REVISION OFTHE AFRICAN CAEClLlAN GENUS SCHISTOMETOPUM PARKER (AMPH IBIA: CYMNOPHIONA: CAECILI IDAE)
BY RONALD A. NU
AND
MICHAEL E. PFRENDER
MISCELLANEC JS PUBLICATIONS MUSEUM OF ZOOLOGY, UNIVERSITY OF MICHIGAN, NO. 18Fb; '
Ann Arbor, September 2 7, 1 998 ISSN 076-8405 MIS(:ELIANEOUS PUBLICATIONS MUSEUM OF ZOOLOGY, LJNTVERSITY OF MICHIGAN NO. 187
The publicatioils of the M~~sclunof Zoology, The [Jniversity of Michigan, consist PI-irnarilyof two series-the Occasion:~lPapers allti the Miscellaneous Publicatio~ls.Both series were founded by Dc Bryant Walker, Mr. Rradshaw H. Swales, anti Dr. W.W. Newcornb. Occasionally the Museuni publishes contributiorls outside of these series; begirlnirlg in 1990 these are titled Special Publicatio~lsa~ld arc numbered. All submitted ~n;inl~scriptsreceive external review. The Misccllarieous Publications, which include ~l~ollographicstltdies, papers on field and ~II- seuln techniques, and other contributions 11ot within the scope of the Occasio~lalPapers, are pl~b- lishcd separately. It is not intended that they be grouped into volumes. Each 11r11nberhas a title page and, when necessary, a table of co1itelits. Tllc Occasional Papel-s, publication of which was begun in 1913, servc as a medium Sol- original studies based prirlcipally upon the collections in the Museurn. They are issurtl separately. MThen a sufficient number of pages has hcen printed to niakc a volume, a title pagc, table of contenb, and an index are supplied to libraries and individuals on the mailing list for the series. A cornplete list of publications on Birds, Fishes, Insects, Mammals, Moll~~sks,Rcpdles and Amphib- ians, and other topics is available. Address inquiries to the Directt)r, Muse~unof Zoolohy, The lir~ivcr- sity of Michigan, Ann Arbor, Michigarl 48109-1079.
RECENT PUK1,ICATIONS
Webb, S.A. a~rdR.R. Millen 1998. Zoogonetict~stequila, a new goodeid fish ((:yprinodontiformes) fi-om the Arneca drainage of Mexico, ;~nda rediagnosis of the gcnus. Occ. I'ap. 725. 23 pp, 5 figs. $2.00. N~~ssbaum,R.A., C:J. Raxworthy, and 0. Prorik. 1998. The ghost geckos of Madagascar: a furthur revision of the Malagasy leal'toed geckos (Kcptilia, Squamava, Gekkonidac) . Misc. Pl~bl.186. 26 pp., 25 figs., 5 tables. $9.00. Dyer; B.S. 1997. Phylogenetic Revision of Atherinopsinae (Teleostri, Atlieri~lopsidac),with <;om- lllerlts on the Systenlatics of the So11th American Freshwater Fish Genus Ba.~ilich/hysGirard. Misc. P~tbl.185. 64 pp., 46 figs. $15.00. O<:onnol; R.M., R.K. Colwell, and S. NaCeln. 1996. The Flowel- Mites of Trinidad 111: 'l3e Gerll~s Kl~inoseius(Acari: Ascidac). Misc. Pllbl. 184. 82 pp, 70 figs. $10.00. IUuge, A.G., and R.A. Nussbauln. 1995. A Review Of Afi-ican-Madagasca~~Gckkonid Lizard Phylog- eny And Biogeography (Squa~nara).Misc. Publ. 183. 20 pp, I1 figs. $8.50. Kohlf, F:]. & 17.1,. Kookstei~l(eds.). 1990. Proceedings of the Michigan Morplion~etricsWorkshop. Spec. Ptrbl. 2. 380 pp. With software, $25.00. Without soft~rarc,$17.50. Alexande~;R.D. 1990. How did humans evolve? Reflections on the u~liquclylnliqlre species. Sprc. Publ. 1. 88 pp. $4.00. Raxwortlly, C.J. Xc R.A. Nussbaum. 1994. A review of the Madagascan Snake (;enera heudoxyrhoflus, Pamrf~,adinara,and Heteroliodon (Squamata: C:olubridae). Misc. Publ. 182. 37 pp, 25 figs. $1 1.50. Gosline, W.A. 1993. A sur-vey of' upperjaw nlrlsculature in higher- teleostean fislies. Occ. Pap. 721. 26 pp, 9 figs. $2.20. Duellman, W.E. &J.A. Campbell. 1992. Hylid frogs of'tlle genus Woclro/iyln: Systr~naticsand phyloge- netic relationsl~ips.Misc. Publ. 181. 88 pp, 21 figs. $9.10. McKitrick, M.C. 1991. Phylogenetic analysis of avian hindlimb musculature. Misc. Publ. 179. 89 pp, 3 figs. $9.60. Kluge, A.G. 1991. Koine snake phylogeny and research cycles. Misc. l'ut~l. 178. 62 pp, 14 figs. $6.00.
THE WGENTS OF THE UNIVERSITY
Laurence B. Deitch, Bloon~fieldHills Daniel D. Homing, (;rand Maven Olivia P. Maynartl, Goodrich Shirley M. McFee, Battle Creek Rebecca Mc<;owan, An11 AI-bor Andrea Fischer Newman, Ann Arbor Philip H. Power, Ann Arbor S. Martin Taylor, Grossc Pointe Farms Lee C. Bolliilger, ex oflicio Revision Of The African Caecilian Genus Schistometopum Parker (Amphibia: Gymnophiona: Caeciliidae)
Ronald A. Nussbaurn
Division of Herpetology Museum of Zoology The University of Michigan Ann Arbor, Michigan 481 09-1079, USA
and Michael E. Pfrender
Department of Biology University of Oregon Eugene, Oregon 97403
MISCELLANEOUS PUBLICATIONS MUSEUM OF ZOOLOGY, UNIVERSITY OF MICHIGAN, NO. 18'7
Ann Arbor, September 27, 1998 ABSTRACT Nus,sbn~u~m,K. A. and M. E. P'frcnder. 1998. Rr~isioizq/' the Ajrican caccilic~ngrizus Schistometopum l'nrlier (AmphiOicl: Cgrr~nophiona:Caeciliidne). Misc. Publ. Mus. Zool. Univ. Michigan 187:l-32, 15/igs., 15 tables, 2 color plates. We revised African caecilians of the cacciliid genus Sclzistornetopum Parker based on univariate and multivariate analyses of morphometric characters and coloration. S. brevzrostrr (Peters) and S. cpl~eleTaylor,arcjunior synonyms of S. thvmense (Barboza du Bocage), and S. garzonhqdti Taylor and Salvador is a junior synonym of Geotrypetes srraphini (A. Durniril). The two valid species oS Schistometopum occur on opposite sides of the African continent: S. grrgorii (Boulengcr) in coastal Kenya and Tanzania, and S. thomense on S5o Tom6 island in the Gulf of Guinea. Two records for S. thomenseon Principe Island and in the Ruwenzori region of central, equatorial MI-ica arc problematic and require verisication. Geographic variation in color and morphometric characters of S. thomense is pronounced and is remarkable considering the small range ofthis species on an island with seemingly uniforrn habitat. The discovery of pronounced, clinal, microgeographic variation in S. thomense and the erroneous taxonomic concl~~sionsderived from it by Taylor is viewed as a strong argument against the trend to elevate "diagnosable" allopatric populations to species status without careful study both in the Geld and in the laboratory. Significant sexual dimorphism in head size is documented for both species oS Schistometopum, with males having the larger heads in both cases. This same pattern of' sexual dimorphism in head size is known for other caecilian species. Sexual dimorphism in head size may result from selection for sexual resource partitioning or for male combat advantage or both. Scarring and fresh bite marks on individuals from both field and laboratory populations suggest that combat may be part of the sociobiology of S. thomense, but the context of this aggression is unknown. S. thom~n,seis abundant in disturbed habitats and does not appear to be threatened with extinction. S. gregoriiwas also abundant in anthropogenic habitats in 1934, but the species has not sincc been reported. Key words: Gymnophiona, Caeciliidae, Schisto.metopum, Systematics, Africa. CONTENTS PAGE
Illti-oduction ...... 1 Acknowledgments ...... 1 Materials and Methods ...... 1 Nomenclatural History ...... 3 Status of Schistometopum thomense and S . ephele ...... 5 Coloration ...... 5 Morphometric and meristic comparisons ...... 6 Taxollomic conclusions ...... 8 Statlis of Schistornetopum brevirostre ...... 8 Status of Schistometopum garzonheydti ...... 10 Status of Schistornetopum gregorii ...... 20 Synopsis ...... 25 Schistometopunc Parkcr ...... 25 Schisto?nelopum gregorii (Boulenger) ...... 25 Schislo~netopumthomense (Bocage) ...... 26 Discussion ...... 29 Literature Cited ...... 31 ILLUSTRATIONS
FIGUFU? PAGE
Cover illustration: skull of Schislometopum gregorii. dorsal view Map of Gulf of Guinea islands ...... 2 Map or Siio Tom6 and Rolas islands ...... 4 Color morphs of Schistorn~topurnthommse on Sfio Tome ...... 6 Intrapopulation color variation in Schistometopum thomenne ...... 7 Female Schistometofium tho.mm.se and newborn ...... 7 Clinal geographic variation in dorsal coloration of Schiston~rtopun~th~m~nse ...... 7 Clinal geographic variation in ventral coloration of Scizislometopurn. LhorrlIensP...... 9 Relationship of head sizc to total length by sex in two populations of Schiston~etopumthonzense ...... 13 Multivariate relationships of populations of Schistometopum thomm.rr ordered geographically ...... 17 Sexual dimorphism in head size of Schistometopum gregorii ...... 19 Canonical variate distribution by sex of Schistometopum ...... 21 Canonical variate distribution of female Schistometof)um ...... 24 Sexual dimorphism in head size of Hypogeophis rostsatu.~...... 28 Southeast coast of Siio Tom6 Island ...... 30 Coastal plateau at Ribeira Peixe. S3o Tome Island ...... 31
TABLES TABLE PAGE
Schi5to.rnetopum sampling localities on Sfio Tome Island ...... 3 Geographic variation in dorsal coloration of Schi.rlometof~umthomensr ...... 6 Geographic variation in ventral coloration of Schi.rtometopunz thomens~...... 8 Geographic variation in mid-ventral coloration of Schistometopum thomm.te...... 10 Geographic variation in size of Schisto?nelopunz thomensr ...... 10 Morphometric and meristic comparison of male and female ScAistometof)um thom~7r.c~from Pousada, Sfio Torn6 ...... 11 Morphometric and meristic colnparison or male and female Schistometof~umthomense from Ribeira Peixe. Siio Tomb: ...... 12 Morphometric and meristic comparison of Pousada and Ribeira Peixe males of Schislorr~etopumthorr~ense ...... 14 Morphometric and meristic comparisoll of Pousada and Ribeira Peixe females of Schistometopum thomense ...... 1.5 Discriminant functions analysis of Schistometopum gregorii and two populations oP S . thommse ...... 16 Discriminant functions analysis of two populations of Schistonzrtopum thomms~by sex ...... 16 Morphometric and meristic co~nparisonsof Schislomelopun~ holotypcs ...... 17 Morphometric and meristic cornparison of males and females of Kenyan sample of Schistometopum gregorii ...... 18 Morphometric and meristic comparison of males of Schi.ctometopum gregorii and S . thornensr ...... 22 Morphometric and meristic comparison of females of Schistomrtopum g.1-egorii and S . thommse ...... 23 INTRODUCTION (MSNG) , Barry Clarke (BM), E. Crespo (MBL) , R. Giinther (ZMB) , H.-W. Koepcke (ZMH), E. Kramer (MHNB), M. Louette (MRAC), The African caecilian genus Schistometop,u*~,with five cur- B. Sanchir (MNCN), F. Tiedemann (NHMW). This research was funded in part by grants to RAN from the Natio~lalScience Foun- rcntly recognized species, has a broadly disjunct distribu- dation, the National Geographic Society, and the Vice President tion,in western and eastern equatorial Africa. Four spe- for Research, University of Michigan. MEP was additionally funded cies (brevirostre, ephele, garzonheydti, thomense) are restricted by a National Science Foundation training grant (BIR 94-13223) to islands in the Gulf of Guinea, and the fifth species in evolutionary genetics to the University of Oregon. John (~regorii)occurs in coastal Kenya and Tanzania (Taylor, Megahan helped with the preparation of the maps. We are greatly 1968). These two centers of distribution are separated by indebted to Mr. Jos6 Luis Xavier Mendes of the Ministkrio da about 3,500 km, which include the vast rain Sorests of the Agricultura e Pescas, Republics Democratica de S2o Tom6 e upper Zaire (Congo) drainage, a region where caecilians Principe for providing clearance to do field work in his country are expected, but none has definitely been found. and for providing export permits. The fo111- western species are recorded from three of the Sour main islands oP the Gulf of Guinea: Bioko (Fernando Po), Principe, Siio Tomk, and on Rolas, a satellite of Siio MATERIALS AND METHODS Tome (Fig. 1). Caccilians have not been reported on Pagalu (Annobon). Schi.rlometopum brevirostre is reported on Siio We visited Sao Tome, Rolas, ancl Principe islands from 1 June through 4 July 1988 in order to determine the microgeographic Tom6 and Rolas; S. ephele on Siio Tome; S. garzonheydti on relationships of Schi.rtometofiumhrcvirostrr, S. efih,ele,and S. thomrnse Bioko; and thomense on Siio Tome/Rolas and Principe S. and to obtain specimens for anatomical, morphometric, and life (Taylor, 1968). The record of the latter species on Principe history studies. Samples werc grouped into ten localities (Table is questionable and requires confirmation. 1, Fig. 2). The single eastern species, Schistometopum gregorii, is We also examined most of the specimens of Schistometopum in known from the region near the mouth of the Tana River the major museums of the World, including the British Museum north of Mombasa, Kenya, and from two localities of Natural History (BM); Musee Royal de 1'Afrique Centrale, (Bagamayo River and Rufigi Kiver) in coastal Tanzania, Tervuren (MRAC); Museo Civico di Storia Naturale di Genova (MSNG); Museo National de Cienias Natnreles, Madrid (MNCN); 500-600 knl south oS the Kenyan locality. Museum of Comparative Zoology, Harvard (MCZ); Museum The distribution of Schisto,rnetopu,rn, as currently consti- d'Histoire Naturelle, Bile (MHNB); Museum of Zoology, Univer- tuted, poses sevcral questions. Why, for example, are the sity of Michigan (UMMZ); Natural History Museum, Wien western species restricted to islands when seemingly good (NHMW); Univcrsidad Lisboa, Museu Bocagc (MBL); University habitat exists on the nearby mainland? What are the of Florida (UF); Zoologisches Museum und Institut fur Spezielle microdistributiorlal (niche) relationships of the three spe- Zoologic, Berlin (ZMB); Zoologishcs Museum, Hamburg (ZMH). cies that are reported to occur together on Siio Tom6 is- Specimens collected on S2o Tom6 were color-coded in the field land? Do the western species form a monophyletic group while alive, and representative individuals were photographed. Specimens were assigned to one of four categories in regard to with rcspect to the single eastern species, and how closely dorsal coloration: 1) yellow; 2) dark yellow, no brown flecks; 3) related are the eastern and western forms? dark yellow, light to moderate brown flecking; 4) dark yellow, 111 addition to these cluestions of relationships, there are heavy brown flecking . The ventral coloration is white with vari- some contradictions and mistakes in the literature that able amounts oC yellow and brown bars. The colored bars tend nced to be addressed. For example, if the original de- to be restricted to individual primary folds and thus are trans- scription of Schistometopum garzonheydti Taylor and Salva- versely disposed. The amount of brown bars on the venter is dor, 1978, is accurate, then the single specimen upon which correlated with the amount of brown dorsal flecking. Ventral the spccies is based cannot possibly be a Schistometopum coloration was scored by selecting 20 contiguous, mid-ventral, pri- (see below). Parker (1941) described and illustrated a mary folds and counting the number of white, yellow, and brown folds in the 20-fold segment. The mid-ventral line was also scored skull or Schistonzetopum indicating that the orbit is in the as 1) continuously white; 2) white interrupted by yellow or brown; squamosal bone. Taylor (1968:678) reproduced Parker's and 3) no white mid-ventral line. illustratioil oS the skull, which clearly shows the orbit in The sex and state of maturity of all specimens were determined the squamosal. In spite of this, Taylor (1968:663) listed by direct examination of the gonads. All measurements are from one of the generic characters oPSchistomelopum as "eye in preserved specimens and are recorded in mm. A ruler was used sockct piercing maxillary bone". to record total length to the nearest 1.0 mm. Dial calipers were In this paper, we review the status ol: the species of used for all other measurements. Body width was measured to Schistometop,iirn, present new information on morphologi- the nearest 1.0 mm, all remaining measurements to the nearest 0.1 mm. The followir~gmeasurements and cou~ltswere recorded. cal variation, and present a revised classification of the Total length; body width, at midbody; head length, from tip of genus. snout to first groove of first collar on side of neck; head width, at level of postel-ior corner of mouth; intel--ocular distance, between medial borders of eyes; inter-narial distance, between medial bor- ACKNOWLEDGMENTS ders of nares; eye-naris distance, anterior margin of eye to poste- rior margin of naris; eye-tentacle distance, anterior corner of eye We thank the followirlg curators for loaning specimens under to midpoint of tentacular aperture; tentacle-naris distance, mid- their carc: P. Albcrch (MCZ), D. Auth (UF), L. Capocaccia point of tentacular aperture to posterior margin of naris; primary GULF OF GUINEA
6" 0 PAGALU So lo0
Figure I. Major islands in thc (;r~ll ol Gu~nca,wcstcrn cquator~alAtrrca: Kolas IS shown as the tlny Nand on the equator,just south of Sao To~nC . annuli, number of prirnary anni~liaStcr sccond collar; vent an- nulus counted in thr dorsal region; primary annuli with scales, nuli, nu~nberof primary annuli intel-I-uptcdby cloaca; secondary number of primary annuli conraining scales. annrlli, number of partial and complete secondary annuli; corn- Teeth wcl-e counted only on borrowed spcci~nensin which the plcte secondary annuli, nu~nbcrof secondary annuli that con- jaws had already been cut. This includecl most ofthe Schistom,elopum plctcly encircle body; vertebrae, total number of vertebl-ar in- ~regorii. 111 addition, teeth wel-e counted on a subsample of S. clrtding thc atl;~scomplex, cou~ltedfrom x-ray plates; scale rows, tl~omrnseto determine whether significant sex~~aldifferences ex- ni~nlberof overlapping rows of scales in a posterior primary an- ist. The premaxillary-maxillary teeth were counted as a unit; Table 1. Sampling localities for Schistomrtopum on S2o Tom6 Island (see Fig. 2) and sample sizes used for multivariate comparisons without tooth count data.
Sample size
Locality elevation (m) males females
Ponta Furada banana plantation Pousada plantations, Casa de Repousa Road to Pousada" between km posts 5 and 6 Pcdroma Pedroma cacau plantation Rio Abade along thc river in cacau Alto Douro I 1 km NW Ribeira Alfo~lso Col6nia Aqoreana plantation 1 km NE Col. Aqo. Ribeira Peixe secondary coastal vegetation Alto Douro I1 hillside 2 km NW Porto Alegre Rolas rocky fields near village
*: not included in multivariate comparisons vomcropalatine teeth were also counted together; and dentary NOMENCLATURAI, HISTORY and splenial teeth were counted. For all four series of teeth, right and left sides werc counted together. Barboza du Bocage (1873) described Siphonops thomensis Morphometric comparisons between scxcs and among on the basis of two specimens donated to the Lisbon Mu- pop~~lationswerc made using a covariance model with means ad- seum by Craveiro Lopes, a naval officer and sometime gov- justccl to body size (total length). The equality of the regression coefficients was assessed by thc significance of the interaction ernor ofAfrican Portugese possessions. Barboza du Bocage between group and the covariate (total length) using type 111 sum did not observe living specimens and, therefore, missed of sqllares. In comparisons with equal regression coefficients the most striking feature of the species, the bright yellow- (slol~cs),standard t-tests with ScheffC allowances for multiple com- orangc coloration (rapidly lost in preservative). He placed parisons were used to test for significant differences between ad- the species in Siphonops based on thc position of the ten- justecl means. Covariance analysis was done with the general lin- tacle, which is close to the eye, and remarked that this is ear ~noclel(GLM) procedure on the statistical package of SAS, the first species of Siphonops found outside of tropical version 6.1 2. America. Peters (1874) established Siphonops brevirostris Multivariate rnorphometric analysis was done using discrimi- based on a single specimen of imprecise locality, "von der nant functions with SPSS for UNIX, release 6.1. All variables were stanclardizcd prior to analysis with Fisher's Z-transformation to Westkuste Africas (Guinea)",that was presented by an ani- equalize the variances. The maximum resolution in all cases was mal dealer. Peters did not compare S. brevirostris to S. achieved by setting the prior probability of inclusion equal to thomensis as he apparently was unaware of Barboza du the frequency of occurrence. In all cases, a stepwisc inclusion of Bocage's very similar species described one year earlier. variables resulted in lower resolution than inclusion of all vari- Likc Barboza du Bocage, Peters did not see the specimen ables. Fonr separate analyses were done. The first was to deter- alive, and he described the color as bluish gray. mine whether males and females could be discriminated Peters (1879) published the first major revision of morpl~omctrically.This analysis was done separately for a popu- caecilians in which he attempted to define genera and ar- lation of' Scl~i,stornetoljurnIhornense (Pousada), a putative popula- tion of S. epl~ele(Ribeira Peixe), and the Kenyan sample of S. gregorii. range them hierarchically on the basis ofseveral morpho- The second analysis was to compare populations by sex of speci- logical characters. His primary division was the presence mens reccntly captured by us on Sao TomC. Thirdly, we did a or absence of dermal scales. This resulted in an artificial , A multivariate comparison by sex of the Pousada and Ribeira Peixe classification in which the scaleless Siphonops and samples from S2o Tom6 ancl the Kenyan sample of S. pegorii. In typhlonectids were placed together in Peters' the latter analysis, we included two specimens of S. gregorii from "Gymnocaeciliae". Because "Siphonops" thomensis and "S'. Tanzania and a single specimen of S. thomense purportedly from brevirostris both have scales, they had to be transferred to the liuwenzori region of central Africa, possibly eastern Zaire a different genus, and Peters placed them together in the (recently Democratic liepublic of Congo). Fourthly, we did an o tropical American genus Dermophis on the basis of having poslrriori attribution of all specimens from Siio Tom6 to either the Pousada (thomense) or Rebeira Peixe (ellhelp) populations based scales, visible eye in a socket, a circular tentacle pit in front on cliscriininate functions calculated from the Pousada and the of and below the eye, a globular tentacle, and two rows of Rebeira l'eixe populations. teeth in the lower jaw. For both the univariate and the multivariatc analyses of west- Peters (1880),having received a specimen of Dermojhis ern Schlslomelopum, we used only those specimens recently cap- thomensis from Barboza du Bocage, decided his Dermophis tured by us on Sao Tomi., as many of the older museum speci- brevirostris was the same, and this synonymy was followed rncns were in poor condition and could not be measured accu- by Boulenger (1882, 1895), Nieden (1913),and Gorham rately. The univariate and multivariate comparisons are based (1962). Gorham inexplicably corrected the type locality on the same samples and measurements. For the East African Schis/omrtopumgregorii, no recent material was available, so we were of D. brevirostris to Rolas Island. Apparently there is no forced to rely on older museum material. justification for this change. 0'20' -
070' -
0 O- ILHEU DAS ROLAS 0 5 -km
690' 6'40'
Figurc 2. Map of Sio Tome alrtl Rolas Islailds; solitl square intlicate locatiolrs ofsarnplc sitcs ('I'ablc 1); thr solitl circle indicates 111e type loc;iliry (Agu;i 1x6) or Scl~~.s/ometr,j~tcrrr/:,f,/rcG Taylor. Giinthcr (1894) listed a specimen from Ngatana, Tana that the tentacle opening is closer to the naris than to the River, Kenya Colony as Dermophis thomensis, but in the same eye ("Entfernung Tentakel zum Auge 3 mm, Tentakel zum year, Boulenger (1894) described a new species, Dermophis Nasenloch 1,2 mm"), a relationship not known for any spe- gregorii, based on the same specimen. Boulenger noted cies of Schistometopum, and generally considered to be of that D. gregorii has a longer, blunter snout and a more generic distinction. anteriorly placed tentacle than D. thomensis. Taylor and Salvador (1978) indicated that Schistometopum Two additional species, Boulengerula denhardti Nieden gnrzonheydti belongs to the subfamily "Dendrophiinae". It (1912) and Bdel1of)lzis unicolor Boettger (1913), from the is not clear whether this is a misspelling of Dermophiinae lower Tana River of Kenya, were synonymized with Dermophis or an attempt to establish a new subfamily. We assume it gregorii by Loveridge (1936). Loveridge based these is a misspelling. synonyinies on new material obtained by him at the type Several lesser but important errors occur in Taylor and localities of the lhree "species". Taylor (1968) has since Salvador (1978). For example, they stated in the diagno- supported Loveridge's conclusions. sis that Schistometopum gnrzonheydti has 36 secondary an- Parker (1941) recognized that Dermophis, in the sense nuli, but two paragraphs below, in the description of the of Werner (1931) and earlier authors, was artificial, and holotype, the number of secondary annuli is given as 26. hc placed some of the included species in the existing genus There is confusion in the literature concerning the spell- Hypogeoj~hisand created a new genus, Sch,istometopum, to ing of both Schistometoj~umbrevirostre and Schistometopum receive Dermophis greg0n.i and D. thomensis. He designated thomense. For examples, Taylor (1965) incorrectly referred D. gregorii as the type species of Schistometopum. Parker to the former as Schistometopum "brevirostris", and Els (1963) distinguished Schistometop,um from Dermophis on the basis used the incorrect spelling Schislometopum "thomensis" for of (I ) the fronlals, which are small and separated mesially the latter. The original names were Siphonops brevirostris in ScAistometoj~u,m(see cover illustration of dorsal view of and Siphonops thomensis (not Siph,onops "brevirostre" and skull of S. gregorii bascd on UMMZ 177011) versus large Siphonops "thornense", as stated by Wake, 1985:631). The with a rnedian suture in Derrnophis; (2) the pseudotemporalis original spellings were correct, because Siphonops (and later muscle, absent in Schisto~metopumand present in Dermophis; Dermophis) is a non-neuter noun. Schistometopum, however, and (3) the "cctoptcrygoids", present in Schistomelopum and is neuter and requires the spellings "breuirostre" and absent in Dermophis. Parker's action restricted Dermophis "thomense". to the Neoptropics and Schistometopum to Africa, with one western (S. thornense) and one eastern (S. gregorii) species. Parker implied that Dermophis and Schistometopu,m are sis- STATUS OF SCHTSTOMETOPUM ?'HOMENNCE ter genera. (BARBOZA DU BOCAGE) AND S. I:'PZTEIIE TAYLOR Taylor (1965, not 1964 as is often cited) described Scl~isto~r~tounephele frorn Sao Tom6, which he distinguished Coloration.-Surveys on SZo Tom6 and Rolas Islands re- from S. thomense from the same island in having a smaller, vealed microgeographic variation in color characteristics morc pointcd head and diflerent coloration. S. ephele was which, according to Taylor (1965), differentiate described (Taylor, 1965:297) as "light yellowish-brown with Schistometol~umthornense (immaculate bright yellow) from dark lilac-brown flecks over dorsum and sides or body; fewer S. ephele (yellow with brown freckles). There is no detect- flecks on vcntcr chin and throat". By contrast, Taylor able sexual dichromatism in any of the populations, there- (1968:679) described S. thornense as "uniformly bright yel- fore the sexes were combined for comparison. Two low". In the same paper, Taylor (1965) resurrected S. populations representing the extremes of color variation bre?~iro.slr%s(sic) from the synonymy of S. thomense. With- werc chosen as standards for comparing the other out having seen a living specimen, Taylor (1965) believed populations. Specimens from Pousada and nearby Road that S. b,revirostre differs in color ("bluish gray") and head to Pousada do not differ in coloration and were combined shape lrom S. thornense. I>oumont (1992:40), citing as "Pousada" for purposes of color comparisons (for "Nussbaum, communication personelle", again listed S. morphometric comparisons, Pousada does not include the 0ravi.rost.re as a synonym of S. thomense, as did Schatti and sample from Road to Pousada). Pousada (population 2) Loumont (I992). has the highest percentage of pure yellow (thomense) indi- Taylor and Salvador (1978) established the most recently viduals, and the population at Ribeira Peixe (8) has the dcscribed species of the group, Schistometof~umgnrzonheydti, most heavily freckled (ephele) individuals. Ribeira Peixe bascd on a single specimen from Bioko Island. The most is about 18 linear km SW along the coast from the type serious of the scveral mistakes in this paper is that, if thc locality (~guaIz6) of S. ephele (Fig. 2). description is accurate, the unique holotype of S. The dorsal coloration of the immaculate yellow individu- garzon,heydti cannot be a Schistometopum. Taylor's (1968) als from Pousada and the heavily freckled specimens from own list oP diagnostic characters for Schistometol,um includes Ribeira Peixe are illustrated in Fig. 3a. Intrapopulation a smooth tongue "without narial plugs" (p. 663). How- variation in the intensity of dorsal brown freckling for ever, Taylor and Salvador (1978:60) wrote that the tongue Ribeira Peixe is shown in Fig. 4. Variation in dorsal col- oSS. gnrzo,nkeydti has narial plugs: "Zunge mit zwei kleinen oration among populations is summarized in Table 2. There NasalzapPen". In additio~~,Taylor and Salvador indicated is obvious clinal geographic variation in dorsal coloration Mrsc. PUBL.Mus.. ZOOL.,UNIV. MICH., NO. 187
Table 2. Dorsal coloration ofSc1~istometof)umfrom populations on S5.o TomC: frequency of individuals in populations with different dorsolateral coloration. Pousada to Alto Douro I1 are geographically ordered castern localities; Ponta Furada is an isolated western locality; Porrsada includes both Pousada and Road to Pousada.
(1) (2) (3) (4) light yellow, dark yellow, dark yellow, dark yellow, no brow11 no brown light heavy freckles heckles freckling freckling
2 Pousada 4 Pcdrorna 5 Rio Abade G Alto Douro I 7 Col6nia Aqoreana 8 Ribcira Peixe 9 Alto Douro I1
1 Porlta Furada between the two extremes as is indicated by comparing distance, and the number of primary annuli with scales. individuals from Rio Abade and Col61lia A~oreana(Fig. Many females at Ribeira Peixc were gravid, which explains 6). Kio Abade is only 1-2 krn N of' ~guaIzC, the type lo- their greater body width. The greater interorbital distances cality of' Schistometopum ephele; Colonia A~oreanais further of males is again correlated with their largcr head size. south, about half way between Pousada and Ribeira Peixe. The sexual difference in primary annuli with scales is re- Ventral coloratiorl shows the same pattern of variation versed in this population compared to Pousada, with males (Table 3) as dorsal coloration. Ventral views of Pousada having more annuli with scales than females. and Ribeira Peixe specimens arc depicted in Fig. 3b, and A comparison of males between Pousada and Kibeira clinal variation is demonstrated in Fig. 7. Peixe (Table 8) reveals small but significant differences The mid-ventral line of Schistometopum thomense may be in measurements associated with the head, with Ribeira white, white broken by brown or yellow, or completely ob- Peixe males having greater distances between structures. scured with brown or yellow. Only the Ribeira Peixc and Kibeira Peixe males also have more vomeropalatine teeth Alto Douro I1 populations have high percentages of indi- than males from Pousada. Pousada males have more pri- viduals with broken or completely obscured mid-ventral mary annuli and vertebrae than males from Ribeira Peixe, white lines (Table 4). a relationship also observed in f'emales (Table 9). This Morphometric and meristic comparisons.-Because all could be a developmental phenomenon associated with measurements and some counts vary with body size, it was higher elevations (Pousada) and lower temperatures, as necessary to use analysis of covariance with total length vertebral number has been shown to be developmentally as the covariate in all univariate comparisons. There is sensitive to temperature in other ectothermic vertebrates no apparent sexual dirnorphism in total length within (Fox, 1948; Fox et al., 1961; Lindsey, 1966; Peabody and populations (Table 5),but sexual dimorphism is evident Brodie, 1975). Females from Ribeira Peixe also seem to in head size (head length and width), and in several other have largcr heads than those from Pousada (Table 9), but measurements associated with the head, in both the the dif'i'erences in head-associated measurements are not Pousada and Kibeira Peixe populations (Tables 6 and 7). as great as for males. Females at Pousada have signifi- Because the regressions in most of these cases are signifi- cantly more primary annuli with scales than those from cantly dif'ferent, neither the means nor adjusted means Ribeira Peixe. can be compared. The regressions Por head length and Because of the signif'icant sexual dimorphism detected head width are shown in Fig. 8. In both populations, males by univariate analysis, discriminant function analysis was have longer and wider heads at all but the smallest body also used to examine sexual dimorphism at the multivariate sizes, and the sexual difference is greater in the Ribeira level. Among Pousada individuals, with all variables ex- Peixe population. In the Pousada population, for those cept tooth counts included, 64.7% of males and 90.9% of characters with equal size regressions, only internarial females were accurately classified to sex a posteriori (Table distance and the number of primary annuli with scales var- 10). The percentages are higher, 94.4 and 100.0%, for ied sexually. Greater internarial distance in males prob- the Ribeira Peixe population, probably reflecting the ably is a manifestation of their larger heads. In the Ribeira stronger sexual differences in head size in this popula- Peixe population, characters with equal size regressions tion. that were sexually different are body width, interorbital The morphometric and meristic characteristics of the
Fig11re 3 (facing page). Dorsal (A) and ventral (B) views of two color types of Schistom~topumth,omen~e from S2o Tomi.. In A, the upper, i~nnlaculateyellow individual is a Cen~alc(small hcad) frorn Pousada, and the lower, fi-eckled individual is a male (large head) from Ribeira Peixe. In 8,the upper individual is from Ribeira Peixe and the lower from Pousada.
Pousada n = 54
Rio Abade n = 43
Colbnia Acoreana
Ribeira Peixe n= 193
Color Category
I"igurc (i. C:lin;rl gcograplric wrriatio~iin tlorsal colol-atior~of Stlti\lornc,/o/~~irr,lILomen\t,. C:olot- c;lrcgov (1) bright yellow, irnn~aculatc;(2) dark ycllow, ilrr~r~ac~rlatc;(3) tbrk yellow, light 10 motlerate brow~lflecking; (4) dark yellow, heavy brown flrcking. The lrislogra~nsare ordered by gcog~.;rphicpl-oxi~r~ity (scc. IGg. 2).
I7igr11-c4 (l':rci~lgIXI~C). Fivc Schi~tornrlo/~rc~~illro~rirnsr Irom liibei~-aPvixe illustra~ingintrapopnlation variation in dot-sal coloration ~\~itliin tlrc /,/)ll~lfcolor type. Fig111.c-5 (Sircillg pirpc). A l'c~~lillei~chit1011ic210/~llnl thomcnsr from Porrsada with her ftrll comple~nentof oMSprilig hot-n within a few ho~rr-sof tlrc time the pl~orograplxwas txken. The adult colol- pattern is present at birth in both the Por~sadaand Ribeil-a Peixe (not shown) populations. Table 3. Ventl-a1 coloration of Schistornetopum SI-omSao Torn6 by population. Means and standard errors of the numbers of annuli colored yello~v,brown, and white along a I-andomlyselected midvesitral scgment of 20 annuli. Ponta Furada is all isolated western locality; Pousada includes both I'ousada and Road to Potrsada.
Ventral Annulus Color
Yellow Brown White
I'olnsllntion 11 Mean SE Mean SE Mean SE
2 Pous;~da 54 8.76 1.22 0.00 0.00 11.24 1.22 4 Pcdroma 34 6.85 1.08 0.00 0.00 13.15 1.08 5 Rio Abade 43 6.98 0.91 0.12 0.07 12.91 0.90 6 Alto Douro I 14 5.93 1.34 1.36 0.63 12.71 1.62 7 Col6nia Aqoreana 74 8.05 0.67 2.54 0.28 9.43 0.79 8 Ribcira Peixe 193 10.03 0.27 9.77 0.27 0.22 0.12 9 Alto Dotu-o 11 26 14.92 0.63 5.00 0.64 0.08 0.08
1 Ponta Furado 23 4.57 0.94 0.09 0.09 15.35 0.96
other populations wcre also examined, but, because no Taxonomic conclusions.-Taylor (1965) distinguished striking differences wcre Sound, and in the interest of space, Schistometopurn ephrle from S. thom,ense on the basis of brown the univariate results are not presented here. Instead, the freckling and smaller and more pointed heads in the multivariate technique of discriminant functions analysis former. We demonstrated clinal geographic variation in was used to compare these populations to the Pousada and coloration between immaculate yellow (thomense) and brown Ribcira Peixe populations, the two populations that are freckled (eplzrle) populations. We also demonstrated clinal maximally difkrent in coloration. Discriminant functions variation in morphology between these populations. Taylor were calculated for the Pousada and Ribeira Peixe (1965, 1968) apparently was unaware of sexual dimorphism populations, and the data from specimens of the remain- in head size and shape in both the thomense and the ephele ing populations were applied to the functions to deter- morphs, as we demonstrated herein. Taylor did not iden- mine wliicli of the two base populatioris they would most tify the sex of the holotype of S. c!phele, but based on his likely bc assigned to based on their overall characteris- comments about the small and pointed head, also indi- tics. cated in his accompanying photographs, we predicted ac- Because of the strong sexllal signal in the m~~ltivariatc curately that it would prove to bc a female. We conclude data, discrinlir~antfunctions analyses for cornpal-irig thcre is no justification for recognizing S. ephele and con- populations were done separately by sex. The results are sider it to be ajunior synonym of S. thomense. su~nmai-izcdin Table 11. The percentage of individuals within geographically intermediate populations assigned a po~l(~io,rito either Pousada or Ribeira Peixe is related to STATUS OF thc geographic distance of the population from Pousada SCHISTOMTOPJM I1Rl:VIROSTRE (PETERS) and Ribeira l'eixe (Fig. 9), suggesting clinal variation in morphometric and meristic characteristics. Three other From 1880 (Peters) until 1965 (Taylor), Schistometopum populations, Rolas, Alto Douro 11, and Ponta Furado, are brevirostre existed only as a junior synonym of S. thomense. not positioned geographically between Pousada and Ribeira Neither Peters nor Taylor clearly explained their reasons Peixe. The two individuals from Rolas arc split between for, respectively, synoriyrriizing and resurrecting S. brevirostre. assignments to Pousada and Kibeira Peixe, even though Taylor (1965) was the first to assign specimens other than liolas Island is much closer to the latter. Alto Douro I1 is the holotype to S. 11re.oir-ostre,but his determinations were close to, and southeast of, Ribeira Peixe, and specimens obscure. He believed S. brevirostre is bluish gray compared from there are colored similarly to the freckled Ribcira to the yellowish S. thornense, but Peters (1880) had already Peixe anirnals. Even so, 30.0 and 12.5% oS males and fe- pointed out that the color of the type of S. brevirostre re- males, respectively, O-om Alto Douro I1 are assigned to dis- sulted from fading. Taylor also believed the head was some- tant Po~tsada.The Ponca Furado population is on the west- what elongated compared to S. thomense, however he was ern side of the island, and specimens Sl-om there are in- unaware of the sexual and geographic variation in head termediate in dorsal coloration between Pousada and shape and size. He also believed S. brevirostre has fcwer liibcira Peixe (Table 2), but more similar to Pousada in splenial teeth, "4-5", (1965:299), but one and two pages ventral coloration (Table 3). Nevertheless, a higher per- later in the same publication he lists splenial tooth counts centage of both males and females from Ponta Furado are of "9-9" and "10-1 1" for the specimens he assigned to this assigned to Rebeira Peixe than to Pousada based on species. multivariate morphometrics (Table 11). Taylor (1965, 1968) indicated that the holotype of Pousada n = 54
Rio Abade 0.8 - n = 43 0.6 -
0.4 -
>r 0.2 - 0 a, 0.0-,-, , , , , , , , , , , , , , , , , , 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 U
Colonia A~oreana
Ribeira Peixe n= 193
# of Brown Annuli
I7igrl1.c 7. (:litla1 gcogl.aphic variation in vcntral coloration of Schistomctopum Ll~ott~cxsr.Tllc Ilistogmrns sllolv tllr number of brown ventral :II~I~II~~(sCcIGg. 312 ant1 nlcthotls) along ;i I-andoirlly chosen 20-anntilrls segnlent of tlrc vcntei-. The histograms are ortlered by gcograpllic proxilliity. Mrsc. PUBL.Mns.. ZOOL.,UNIV. MICH., No. 187
Table 4. Midventral coloration of Schistomelop~~mfrom Siio Tome by population: frequency of individuals in populations with different amounts of white coloration along the midventral line. Pousada to Alto Douro I1 are geographically ordered eastern localities; Ponta Furada is an isolated western locality; Pousada includes both Pousada and Road to Pousada.
(1) (2) (3) white scattered none n
2 Pousada 4 Pedro~na 5 Rio Abade 6 Alto Douro I 7 Colonia A~orcana 8 Kibeira Peixe 9 Alto Douro I1
1 Ponta Furada
Table 5. Size (total length, mm) variation in Scllirlomelopum.
Males Females
Mean SE Range N Mearl SE Range N
S. thomerrse (all) 1 Ponta Furada 2 Pousada 3 Road to Pousada 4 Pedroma 5 Rio Abadc 6 Alto Douro I 7 Col6nia A~oreana 8 Ribeira Peixe 9 Alto Douro I1 10 Rolas S. gregorii (Kenya) S. grego'7'i (Tanzania)
Sipi~onopsbrevirostris was not available for examination when STATUS OF SCHISTOMETOPUM GARZONHEYDTI he visited the Berlin Museum in 1961. Wake (1985) ap- TAYLOR AND SALVADOR parently interpreted this to mean the type is "lost". One of us (RAN) borrowed caecilians from the Berlin Museum Based on the original description of Schistometopum in 1986 and among them identified two very similar speci- gurzonheydti Taylor and Salvador, 1978, it was apparent that mens (ZMB A600 and 491 1) one of which, presumably the this form could not be a Schi~tometopum. This was surpris- latter, was likely to be the holotype of S. brevirostris. In a ing, because Taylor was the foremost authority on caecilians letter to RAN dated 27 October 1992, Dr. Rainer Giinther and had described many new families and genera and pro- confirmed that ZMB 491 1 is the holotype of S. brevirostris. vided diagnoses for all genera (Taylor, 1968). RAN exam- Dr. Gt~ntherwrote that Peters (1874) had incorrectly listed ined the holotype, MNCN 1239, in 1987 and found that it the holotype as number 471 1, which led to the confusion was badly dried and brittle with a broken body and cut and suggestion that the type was lost. jaws. The color was generally yellow-brown resulting from When examined by RAN in 1986, the holotype was grayish drying and perhaps bleaching, as indicated by the rather brown with somewhat lighter annular grooves. The col- irregular and asymmetrical intensity of the yellow-brown oration is typical of specimens that have faded with pre- coloration. The annular grooves are clearly edged in white, servative and age. Our morphometric and meristic data although in 1987 this was only faintly indicated on some recorded from the holotype (Table 12) differ only slightly grooves, resulting from fading. The tentacles are far for- from that of Peters (1874), and demonstrate that, in agree- ward, nearly even with the anterior margin of the mouth, ment with Peters (1880) and contrary to Taylor (1965, much closer to the nares than to the distinct eyes. This is 1968), there is no basis for recognizing Schistometopum unlike Schistometopum, in which the tentacular apertures brevirostre. We consider it to be a junior synonym of S. are much closer to the eyes than to the nares. Unlike thomense. Schistometopu,m, the tongue bears a pair of narial plugs Table 6. Analysis of covariance of morphometric and meristic characters of males and females of Schirtometopum thomense: Pousada sample only, means adjusted to size (TL), measurements in mm, variables untransformed.
F-test (p) Scheff6 (p) Adjusted mean equal slopes difference
Body width males (n = 17) fernales (n = 33) I-lead length males f'emales Head width males females Interorbital distance males females Internarial distance males lemales Eye-naris distance males females Eye-tcntacle distancc males females Tentacle-naris distance males females Primary annuli males females Vent primary annuli males fcmales Secondary annuli males females Complete secondary annuli males females Max-premaxillary teeth males1 females' Vomeropalatine teeth males' fcmales' Dentary teeth males' females' Splcnial teeth males' females' Vertebrae males females Posterior scales rows males females Primary annuli with scales
females
* significant difference in mean and/or slope samples sizes for tooth counts differ from other characters: males (n = 17), females (n = 15) a p < 0.001 b p < 0.01 c p < 0.05 Table 7. Analysis of covariance of morphomrtric and meristic characters of ~rlalrsand f.r~~lalrsoS Schrstornrlo/~~~~~~ttlomr71sr: Ribeira Peixe sample only, means adjusted to size (TL), measurements in mrn, variables ~untransformed.
F-test (p) ScllefiG (p) Adjusted mean equal slopes difference
Body wicltll males (n = 18) females (11 = 32) Head length males females Head width males females
males
Internaria1 distance malcs females Eye-naris distance males females Eye-tentacle distance males Se~nales Tentacle-naris tlistarlce males trtnales Primary annuli males females Vent primary annuli rliales fcmalcs Secondary a1111~1li ~nales f'cmalcs Complete secondary a~~nuli malcs females Max-premaxillary teeth ~nales' females' Vorncropalatine teeth males1 females' llcntary teeth malcs' females' Splenial tccth nlales' females1 Vertebrae males lrmales Poslerior scales rows males females I'rimary anni~liwith scales males fernales
'"igniticant difference in mean and/or slope samples sizes for tooth counts difrcr from orher characters: males (n = lo), females (11= 10) ;1 p < 0.001 b p < 0.01 c p < 0.05 NUSSBAUMAND PPKENDBR
Schistometopum thomense
Pousada Ribeira Peixe
9- /
8 -
7 -
6-
5 - 5 Males y = ,024~+ .74 Males y = ,029~- .08 Females y = .019x + 1.57 //7: Females y = .016 + 2.16 4 I I I I
Total Length (mm)
Figure 8. Regression of head length and head width on total length for males and fernales of the Pousada and Riheira Peixe populations of Scl~i.\lom(~/oj~~rmll~omense. Adult males have larger heads than females in both populations. The sexual dimorphism is greatest in the liihcira l'cixc pop~~lation. Table 8. Analysis of covariance of morphometric and meristic characters of Pousada and Ribeira Peixe populations of Schislometof)urn thommsr: male sanlples only, means adjusted to size (TL), measurements in mm, variables untransformed.
F-test (p) Schef'fe (p) Adjusted mean equal slopes difference r2
Body width Pousada (n = 17) Ribeira Pcixe (n = 18) Head length Pousada Ribeira Peixe Head width Pousada Ribeira Peixe Intcrorbital distance Pousada Ribeira Peixc lrlternarial distance Pousada Ribcira Peixe Eye-naris distance Pousada Ribeira Peixe Eye-tenracle distance Pousada Ribeira Peixe Tentacle-naris distance Pousada Ribcira Peixc Primary annuli Pousada Ribeira Peixe Vent primary atlnuli Pousada Ribeira Peixe Secondary annuli Pousada Ribeira Peixe Complete secondary a~lliuli Pousada Ribeira Peixe Max-premaxillary teeth Pousada' Ribeira Peixc' Vomeropalatine teeth Pousada' Ribeira Peixc' Dentary teeth Pousada' Ribeira Peixe' Splcnial teeth Pousadal Ribeira Peixc' Vertebrae Pousada Ribeira Peixe Posterior scales rows Pousada Ribcira Peixc Primary annuli with scales Pousada Ribeira Peixc
" significant difference in mean and/or slope I samples sizes for tooth counts diffcr from other characters: Porlsada (n = 17), Ribeira Peixe (n = 10) a p < 0.001 b p < 0.01 c p < 0.05 Table 9. Analysis of covariance of morphometric and meristic characters of Pousada and Ribeira Peixe populations of Schistonz~toflumthomensc female samples only, means adjusted to size (TL), measurements in mm, variables untransformed.
F-test (p) Scheffk (p) Adjusted mean equal slopes difference r2
Body width Pousada (n = 33) Ribeira Pcixe (n = 32) Head length Pousada Ribeira Peixe Head width Pousada Ribeira Peixe Interorbital distance Pousada Ribeira Peixe lntcrnarial distance Pousada liibeira Peixe Eye-naris distance Po~~sada Ribeira Peixe Eye-tcntacle distance Pousada Ribeira Peixe Tentacle-naris dis~arice Pousada Ribeira Peixe Primary annuli Pousada Ribeira Peixc Vcnt primary annuli Pousada Ribeira Peixe Secondary annuli Pousada Ribeira Peixe Complete secondary annuli Pousada Ribeira Peixe Max-premaxillary teeth Pousada' Ribeira Peixel Vorneropalatine teeth I'ousada' Ribeira Peixe' Dentary teeth Pousada ' Ribeira Peixel Splenial teeth Pousada' Ribeira Peixe' Vertebrae Pousada Ribeira Peixe Posterior scales rows Pousada Ribeira Peixc Primary annuli with scales Pousada Ribeira Peixe
"significant difference in mean and/or slope 'sample sizes for tooth counts differ from other characters: Pousada (n =15), Ribeira Peixe (n = 10) ;' p < 0.001 " p < 0.01 ' p < 0.05 Table 10. Discriminant filnctions analysis: apost(,riori classification of all inclividuals of Schi.\tome/opum thomrnsefrom Pousada and Ribcira l'eixe arld S. gr.rgo~.iifrom Kenya as either male or female; two sets of discriminant fi~nctionswere calculated Sor S. /h(rrn~,r~sr,the first included the total sample without the lour tooth count variables and the second included a subset of the samples with the fotu- tooth count variables; values are lumbers and percent of'individttals.
Predicted Group Membership
Actual group 11 females
Schislornelof~on2tl~omrnsr A. without tooth colunts Pousada males females Ribeira Peixe males females R. with tool11 counts Pous;tda malcs I'elnales Rihcira l'eixe males fernales Sc/~i.\/orr~ctoj)~~m gre.go?-ii A. with tooth counts males females
Table 11. Discriminatlt f~lnctionsanalysis: apostwior; assignmerlt of all individt~alsSrom Sao Tom6 poptllations of Srhistomrtopurn, to eitliel- the Pousada or the Iiibeira Peixc populatiolis; all variables includccl; data are 11umbe1-of individuals and percentage of' total population.
Pledicted (;roup Membership
11 2 Pousada 8 Rlbe~raPeixe
Actual group males lcmales males fcn~ales male5 fernnles
1 Ponh Furado
2 Pousada*
4 Pcdroma
5 Rio Abatie
6 Alto Do11ro I
7 Col6nia ACOI-cana
8 liiheira Peixc
9 Nto Douro I1
10 Iiolas
" missing case disall~wcdfor one specimen Fig111-c9. Similarities betwecn adjacent populations of'Schistomc/@um thomrnse bascd on multivariate analysis of morphometric and meristic data. The graphs show the percent of tnales and females of geographically ordered popr~latioristhat are assigned a,hostm.ori through discriminant fiinctions analysis to either the l'ousada or the Ribeira Peixe popnlations. Population sample sizes given in Table 1.
Table 12. Morphomctric and meristic comparisons of holotypes of Schis/ometupum (~neasurcmentsin mm).
brri~irn.~tre ephek gorzonheydti prgorii ZMB MSNG MNCN BM 491 1 8773 1239 1946.9.5.53
-- Scx fernale female unknown unknown Maturity adult adult adult (?) adult (?) Total Icngtlr 208 185 225 289 Body width 7 8 10 10 Head lellgth 7.6 7.3 7.7 11.0 Head width 5.2 5.7 6.5 7.3 Interorbital c1ist:lrice 3.3 3.1 3.6 4.9 Intcrnarial distance 0.8 0.9 1.6 2.6 Eye-naris distance 3.5 3.4 3.2 4.5 Eye-tentaclc distance 0.7 0.7 3.1 1.7 Tcnraclc-naris distance 2.8 2.6 1.2 3.4 Primary anntlli 97 96 90 114 Vent primary annuli 2 2 2 3 Secondarv annt~li 34 40 42 46 Completc secondary annuli 5 8 7 11 Max-premaxillary teeth 29 29 40 32 Vorricropalaline tecth 35 36 27 Detitary teeth 23 30 26 Splenial teeth 16 34 2 Vertebrae 100 95 118 Poslcrior scale rows 3 4 5 Prirnary annuli with scales 71 40 94
"vomcropalatine teeth uncountable Table 13. Analysis of covariance of morphometric and meristic characters of males and females of Schistometopum gregorii: Kenya sample only, means adjusted to size (TI,), measurements in mm, variables untransformed.
F-test (p) Scheffk (p) Adjusted mean equal slopes difference r2
Body width males (n = 23) females (n = 22) Head length males females I-lead width males females Interorbital distance males females Illternaria1 distance males females Eye-naris distance males females Eye-tentacle distance males females Tentacle-naris distance males females Primary annuli males femalcs Vent primary annuli males females Secondary annuli males
Complete secondary annuli males femalcs Max-premaxillary teeth males females Vomeropalatine teeth males females Dentary teeth males females Splenial teeth males females Vertebrae males females Posterior scale rows males remales Primary annuli with scales males females
" significant difference in mean and/or slope a p < 0.001 b p < 0.01 c p < 0.05 NUSSBAUMAND PFRENDER
Schistometopum gregorii
12 -
11 -
Males y = ,025s + 2.45 Females y = .02x + 3.06
100 200 300 400
9-
8-
vE 7- 9
3 6-
5-
Males y = ,017~+ 2.07 = / Females y .014x + 2.62 4l I I I 100 200 300 400
Total Length (mm)
Figure 10. Scxual dimorphism in head size of Scl~islometopumgregorii parallels that observed in S. thommse. Graphs show scatter and least squares regression line. Sample sizes given in Table 13. 20 M~sc.Pr~ul.. Mus.. Zoor-., UNIV.MICH., No. 187
(smootl~in the former). Thc teeth of all series are bicus- being cataloged at the sarnc time in the Madrid Museum. pid, especially noticeable in the smaller more posterior It seems most likely that the holotype of S. garzonheydti tecth. The spleilial teeth are very small. No attempt was (MNCN 1239, = Geotryprlrs srraphini) and the specimen of made to determine the sex arid condition of the gonads C. tentaculata (MNCN 1237) wcre both wrongly recorded because of the fragility of the specimen. Gom Fernando Po at the same time and for the same (un- Morphometric and meristic data for the holotype of known) reason. Until additional specimens of C. trntaculata Schistomrto;~umgarzonheydti arc summarized in Table 12. and/or G. seraphini are found on Bioko, they should not These data differ only slightly from those presented by be considered to be part of the fauna of that island. Siini- Taylor and Salvador (1978). The largest discrepancy is in larly, the record of H. squalostoma for Fernando Po is sus- the number of secondary annuli, stated to be 26 and 36 pect. by Taylor and Salvador. The Sormcr value is probably a typographical error, as their 36 is much closer to RAN's count of 42 secondary annuli. STATUS OF The holotype cannot be distinguished from Geotrypetes SCHISTOMl
S. thomense (Pousada) 4 S. thomense (Ribeira Peixe) o S. gregorii (Kenya)
Females
Canonical Variate I
Figure 11. Distribntior~oT individnals of Scl~islornetopurr~grego~ii (Kenya sample only) and S. thornense by sex on the first two canonical variates based on all rnorpho~netric;ind meristic characters exccpt tooth counts. The first canonical variate accounts [or 94.61 (males) and 91.97 (fe- males) 76,and thc second canonical variate fol. ttle remaining 5.39 (males) and 8.03 (females) %, of the total variance. The results demonstrate ~nicrogeographicvariation in morphometric and meristic characters of S. thom~nsethat parallel rnicrogeographic variation in color. Table 14. Analysis of covariance of morphometric and meristic characters of Sclzistometopurn thomense (Pousada and Kibeira Peixe combined) and S. gregorii: males only, means adjusted to size (TL), measurements in mm, variables untransformed; sample size same as indicated for body width throughout, except interorbital distancr for S. grpgorii sample.
F-test (p) ScheffP (p) Adjusted mean equal slopes diffeferences r2
Body width thomense (n = 27) gregorii (n = 23) Mead length thomense gregorii Head wid~h tholnen,se gregorii Interorbital distance thommse gregorii (n = 22) Internarial distance thomense gr-egorii Eye-naris distance thomrnse gregorii Eye-tentacle distance thonlense gregorii Tentacle-naris distance thovnm.se gregi~rii Primary annuli thor~~ense gregorii Vent primary annuli thonlense g~egorii Secondary annuli thomense gregorii Complete secondary annuli lhomense gregorii Max-prcmaxillary teeth thomenst gregorii Vomeropalatine teeth thommse gregorii Dentary teeth thomense gregorii Splenial teeth thomtn.se greprii Vertebrae thomense gregorii Posterior scale rows thomense gregorii Primary annuli with scales thomense gregorii
* significant slope and/or difference " P < 0.001 "< 0.01 ' p < 0.05 Table 15. Analysis of covariance of morphometric and meristic characters of Schistometopum thomense (Pousada and Ribeira Peixe combined) and S. gregorii: females only, means adjusted to size (TL), measurements in mm, variables untransformed.
F-test (p) Scheffk (p) Adjusted mean equal slopes differences
Body width tliomense (n = 25) gregorii (n = 22) Head length
gregorii I-Iead width thomenue gregorii Iilterorbital distance tliomense gregorii Intcrnarial distance lhomcnse gregorii Eye-naris distance thomense gregorii Eye-tentaclc distance
grego rt i Tcntaclc-naris distance thomense gregorii Primary annuli thomense grprii Vent primary annuli tliomense gregorii Secondary annuli thomense grrgorii Complete secondary annuli thomense gregorii Max-premaxillary tceth Lhomc,nse pegorii Vorneropalatinc tceth
grego t-ii Dentary teeth thomcnse greg0n:i Splenial teeth lhom~nse gregorii Vertebrae thomrnse pegorii Posterior scale rows ll~omense gregorii Primary annuli with scales thomense grpgun'i
* significant slope and/or difference " P < 0.001 " p < 0.01 ' p < 0.05 Mrsc. P~JBL.Mus.. ZOOL.,UNIV. MICH., NO. 187
5- S. thomense (Pousada) - + S. thomense (Ribeira Peixe) o S. gregorii (Kenya) - A S. gregorii (Tanzania) Ruwenzori !=I d7 0 +-, OH 000 0 .s 1- 08 0 8 *+ cd oO 0 0 A00 - a Cl 40 + -1 - 0 8 0 u - A 6 + 0 -3 - + + +
-5 ....,....lr.mmlmrm.Imm.*InmrvI -15 -10 -5 0 5 10 15
Canonical Variate I
Figure 12. Distribution oS Senlale S~~L~.~~OII~OJLIIIgr(!gol%iarlcl S. lhornrilsr on tllr first two carlorlical variatcs based on all morphornetric and meristic data inclrlding the Sour tooth counts. The first canonical variable accounts fox- 96.35 %, and the seco~ldca~lorlical variable for the remaining 3.65 %I, 01' tlle total variance. All gl-oups are clearly distinguished, as in the analysis without tooth courrts (Fig. I I). The two Tanzanian ferrlales of S. pegorii and the putative Ruwenzori fernale of S. ll~o~trmsoarcisolateti. phic. Males have more dentary teeth and more scale rows seen macroscopically and recorded. in the posterior annuli than females (Table 13). A postel-L- There is no ontogenetic nor sexual variation evident in ori assignn~etltof individuals to sex indicates that discri- coloration, but all of the available specimens are old, and minant fr~nctionsanalysis using all morphometric cliarac- their color in life is unknown, except for Loveridge's ters accurately distingrrishes 90.5% of males and 95.5% of (1936:375) brief indication that they are "Uniform glossy females, which suggests the overall degree of sexual di- black resulting in a lacquered appearance." Sexual di- morphism is greater than in the Pousada sample of S. chromatism is unreported in caecilians and apparently does thotnense, but less than the Ribeira l'eixe sample ofthc lat- not occur in this species. ter species (Table 10). There arc only two specimens, both females (MCZ 27901, As expected, there is an ontogenetic increase in tooth 28601), available from southern coastal Tanzania that have number on all four rows for both sexes and no ontoge- been assigned to Schistometofium
Hypogeophis rostratus
91 II Males y = 0.023~+ 0.007 r2 = 0.965
Total Length (rnm)
Figure 13. Regression oC head length and head width on total length for a population of the caeciliid, I3ypogeof1kis rostr(~tus from Frkgate Island, Republic of' Seychelles. The same pattern of sexual dimorphism observed in Schi.sto?netopum Ll~o.me?~.se and S. gregorii (males with larger heads) is obscrved in this species. closc to the eye, and it has a higher number of maxillary- In addition to the taxonomic changes derived from this premaxillary teeth (38) than recorded for S. thomense, which study, there are two additional results that are of particu- is especially noteworthy given the relatively small size of lar interest. Firstly, the results dispute the attitude of some the specimen. The other tooth counts are either near the proponents of the evolutionary and phylogenetic species rnaximum (voineropalatine) or minimum (dentary, sple- concepts (ESC and PSC), who argue that if an allopatric nial) recorded for S. tl~omensefromSio TomC. The number group of individuals is "diagnosable" then it is a species of primary and secondary annuli and vertebrae fall well (Cracraft, 1983, 1987; Rosen, 1978, 1979; Highton, 1990; within the range of S. thomense. The specimen represents Howard et al. 1993; Ridley, 1989), and argue against the either a remarkable range extension for S. thomense, or a wholesale elevation of distinguishable, closely related, cryptic species closely related to the latter, or perhaps it is allopatric populations to species level without careful field a mislabeled specimen actually from Sio TomC, although studies, as was suggested by Collins (1991, 1992). Secondly, there is no apparent reason to doubt the record. sexual dimorphism has rarely been reported in caecilians, and the discovery of marked sexual dimorphism in both species of Schistometopum raises questions about the social DISCUSSION interactions of these and other caecilian species. There is a trend in herpetology to name new species based Taylor (1965) was intemperate in describing Schi.stometopum on minimal character differences between obviously closely "e1)helen because he had access to information that should related forms that are allopatric. This is especially true have invoked caution and because he had not done any among molecular systematists who often use genetic dis- field work aimed at resolving the status of the forms of tances based on allozymes or DNA sequences as evidence Schistoni~etopumon Sio TomC. S. "ephele" differs, according that polytypic species should be resolved into two or more to Taylor (1965:297), from S. thomense only in that the former species. Often, the ESC or PSC is cited to justify these has a "proportionally small, more pointed head than proposals (e.g., Collins, 1991, 1992; Daugherty et al., 1990; thornen.seXand has a dorsolateral yellowish coloration "with Howard et al., 1993). These investigators seem not to have dark lilac-brown flecks", whereas S. thomense is immaculate noticed that there is nothing axiomatic in the ESC nor in yellow. Had Taylor recorded the sex of specimens and care- the PSC that unquestionablyjustifies these actions. There- fully studied variation in the sample of S. thomense that was fore, to accept diagnosability as justification for naming available to him, he would have understood the significance allopatric species is to accept a conventional and/or au- of head size and shape in the holotype of S. "ephele" and thoritarian solution to the complex issue of species recog- realized that its head is small and pointed because it is a nition. We fear that acceptance of simplified, convelltional Pernale, ~ndnot because it is a different species. In regard solutions to the species problem, such as is advocated by to color pattern, the range of variation in specimens from some extreme proponents of the ESC and PSC, will ob- Sio Tom6 examined by Taylor should have given him pause, scure much interesting biology and worse yet may reduce and apparently he was uncertain about the status of some the amount of field research that goes into studying distri- individuals as indicated by his remark (1965:297) that a butions, geographic variation, speciation, and biodiversity. second specimen with the same catalogue number as the For example, in response to one of his critics, who sug- type 01's. ephele "may likewise belong here". gested that additional field work might resolve the status We suspected Sclzistometopum ephele was invalid, but deemed of some allopatric populations, Collins (1992) replied that it necessary to conduct systematic field surveys on S2o Tom6 the possible "discovery of new records [that might show and Principe Islands to resolve with certainty the taxonomic intergradation] is speculation and has no bearing on the status of this form. Our Geld work revealed intergradation situation". By this we presume Collins means that allopatry between nominal S. thornenseand the ephele morph, and more and diagnosability alone are enough under the concept of interestingly showcd that dramatic geographic variation the ESC to justify taxonomic revision, even before field work within a species can occur over very short distances in seem- and careful analysis of variation are done. We disagree ingly uniform habitat, even within a small island system. with Collins, and believe that exhaustive field work and Similar unexpectedly high amounts of microgeographic extensive data analysis should precede any taxonomic re- variation in island populations was recorded for three liz- vision that involves uncertain status of allopatric ard species (Brown and Thorpe, 1991; Brown et al., 1991; populations. Thorpc and BBcz, 1987; Thorpe and Brown, 1989), which Caecilians are poorly studied from most aspects, so it is suggested to the authors that strong natural selection along not surprising that sexual dimorphism in this group is largely cnvironmcntal gradients and/or sexual selection, rather unknown. It has long been known that males of many spe- than hybridization in secondary contact zones, might be cies of most families have a pair of "anal glands" anteriorly respollsiblc for the sharply clinal variation that was docu- on either side of the cloacal opening and that these glands mented. However, the reasons for strong microgeographic are absent or very rarely present in females (Taylor, 1968; variation within these island populations of lizards and of Nussbaum, personal observations). Sexual dimorphism in S. thomense are unknown and invoke questions about evo- the size and shape of the "anal plate", or the area around lution and the process of speciation that transcend in im- the cloacal opening has been documented in typhlonectids portance the issue of species definitions. (Gon~alves,1977; Taylor, 1968; Wilkinson, 1989). Nussbaum 3 0 Mrsc. PCJUL.Mus.. ZOOL.,UNIV. MICH., NO. 187
(1985) reported sexual dimorphism in the nnlnber of ver- prone to bite, althongh it is clear that individuals of both tebrae and primary annr~liin all thrcc currently recognized sexes are bitten, both in the field and in laboratory colo- species of Scoleconrorph,us. The pattern is particularly strik- nies. This is evident from the signature scars present on ing in S. ulugurue?zsis, the only spccies in which a large ho- the bodies of both freshly canght and captive individuals. mogeneous sample exists. In S. uluguru~nsis,there is vir- Teodecki et al. (in press) demonstrated that females of tually no overlap in the numbel- of vertebrae and primary Schistometopurn thomense, with their smaller and more pointed annuli, with fcmales, the larger sex, having greater num- heads, burrow significantly faster than the larger and blunter bers of both. Nussba~~m(1985) suggested that more body headed males, which suggests a cost to males of having larger segments in females may be related to reproductive po- heads. These observations, while intriguing, are limited tential, in that added segments might provide additional and subject to various interpretations. Detailed and clev- space for developing embryos in these viviparous species. erly designed studies are needed to reveal the mating sys- In regard to head size, Nussbaum (1985) found that males tem and other aspects of the social lives of caecilians, so of Scoleco,rnorphus ulug.ururn.si.s have significantly longer (p that secondary sexual dimorphism in head size and the < 0.00) and wider (p < 0.000) heads than females. In S. number of body segnlents might be understood. kirkii, the adjusted mean head length and width of males The life history and ecology of Schi~tomelopvrnpegorii and and females could not be compared, because the slopes of S. thomense have not been studied in detail. Arthur head size regressed on total length werc significantly dif- Loveridgc, who obtained most specimens of S. Figure 14 (facing page). View of southcastcrn coast of Sfto Tom6 island looking to the southwest. Thc rocky, volcanic shore line and highly modified vegelation typify tl~rarea where most silmples of Schislorn,elopum thomrnsr werc collected. mcrous occasions. In these cases, the caecilians were ei- is unusual in that most other caecilians are not brightly ther mating or feeding in the water; their eggs are depos- colored, although some species of the South American ge- ited on land. H. rostratus are attracted to dead fish placed nus Epicrionops and the Asian genus Ichthyophis have yellow, in the water by Seychellois for cooling, or to offal discarded lateral stripes. Bright yellow is usually associated with in water, both of which they attempt to eat. It would not aposematic or pseudoaposematic coloration, but during our be surprising if's. 'gregorii, which apparently is semi-aquatic, brief stay of about one month on Sio Tome, we observed no occasionally feeds on frog and/or fish eggs, which might interactions suggesting a function for the bright coloration. give natives the impression these eggs are deposited by The only potential models observed were millipedes. caecilians. Lovcridge (1936) reported only termites in the Both Schistometopum gregorii and S. thomense, although known stomachs of the specimens taken from more terrestrial sites from relatively few specimens from small geographic areas, and only earthworms in stomachs of specimens from the appear not to be threatened with extinction. S. gregorii is shallows of Lake Peccatoni. apparently abundant on the Tana River delta, an area heav- We observed Schistometopum thomense, or "cobra bobo" (silly ily used for agriculture, and they occur in muddy areas around or foolish snake), from sea level up to about 1,000 m el- villages and in plantations. Similarly, S. thomense is common evation on S5o Tomi, and they almost certainly occur at and does not seem to be adversely affected by deforestation higher elevations. They were collected in June during the and other habitat disturbance. Although the current status dry season, in relatively undisturbed forest, but they are of S. b-regorii is uncertain (it hasn't been seriously sought re- common in a variety of habitats, including highly hctero- cently), both species of Schistometopum thrive in anthropo- geneous, rocky, coastal, habitats (Fig. 14) and completely genic habitats, as do some other African species such as denuded and degraded habitats (Fig. 15) such as cacau Boulengerula taitanus (Nussbaum and Hinkel, 1989), some plantations, sugar cane plots, and trash piles. They were Indian species such as Gegeneophis ramas7uamii (Nussbaum found in wet soil, under discarded cacau husks, under stones and Gans, in preparation), and some species of Seychellearl and logs, and in litter. They were lrsually found at consid- caecilians (Nussbaum, unpublished). For these species, forest erable depth in rnincral soil at drier sites and closer to cover is not important, but rather the critical factors are an the slirface at wetter sites such as in seeps and muddy talus abundance of moisture, soil and litter, and abundant food at edges of streams. I11 a cacau plantation at Pedroma (220 in the Sorm of soil and litter-layer invertebrates. Deforesta- m), smaller specimens were fonnd closer to the surface tion and traditional native agricultural practices may actu- than larger individuals. Early afternoon soil temperatures ally benefit these species by increasing the amount of moist at 12 sites where individuals were collected 8-17 June at soil (rice paddies, banana plantations) available to live in 850 In in degraded habitat ranged from 21.4 to 25.2'' C. and the amount of decaying organic matter at the soil sur- They tended to be concentrated at some sites, but it is face, which results in higher productivity of invertebrate prey ~rncertainwhether this was due LO social interactions or available to the caecilians. A switch from traditional to mod- local habitat conditions. Occasionally, two large ern agricultural practices involving the heavy use of pesti- individuals werc found very close together, nearly touch- cides and synthetic fertilizers could, however, change this ing. Many had bite marks on them, which we confirmed picture. as conspecific bites by later laboratory observations. Sciristometopu,rr~thomense is viviparous, with 2 to 7 young being born 11s11allyat the same time in captivity. The young LITERATURE CITED (11 = 27) are 91-118 mm (mean = 105.6) total length at birth, Amadon, D. 1953. Avian systematics and evolution in the Gulf of weigh 0.90-1.99 g (mean = 1.52), and are Sully formed with Guinea. The J. G. Correia collection. Bull. Amer. Mus. Nat. Hist. 110 evidence of' gill scars and with essentially adult colora- 100:393-451. tion (Fig. 5). In June, most of' the adult females we cap- Barhoza d11 Bocage, J. V. B. 1873. MClangcs crpetologiques. 11. Sur tured were pregnant, nlany giving birth within a few days quelqucs reptiles et hatraciens uouveaux, rares 011 peu connus oP capture. Stomach contents of the few specimens exam- d'Afriqne occidentale. 1. Scienc. Math. Phys. Nat. 43209.227. incd contained sand (probably from earthworms), earth- . 1897. Mammiferos, reptis e batrachios d'Africa de que cxistem exeniplares typicos no Muscu de Lisboa. Jornal de Sciencias, worms, and insects. Mathamaticas, Physicas e Natnraes, 2ed series, 4(16):187-206. Occasionally dcad specimens were found under litter, es- Bauer, A. M., Good, D. A., and I<. Giinther. 1993. An annotated type pecially under cacau hrrsks. At Pedrorna, three dead indi- catalogne of the caecilians and salamanders (Amphibia: viduals were found with pustules on their skin. They were Gymnophiona and Caudata) in the Zoological Museum, Berlin. being caten by ants and other insects, which arc particu- Mitt. Zool. Mus. Berl. 69:285-306. larly abundant in the litter and piles ofcacau husks in plan- Boettger, 0. 1913. Reptilien und Amphibien von Madagaskar, den tations. It was unclear whether the insects had killed the Inseln und dem liestland Ostafrikas. pp. 269-375. In: Voeltzkow, A,, Keise in Ostafrika in den Jahren, 1889-1895, 1903-1905. Stutt- caccilians. We oSten Sound healthy and uninjured caecilians gart, Bd. 3, Heft 4. in litter with dense pop~rlationsof biting ants. Boulenger, G. A. 1882. Catalogue oC the Batrachia Gradientia s. Caudata The brilliant yellow coloration of Schistometopum thomense in the collection of the British Museurn. Second ed. Taylor and Figr~rc15 (facing pagc). <:oastal plateau at Ribeira Peixe. This region was being cleared for planting oil palm in 1988. Caecilians were fonnd in the degraded forcsts shown in the background. Francis, London, i-viii + 123 pp., 9 plates. . 1913. Gymnophiona (Amphibia apoda). Das Tierreich 37, , 1894. Third report of additions to the batrachian collec- i-x + 31 pp. tion in the Natural-History Muserrm. Proc. Zool. Soc. L.onr1. 5:640- Nusshaum, R. A. 1985. Systematics of caecilians (Anrpliibia: 646 + 2 plates. Gyrnrlophiolla) of the filllily Scolcconrorphidac. Occas. Pap. Mus. 1895. A synopsis of the genera and species of apodal Zool. Univ. Michigan 71 3: 1-49. batrachia~~s,with tlescriptions of a ncw gcnus and species (Udellophn , and I-I. I-Iinkel. 1994. Revision of East African caecilians of oi//r~t,us).PI-or. Zool. Soc. 26:401-414. tlre genera A/rocol:cilia Taylol- and Boulengerz~laTOI-nicr (Amphibia: Brown, R. P. and R. S. Thorpe. 1991. Within-island microgeographic Gynrnophiona: Caeciliaidae). Copcia 1994:750-76. variation in body dimensions and scalatior~of the skink Chalcidr\ Parker, H. W. 1941. The caecilians of the Seychelles. Ann. Mag. .sc,xlineat~~s,with testing of causal hypotheses. Biol. J. I.inn. Soc. Nat. Ilist. Ser. 11, 7:l-17. 44:47-64. Peabody, R. B. and E. D. Broclie, Jr. 1975. Effect of temperature, , and M. B8cz. 1991. Parallel within-island salinity, and photoperiod on the number of trunk vertebrae in rnicroevolution of lizal-ds on neighboring islands. Nature 352:60- Ambystoma rnaczclatunl. Copeia 1975:741-746. 62. Peters, W. 1874. ~berncue Arnphibien ( (;ymriopi.s, Sif~h~riops,IJolypedatr.7, Collins, J. T. 1991. Viewpoint: a ncw taxono~nicarrangement for Rl~arophorus,Hylo, Cyclorlus, Euprepr.7, Clemnzys). Monatshericht some North Arnericair ;t~r~phibiansand reptiles. Herpetol. Review Konigl. Akadernie Wissenscliaften Berlin 1874:616-625 + 2 plates 22:42-43. . 1879 (1880). her die Eintheilung der Caecilien und . 1992. The evolrltionary species concept: a reply to Van insbesondere iiber die Gattungen liliinatrema und Gymnopzs. Lkvender c.t al. and Montanucci. Herpctol. Review 23:43-46. Monatsbericlrt Konigl. Akademie Wisscnsclrafirn Berlin, 1879:924- c..,tc~aft, . .. J. 1983. Species concepts and speciation analysis. Pp. 159- 943 + 1 plate. 187. In: Johnston, R. F. (Ed.), Current Ornithology 1, Plenum, . 1880. ~bernetre oder wenigel- bekanrrte Amphihien des New York. Berliner Zoologischcn Museums (Leposome dispnr, Morrofieltis . 1987. Spccics concrpts and the ontology of evolution. Biol. (1'Arartogonu.s) ji~gularis,7yphlops drf)rr.ssus, Lef)toralnmus trilineatus, I'lrilos. 2:329-346. Xrn,odon punctatics, I