienta- breakage. Before discussing the mecha- tation of the processes of the first caudal nism of breakage, a brief review of the vertebra contributes further to segment- morphology of the tail base region of spe- shortening. cies with constricted-based tails is neces- The insertion of the intermyotomal sep- sary. Throughout the discussion, emphasis tum into the subcutaneous connective tis- will be placed on the most highly special- sue is almost perpendicular in primitive ized of the genera that display constricted salamanders. Measurements have been tail bases, Hemidactylium, Ensatina, and made on frontal sections through the cen- Bolitoglossa. Despite numerous differences, tra of a variety of salamanders to deter- all three genera share at least some fea- mine the angle formed by the intermyo- tures. In all genera, the centra of the first tomal septa and the connective tissue caudal vertebrae are markedly shorter sheath of the tail base region. The septum than those of immediately surrounding between the last caudosacral and first cau- centra. All have a haemal arch on the first dal segments has been pulled far forward, caudal vertebra that is more vertically and the angle between the septum and the oriented than are those of surrounding fibrous connective tissue sheath is mucb vertebrae. The arch has relatively short more obtuse in those forms in which tail anterior-posterior dimensions and either constriction is marked than in more gener- lacks a median keel or has but a low ridge alized species. In Pseudotriton, the angle is on its broad, anterior face. This surface is something over SO". In Desmognathus in close proximity to the broad posterior ochrophaeus, a species with a rather slen- flange of the haemal arch on the last cau- der-based tail, the angle is about 130". In dosacral vertebra. Zygapophysial articular Hemidactylium the angle approaches 150" surfaces of the last caudosacral and first -155", and in Bolitoglossa the angle may caudal vertebrae have smaller dimensions be as large as 165". Increasing the angle than those of surrounding vertebrae. The has functional significance in increasing anterior part of the neural arch of the fist the efficiency of the tail breakage pheno- caudal vertebra tends to be raised, and the menon and especially the wound-healing transverse processes arise very near the specialization. As the angle becomes large, anterior margin of the vertebra. The proc- the connective tissue of the intermyotomal esses tend to be oriented more perpendicu- septum forms almost a straight line with larly relative to the longitudinal body axis the subcutaneous connective tissue (fig. than in less specialized forms, and in 8). If the angle is measured relative to a many species, the processes are directed line laid along the connective tissue sur- anteriorly. rounding only the second caudal segment Muscle fibers in the first caudal segment rather than the whole tail, it is very close are noticeably shorter than those in sur- to 180". An additional factor is the angu- rounding segments. The decreased length lar relation of the subcutaneous connec- of the first caudal vertebra is not sufficient tive tissue of the first segment to that of the to account for the rather dramatic muscle second and to the intermyotomal septum. fiber shortening. Major contributing fac- Orientation of the septum and the connec- tors are the movement of the transverse tive tissue in the anterior part of the sec- processes of the first caudal vertebra to a ond segment results in essentially a straight more anterior position and a more vertical line of fibers. The fibers in the posterior orientation of the haemal arch on that part of the first caudal segment join this vertebra than is found elsewhere. Since line and form an angle of more than 50" the intermyotomal septum between the with the septum. Such an arrangement first and second caudal segments attaches would favor a break on the cranial rather to the process and the haemal arch, the than the caudal side of the attachment of change in orientation of these structures the septum to the skin. Anterior orienta- TAIL AUTOTOMY IN SALAMANDERS 283 tion of the intermyotomal septa through- between the muscle and the skin. These out the tail, with an acute cranial and an cells approach the epithelium at the caudal obtuse caudal angle relative to the skin of end of the segment and effect a complete the tail as a whole, occurs in all species separation of the fibrous layers of the two that have wound-healing specializations. segments. Attachment of the muscle fibers Regional differentiation with a great exag- of the first caudal segment to the anterior geration of the trend in the first two caudal intermyotomal septum is weaker than in segments occurs only in forms with con- any other surrounding segments. Finally, stricted tail bases. the amount of fibrous connective tissue The gross organization of the tail dis- between vertebrae in the area of constric- cussed above forms a clear morphocline, tion is much less than that found else- with the most primitive condition having where, and no fibrous tissue has been intermyotomal septa almost perpendicu- found between the centra of the last cau- larly oriented relative to the skin. There dosacral and first caudal vertebrae. is a gradual transition to forms that have The areas of greatest structural weak- all septa extending anteriorly from the ness correlate exactly with the break pat- skin to the axial skeleton. In the most ex- tern, as determined experimentally. We treme condition, such as is found in Pleth- have broken the tails in individuals of odon and Batrachoseps, wound-healing Bolitoglossa subpalmata, B. yucatana, En- specializations are very well developed satina eschscholtzii, and Thorius dubitus. and breakage readily occurs at virtually In all the break pattern is similar to that any point in the tail. At the derived end of illustrated in figure 3. The skin breaks at the morphocline is the group of species the posterior end of the first caudal seg- that have constricted tail bases. Here the ment and the muscle separation is at the anterior septa are sharply directed anter- anterior end of the same segment. Thus the iorly from the skin, but more posteriorly route extends through the few subcutan- along the tail, the septa become increas- eous fibers that join the skin of the first ingly perpendicularly oriented. In this caudal segment to the underlying muscul- group, features associated with wound- ature. More internally, the break extends healing are concentrated anteriorly and through the paravertebral fat deposits and breakage of posterior parts of the tail is between the first caudal and last caudo- uncommon. A cutaneous connective tis- sacral vertebrae. The notochord, blood ves- sue defect on the cranial side of the first sels, and finally, the spinal cord are also caudal intermyotomal septum concentrates severed at that point. breakage tendencies in the area of con- All plethodontid salamanders exhibit a striction. Additional features involved in characteristic behavioral response when tail constriction are the decrease in dorsal restrained by their tail. The animal twists skin glands in the basal tail region and a the body sharply and violently in a 360" decrease in the number of muscle fibers, arc around the body axis. Often the result and hence in the thickness of the segmen- is tail breakage at a point in advance of tal muscle layer, in the first caudal seg- the point of tail immobilization. In forms ment, relative to immediately more pos- with basal tail constriction, the entire tail terior parts of the tail. is lost except when the extreme caudal tip In earlier sections of this paper, we dis- is held. In oder to determine the sequence cussed the several unique histological fea- of events during tail breakage, we have tures of the first caudal segment. These in- performed a number of simple erperiments clude diminution or complete absence of and have obtained sufficient information the cutaneous and subcutaneous connec- to be able to outline the gross aspects of tive tissue at the caudal end of the first tail breakage. segment, immediately anterior to the inter- Tail breakage in forms with slender- myotomal septum. The skin glands may based and constricted-based tails occurs be absent from a small area just anterior rapidly as part of a violent behavioral pat- to this septum. In addition, the epidermis tern. It seems likely that mechanical fac- may be thin in this region. Several forms tors alone are involved in the species with have a layer of specialized cells inserted slender-based tails, the unrestrained parts 284 DAVID B. WAKE AND IAN G. DRESNER of the animal simply being twisted from spinal cord are separated in the same mo- the restrained parts. In the forms with tion. slender-based tails there is no evidence We have never observed completely vol- that autotomy, in the sense of voluntary untary tail breakage, or true autotomy. tail loss in the absence of mechanical fac- One individual of Bolitoglossa subpatmata tors, can occur. was restrained by its tail, just posterior to Apparently, a combination of voluntary the area of constriction. The segments con- and mechanical factors are involved in tracted violently for a few seconds, then a the breakage of tails at points of constric- sinusoidal wave was produced down the tion, at least in the most highly specialized back musculature and onto the tail at the species. When an animal is in an excited same time as the animal propelled itself state, as a result of being rolled or shaken, forward with its limbs. Basal breakage oc- and is held in the tail base region, one can cured and the animal walked away from detect the violent contraction waves in the the tail. A young Bolitoglossa yucatana caudal myotomes. Our hypothesis is that lost its tail as is was being rolled in a hu- the first event in “autotomy” in these forms man hand. The tail was broken readily and is the separation of the musculature of the animal had no opportunity to twist its the first caudal segment from the inter- body from its tail. It was not possible to myotomal septum that joins the first cau- detect the sequence of events during break- dal to the last caudosacral segment. We age because of the speed of the movement. have demonstrated above that this area Very little information can be found in is one of relative weakness. Immediately the literature concerning the mechanism following its separation from the septum, of tail breakage in forms with basal tail the muscle contracts in the direction of constriction. Almost all of the information the intermyotomal septum joining the available deals with Ensatina and Hemi- first and second caudal segments. These dactylium, and it is very scanty. Hubbard freed, contracted muscle fibers orient (’03) states that only under the most ex- themselves at right angles to the intermyo- treme provocation (being eaten by a snake, tomal septum to which they remain at- dropped into “killing fluid”) will Ensatina tached. Because this latter septum is ori- lose its tail. Esterly (’04), on the other ented at angles from 140” to 160” relative hand, noted that an individual of Ensa- to the tail surface, the diameter of the con- tinn that had been inadvertantly trapped stricted segment will suddenly be greatly among debris in a terrarium became very increased. The result is a greatly increased agitated upon gaining its freedom and be- force directed on the skin of the poster- gan moving about “swinging its tail from ior part of the first caudal segment, par- side to side like an angry cat.” After five ticularly at a point immediately craniad to minutes of such behavior Esterly “merely the point at which the intermyotomal sep- touched the tail” and it was suddenly tum between the first and second caudal “thrown off.” Stebbins (’51) stated, in re- segments joins the connective tissue of the gard to tail loss in Ensatina, “the tail may skin. This is precisely the area that we shed but usually only under stress, com- have shown to be structurally weak. It is monly breaking at the basal constriction.” likely that, in the most highly specialized In 1954, Stebbins added that the tail can forms, breakage of the skin at the poster- be shed under stress or when injured. He ior end of the first caudal segment follows did not elaborate further on the process of immediately upon separation of the muscle breakage, but did discuss the route of sep- of the segment from its anterior attach- aration. This discussion, although very ment. In most forms, twisting of the body short, is the most complete to be found in from the tail produces the actual skin the literature. It should be modified in re- break. Intervertebral attachments are gard to two points. The break, if perfect, weak and persumably the violent move- leaves no exposed musculature on the end ments of the partially detached tail are of the body, but only the connective tissue sufficient to effect final separation, al- of the intermyotomal septum and skin in though when the body is twisted, the skin, the areas away from the nerve cord, noto- vertebrae, notochord, blood vessels and chord, blood vessels and bone at the center TAIL AUTOTOMY IN SALAMANDERS 285 of the tail. Also, the cloaca is functionally The senior author has had considerable part of the trunk and is protected from experience with Ensatina in California, tearing because it is attached to the spe- Oregon and Washington. Animals have cialized last caudosacral vertebra, and not never broken their tails in the field, unless because it is protected by a specialized found partially pinned in rotted logs that caudal vertebra. In the same paper, Steb- were being torn apart. Individuals have bins (’54) reported that a good indication frequently been picked up by their tails of the critical maxium temperature of En- and it has been noted that tails of northern satina was that temperature at which animals appear to be lost more readily “spasmodic twitching, and sometimes than those of the southern animals. Two autotomy of the tail“ occurred. animals collected in Tacoma, Washington in June, 1966, broke their tails, but only Gnaedinger and Reed (’48) reported ob- after being allowed to twist their bodies servations on Oregon Ensatina that are at around the long axis while the tail was be- some variance with those of the workers ing held. cited above, all of whom worked with cen- Much less information is available con- tral and southern California populations. cerning Hemidactylium than for Ensa- They found that salamanders in their lab- tina. Bishop (’20) stated “a specimen in oratory lost tails if handled at all roughly captivity was induced to cast off the tail and that tail loss would “invariably result by touching the back with a drop of dilute within a short time if an animal were held acetic acid.” He reported that a “single suspended by the tail”. Stebbins (’54) de- violent wrench of the body sufficed to sep- liberately tested the degree of “caudal arate it from the tail,” but added that fragility” by picking up animals in the breakage was uncommon in specimens field by their tails. He reports that only one handled in captivity. Noble (’31) stated of some forty individuals lost its tail and, that Hemidactylium has a basal constric- in this animal, the tail was pinched as the tion that is a ‘‘provision for quick autot- salamander hung in a suspended position. omy.” Bishop (’41) stated, concerning Charles Brown (personal communication) Hemidactylium, “perhaps the most striking reports that none of hundreds of Ensatina habit is the voluntary dropping of the tail,” that he has collected in California over but did not elaborate. Other workers repeat several years have voluntarily broken their these statements. We have had only two tails at the time of collection. living Hemidactylium available for this We have been unable to discover any study. Both were frequently picked up by accounts in the literature of the precise be- their tails but they were not given an op- havior and movements of the animal im- portunity to twist their bodies while the mediately prior to tail breakage. In our tails were held. Several times they were experience, twisting the body around the released about midway through the twist, axis is the only way in which the tail can but neither animal lost its tail. be broken in Ensatina. We have conducted There is virtually nothing in the litera- a number of experiments with California ture concerning tail breakage in neotropi- Ensatina, including placing the animal in cal salamanders of the supergenus Bolito- deep and shallow water and raising the glossa apart from general statements con- temperature slowly above the critical ther- cerning the nature of the basal tail mal maximum, dropping the animals into constriction and allusions to its possible water heated to the critical temperature, use. Dr. L. C. Stuart (personal communi- dropping animals directly into 10% for- cation) reports that, by placing members malin solutions, placing a drop of dilute of the genus Bolitoglossa in water and acid on the back and tail, feeding the ani- rapidly raising the temperature above the mals to water snakes (Natrix), pricking critical thermal maximum, he has been the tail with a variety of instruments, and able to induce autotomy in some individ- suspending the animal from its tail and uals. In these, the tails simply appeared to pinching the tail. Unless the individuals drop off. were able to twist the body around the From our studies on large numbers of long axis, breakage did not occur. museum preserved specimens, it is appar- 286 DAVID B. WAKE AND IAN G. DRESNER ent that the muscle separation must pre- the first caudal vertebrae differ rather cede the skin break. Most workers prepare sharply in the three groups. Tails tend to animals for fixation by immersing them be rounder in the supergenus Bolitoglossa first in a solution of a narcotizing agent than in Ensatina and Hemidactylium. The such as chloretone, in warm water, or in extremes of segment shortening and of a weak alcohol solution. Some simply drop specialization in the attachment of the the living animal in the fixative. In most muscle fibers of the first segment found instances there is no opportunity for the in members of the genus Bolitoglossa are organism to wedge itself against an object not closely approached by Ensatina or to break the skin. Occasional specimens Hemidactylium. These and other charac- are found that have unbroken tails but in ters tend to support the hypothesis, based which the constriction of the anterior part on other features (Wake, '66), that tail of the first caudal segment is much more constriction specializations have evolved pronounced than in other individuals, and separately on three occasions. in which the posterior part of the same Basal tail constriction has evolved only segment is bulged out. Dissection reveals in forms that are largely or completely that the musculature is separated from the terrestrial in habits, since, in aquatic and intennyotomal septum at the anterior end semiaquatic habitats, tails have a very im- of the segment. This observation demon- portant propulsive function. It is likely strates that, at least in these individuals, that selective factors which led to the evo- the muscle separation precedes the skin lution of tail constriction specializations separation and suggests that this is the were in some way associated with the ac- normal sequence of events during volun- quisition of terrestrial habits or of adapta- tary tail breakage. Presumably muscle sep- tion to some particular aspect of the ter- aration is partially voluntary and skin and restrial way of life, such as predation by vertebral separation are strictly mechani- terrestrial organisms. Hemidactylium is cal, as a result of expansion of muscula- the most aquatic of the genera that have ture, application of external force, or other basal tail constriction, yet it is the most factors. terrestrial member of its tribe (Wake, '66). Evolutionary implications. It is evident Although it has an aquatic larval stage, that basal tail constriction and associated metamorphosis proceeds early and rela- phenomena have evolved in parallel in tively rapidly. The organisms are terres- three separate phyletic lines (Wake, '66). trial, woodland-inhabitants throughout life Ancestors of the three groups probably had except during egg-laying and brooding wound-healing specializations. Concentra- periods (Blanchard, '23; Bishop, '20, '41). tion and accentuation of these specializa- Ensatina is as terrestrial as any member tions in the basal tail region led to the of the family Plethodontidae and is prob- evolution of the extremes of specialization ably the most terrestrial and xeric adapted reported in this paper. member of its tribe (see also Stebbins, One can distinguish subtle differences '54). All members of the supergenus Boli- between the three groups that display basal toglossa lack aquatic larval stages; their tail constriction. For example, the short- eggs undergo direct terrestrial development ened vertebra is the fourth post sacral in (Storer, '25; Brame and Wake, '63). Hemidactylium and Ensatina, but the third Patterns of evolution in the supergenus in all members of the supergenus Bolito- Bolitoglossa require special comment. This glossa. Both first and second caudal verte- group includes over two-thirds of the spe- brae are specialized in structure and both cies of plethodontid salamanders. Many first and second caudal segments are short- species have a very large number of spe- ened in Hemidactylium, but not in the cial adaptations. Within this adaptive ra- other genera. The cutaneous groove at the diation are found genera with bizarre body caudal end of the first caudal segment of fonns. Among these are Lineatritcm, an Hemidactylium and Ensatina is never as elongated, poorly known, monotypic genus, sharply defined as in some members of the and Oedipina, an elongate, semiburrowing genera Bolitoglossa, Chiropterotritcm, and group of about 20 species. Both groups Pseudoeurycea. Details of structure of demonstrate only slight constrictions of TAIL AUTOTOMY IN SALAMANDERS 287 their very long tails. The genera Thorius lost their tails.” They stated that 14 others and Paruimolge include diminutive species had partially regenerated tails, some re- that are the smallest species of salaman- generating from the base but others from ders. These species do not have conspicu- more distal points. The figure for basal ous tail constrictions. Members of all of tail loss thus lies somewhere below 13% these genera have at least some features for their sample. Stebbins (’54) reported that we associate with tail autotomy (short- that 46 of 603 animals (7.7% ) studied by ened first caudal segment, specialized first him had regenerated all or part of the tail. caudal vertebra). Thus we suggest that the In an attempt to obtain information con- ancestral stock of the supergenus Bolito- cerning the natural incidence of tail loss glossa combined the features related to tail in plethodontid salamanders we have ex- autotomy encountered in primitive mem- amined large series of preserved animals bers of the modern genera Chirqterotri- in the Field Museum of Natural History, ton, Bolitoglossa, and Pseudoeurycea, the the Museum of Zoology, University of most generalized members of the group Michigan, and the Carnegie Museum. An (Wake, ’66). All members of these three attempt was made to use large samples genera now have basally constricted tails; collected at one time and place, but it was this was probably true of ancestors of the impossible to be assured of nonselective supergenus and thus basal tail constriction collecting (i.e., the collection of all indi- may be considered a primitive condition in viduals regardless of size and condition of these Neotropical salamanders. Selection tail). The collector of the series of 349 has apparently been for increased tail Bolitoglossa morio was paid for every ani- length and possibly for increased tail func- mal collected and apparently took every- tion in Lineatriton and Oedipina. Under thing that he encountered. Most of the such conditions, continued selection for other series were also collected by individ- tail autotomy specializations could be ex- uals or groups such as classes that were pected to be relaxed, and in these genera thought to be nonselective. Most animals only the grosser features of the specidiza- counted were adults. In general, the figures tions have been retained. presented represent minimal values. One would expect that organisms such Data presented in table 2 are based on as Hemidactylium, Ensatina, and Bolito- examination of museum collections. Break- glossa, which have localized areas clearly age figures for species with constricted- specialized for facilitating tail breakage, based tails were obtained by counting in- would have a high proportion of broken or dividuals with obviously regenerated tails regenerated tails in nature. Since salaman- or with tails that were broken prior to ders regenerate virtually all parts of the death. If breakage occurs after fixation, tail, including vertebrae but excluding skin in the break area does not close over notochord (Holtzer et al., ’55), it is im- the wound, and such individuals were possible to determine whether or not a eliminated. It is not as simple to detect slightly shorter than normal tail is regen- breakage in species with slender and thick erated without making histological prep- tail bases, but we accepted as broken only arations. Therefore all data presented by individuals in which approximately one- us and others must be considered to be half of the tail or more had been lost or minimal estimates. However, since Steb- regenerated. Because the patterns of break- bins (’54) has shown that salamanders age in species with thick-based tails differ take over two years to regenerate a tail greatly (rarity of basal breaks, no wound- and, since most salamander species seem healing specializations) from those with to live on the order of five to seven years, constricted-based tails, we have not at- it is unlikely that a significant number tempted any statistical analysis. The spe- of individuals with complete regeneration cies with slender and constricted tail bases have been missed. We have found data in are comparable to a greater degree, and the literature only for Ensatina as far as we have compared the combined data for natural incidence of tail breakage is con- the two groups for statistical si@cance. cerned. Gnaedinger and Reed (’48) report Chi Square analysis reveals that the figures that 4.9% of 169 Ensatina had “recently of 23% with broken tails for the slender- 288 DAVID B. WAKE AND IAN G. DRESNER

TABLE 2 Broken Species Number tails % Thick-based tails Desmognathus fuscus 239 33.4 Desmognathus monticola 169 11.2 Desmognathus wrighti 420 6.9 Eurycca bislineata 96 8.3 Slender-based tails Plethodon cinereus 273 13.0 Plethodon jordani 455 28.2 Plethodon ouachitae 147 19.8 Totals (SBT) 875 23.0 Constricted-based tails Ensatina eschscholtzii 62 9.7 Hemidactylium scutatum 139 4.3 Chiropterotriton chiropterus 397 7.1 Ckiropteratriton multidentatus 214 5.1 Pseudoeurycea leprosa 75 7 7.4 Bolitoglossa altamazonica 46 10.9 Bolitoglossa morio 349 18.9 Bolitoglossa platydactyla 269 7.1 Bolitoglossa rostrata 202 11.4 Bolitoglossa rufescens 275 6.2 Bolitoglossa subpalmata 142 7.7 Thorius macdougalli 184 5.4 Thorius narisovalis 56 17.8 Totals (CBT) 3092 9.5 based species and 9.5% with broken tails However, the tail is a very important organ for the constricted-based species differ sig- with locomotor, food storage, behavioral, nificantly at the level, p = < 0.001. Thus and respiratory functions. Tail loss in spe- tail loss is less frequent in our samples of cies with basal constriction is more com- species with constricted tail bases than in plete but less frequent than in their less those with slender tail bases at a level of specialized relatives. A complete tail, dis- high significance. carded and twitching violently, is probably There are many difficulties with data more efficient in distracting a predator, or such as these. We have no information, acting as a decoy, than part of a tail. It is for example, concerning patterns of vari- thus to be expected that morphological ation in a single species from different specializations that facilitate tail breakage habitats, different parts of its range, or in would be localized in the tail base region. sampIes collected at different times of the Some control over tail loss would probably year. No data are available for different be necessary before selection for localiza- size classes. Figures from three of the four tion and accentuation of breakage special- samples of species with thick tail bases izations could become effective. Because (table 1) do not differ significantly from the tail has many important functions and the figure for species with constricted tail because tails regenerate slowly, those or- bases, yet the fourth has the highest inci- ganisms that have some control over tail dence of tail loss encountered. Neverthe- loss would be favored over those in which less, it is apparent that tail constriction tail loss was indiscriminate. does not lead to an increase in the inci- Tail loss occurs only under conditions of dence of tail loss. extreme stress in most animals with basal Tail breakage can occur in all salaman- tail constrictions. This may be the result of ders and is so frequent in many species an evolutionary compromise in which an (e.g. Batrachoseps; Hubbard, '03) that it important organ has become highly spe- must be important in increasing survival. cialized in a way which may directly in- TAIL AUTOTOMY IN SALAMANDERS 289 fluence survival of the organism. Use of Bodian, D. 1936 A new method for staining the specializations by the organism results nerve fibers and nerve endings in paraffin sec- tions. Anat. Rec., 65: 89-97. in loss of all other functions. In salaman- Brame, A. H., Jr., and D. B. Wake 1963 The ders with basal tail constriction and asso- salamanders of South America. Los Angeles ciated morphological specializations, se- Co. Mus. Contrib. Sci., 69: 1-72. lection has resulted in highly adapted Etheridge, R. 1967 Lizard caudal vertebrae. organisms that take full advantage of the Copeia, in Press. Esterly, C. 0. 1904 The structure and regen- primitive functions of the tail. However, eration of poison glands of Plethodon. Univ. they have also evolved a high degree of be- Calif. Publ. Zool., l(7): 227-268. havioral control over loss of the complete Gnaedinger, L. M., and C. A. Reed 1948 Con- tail, a function that under certain condi- tribution to the natural history of the pletho- tions is of possibly crucial significance for dont salamander Ensatina eschscholtzii. CO- survival. peia, 1948(3): 187-196. Gomori, G. 1950 A new stain for elastic tissue. Amer. J. Clin. Path., 20: 665-666. ACKNOWLEDGMENTS Hay, E. D. 1966 Epithelium. In: Histology. Materials for this study have been pro- R. 0. Greep, ed. McGraw-Hill, New York, pp. vided by a number of individuals, includ- 87-88. Highton, R. 1962 Revision of North American ing R. Altig, A. H. Brame, Jr., C. W. Brown, salamanders of the genus Plethodon. Bun. M. M. Stewart, J. Kezer, R. Worthington, Florida State Mus. Biol. Sci., 6(3): 235-367. and H. Voris. Specimens have been loaned Holtzer, H., S. Holtzer and G. Avery 1955 An by Alan Leviton of the California Acad- experimental analysis of the development of emy of Sciences, Charles F. Walker of the the spinal column. IV. Morphogenesis of tail vertebrae during regeneration. J. Morph., 96: University of Michigan Museum of Zool- 145-171. ogy, Hymen Marx and Robert F. Inger of Hubbard, M. E. 1903 Correlated protective de- the Field Museum of Natural History, and vices in some California salamanders. Univ. C. J. McCoy, Jr. of the Carnegie Museum. Calif. Publ. Zool., l(4): 157-171. We are very grateful to these individuals Lillie, R. D. 1954 Histopathologic Technique and Practical Histochemistry. Blakiston Div., for their assistance. McGraw-Hill, New York, pp. 346-351. Peter P. H. DeBruyn has kindly allowed Mallory, F. B. 1938 Pathological Technique. us access to his photomicrographic equip- W. B. Saunders Company, Philadelphia, p. 351. ment, and has provided technical assis- Moffat, L. A., and A. D'A. Bellairs 1964 The regenerative capacity of the tail in embryonic tance. Some of the histological work was and post-natal lizards (Lacerta vivipara Jac- done by Mrs. Addye C. Brown. William L. quin). J. Embryol. Exp. Morph., 12(4): 769- Doyle and Ronald Lawson have read the 786. manuscript and made valuable sugges- Noble, G. K. 1931 Biology of the Amphibia. McGraw-Hill, New York, 577 pp. tions. Support for this work has been pro- Piersol, W. H. 1910 The habits and larval state vided by a National Science Foundation of Plethodon cinereus erythronotus. Trans. grant (GB-3868) to the senior author. The Canad. Instit., 8: 469-493. junior author is supported by a United Quattrini, D. 1952a Ricerche anatomiche e sperimentali sulla autotomia della coda delle States PubIic Health Service Traineeship lucertole. I. Dinamica dell'autotomia e conse- (T1 GM 94). quenze nel tegumento. Arch. Zool. Ital., 37: 131-170. LITERATURE CITED 1952b Ricerche anatomiche e speri- mentali sulla autotomia della coda delle lucer- Barrnett, R. J. 1966 Muscular Tissue. In: His- tole. 11. Musculotura adipe sottomuscolare e tology. R. 0. Greep, ed. McGraw-Hill, New scheletro. Arch. Zool. Ital., 37: 465-515. York, p. 188. Bishop, S. C. 1920 Notes on the habits and 1954 Piano di autotomia e rigen- development of the 4-toed salamander, Hemi- erazione della coda nei Sauri. Arch. Ital. Anat. dactylium scutatum (Schlegel). New York Embr., 59: 225-282. State Mus. Bull., 219,220: 251-282. Stebbins, R. C. 1947 Tail and foot action in 1941 Salamanders of New York. New the locomotion of Hydromantes platycephalus. York State Mus. Bull., 324: 1-365. Copeia, 1947(1): 1-5. Blanchard, F. N. 1923 The life history of the Stebbins, R. C. 1951 Amphibians of Western four toed salamander. Amer. Nat., 62: 156- North America. Univ. Calif. Press, Berkeley 164. and Los Angeles. 290 DAVID B. WAKE AND IAN G. DRESNER

1954 Natural History of the salaman- Wake, D. B. 1963 Comparative osteology of the ders of the plethodontid genus Ensatina. Univ. plethodontid salamander genus Aneides. J. Calif. Publ. Zool., 54(2): 47-124. Momh., 113(1): 77-118.

Storer, T. I. 1925 A synopsis of the Amphibia ~ 1966 Comparative osteology and evo- of California. Univ. Calif. Publ. Zool., 27: 1- lution of the lungless salamanders, family 342. Plethodontidae. Memoirs, So. Calif. Acad. Sci., 4: 1-111.

PLATE 1

EXPLANATION OF FIGURE

6 Epidermis (E) and dermis (D) of Desmognatkus quadramaculatus. The intermyotomal septum (SM) is continuous with the thick sub- cutaneous connective tissue (SC). X 293. TAIL AUTOTOMY IN SALAMANDERS PLATE 1 David B. Wake and Ian G. Dresner

29 1 PLATE 2

EXPLANATION OF FIGURES

7 Section of skin between third caudal segment (SB) and second caudal segment ( Sz) of Bolitoglossa subpalmata. Note the continuity of the thick fibrous layer between segments. X 150.

8 Section of skin between first (S1) and second (S2) caudal segments. The dermal connective tissue (D) is much reduced anterior to the tail constriction. CP, cell partition. X 150.

292 TAIL AUTOTOMY IN SALAMANDERS PLATE 2 David B. Wake and Ian G. Dresner

293 PLATE 3

EXPLANATION OF FIGURE

9 Magnified view of the tail constriction in figure 8. The break in the skin will extend from B- to -B. Note the epidermis is in contact with the dermal connective tissue. SI, first caudal segment; SZ,second caudal segment; M, mucous gland; P, poison gland; DI, dermis of first caudal segment; Dz,dermis of second caudal segment. X 683.

294 TAIL AUTOTOMY IN SALAMANDERS PLATE 3 David B. Wake and Ian G. Dresner

295 PLATE 4

EXPLANATION OF FIGURES

10 Section of skin between second ( Sp) and third ( SI) caudal segments in Ensatina eschscholtzii. The dermal connective tissue is thick and continuous between segments. X 150. 11 Section of skin at the tail constriction in Ensatina eschscholtzii be- tween first (&) and second (SZ) caudal segments. The dermal con- nective tissue is absent (A) anterior to the constriction. X 163.

296 TAIL AUTOTOMY IN SALAMANDERS PLATE 4 David B. Wake and Ian G. Dresner

297 PLATE 5

EXPLANATION OF FIGURES

12 Frontal section between third (S3) and fourth (S4) caudal segments of Hemidactylium scutatum. x 75. 13 Frontal section between second (SZ) and third (SS) caudal segments in Hemidactylium scutatum. x 75. 14 Magnified view of the tail constriction from figure 13 between second (SP) and third (S3) caudal segments. The dermal connective tissue is absent (A) anterior to the constriction. X 390.

298 TAIL AUTOTOMY IN SALAMANDERS PLATE 5 David B. Wake and Ian G. Dresner

299 PLATE 6

EXPLANATION OF FIGURES

15 Section demonstrating the cell partition (CP) extending beneath the skin in the first caudal segment (S1) in Bolitoglossa subpalmata. D, dermis. x 357. 16 Section at the tail constriction of Bolitoglossa subpalmata demon- strating the cell partition (CP) as it nears the constriction. D, dermis. X 260. 17 Section of skin at the tail constriction in Bolitoglossa subpalmatu showing the cell partition extending (CPE) almost to the epidermis. CP, cell partition; D, dermis; SM, intermyotomal septum. X 420.

300 TAIL AUTOTOMY IN SALAMANDERS PLATE 6 David B. Wake and Ian G. Dresner

301 PLATE 7

EXPLANATION OF FIGURES

18 Muscle attachment to an intermyotomal septum in Desmcgnathus quadramaculatus. X 260. 19 Muscle attachment to the intermyotomal septum in Bolitoglossa sub- palmata. Note the specialized attachment to the septum in the first caudal segment (S). CS, last trunk segment. x 260. 20 Muscle contracted away from the intermyotomal septum in Pseudo- eurycea Zeprosa. Note the great proliferation of cells (PC) which follows. S, first caudal segment. X 260. 21 End of muscle fibers after breaking away from the intermyotomal septum in Bolitoglossa subpalmata. x 420.

302 TAIL AUTOTOMY IN SALAMANDERS PLATE 7 David B. Wake and Ian G. Dresner

303 PLATE 8

EXPLANATION OF FIGURES

22 Junction between the last body vertebra and first tail vertebra show- ing absence (A) of binding connective tissue in Hemidactylium scutatum. x 163. 23 Comparable section at the junction between first and second caudal vertebrae showing large fibrous bundle (F) binding vertebrae in Hemidactylium scutatzrm. x 163. 24 Anterior end of a broken tail of Thorizcs dubitus. x 138.

304 TAIL AUTOTOMY IN SALAMANDERS PLATE 8 David B. Wake and Ian G. Dresner

305