Phylogenetic and Evolutionary Perspectives on an Enigmatic Organ

7/29/2019 Phylogenetic and Evolutionary Perspectives on an Enigmatic Organ

1/16

ZOOLOGY00l0gy 101(1998), 107-1 230 y Gustav FischerVerlag idysis of Complex System

Phylogeneticand evolutionary perspectiveson an enigmatic organ:the balancer of larval caudateamphibiansAndrew J. Crawfordl and DavidB.Wake2

Committeeon Evolutionary Biology', Universityof Chicago, IL60637, SA;Tel.: 773 702-9638,Fax: 7737024699, e-mail:crawfordBuchicago.eduMuseum of Vertebrate Zoology2,Universityof California, Berkeley, CA 94720-31 0, USA;Tel.: 510 642 3567, Fax: 510 6438238, -mail: wakelabQuclink4.berkeley.edu

Received:10. 1. 1998; revised version: 1.6. 1998;accepted:26. . 1998Key words: Collagen, evolution, homoplasy, immunohistochemistry,morphology, phylogenetic analysis.

SummaryThe balancer is a larval appendagewhose taxonomicdistribution s limitedto approximately half of the speciesfound in three of the ten families of salamanders. Theseorgans are small projections from the region of the jawjoint that are present before limbs develop; they appar-ently function in mechanical support and chemicaladhe-sion. Balancersare typically associatedwith speciesthatare pondbreeding, as opposed to stream breeding, andare therefore assumed to be adaptive in still water andnon-functional in flowing water. However, many excep-tions to this generalizationexist. We explain this unusual

distributionby combining morphologicaland naturalhisto-ry data with a phylogenetic analysis. Because the bal-ancer has been littlestudied, we summarizethe literatureon morphology,development, variation, function, ontoge-netic fate, ecological distribution, and homology. Usingimmunohistcxhemical methodswe find that the balanceris a complex organ. The balancercontainsa layer of typeIl collagen that may function as a non-bony skeleton andis innervatedby the ramus mandibularisof the fifthcranialnewe. We hypothesize that balancers evolved as asynapomorphy of advanced salamander families, afterthe divergence of the Sirenidae, and that they have beenlost independently many times. Homoplasy is unidirec-tional involving loss of the balancer alone, according tothis hypothesis.Introduction

The caudate balancer is a paired, larval organthat isslender and rod-like with slightly bulboustips.l ) Correspondence o: A. J. Crawfordz, Reprint ordersto: D.B. Wake

Each balancerprojects laterally and posteroventral-ly from a point slightly posteroventral to the eye ofthe hatchling, dorsal to the jaw joint. It is a morpho-logical novelty of Caudata (Amphibia), putativelywithout homologue (Lieberkind 1937). Balancersare found in a limitednumber of caudate taxa, andonly in mid- to late embryonic and early hatchlingstages of development. The organ has largely es-caped the recent attention of herpetologists andevolutionary morphologists. In this paper we reviewthe literature on balancer morphology and reportnew, comparative irnmunhistochemical data. Wefurther provide a comprehensive summary of cur-rent knowledge about development, variation, func-tion, ontogenetic fate, ecological distribution, andhomology of the balancer.We then present a phylo-genetic analysis of the distribution of balancers,evaluating their origin and evolution in light of ourunderstandingof morphologyand life history.Balancers are typically found in those caudatespecies that possess a pond-type lawal morpholo-gy and life history (Noble 1931). These organs arethought to confer physical support and stability of alimbless larva at rest (e.& Pough et al. 1998; butsee below). To explain its patchy taxonomic distri-bution, Noble (footnote in Harrison 1925) proposeda working hypothesis that ?he balancerdevelops inall forms where it can function," implying that thisorgan may function in a still pond, but not in astream or swift current. Noble also acknowledgedsome exceptions, and '?he absence of balancers insome pond-type larvae (e.g., sirenids) is unex-plained at present" (Duellman and Trueb 1986, p.157). Further, "thecurious absence of a functional

Zoology 101 (1S 8 )2 107
http://crawfordbuchicago.edu/http://wakelabquclink4.berkeley.edu/http://crawfordbuchicago.edu/http://wakelabquclink4.berkeley.edu/


2/16

balancer in at least two races of Ambystoma tigri-num does not appear explainable at this time?(Salthe and Mecham 1974, p. 455). Commentssuch as these suggest that a phylogenetic analysisof balancer evolution is needed.Balancers were first described by Spallanzani(1784), who mistook them for ?smallerbundles ofgills? (p. 59). These organs have been identified byvarious terms, including crochets de Rusconi (vanBarnbeke 1880),Kiefetiuttsatz and Fortsatz desKiefehogens (Maurer 1888), SaugnApfe (Wieder-sheim 1875), Rusconi?sche Hdkchen (Wunderer19 10), Stiifrorgane (Egert 1973), Balanzierstange(Mu ayama 19281, balanc ng st ks (0ama1929b), aftfaden Mangold 1931), nd Haftorgane(Lieberkind 1937). The term ?balancer?has beencredited erroneously to Clarke (1880) who, in fact,stated that the term was already in use, and notedthat its origin derived from resemblance of the or-gans o the balancers (i.e., halteres) of dipteran in-sects.Morphology

The general morphology of balancersshows re-markable constancy. While anatomical studies ofbalancers have been conducted in only a fewspecies, taxonomic diversity has been well repre-sented: Ambystoma jeffersonianum (Anderson andKollros 1%2), A. maculatum (Latta 1919; Harrison1925), A. opacum (Latta 1919;Anderson and Kollros1962), A. tigrinurn (Nicholas 1924b3, Hymbiius ni-grescens (Murayama l 28), Notophthalmus Virides-cens (Bell 1907), Pleurodeles walrl (Fox 1985; Lie-berkind 1937), T&ws ajpestris (RollhBuser-terHorst 1977), T: cristafus (Lieberkind 19371, and T;vulgaris (Maurer1888;Egert 1913;Lieberkind1937).Balancers are vascularized by branches of thehyomandibular artery and jugular vein, which to-gether form a single loop within each organ {Harri-son 1925; Maurer 1888). Each balancer is lined bytwo layersof epithelialcells, an outer cuboidal layerandan inner columnar layer, which cover the base-ment membrane (Harrison 1925; Le., the basementlamella of Fox 1985). Filaments of collagen (Fox1985)compose the basementmembraneof the bal-ancer and lack the high degree of orientationfoundin the skin (Anderson and Kollros 1962). Below thebasement membrane lies an undifferentiated mes-enchyme matrix (Harrison 1925; Anderson andKollros 1962). A branch of the fifth cranial nenresupplies the balancer (Harrison 1925), and the bal-ancer epithelium is more extensively innenratedthan the surrounding tissue (Fox 1985).The bulbous tip of the balancer contains manysmall secretory cones, each emanating from a sin-gle epithelialcell, that excretea sticky mucous [Am-108 Zoology 101 (1998)2

bystoma maculatum (Harrison 1926); Hymbius ni-grescens (Murayama 1928); riturus alpestris(Greil1906)). Bell (1907) may have obsewed these samestructures, which he calledthick, sucker-likeprojec-tions, in Nofophthalmus viridescens. Investigationof the cells of secretory cones by etectron micros-copy has revealedan extensive, granular endoplas-mic reticulum and well developed Golgi complexesthat are involved in the synthesis of a muco-pro-teinaceous substance (Fox 1985).Dunn (1923a) and Noble (1927) reportedthe oc-currence of ?double balancers? in Hynobius m e -vius. These reports are second hand accounts ofTago?s 1907) original study published in Japanese;they have never been verified. In contrast, Oyama(1929a) reporteda complete absence of balancersin H. naevius. The photographs of H. naevius inObara (1984) also show no balancers, standard ordoubled, in the embryos and hatchlings. Accordingto M. Matsui (pers. comm.), the species Tag0 stud-iedwas in fact H, tokyoensis,which has typical bal-ancers. Perhaps the Japanese term for ?pairedbal-ancers?was erroneously translated as ?doublebal-ancers.?Noble (1927, p. 37) lso reported that dou-ble balancers were found once in Ambystoma, buth e provided no evidenceor citation for this claim.ImmunohlstochemicalAnalysis

Balancers of Ambystoma gracile (Ambystomati-dae) and Taricha t o m a (Salarnandridae) werestudied using whole mount immunohistochemistry.Eggs of the two species were collected in naturallyoccurring ponds in Del Norte and Monterey coun-ties, California, respectively, allowed to hatch in thelaboratory, and the larvae fixed in Dent fixative (onepart dimethyl sulfoxide in four parts 100% metha-nol). Following standard techniques (Klymkowskyand Hanken 1991 y whole mount preparations weremade usingtwo kindsof antibodies: type II collagen(llsB3/15A4; Linsenrnayer and Hendrix 1980) whichstains for dense collagen matrices, and alpha-tubu-lin (6-IIB-I; Piperno and Fuller 1985) which stainsfor peripheralnerves.Balancers are closely similar in the two species.They are attached to the side of the head, just be -hind and below the eyes, and are directed posteri-orly in the preserved specimens. They are wellstained by type I I collagen, especially near theirbase but also distally (Figs. 1and 2). There is a rel-atively swollen basal region that is apparently anoutgrowth of the outer surface of the developingpalatoquadrate cartilage, immediately above theposterior end of Meckel?scartilage. Also visible atthis stage of development are incomplete but pairedMeckel?s cartilage, trabecula, suspensorium, oticcapsuleand parts of the hyobranchial apparatus,as


3/16

are proximal elements of the forelimb and shouldergirdle. The surface of the balancer appears to befuzzy, especially distally, in these whole-mounts.The apparent fuzziness may be due to secretorycones protrudingfrom thesurfacenear the tip (Har-rison 1925). Proximal portions of the balancer aredensely stained. This stained region is relativelythick and completely encircles the base of the bal-ancer. Although it is thinner distal to the base, thedensely stained material is clearly present through-out the length of the organ as a distinct layer, whichis a continuation of th e basement membrane (thebalancer membrane of Harrison 1925). The rigidityof the balancer is likely due to this thick layer, whichmay serve as a compact, non-bony skeleton. Harri-son (1925, p. 264 andsee ig. 49) found that a sev-ered balancer with its associated epithelium andmesenchyme removed by digestive enzymes re-tains its shape end appears similar to a long nar-row test-tube. However, type I1collagen does notnecessarilyserve 8 skeletal function, and other fac-tors may contribute to the balancers rigidity.Balancers in both species are supplied by a dis-tinct nerve that branches from the ramus mandibu-laris of the fifth cranial (trigeminal) newe. The bal-ancer new8 iswell developed and branches exten-sively (Figs. 3 and 4). The main nerve forms a longloop within the balancer that fades into smallbranchesas t returns to the base of the organ.Whole mount immunohistochemistry enablesthree-dimensional visualization of structures previ-ously visible only by using histological methacis in-volving sectioning. The innervation is especiallyclear using this approach, and in general our resultsfully support Harrisons (1925) observations con-cerning the morphology of the organ.Development

The classic study of balanwr development wasconducted on A. maculatum by Harrison (19251,whose results we briefly summarize. Internal, visi-bledevelopmentof the balancer starts at Harrisonsstage (H)33, hen the inner ectodermal layer thick-ens as its cells take on a columnar shape. The bal-ancer first appears exteriorly in H 34 as a faintrounded bulge on the mandibular arch, near thepalatoquadrate cartilage. At H 36, microvilli and afew cilia appearon the surface ofthe now cylindricalbalancer. Cilia are not found on balancers of olderstages (Fox 1985). Secretory cones develop at H38, accompanied by elongation of the balancerthrough change in cell shape. The balancer is fullydeveioped at H40-41 . The balancer is lost by H46.At this point each forelimbpossesses hree digits.By transplanting tissues between individuals ofdifferent ages, Harrison (1925) showed that the life-

time of the balancer is determined intrinsically(donor specific), and that the developmental fate ofthese ectodermal cells is determined perhaps asearlyas the latergastrulastage. Also, transplant ex-periments demonstrated that thedevelopmental ori-gin of the balancer is ectodermal, Le., only ecto-derm is required for the initiation of balancer forma-tion. Further evidence of ectodermal control of bal-ancer deVdOpm8ntwas revealed by heterospecifictransplantatlon experiments between Triturus vul-garis, a species possessing balancers, and Am-bystoma mexicanurn, a species without balancers(Mangold 1931). These experiments showed thattransplanted mandibular arch tissue from late gas-trula or neurula embryos developed th e phenotypeof the donor, but not of the host species. Regenera-tion of balancers is limited or nonexistent (MI11907).

Inter- and intraspecific variation in the develop-ment of balancers scarcely has been investigated.In one of very few comparative studies of balancerdevelopment, Oyama (1929b) found that the bal-ancers of Cynops pyrhogaster become shorterafter hatching, whereas those of Hymbius nebulo-sus continue to grow somewhat after hatching andhavea longer lifetime than those of C, yrrhogasfer.Regarding intraspecific variation, Clarke (1880) ob-served both of the preceding patterns within Am-bystoma maculatum. Clarke claimed that in exam-ining a large number of [A. maculatum] specimens,it is at once seen that there is great variation in theprogress of development. The position of the bal-ancers too, varies considerably indifferent individu-als of the same age (p. 115).Unfortunately, Clarkeprovided no data to back up this provocative andunique claim.Balancers are not always fully developed inspecies that possessthem, norare they always ab-sent in all species said to lack them. Rudimentarybalancers are described as part of the normal de-velopment of certain salamander species, e.g.,Salamandm afm (Wunderer 1910)and chinotritonandersoni (Utsunomiya and Utsunomiya 1977). Inthese species the balancers first appear at roughlythe same stage in ontogeny as functional bal-ancers, but they grow to only a fraction of the ex-pected size and are quickly resorbed. individualswith rudimentary balancers may also m u r nfre-quently in a species that otherwise lacks balancersthroughout development. Y. Misawa (pes.comm.)foundone unpaired yet well-developed balancer outof 1,200 embryos of Hymbius kimurae. Maurer(1888) reported one paired balancer in 60 individu-als of Ambystoma mexicanurn. Rudimentary bal-

Zoology 101 (1998)2 109


4/16

110 fwlogy 101 (1998) 2


5/16

ancers have also been reported in A. tigrinurn.Nicholas(1924a, b) reportedrudimentarybalancersin as many of 56 of 61 individuals of A. trigrinumfrom Wisconsin or North Dakota (exact source lo-cality unknown), whereas Harrison (1925) foundnone inA. tigrinurnfrom Illinois, Kansas, Minnesota,and perhaps North Dakota. Lastly,Twitty (1936) re-ported rudimentary balancers of various sizes in42% of Taricha n'vularis larvae.Function

Five nonexclusive functions of balancers havebeen proposed. The first three involve structuralsupport. First, balancerskeepthe limbless larvaup-right while it rests on the substrate (Egert 1913).Second, balancers also function in preventing thelawa from sinking into the sediment, and concomi-tantly serve to elevate the head and gills above thesubstrate into a zone where the water is cleaner{Clarke 1880). Latta (1919) tested Clarke's hypoth-esis by amputating the balancers of Ambystomarnaculatum. He observed that when approachingthe substrate these larvae "sink into the ooze whichso cover the gills that respiration is necessarily hin-dered.* These first two functionsare the most wide-ly cited (e.g., Duellman and Trueb 1986; Pough eta/.1998). Third, Clarke (1880) pointedout that notonly are the gills elevated by the balancers, but sotoo is the pericardial region. He suggests that this isan adaptation to reduce interference to the heart-beat from the substrate.The fourth proposed function of balancers is ad-hesion. This function may be more important than is

recognized currently. The distal, bulbous ends ofbalancers, as mentioned above, are covered withtiny mucous secreting cones, and Fox's (1985) in-vestigation of balancer ultrastructure lends furtherindirect support to the hypothesized mucous-se-creting and adhesion function of balancers. Harri-son (1925) found that 'when a needle is applied tothis region, it will stick." Bell (1907) observed that"he balancersare made use of at once for holdingto the sides and bottomof the dish and for support-ing the body" (and pers. obsv.). Similarly, Lath(1919) noted that larvae use their balancers tohangon blades of grass, etc. Balancers may also beused to "adhere" from below to the water's surface.Captive Saiamandrella keyerlingii larvae were ob-served "hanging" vertically by their balancers whilefeeding at the water's surface, before diving backdown o the bottomof the aquarium (E. Vorobyeva,pers. comm.). We believe that the adhesion functiongains support from the morphology of the organ andthe presence of secretory cones.Adhesion appears to us to be the one functionthat would be present in hypothesized ancestralrudimentary structures when the organ made itsfirst evolutionary appearance.Lastly, balancerscould function as mechano- orchemo-sensory organs. This idea had previouslybeen dismissed by Maurer (1888) who noted an ab-sence of any differentiated sensorycells in the bal-ancer epitheliaof T~iurusu/garis.Fox 1985), how-ever, has argued for the possibility that balancersindeed have a sensory function. He described thebalancer of Pleurodeles waltl as being extensivelyinnervated by non-myelinated neun'tes. We haveconfirmed the presence of extensive innervationof

Fig. 1. Dorsal view of the head and anterior trunk of a larval salamander, Taricha toma,stained wlth antibody to type11 collagen. This is a 10.0 mm (total length) larva, stage 37 or 38 (Twitty and Bodenstein19#}, collected from BottlePond, Hastings Natural HistoryReservation, MontereyCo., A. Strong stainingof the balancer membrane,especiallybasally, indicates that there is a skeletal component to the relatively rigid balancers. Note the lightly stained otic cap-sules and the even more ightly stained olfactory capsules. Scale bar equals 0.5 mm.Fig. 2. Lateral view of the left side of the head and anterior trunk of a larval salamander,Ambystoma gracile,stainedwith antibody to type IIcollagen.This is a fresh hatchling,15.3mm (total length), from eggs collected from a small pondnear the lower reaches of Wilson Creek, DelNorteCo., A. It sapproximatelystage 43, using the normal table for Am-bystoma maculafum (Harrison 1969). Notethe deeply stained balancer and its connection to the distal end of the sus-pensorlum, just above the articulation with Meckel's cartilage.Note also the weakly stained otic capsule, the develop-ing trabecular cartilages, parts of the hyobranchialapparatus, and elements of the developing forelimb. Scale barequals 0.5 mm.Fig. 3. Lateral view of the right side of the head and anterior trunk of a larval salamander, Taricha tomsa,stained withantibody to alpha-tubulin. Same data as for Fig. 1.Note the fine nerves in the balancers. These nerves are branchesof the ramus mandibularisof the trigeminalnerve.Outlinesof th e developingbrainand the fourth ventrlcle are alsovls-ible. Scale bar equals0.5 mm.Fig. 4. Lateral view of the lefl slde of the head of a larval salamander, Ambystoma grach, stained with antibody toalpha tubulin to reveal peripheralnerves. Same data asRg. 2, except this approximately stage 42 (Harrison1969) andis 16.2 mm (total length).Note the fine nerve supplying the balancer,which forms a loop. This nerve is a branch of theramus mandibularisof the trigeminal nerve. The eye is well developed inthis individual, as are allof the cranialand pa-ripheral nerves.Patchesof epithelial cilia arealso stained. Scale bar equals0.5 mm.

Zoology 101 (1998)2 111


6/16

the balancer (see Figs. 3 and 4). Whether all theabove functions are exhibited, or are adaptive in alllarvae that possess balancers, is not known.Ontogenetic Fate

Balancers are ephemeraf organs; they alwaysare lost well before metamorphosis. Three modesof balancer disappearance have been recorded: 1)they may graduallyshrivel and be resorbed (autoly-sis}; 2) they may simply break off due to injury orstructural weakness; and 3) they may breakoff dueto an inward, constrictinggrowth of ectoderm at thebase (autotomy), with or without the formation ofwhat has been described as an internal epithelialplug. Harrison (1925) observed an external con-striction at the base of the balancer concomitantwith the formation of a so-calledepithelial plug, pre-ceding autotomy of the balancer in Ambystomamaculatum. Anderson and Kollros (1962) showedthat constriction without plug formulation precedesbalancer loss in A. jeffersunianum. Accordingly, bal-ancers may be lost by a combinationof modes.Oyama (1929b) omparedthe loss of balancersin Cynopspyrrhogasferand Hynobius nebuiosus. Inthe former species, balancer loss occurs via autoly-sis, involving the inward growth of the epithelial plugthat loosely constricts blood flow to the balanmrand promotes internal disintegration and resorption.In the latter species, however, balancer loss is viaautotomy, in which the epithelial plug grows tightlyinto the base of the balancer, cutting off blood flowcompletely, and causing the balancer to decay andfall off. Hynobius nigrescens (Maruyama 1928) 0 x -hibits a mode of balancer loss similar to that of H.nebulosus. The balancers of Notophthalmus viri-descens (Bell 1907), Ambystoma maculatum, andA. opacum (Lam 1919) are reportedly last by auto-tomy. Liebekind (1937) found that autolysiswas themode of balancer loss in fleumdeles waiti, Trituruscristatus, and T: wuigaris.A superficial view of autol-ysis is illustrated for T: helveticus in Gallien andBidaud (1959). Fox (1985) described what may besmall primary lysosomes involved in balancer auto-lysis in F! waltl.Of the eleven species summarized above, onefinds three cases of balancer loss by autotomy inambystomatids, tw o cases of autotomy in hynobi-ids, and in salamandrids one case of autotomy andfive cases of autolysis. The occurrence of differentmodes of balancer loss across species may exhibita phylogenetic pattern, as suggested by Salthe andMecham (19741, but more data are needed. Fur-thermore, different modes may occur within a singlespecies. Autolysis necessarily involves a shrinkingand weakening of the balancer, thus increasing thelikelihood that the balancerwill simplybreakofl (Fox

1985). Perhaps one mode serves as a backup foranother. For example, when Kollros (1940) experi-mentally increased the life span of the balancersofAmbystoma maculafumbeyond the normal periodof autotomy, these organs were lost via resorption.Clarke (18801, who also described a constrictionand concomitant reductionof blood low at the baseof the balancers of A. maculatum,observeda larvaactively breaking off itssenescing balancers.Ecologlcal Oistribution

Salamander larvae have been categorizedbased upon morphology and life history as pond,stream, and mountain-brook types (Valentine andDennis 1964; see also Ouellman and Tnreb 1986;Salthe 1969;and references therein). In an archety-pal pond-type species, newly hatched larvae pos-sessa pair of balancers, long gill rami and fimbriae,and limbs still in early stages of development. Theanterior limit of the caudal fin in these larvae isgreater, and the caudal musculature is less devel-oped, than in stream- and mountain brook-type lar-vae. Embryonic and larval growth exhibit an exag-gerated anterior to posterior gradient as comparedto stream-type and mountain-brook-type lawae.Older pond-type lanrae also typically possesslonger toes and a lateral line system. Inaddition tothese morphologicaldifferences, pond-typespeciesalso exhibit differences in life history mode: totalclutch sizes are larger and eggs are often laid inclusters that are exposed, ova are smaller and pos-sess melanin in the animal hemisphere, hatching isprecocial, and parental care of clutches is not ob-served. Streamand mountain brook larvae lack bal-ancers, have short gills, have a fin restricted to thetail, and hatch with well developed limbs.While balancers generally occur inthose specieswhose lanrae lack functional forelimbs upon hatch-ing, it may be more accurate to say that balancersare found in those species whose larvae beginfeedingbeforeor concomitant with the developmentof functional limbs. Some caudate species com-pletely lack functional appendages upon hatching,having no balancers or adequately developedlimbs. Among these species, there may exist a pre-viously unreported correlation between the absenceof balancers and thepresence of a well provisionedyolk supply. A key difference between these hatch-lings and the typical pond-type larval life history, isthat upon hatching these species are still utilizingtheir maternally provisionedyolk supply and are notrequired to obtain their own energy. This suggeststhat the occurrenceof balancersmay be more tight-ly correlated with the state of limb developmentnot at the time of hatching, but at the time of firstfeeding.

1I2 Z#lOgY lOl( l998) 2


7/16

Examplesof this novel lawal life history stmtwyinclude the following. Within Hynobiidae, Hynobiusformosanus, H. kimurae, H. naevius,and H. onmiall lack balancers and appear well provisionedwithyolk (Oyama 1929a; Kakegawa et al. 1989;Matsui1993).Size of gills, anterior extent of the tail fin, andthe steep anteroposterior gradient of developmentin these species seem intermediate between pondand stream types. The larvae hatch out withoutfunctional limbs or balancers, but they still possessa large amount of yolk, The yolk supply remainsuntil the forelimbs have developed into functionalappendagesand the lanrae begins feeding.Such observationsare not limited to the hynobi-ids. Among the samalandrids, chinotritonhatch-lings are well provisioned with yolk (Utsunomiyaand Utsunomiya 1977, plate 11). Chioglossa lusita-nica hatchlings also lack a balancer and limbs andseem well endowed with yolk (Figs. 5 and 6 inGonqalves 1962). The same may be true of an un-described species of Pachfldtun whose develop-ment is reported by Thiesmeier and Hornberg(1997). Further evidence of this correlationis foundin species of the plethodontid tribe Hemidactyliini.The somewhat pond-type hatchlings of Stere-ochilus are apparently without functioning limbs,and "when resting the lawae lie on the bottom ontheir sides," and "ithe mouth is apparently unopenedat hatching, and much yolk material is present"(Schwartzand Ethsridge 1954).Homology

The observation that balancers develop from ec-toderm that is associated with th e mandibular arch(based on patterns of innervation) suggests that,contra Noble (1931; p. 231, they are not evolution-ary homologues of the adhesive organs of Anura(Lieberkind 1937), which develop from the hyoidarch. For the same reason, balancers cannot beconsidered serial homologues of the gills (Latta1919). Thus, balancers are a novelty of Caudata,though not all species have them. The question ofwhether balancers arose singly, and are all homo-logues, or multiply, representingone or more paral-lelism within Caudata, is discussed below.Phylogenetic Oistribution

In addition to the ecological component of thedistribution of balancers here is a historicalcompo-nent as evidenced by the phylogenetically non-ran-dom distribution of pond-type larvae without bal-ancers. Herewe present the first phylogeneticanal-ysis of the distribution of balancers and evaluatetheir evolutionary history in light of what is known

about the ecology of each species. We concludethat the evolution of balancers has been non-parsi-monious and has involved extensive, unidirectionalhomoplastic loss.There are two limitations to our phylogeneticanalyses.First, a phylogenetic analysisof characterevolution is only as reliableas the phylogenetic hy-potheses on which it is based. While significant ad-vances have been made recently to our under-standing of the phylogeny of salamanders at di-verse taxonomic levels (e.g., Shaffer et al. 1991;Larson and Dimmick 1993; Titus and Larson 1995,1996; Jackman et al,1997),further work is neededon several taxa, e.g., the Hynobiidae, Plethodonti-dae, and certainsalamandrids. Second, knowledgeof balancer presence or absence is not recordedformany species. Because balancers begin develop-ment in the embryo and are quickly lost, their pres-ence or absence is often not mentioned in speciesdescriptions and natural history accounts. The Ap-pendix providesa compilationof data from the liter-ature (and a few previously unpublished observa-tions) concerning the presence or absence of bal-ancers in members of the three families in whichthey occur. The possibility exists that the larvae of aspecies may lack balancers, while the embryo maypossess rudimentary balancers that are resorbedbefore hatching or parturition.Balancers are known from three caudate families:Hynobiidae, Salamandridae, and Arnbystomatidae(Peters 1964; see Fig. 5). Balancers are absent inthe phylogenetically basal Sirenidae, yet larvae ofthis family typically display a pond-type life historyand adults retain larval morphology and live only inpond habitats. Hatchlingsof Pseudobranchusaxan-thus have extensive dorsal and ventral fins al-though, unlike most pond-type species, they alreadypossess three digits (Goin 1947). The newlyhatched lawa of Siren lacertina exemplifiesa pond-type morphology (Goin 1947),except for its lack of abalancer. The same holds true for S. infermedia,whose larva newer shows even a vestigial balancer(Noble and Marshall 1932). As Noble (footnote inHarrison 1925) pointed out, 'This little fellow was insad need of a balancer for h e could lie only on hisside." Noble recognized but could not account fo rthis notableexception to this functional or phyloge-netic hypotheses of the taxonomic distribution of bal-ancers. The absenceof a balancer in the pond-typel a m e of Sirenidae, as in the case of pond-typeplethodontids, suggests that this organ does not fol-low a strictly ecological or functional distribution.Strearn-type arvae are found throughout the fam-ilies Cryptobranchidae, Dicamptodontidae (Noble1925,19271, and Rhyacotritonidae(Good ndWake1992).The absence of balancers in these taxa is ex-pected, concomitant with the absence of pond-typelarvae, and no balancers have been recorded in the

ZmlQgy101 (1998)2 113


8/16

AmphlummePkthodonttdas

b D h m p l o d o n t bFig. 5. This phylogenetic hypothesisof family level rela-tionships in the order Caudata is derived from the strictconsensus tree of Carson and Dimmick (1993), inwhichthe relationshipsof Rhyacotritonidaeare unresolved. Themurrence of balancerswithin families s shown in black.Hatched lines indicate that the ancestral character statein this lineage remains unresolvedunder the criterion ofparsimonious evolution (Le., minimizingthe total numberof character state changes). The balancer or its homo-logue is unknown outside this clade. References aregiven inAppendix.

literature. Species or Amphiuma pnphiumidae) laytheir eggs on land, larvae hatch at an advancedstate of development, completewith functional limbs(Hay 1888; Weber 1944), and balancers ar e notknown within this group (G. K. Noble, footnote inHarrison1925). Although he did not publish a devel-opmental series, Latta (1919) reported having foundno balancer among several series of sections of em-bryos and larvae of Am phi m a sp. The existence ofrudimentary balancers in this family should not beruled out until complete developmental sequencesare available for at1 species.Balancers are unknown within Proteidae. Com-plete developmental sequences are available forNecturus maculosus (Eycleshymer and Wilson1910) an d Proteus anguinus (Briegleb 1962; Du-rand 1971). However, the existence of rudimentarybalancers within other species of Nectunrs shouldnot be ruled out until complete developmental se-quences are available. N. maculusus usuallybreeds in streams but often breeds in lakes as well(Bishop 1941). This species displays an intermedi-ate larval morphology (Bishop 1941, fig. 7; Smith1911, fig. 2).Relative to a strearn-type lanral mor-phology, the hatchlingof this species displays rather

poorly developed limbs but well developed gills.Similar to a strearn-type larva,however, the anteri-odorsal limit of the caudal fin stops at the pelvis.Relative to N. rnacuiosus, the hatching of Proteusappears more pond-type with its poorly devetopedlimbsand anteriorly extensive dorsal fin, but has re-duced gills (Briegleb 1962; Dorand1971).The plethodontids include a small number ofspecies possessing pond-type larvae,yet balancersare unknown within this group. We have been un-able to substantiate the claim of rudimentary bal-ancers in plethodontids (Duellmanand Trueb 1986,p. 758). Plethodontids with pond-type larvae arelimited to the tribe Hemidactyliiniand include Hemi-dactylhim scutafum (Noble 1931, p. 59), Stereo-chilus marghatus (Schwartz and Etheridge 1954;see above) and Eurycea guadridigjitata (DuellmanandTrueb 1986).Larvae of Hernidactylium show anumber of pond-type characteristics, such as largegills and a strong anterior-posterior gradient of de-velopment, but they hatch with forelimbs fullydevel-oped (Bishop 1919). Another plethodontid knowntohatch before the developmental of functional ap-pendages is Pseudotriton montanus floridanus. Incaptivity these hatchlings rest usually on theirbacks or sides (Goin 1947), but otherwise conformto a stream-type morphology.Noble (1925) interpreted the absence of bal-ancers in plethodontids by invoking evolutionaryhistory. He pointed out that balancers are lacking inany pond-type species which have been derivedfrom terrestrial or mountain-brook forms. Pond-typecharacteristics of Hemidactyljumand Eurycea,suchas increased clutch size and enlarged caudal finand gill rami, represent a derived condition withinPfethodontidae (Noble 1931), or a reversal withinCaudata (Dunn 1923a; Larson and Dimmick 1993,if oneassumes that sirenids represent the ancestralcondition).Caudata

Mapping the occurrenceof balancerson the fam-ily level caudate phylogeny motivates several hy-potheses of their evotution {Fig. 5). In light of sever-al phylogenetic hypotheses of the relationships ofsalamander families in which sirenids are the basalextant taxon of Caudata (Duellmanand Trueb 1986;Larson and Dimmick 1993),coupled with the obser-vation that sirenids conform to an otherwisearchetypal pond-type morphology, we interpret th eprobable absence of batancers in sirenids (and alloutgroup taxa) as evidence that these organsevolved after the split between Sirenidae and thecommon ancestor of all other extant caudates. It isof course possible that sirenids lost balancers,since we argue that loss has been common (seeI 4 Zoology 101 (1W8)


9/16

below), but given that sirenids use the only habitatin which balancers appear to function, we considerloss to be unlikely.There are two equally parsimonious hypotheses(he, inimizing the amount of evolution necessary)concerning the evolution of balancers: they evolvedthree times (ancestors of the hynobiids, the sala-mandrids, and the ambystomatids), or twice (oncein the ancestor of Hynobiidae and once in a com-mon ancestor of Salamandridae, Dicamptodonti-dae, and Ambystomatidae) with subsequent loss inthe dicamptodontidancestor. It is difficult to evalu-ate which of these tw o hypotheses is more likely,unless one argues that losses should be more like-ly than gains, in which case the latter scenariowould be favored.We argue, however, that the evolution of balancerorgans involved more homoplasy than a prior as-sumptions of parsimonious character evolution sug-gest. Balancers are complex structures and are aunique morphological feature without homologue out-side of Caudata. These observations lead us to be-lieve that balancers did not originate more than oncewithin the lineage, but more likely evolved one timean d have experienced numerous secondary losses.Thus, we hypothesize that the evolution of balancerorgans at the family level consisted of a single originand from four to six subsequent losses. Even if the re-lationships among the taxa of less well-supportednodes were altered to minimize homoplasy in bal-ancer evolution, given a single origin of this characterone would have to hypothesize an independent lossoccurring in the ancestor of each of the taxa Crypto-branchidae, Dicamptodontidae, andProteidae.Hynobiidae

The most basal taxon that possess balancers isthe eurasian family, Hynobiidae (Fig. 5), but not allhynobiids have balancers. The ancestral conditionfor the family is ambiguous (Fig. 6), but we hypoth-esize that balancerswere present. Noble (1927) be-lieved Hynobiidae to be the most basal of all cau-dates, whose common ancestor was a pond-typespecies, and from this stock the stream-breedingtaxa evolved secondarily. Subsequentdata suggestthat sirenids are the basalcaudates (Fig. 5), and ifthis family is assumed to represent the ancestralcaudate condition, then the hynobiid ancestor wasstill most likely a pond-type species, like that ofsirenids, but with the addition of a balancer.If the presence of balancers is plesiomorphic forHynobiidae, then one might weight character stategains over losses and hypothesize that the loss ofbalancers across genera represents independentevolutionary events, likely correlatedwith the adap-tation of each lineage to a stream or mountain-

Flg. 6. This phylogeny of hynobiid salamanders repre-sents the hypothesis of generic level relationships byfhao et aL (19881, with the addition of those species forwhich data on the presence or absence of balancers areavallable. lntrageneric relationships are unresolved.Speciesknown to possess a balancer are representedinblack. Data are unavailable for Liua. References aregiven nAppendix.

brook habitat, as Noble (1927) suggested. Indeed,Ranodon and Onychodacfylus are mountain-brookforms (Dunn 1923b),as are most species of Batfa-chuperus (Liu 1950).Batrachuperus muslersi pro-vides an interesting exception that Reilly (1983)classified as an intermediate form between moun-tain-brook and stream types. This species has bal-ancers at hatching, but they are soon ost (Nawabi1965). Its tail fin is shortened relative to its con-geners, but the gills are larger, clutch size larger,ova smaller, and larvae tend to live in pools (Reilly4 983).Alternatively, one could make no assumptions ofthe outgroup character state, inwhich case the bal-ancer evolved independently at least three times,once in each genus in which it is found. Evidenceagainst this hypothesis is the observations (Y. Mis-awa, pers. comm.) of a single,unpairedbalanceronone in 1,200 embryos of Hynobius kirnurae fromKyoto, Japan. We interpretthis anomalousbalancer

Zoology 101 (lg98) 2 115


10/16

Curiously, two species within this balancerlessclade breed and lay eggs in streams. Ambystomamsaceum breeds in slow-moving mountainstreams, but, like Saiamandrina, shows no larvaladaptations to this habitat (Anderson and Webb1978). Ambystoma ordinarium, on the other hand,breeds in fast-moving mountain brooks, and, likeTaricha rivularis, has evolved a lawal morphologyintermediate between pond- and stream-types (An-derson and Worthington 1971). Hatchlings of thisspecies have smaller gills and a fully developed tailfin that regresses quickly to a stream condition.Ovum size is small as in pond-type species, butclutch size is reduced and limb development is ac-celeratedsimilar to a streamtype species. Inaccor-dance to the phylogenetic hypothesis above, how-ever, the lack of balancers in these two ambystom-atid species should not be regardedas an adapta-tion to the stream habitat, contra Anderson andWorthington (1971), but as the ancestral conditionof this lineage.Ambystomabatbourioffers a second example ofa species that has colonized running water but re-tains a balancer (J. Petranka, pers. comm.). Differ-ences between the intermediate stream-type A,barbouri and its sister taxon, the pond-type Am-bystoma fexanum, include the following. A. barbourilays its eggs singly on the undersides of rocks instreams, clutch size is two imes smaller, eggs aretwo o three times larger, and larvae hatch roughlytwo Harrison stages later than those of A. fexanum(Kraus and Petranka 1989; Petranka 1982). Phylo-genetic considerations allow the assignment of aYypical Ambystoma pond-type morphology andlife-history to the ancestor of A. &houri (Fig. 8).Therefore, in the evolution of A. bahouri from itsmore pond-type ancestor, numerous life historycharacters have approached stream-type stateswhile the balancerpersists. This phylogenetic iner-tia of the balancer supports our contention, basedon morphological considerations, that these organsare not phenotypically plastic.Overview and Directions for FurtherResearch

Simple obsenration of the phylogenetic distribu-tion of balancers suggests that these organs haveundeQon8 numerous independent acquisitions(Figs. 5 and 6) and losses (Figs. 7 and 8). Weargue, however, that the most likely evolutionaryhistory of the balancer is less parsimonious. Bycombining morphological and ecological data withthe phylogenetic data, we conclude that the bal-ancer is a synapomorphy of thederived salamanderfamilies, Le. this organ evolved only once, subse-quent to the split of the sirenid lineage from the an-118 Zoology 101 (1998) 2

cestor of all other extant caudates (Fig. 5),and thatall subsequent character state evolution has in-volved only the repeated, homoplasious loss of thebalancer.While it is difficult to quantify morphologicalcom-plexity, we believe that the balancer, and its compo-nent genetic and biochemical machinery, are suffi-ciently complex to make unlikely the possibility thanthese organs evolved more than once. The ecolog-ical distribution of balancers is widely consistentwith our hypothesis of a single origin with severalsubsequent instancesof loss. We note that th e re-duction or absenceof balancers in any derivednon-sirenid)taxon is oorrelatedwith the evolutionofa derived larval life history mode of one kind or an-other, with the singular exception of the tiger sala-manders. This observation holds both within andbetween families. Phylogenetic arguments favor apond-type life history mode as being ancestral forboth Salamandridae and Ambystomatidae. The ev-idence is less strong for Hynobiidae, but any hy-pothesis of evolutionary history within this group re-mains speculative pending a rigorous phylogeneticsystematic study of the group. Also needed aremore data on the occurrence of balancers, espe-cially in embryonic development.Critical to any hypothesis of the evolution of bal-ancers is a general understandingof their functionalmorphology and ecologicat role. While no data existon how variation inbalancer morphology may bere-lated to fitness, the organs do keep larvae upright,slightly elevated, and allow them to adhere to sur-faces. However,such obsenrationsdo notaprio# re-veal in what functional and ecologicalcontexts theseorgans evolved, are adaptive, or are lost. The exis-tence of six taxa (Sirenidae, Hemidactyliini, Hyno-bius, Chiogiossa, Echinotriton, and Ambystoma) inwhich hatchlingsmay lackbotha balancerand func-tional limbs raises questions concerning thO sup-posed functional consequences to larval sunrival ofhaving no anterior appendages. Clearly, a caudatelarva without appendages may survive outside theovum. Of these six taxa, however, only one, the tigersalamanders (A. tigdnum clade), has an ancestorhypothesized both to have balancers and to lackfunctional limbs at the time of first feeding. The pres-ence of balancers, therefore, may be more tightlycorrelated with the extent of larval development atfeeding rather than at hatching. If this correlationhasbiological reality, the evolutionary significanceof bal-ancers might be found by looking at their functionalrole while feeding (recall the above obsewations byE.Vorobyeva), notwhite resting.This study of balancer evolution raises the largerissue of the evolution of pond and stream type mor-phologiesand naturalhistories,as well as the moregeneral question of whether the pondlstrearn di-chotomy is, in fact, valid (ag., the unusual case of


11/16

Salamandrina terdigiata). With the production ofmore complete and robust phylogenetic hypothesesfor groups of caudates, we are in a better position toaddressquestions concerning the roles of ecologyand phylogeny in promoting or constrainingthe evo-lution of complex and integrated traits (i.e., clutchsite, ovum size,balancers, gills, timing of develop-menta! events along anterior-posterior axis,parental care, etc.). A detailed, comparative analy-sisof whole suites of traits involvedin th e evolution-ary transition between so called pond an d streamtype life histories would reveal the order of charac-ter state changes and the nature and strength ofcorrelations among traits. Such analyses may sug-gest possible mechanisms of evolutionary change.To address such questions, onemight compare theevolutionary transitions found in such pairs of taxaas Ambystoma barboun'and Taricha rivularis, vari-ous genera of salamandrids, or salamandrids andhynobiids. One might also compare the evolution ofpond type traits in the plethodontids, Eurycea,Hemidactylium, and Stereochilus.Forexample, one hypothesis for the evolution ofbalancers might be that salamander populationsmay experience an increase in fecundity selection,leading to an ncrease in clutch size, which indirect-ly muses a decrease in ovum size (Salthe 1969).Smaller ova would necessitate greater nutritionalself-sufficiency on the part of hatchlings and requirethem to hatch and feed at an earlier, limblessstageof development. The imited maternal energetic in-vestment given to pond type larvae also causes anincrease in the anterior-posterior gradient of devel-opment (Schrnalhausen 1925). Thus, an ancestrallineage of pond breeding salamanders that hap-pened o evolve an energetically economicalway of

making forelimbs mightbeataselective advantage.We hypothesize that sucha complex and serendip-itous event never happened in the ancestor ofsirenids, but did happen in the lineage leading to allother caudates.Our challenge to this hypothesis is the fact thatlimbs are not a functional requirement for pond liv-ing, as evidencedby, for example, Pseudobranchus(Noble, in Harrison 1925), Stersochilus (Schwartzand Etheridge 1954), Hynobius naevius (Oyama1929a), andAmbystomacaliforniense Twitty 1941).Instreams, limblessnessmay not bean option for a

feeding larva, although non-feeding limbless lamemay be found in streams (Kakegawa ef a/. 1989).Therefore, a well-supplied ovum is perhaps a con-straint on stream or mountain-brook breedingspecies,whose larvae cannot achieve self-sufficien-cy until at least the forelimbs are functional. Certain-ly, more data are needed concerning the apparentcorrelation between hatchling yolk supply, forelimbdevelopment,and the time of first feeding.Finally, any phylogenetic hypothesis for the evo-lutionary historyof balancers in the salamander lin-eage reveals a great deal of homoplasy, a phe-nomenon that is rampant in this group (Wake 1991).We believe that homoplasy is all in the same direc-tion - oss of balancers. This argument ts based onthe fact that balancers arecomplex organs, and thatthe exact combination of factors making up theorgan would bedifficult to duplicate. How ikely is itthat a relatively complex organ will evolve identicalstructure independently within a single phylogeneticlineage?There are some remarkable similarities infunction between the eye of some mollusks andthose of vertebrates, but many'4undamentaldiffer-ences suggest that they are analogues, asphyloge-netic analysis reveals. In the present case we areunable to find any significant differences n structureamong the taxa having the organ, and accordinglyconclude that homoplastic loss of balancers hasbeen extensive.Acknowledgments

We thank J. Bogart, M. Garcia-Paris, M. Matsui,Y. Misawa, J. Petranka, 8. Thiesmeier, and E.Vorobyeva for sharing their expertise. A. Glue-senkamp and M. Strornberg provided some speci-mens for our study. A. J. C. would like to thank theU.S. Dept. of Education for the GAANN EcologyTraining Grant to the UniversityofChicago for sup-pon during the writingof this manuscript.J. Hankenprovided access to his laboratory, instruction in im-munohistochemical techniques, and photographicassistance. K. Klitz assisted with the illustrations.We thank the following for critical comments on adraft of the manuscript: M.Garcia-Paris, C. Had-dad, M. Mahoney, S. Deban, R. Bello,S. Rollmann,and two anonymous reviewers.

Zoology 101 (1908)2 119


12/16

I I , 'ppendW" ' I, L, L* II1 4 - 0 'A: j f , r m t , T ' L , J ~ ~ ;? irSummaryof the taxonomic distributionof balancers within the threecaudate families known to possess them. Within families, speciesare groupedby presence of standardbalancers, rudimentary balancers, and absence of balancers, then arrangedalphabebically.conflicting amounts; + complete developmental sequenoe known for this species without balancers; d paternalhybrid parent; 0 ma -ternal hybrid parent.

Hy nobiidae4 Balancerspresent in:*BatraChupervsmustersi (Nawabi 1Q65)Hynobius abei (Sato 1943)H. dunni (Sato 1943)H. hidamonfanus (M. Matsui, pers.H. eechii (Sato 1943)H. lichenatus (Sato 1943)*H. aevius (Tag0 1907;contestedbyH. ebulosus (Oyama 1Q29b)H. @res#ns (Murayama 1928; OkadaH. takedei (M . Matsul, pers. comrn.)H. toky#nsis (Sato 1943)H. tsuensis (Sat0 1943)S8/6f??mdrf3llaeyerlingii (Shitkov 1895)

Balancen absent in:Batrachuperus kadschmidti(Liu 1950,5. inchonii (Y. Misawa, pers. cornm.)Hp obi us boulengerl(Sato1934,1943)H. f o m a n u s Kakegawa eta/. 1989)H. kimurae (Matsui 1993)H.naevius (Oyama 1929a;Obara 1984,H. sonani (Kakegawa et al . 1989)onyChod8ctylus fischeri (B. Thiesmeier,+O. epwricus (Iwasawa and Kera 1980)Ranodonsibericus (Snitnikov 1813, cited

comm.)

Oyama 1929a)

1933)

hatchling shown in flg. 13)

figs.2 and 3)

pers. comm.)by Noble in Harrison1925)

SalamrrnclrldaeBalancers present in:CYnqDspyrrhogaster (Oyarna 1929b)Mfophthalmus viridescens (Bell 1907)Pleumd8Ies walfi (Liekrkind 1937)Taricha granulosa (Twitty 1936)T.granulosa 8 x T. rivularis 0 (TwntyT. torosa (Twitty 1936)1936, exslhrfertilization)

T. torosa 8 x T. rivularis 0 (Twitty 1936,Tritum alpestris (Eppetieinand Jungin-T. boscai (Garcla-Paris 19885)T. cristatus (Lieberkind 1937)T. camifex (Fox 1955)T. italicus ( L a n a 1983)T. kare/ini(Fox 1955)T. hehticus (Gallien and Bidaud 1959;van Barnbake 1880)T.marmuratus (Salvador1985)T. vulgaris (Maurer 1888)Tylototriton fslimpmis 8.Thiesmeier,T. vemmsus (Ferrier 1974; ShresthaSalamandrina terdigitata (Lessona 1874)

Rudimentary balancers present in:Echinafritonandersoni(Utsunomiya andSalamandraam (Wunderer 1910)S. salarnandra(Wunderer I910)'Taricha riw/aris Twitty 1936; 42% freq.,size varlable but reducedT. rivularis S x T. t o m 0 (Twitty 1936,size of balancer= O M % f std.T. rivuleris 6 x T. granulosa 0 (Twitty1936, sire = m%413%f std.

BX situfertilization)ger 1982; an Bambeke 1880)

pers. eornrn.)1989)

Utsunomiya1977)

Balancers absent in:Chfoglossa Iusitanica (Gongalves 1962)+Euprochrs ssper Gasser 1964)E. mwrtenLcs (B. Thiesmeier, pers. oomm.)E. pla@cepha/us Alcher 1980)Merfensiell6camsica Schultschik 1994)M. uschani antalyena (bzeti i 79)Neurergussfrauchii(Schrnidtler andPachytriton labiaturn (6.Thiesmeier, pers.Paramesofritondeloustali (Rehlk 1984,

Schrnidtler 1975)oomm.)see figs. 3 and 4)*TwicfH mkgjF..IRi9vaA1%4 I , , , .,

.'td?i ; , , a ' ! : 1 ,> $> L,,>'J

ReferencesAlcher M (1980) Contribution a IYtuae au aeveioppementde r'urodble Euproctus platycephalus (Gravenhorst,

1829).Vie Milieu30: 57-164Anderson E, Kollros JJ (1962) The ultrastructure and de-velopment of balancers in Ambystoma embryos withspecial referenm to the basement membrane.J Ultra-structural Res6: 36-56Anderson JD (1967) A comparison of the life histories ofcoastal and montane populations of Ambystomamacrodacty/um n California. Am Midl Nat 77: 323355Anderson JD, Webb RG (1978)Life history aspects of theMexican salamander Ambystoma rosaceum (Amphib-ia, Urodela,Ambystomatidae). J Herpetol 12: 89-93

AmbystomatidaeBalancerspresent in:Ambystomaannu/aturn(Trapp1959, citedA. barb4uri (J. Petrank, pers. cornm.).A. cingulaium (Palis 1995)A. gracile (Henry and Tw W 1940)A. jeffersonianurn(Harrison 1925; Ander-son and Kollros 1962)A. laterale (J. P. Bogart, pew. comrn.)A. macrodactylurncroceurn (Andenon1967)A. maculahrm (Clarke 1880)A. opacum (Anderson & Kollros 1962)A. tapbideurn (Volpe and Shoop 1963)A. texanum (Brandon f 961

inHutchersonet a 1989)

Rudimentary balancers repofld in:'A. mexicanurn (Maurer 1888, 1 in 60)*A. tigrinurn(Nicholas 1924b, N. Dakota,56 af 61)A. tigrinurn x A. Ieterale hybrid (Bogart eta/. 1987,balancers absent or 1 sideO WA. trig#num x A. Iexanum hybrid (Bogartet a/ . 1987,balancersabsent or 1 sideW)Balancers absent in:A. a n d e m / Krebs and Brandon 1984)A. Oalifomiense (Twitty 1941)A. dumerilii (Brandon 1972)A. mavOrtiurn (Webb and Roueche 1971)+A. mexicanurn (Schreckenberg and Ja-

*A . opacum (Noble 1931, p. 51, *rarelyA. ordinarium (Anderson and WorthingtonA. maceurn (Anderson and Webb 1978)A. tigrinurn (Hartison1925; Bogart et al .

cobson 1975)absenp)1971)

1987)

Anderson JO, Worthington RD (1971) The life history ofthe Mexlcan salamander Ambystoma ordinarilrm Tay-lor. Herpetologica27: 165-1 76Bell ET (1907) On regenerationand transplantationof thebalancers of embryos of Diemycfylus(with a note onthe external gilts). Anat Anz 31: 283-291Bishop SC (1919) Notes on the habits and developmentof the four-toed salamander, Hemidactylium scufatum(Schlegel). New York St Mus Bull 219-220: 251-268Bishop SC (1941) The salamanders of New York. New

Bogart JP, Lowmck LA, Zeyl CW, Mable BK (1987)Genome constitutionand reproductive biology of hybridsalamanders, genus Ambystoma, on Kelleys Island inLake Erie. Can J 200165: 2188-2201

YOrk St MUSBull 324: 1-365

120 Zoology 101 (1998) 2


13/16

Brandon RA (1961) A comparisonof the larvae of fivenortheastern species of Ambystoma (Amphibia,Cau-data). C w i a 1961:377-383Brandon RA (1972) Hybridlzation between the Mexicansalamanders Ambystoma dumerilii and Ambustomarnexfcanumunder labratoryconditions. Herpetologica28: 194-207Briegleb W (1962) Zur Biologie und Okologie des Grot-tenolms (Proteus anguinus Laur. 1768). 2 Morpho16kol Tiere51:19-334Clarke SF 1880) Thedevelopmentof Amblystoma punc-tarurn, Baird, pafl I,external. Stud BiolLabJohns Hopkins Univ11: 105125Duellman WE, Trueb L (1986) Biology of Amphibians.Johns Hopkins Univ Press, BaltimoreDunn ER (1923a) The breeding habits of salamandersand their bearingon phylogeny.Copeia 1923(no. 115):25-28Dum El7 (1923b) The salamandersof the family Hyno-biidae. Proc Am Acad Arts Sci 58: 445-523Durand JP I19711 Recherches sur I'appareil visuel deprotke, Proteus anguinus Laurenti, urodele hypog6.Ann SpMol26:503-624Egert F (1913)Die Kopfanhangeder Amphibien. foolAni! 42: 283-291EppedeinHH,Junginger M (1982) The normal develop-mentof the newt, Trituns a/pstr is. Amphibia-Reptilia2: 295-308EycleshymerAC, WilsonJM (1910) Normal platesof thedevelopmentof Necturusmaculosus. In: Normentafelnzur Entwicklungsgeschichteder Wiheltiere. Vol. 11 .KeiblF (ed.), GustavFischer,JenaFerrierV (1974) Chronologiedu ddvdoppement de I'am-phibien urd& Tflototdmn vemcosus Anderson(Salamandridae).Ann Embryo! Morph7:407-41 6Fox H (1955) Early development of two subspeciesofthe salamander Trirurus cristafus. Copeia 1955:Fox H (1985) Balancer fine structure of the Pleurodeleslarva. Acta 2001 Stockh.)66:97-1 10Gallien L, Bidaud 0 (1959) Table chronologique duddveloppement chez Tihrrus helvetima Razou-mowsky. BullSoc Biol FranceBelgique91:97-1 14Garcfa-ParisM (1985) Losanflbimde Espaha. Ministeriode Agrlcultura, Pescay Alimentaclbn MadridGasser F (1964) bsetvattons sur les stades initiaux dudeveloppement de I'urodble Pyr6nBen E u p m u sasper. BullSoc 2001France89:4 2 3 4 2 8Goin CJ (1947) Notes on the eggs and early larvae ofthree Floridasalamanders. Nat HlstMisc10: 1-4GonqalvesL (1962) A reproduqao de Chioglossalusitan-ica Bocage. Naturo8 72-74Good DA, Wake DB (1992) Geographic variation andspeciation in he torrent salamandersof the genusRhy-amtriton (Caudata: Rhyacotritonidae). Univ CalH Pub2001126: 1-91GreilA (1 906)Uberdie Homologis der Anamnierkiemen.Anat An2 28: 257-272Griffiths RA (1996) Newts and Salamanders of Europe.Academic Press, SanDiegoHarrison RG 1925) The development of the balancer inAmblystoma, studied by the method of transplantationand in relation to the connective-tissueproblem. J Exp200185:33-52

131-133

Harrison RG (1969) Harrison stages and description ofthe normal development of the spotted salamander,Amblystoma punctatum (Linn.). In: Organization andDevelopment of the Embryo. Harrison RG (ed), YaleUniversity Press, New Haven NJ, pp 44-66Hay OP (1888) Observations on Amphillma and itsyoung. Am Nat22: 315-321HenryWV, Twitty VC (1940) Contributionsto the life his-tories of Dfcamptodon ensatus and Ambystoma gra-cile.Copeia 1940: 247-250HutchersonJE, Peterson CL, Wilkinson RF (1989)Re-productive and larval biology of Ambystoma annula-hrm. J Herpetol23: 181-183lwasawa H,Kera Y (1980) Normalstages of developmentof theJapanese lunglesssalamander,Onychdactyhsjqwnjcus (Houttuyn). Jap J Herp 8: 73-89Jackman TR, Applebaum G,Wake DB (1997) Phyloge-netic relationships of bolitoglossine salamanders: ademonstrationof the effects of combiningmorphologi-cal andmoleculardata sets. Mol BiolEvol 14: 88-91Kakegawa M, lizuka K, Kurumi S (1989) Morphology ofeggsacs and larvae ust afterhatching inHymbius su-nani andH. formasanus from Taiwan, with an analysisof skeletal muscle protein compositions. In: CurrentHerpetology in East Asia. MatsuiM, HikidaT, Goris RC

(eds.), Herp SQCJapan, Kyoto,pp 147-1 55KlymkowskyMW, HankenJ (1991)Whole-mount stainingof Xenopus and other vertebrates. Meth Cell Biol36:Koliros JJ (1940) he disappearanceof the balancer inAmbtptom larvae.J Exp 200185:3+52Kraus F, Petranka JW (1989) A new sibling species ofAmbystoma from the Ohio River drainage. Copeia1989:94-110Krebs SL, Brandon RA (1984) A newspeciesof salaman-der (Ambystomatidae) from Michoacan, Mexico. Her-petologica40: 238-245Lanza6 (1983)Anfibi, rettili, Guideper it riconoscimentodelle specie antmali dell aque interne italiane. No 27Larson A, Dimmick WW (1993) Phylogenetic relation-ships of the salamander families: an analysis of con-gruence among morphological and molecular charac-ters. Herpetol Monogr7 77-93Latta JS (1919) The morphology of the so-called bal-ancers incertainspecies of Amb&&tr?a. Anat Rec 17:63-71Lessona M (1874) ote intorno alla ripmduzions dellaSatamand ha pespiciillata. Atti R AccadSci Torino 1047-55Lieberkind I (1937) Vergleichende StudienLiber die Mor-phologie und Histogeneseder lawalen Haftorganebeiden Amphibien.C. A. Reitzels,CopenhagenLinsenmayer TF, Hendrix MJC (1980) Monoclonal anti-bodies to connective tissue macromolecules; type 11collagen.BiochemBiophysResComm 92: 4 4 0 4 4 6Liu CC (1950) Amphibians of western China. Fieldiana:2001Mem2: 1 4 0 0MacgregorHC, SessionsSK, ArntzenJW (1990) An in-tegratlve analysis of phylogenetic relationshipsamong newts of the genus Tiiturus (family Sala-mandridae), using comparative biochemistry,cytoge-netics and reproductive interactions. J Evol Biol 3:329-373

41 9441

1-196

Zoology to1 (1998) 121


14/16

Mangold0 1931) ersuche zur Analyse der Entrvlcklungdes Haftfadensbei Urodefen; ein Beispiel ur die Induk-tion artfremder Organe. Natumissensehaften 19:905-91 1Matsui T (I 993) Japanese Amphibians and Reptiles(6h4.).hougakukan, TokyoMaurer F (1888) Die Kiemen und ihre GefdBe bei anurenund urodelen Amphibien, und die Urnbildungen der bei-den ersten Arterienbogen bei Teleosteiem. MorphJahrb 14: 175-222Murayama T 1928) Uber die Balanzierstange be i Hyno-biuslarven. Folia Anat Japonica 6: 375-388Nawabi S (1965) in seltener Vertreter der Amphibia inAfghanistan: Satrachyperusmustersi.Science (Kabul),Nicholas JS (1924a) A balancer in larvae of Amblysbmatrigrinum. Am Nat59: 191-192NicholasJS (1924b) The development of the balancer inArnblystoma tigrinurn.Anat Re c 28: 317-329NobleGK (1925) An outlineof the relationof ontogeny tophylogeny within the amphibia, II. Am Mus Novit no.166: 1-10Noble GK (1927) The value of tife history data in the studyof the evolution of the amphibia. Ann New York AcadSci30:31-128Noble GK (1931) The Biotogyof the Amphibia. Dover NYNobleGK, Marshall BC (1932) he validity of Siren infer-mediaLeConte, with observations on its ife history. AmMus Novit no.532:1-17Obara K (1984) Development of the salamander, Hym-bius {Hynobius) naevius naevius (Schlegel). Saishu toShiiku(Cotlecting and Breeding) 48:3"Okada Y (1933) he herpetological auna in the vicinity ofNikk6, Japan. Sci Rep Tokyo Univ Lit Sci Sec B 1:1591 3Oyama J (1929a) Some embryologicat notes of Hymbjusnaevius. Copeia 1929 (no. 173).92-94Oyama J (1929b) Balancer in DiemiCtyuspymhogasferand in Hynobius nebulosus. Copeia 1929 (no. 173):103-1 06Ozeti N (1979) Reproductive biology of the salamanderMertensielh iuschani antalayana. Herpetologica 35:193-1 97Palis JG (1995) Larval growth, development, and meta-morphosis of Ambystoma cinguktum on the GulfCoastal Plain of Fforida. FloridaScientist 58: 352-358Peters JA (1964) Dictionary of Herpetology. Hafner, NewYorkPetranka JW (1982) Geographic variation in the mode ofreproduction and larval characteristicsof the small-mouthed salamander (Ambystoma texanum) in theeastantral United States. Herpetologica38: 375-385Pipemo G, uller MT (1985) Monoclonal ant iwies spe-cific for an acetylated form of alpha-tubulin recognizethe antigen in cilia and flagella from a variety of organ-isms. J Cell Biol 101 20852094Pough FH, Andrews RM, Cadle JE. Crump ML,SavitzkyAH , Wells KD (1998) Herpetology. Prentice Hall, UpperSaddle River,NJR e H k I (1984) A study on Paramesufribnddoustali incaptivity, with descriptionsof the egg, the larga, the ju-venite and the adult (Amphibia: Caudata: Salamandri-dae). Vest Cs Spolec 200148: 11&I31Reilly SM (1983) The biology of the high altitude sala-

August: 21-25

mander Batrachupems mustersi from Afghanistan. JRiemer WJ (1958) Variation and systematic relationshipswithin the salamander genus Tatkh&.Univ Calif Pub1Zoo1 56: 301-390Rollhhser-ter Horst, J (1977) Artificial neural nduction inAmphibia. Anat Embryo1151: 317424Salthe SN (1969) Reproductive modes and the numberand sizes of ova in th e urodeles. Am Midl Nat 81:467-490Salthe SN, Mecham JS (1974) Reprcductive and courtshippatterns. In:Physiology of the Amphibia, Vol. II . Lofts B(ed.), Academic Press, New York, pp 309-521Salvador A (1985) Guia de campo de 10s anfibios y rep-tilesde la Peninsula IMrica, lslas Balearesy Canarias.Author's edition, LebnSat0 I (1934) Studies on salamanders from Shikoku. I . Onthe salamanders found at lshizuchiyama and Its vicini-ty (in Japanese). Dobutugaku Zasshi, fool Magazine,Tokyo no. 553: 464472Sato I (1943) A Monographon the Japanese Urodeles (InJapanese). Japan Press, OsakaSchmalhausen II (1925) Uber die Beeinflussungder Mor-phogenese der ExtrerniBten vom Axolotl durch ver-schiedene FaMoren.W Roux's Arch EntwicklOrg 105:48-97Schmidtler JJ, Schmidtler JF (1975) Untersuchungen anwestpersischen Berbachmolchen der Gattung Neu-m@us (Caudata, Salamandridae). Salamandra 1184-98SchreckenbergGM, Jacobson AG (1975) Normalstagesof development of the axolotl, Ambystoma mexicanurn.Dev Biol 42: 391-400Schultschik G (1994) Mertensjella caucasiCa: Haltung,Nachzucht und Freilandbeobachtungen. Salamandra

Schwartz A, Etheridge R (1954) New and additional her-petological records from the North Carolina coastalplain. Herpetologica I O : 167-1 71Shaffer HB, Clark JM, Kraus F (1991) When moleculesand morphologyclash: a phylogenetic analysis of theNorthAmerican ambystomatid salamanders (Caudata:Ambystomatidae). Syst200140: 84-303ShafferHB, McKnight M (1996) The polytypic species re-visited: genetic differentiation and molecular phyloge-netics of the tiger salamander, Ambystoma tigrinurn(Amphibia: Caudata) complex. Evolution50:417-433Shitkov B (1895)Uberdie Fortpflanzung des fsodactyliumSchrenki Strauch. 2001Anz 18: 165-1 68Shrestha TK (1 989) Ecological aspects of the life-historyof the Himalayan newt, Tylototritonverrucosus (Ander-son)with reference to consewationand management.J Bombay Nat Hist Soc 86:333-338Smith BG (1911) The nests and larvae of Nechrrus. BiolSpallanraniL (1784) Dissertations Relative to the NaturalHistory of Animals and Vegetables, Vol. I t [translatedfrom *Dissertazioni di Fisica Animale e Vegatabile,"17761.J. Murray, LondonSteinfartz S (1995) Zur Fortpflanzungsbiolagie von N e wergus c m t u s und Neurergus sfrauchii barani. Sala-mandra 31: 15-32Tag0 K (1907) Study on Urodela of Japan (ConclUdedJ:2001Mag (Tokyo)19:224-248

Hewto117:1-9

30: 61-173

Bull 20: 191-200

122 Zoology 101(lW8)


15/16

Thiesmeier B, HornbergC (1997) Paarung, Fortpflanzungundbwalentwicklungvon Pachytrfton ep. (PachytritonA) nebst Bemerkungen tu r Taxonomie der Gattung.Salamandra33: 97-1 10Titus TA, Larson A (1995) A molecular phylogeneticper-spective on the evolutlonaty radiation of the salaman-der famlly Salamandridae. Syst Biol44: 125-1 51Titus TA , Larson A (1996) Molecular phylogenetks ofdesmognathine salamanders (Caudata: Plethodonti-dag): a reevaluationof evolution n ecology, life history,and morphology. Syst Biol 45: 451472Twitty VC (1936) Corretated genetic and embryologicalexperiments on Trihms, I and II. J Exp 2001 74:234-302Twitty VC (1941) Data on the life history of Ambystomaf i ~ d n umalibmienseGray. Copeia 1 41 1-4Twitty VC, Bodenstein D (1948) Triturus torosus. In: Ex -perimental Embryology. Rugh R (&.), Burgess, Min-neapolis MN, p. 90UtsunomiyaY,UtsunomiyaT (1977) On he developmentof TybbtritOn admsoni. J Fac Fish Anim Husb Hi-roshima Univ 16.65-76Valentine BD, Dennis DM (1964) A comparison of the gill-arch systemand finsof three genera of larvalsalaman-ders, Rh y amm , Gyrinophilus,and Ambystoma. Co-peia 1964: 1-201

van Bambeke C (1880) Nouvelles recherches sur I'em-bryolcgiedes batraciens. Arch Biol (Paris) 1 305-380Volpe EP, SRoop CR (1963) Diagnosis of larvae of Am-bystoma h/poideum.Copeia 1963:444446Wake DB (1991) Homoplasy: the result of natural selec-tion, or evidence of design limitations. Am Nat 138:543-567Wake DB, 6zeW N (1969) Evolutionary relationships inthe famlly Salamandridae. Copeia 1969: 124-1 37Webb RG, ouecheWC (1971) Life history aspects ofthetiger satarnander (Ambystoma tigrinurn mavufi'um) inthe ChihuahuanDesert. Great Basin Nat 31: 193-21 2Weber JA (1944) Observations on the life history of Am-phiuma means. Copeia 1944.61-62Wilder IW, Dunn ER (1920) The correlationof lungless-ness in salamanders with a mountain brook habitat.Copeia 1920:63-68Wledersheirn R (1895) Salamandtina petspicillata andGeotriton fuscus - Versuch einer vergleichendenAnatomie der Salarnandrinen. Inst. Sordo-Muti, GenuaWunderer H (1910) Die Entwicklung der AuOern Korper-form des Alpensalamanders (Salamandra atra Law.).2001J a hh Abt Anat Ontog Tiere 29: 393-414Zhao E, Jiang Y,HuQ, Yang Y (1988) Studies on Chi-nese Salamanders. Swiety for the Study of Amphi-bians and Reptiles, Oxford, OH


16/16

Documents

Phylogenetic and Evolutionary Perspectives on an Enigmatic Organ