©Zoologische Staatssammlung München;download: http://www.biodiversitylibrary.org/; www.biologiezentrum.at

SPIXIANA ©Zoologische Staatssammlung München;download: http://www.biodiversitylibrary.org/; www.biologiezentrum.at

level and possibly are witnesses of climatic shifts of ß. tephraeomystax, as outgroup. in the past, when more humid conditions pre- Sequences were aligned using SEQUENCE vailed also on 's west coast (Nuss- NAVIGATOR (Applied Biosystems). Inclusion of gaps to account for indels was only necessary at a baum & Raxworthy 1998, Nussbaum et al. 1998). Single Position. The aligned sequences were sub- The treefrog Boophis tephraeomystax is a wide- mitted to analysis using PAUP*, Version 4.0 (Swof- spread generalist species occurring in low- to ford 1998). We performed Maximum Parsimony regions throughout mid-elevational Madagas- analyses using the exhaustive search Option and car (Blommers-Schlösser & Blanc 1991; Glaw & Neighbor joining analyses using LogDet distances Vences 1994). As the species apparently is una- which are robust against possible Variation of se-

ble to colonize highland regions, it is likely that quence evolution among lineages (Lockhart et al. the central mountain chain acts as a barrier, 1994). Homogeneity of base frequencies across se- inhibiting gene flow between eastern and west- quences was tested using algorithms implemented in PAUP*. ern populations. Contact zones exist, however, Advertisement calls were recorded in the field around Tolagnaro in southern Madagascar, and using different portable tape recorders and micro- in the Sambirano region in the northwest. phones. Detailed call descriptions will be published Previous studies on intraspecific Variation elsewhere; we here provide only short informations are limited, of Boophis tephraeomystax but Blom- on the main differences that are obvious to the hu- mers-Schlösser (1978) reported on important karyological differences between species from eastern ( and Foulpointe) and south- Results and discussion western Madagascar (Toliara). In the present study, we analysed mito- A chi-square test did not contradict homogene- chondrial DNA sequences from seven Boophis ity of base frequencies across taxa (df = 21; P = 1). tephraeomystax populations throughout its ränge Of the 562 characters included in the analysis, and compared advertisement call data from a 516 were constant, 33 variable but parsimony- total of 11 populations. Our aim was to find uninformative, and 13 parsimony-informative. evidence for phylogeographic structuring (Avise Maximum parsimony analysis resulted in two 2000) of the haplotypes found within the spe- equally parsimonious trees (51 steps; consist- cies, and to discuss the observed pattern in the ency index 0.98, retention index 0.96). The re- light of possible vicariant east-west differentia- sulting strict consensus free (Fig. 1) divided the tion of Malagasy anurans. B. tephraeomystax sequences into two clearly distinct clades, corresponding to the eastern Materials and Methods and Sambirano (Cap Est, Sambava, Ambanja, ) and western (Ankarafantsika, Kir- fragments of the mitochondrial 16S We sequenced indy, Isalo) populations. Within the first clade, rRNA gene of 562 nucleotides (nt) using primers the Ambanja sequence formed a subclade with and protocols given in Vences et al. (2000b). Vouch- the Nosy Be sequence. The Neighbor-joining ers were deposited in the herpetological collections analysis resulted in a similar topology. Boot- of the Universite d' , Departement de strap analyses (Felsenstein 1985) with 2000 rep- Biologie Animale (UADBA), the Zoologisches For- schungsinstitut und Museum A. Koenig, Bonn licates resulted in high support (> 95 %) for these (ZFMK) and the Zoologische Staatssammlung Mün- clades using either method. chen (ZSM). Specimens of Boophis tephraeomystax Genetic differentiation was 0-3 nt among used (Genbank /EMBL accession of sequences and sequences in the eastern /Sambirano clade (the voucher specimen numbers in parentheses) were sequences from Nosy Be and Ambanja were from (1) Nosy Be (ZSM 458/2000, AJ312114), (2) identical), and 1-2 nt among sequences in the Ambanja (ZSM 489/2000, AJ312115), (3) Sambava western clade, but 12-13 nt (2% pairwise se- (UADBA (FG/MV) 2000. 379, AJ312116), (4) Cap quence divergence) between clades. Est, Masoala Peninsula (AF215333; ZFMK 66685), The genetic data therefore clearly indicate (5) Ampijoroa, Ankarafantsika Reserve ZSM 704/ the presence of different haplotypes in eastern 2001, AJ312117), (6) Kirindy (AF215334; ZFMK populations of Boophis tephraeo- 66690), and (7) Isalo (AF 215332; ZFMK 70495). We and western used a specimen of Boophis xerophilus from Kirindy mystax (Fig. 2). In our extensive data set on (ZFMK 66705, AF 215335), which is a close relative Malagasy anurans, we never observed intrap-

80 ©Zoologische Staatssammlung München;download: http://www.biodiversitylibrary.org/; www.biologiezentrum.at

63/63 Nosy Be (1)

Ambanja (2) 97/98

Sambava (3)

Cap Est (4)

Isalo (7) 99/100 Ampijoroa (5)

Kirindy (6)

B. xerophilus

Fig. 1. Strict consensus of two equally most parsimonious trees obtained by Maximum Parsimony analysis oi sequences (562 bp of the 16S rRNA gene) from seven populations previously assigned to Boophis tephraeomystax. Numbers above branches indicate bootstrap support (Felsenstein 1985) of Maximum Parsimony and Neighbor-joining analyses, respectively. Boophis xerophilus was used as the outgroup.

opulational polymorphism of such magnitude served at Nosy Be approaching the breeding in the 16S gene; although we here sequenced pond and mating with two of the calling males, only a Single specimen per population, it can but nevertheless no trill calls were heard in this safely be excluded that the observed differenc- population. On the other hand, at Tolagnaro, es between populations are also present within Ranohira, Kirindy and Ankarafantsika, trill calls populations. The differentiation of eastern and and note series were usually part of the calling western populations of B. tephraeomystax is fur- repertoire. At Ankarafantsika, after a few days thermore corroborated by bioacoustic and ka- of drought, specimens were very shy and little ryological differences. motivated; nevertheless, trill calls were regu- The advertisement call of ß. tephraeomystax larly heard, as in highly motivated choruses was first described by Blommers-Schlösser after heavy rain at this site (though less fre-

(1979), who characterized it as reminding the quently). The sonagram of a call recorded at yelping of a young dog. In a previous publica- Toliara in Blommers-Schlösser (1979) also shows tion (Glaw & Vences 1994) we confirmed this two rapidly repeated yelping notes foUowed characterization, but mentioned that these vo- by a short trill of three pulses. We therefore calization can be emitted as Single notes or as conclude that the observed differences are geo- short note series, or can be foUowed by a short graphically correlated patterns: western popu- click. We also noted that highly motivated males lations of B. tephraeomystax have complex calls from Ranohira (near Isalo) had a more complex with trill notes, while eastern populations usu- call structure, with a faster note repetition rate ally emit only single yelping notes. and including additional trill notes of 5-8 puls- The karyological differences reported by es. During intensive survey work in 1995-1996 Blommers-Schlösser (1978) between southwest- and 2000-2001, we heard choruses of Boophis ern and eastern populations of ß. tephraeomys- tephraeomystax at many additional sites, and tax were significant. All specimens had a kary- found that the observed differences were large- otype of 2n = 26. However, in eastern speci- ly not correlated with the male's State of sexual mens, the chromosome pairs 2, 4 and 6 were motivation. At eastern sites (including the Sam- submetacentric (the remaining pairs being met- birana Region), the calls were generally Single acentric), while this State was expressed in pairs yelping notes repeated after longer intervals. 2, 3, 4, 9 and 10 in the southwestern specimens. This was true in Tolagnaro, Ranomafana, An- A secondary constriction was noted on pair 6 in dasibe, Vohidrazana, Maroantsetra, Sambava, the eastern specimens and on pair 10 in the Antsirasira, Ambanja, and Nosy Be. In several southwestern specimens. These differences led

cases (e.g., on the northwestern offshore Blommers-Schlösser & Blanc (1991) to pose the Nosy Be), the specimens were calling in a high- question whether these two populations may ly motivated chorus. Two females were ob- correspond to different species.

81 ©Zoologische Staatssammlung München;download: http://www.biodiversitylibrary.org/; www.biologiezentrum.at Evaluating the differences summarized

above, there remains little doubt that this ques-

tion can be answered positively . The amount of karyological, bioacoustic and genetic differen- tiation indicates that ß. tephraeomystax actually

is a complex of two allopatrically distributed species. We were unable to detect reliable dif- ferences between these two species in externa! morphology; a detailed morphometric study including large sample sizes is necessary to address this question. B. tephraeomystax was described from Mada- gascar, without precise locaUty; specimens from Nosy Be were later erroneously also consid- ered to be part of the type series (Glaw & Ven- ces 1997). Despite the iincertain type locality, it can be assumed that the type was coUected from eastern Madagascar or the Sambirano re- gion, as most early anuran collections from Madagascar were made in these areas. We there- fore consider the form occurring in these areas as B. tephraeomystax. Origin of the types of the junior Synonyms Polypedates crossleyi Peters, 1874 (Nosy Vola) and Polypedates dispar Boett- ger, 1879 (Nosy Be) are also in the eastern re- pectively Sambirano regions (see Blommers- Schlösser & Blanc 1991). Rhacophorus hildebrandti Ahl, 1925 was described from "north-western Fig. 2. Localities of populations that can be assigned Madagascar"; according to Beentje (1998), the to Boophis tephraeomystax (grey Symbols) and Boo- coUector M. Hildebrandt spent only little time phis doidioti (black Symbols). Circles refer to popula- J. in the Ankarafantsika tions from which DNA sequences were included in during the dry season the present study; Squares refer to populations of region, while he collected during more extend- which call data are available (additionally also from ed periods in the rainy season in the Sambirano

all populations symbolized by circles except for Region. It is unlikely that the type of hildebrand- Toliara); rhomboids mark sites Cap Est, and from ti was collected in the seasonal Ankarafantsika which were sampled for karyological data by Blom- environments during the dry season, and we mers-Schlösser (1978); and the star marks the type therefore suppose that it originales from the locality of Boophis doulioti. Localities 1-7 as in Fig. 1 Sambirano Region (and is therefore a synonym and in Materials and Methods: 1. Nosy Be. 2. Am- of B. tephraeomystax). The only junior synonym banja. 3. Sambava. 4. Cap Est. 5. Ampijoroa. 6. Ki- from western Madagascar is Rhacophorus douli- rindy. 7. Isalo. Other localities: 8. Antsirasira. described from 9. Maroantsetra. 10. Foulpointe. 11. Toamasina oti Angel, 1934. This taxon was (=Tamatave). 12. Vohidrazana. 13. Andasibe. 14. , which is close to our sampling Ranomafana. 15. Tolagnaro (=Fort Dauphin). 16. locality Kirindy. We therefore propose to res- Toliara (=Tulear). 17. Morondava. The cladogram urrect this name as Boophis doulioti (Angel, 1934) (one of the two most parsimonious trees from from the synonymy of Boophis tephraeomystax, Fig. 1; 51 Steps, consistency index 0.95) is superim- and to apply it to the populations in western were adjusted to posed on the map; brauch lengths Madagascar between Ankarafantsika and Fort fit the geography and are not indicative of genetic Dauphin. distances. Bars mark apomorphies in sequence evo- The partition of Boophis tephraeomystax into lution as reconstructed using PAUP* (Swofford separate species, B. tephraeomystax and 1998); thin grey bars are transitions, thick black bars two bio- are transversions. B. doidioti, adds relevant perspective to the . Only few anurans

82 ©Zoologische Staatssammlung München;download: http://www.biodiversitylibrary.org/; www.biologiezentrum.at are known to occur in eastern as well as in (northeast and northwest) and M. corvus (west), western Madagascar (Blommers-Schlösser & betsileo (east) and H. carbonei (west), Blanc 1991; Glaw & Vences 1994; Hawkins 1994; Aglyptodactylns madagascariensis (east) and A. se-

Emanueli & Jesu 1995; Vences et al. 1999): Ptij- cnrifer (west) (Glaw & Vences 1994, Glaw et al. chadena mascareniensis is ubiquituous in Mada- 1998, Vences et al. 2000a). The present study gascar, including east and west; Mantella betsi- adds Boophis tephraeomystax (east and north- leo and Mantidactylus wittei (subgenus Blommer- west) and B. doulioti (west) as a further example sia) show a coastal distribution which is rather to this list. similar to that of B. tephraeomystax; Scaphiophryne A scenario which could explain this pattern mannorata occurs at the western site Tsingy de assumes the existence of a more humid Bemaraha and in eastern rainiorests; Boophis which allowed the respective ancestors of these Intens is known from the east and from Isalo; species pairs to populate continuously eastern Mantidach/his idcerosus, M. biponis, M. cnrtiis and western Madagascar. In a second step, more and M. bctsüeanus (all subgenus Brygoomantis) arid climate arose and isolated the western pop- have been recorded from several eastern, north- ulations during a considerable amount of time western and western sites; Mantidactylus Intens in small humid refuges. During this time of (subgenus Gephyromantis) and M. granulatus isolation, allopatric speciation of the vicariant (subgenus Phylacomantis) are known from the populations occurred. Subsequent shifts back east and from one western site, respectively. to more humid conditions allowed a secondary Of these examples, several are probably due contact between some species pairs. To fest this to wrong determinations, such as the records of hypothesis, it will be necessary to study more the different Brygoomantis which are very diffi- variable genetic markers and to identify in de- cult to distinguish morphologically; so far, all tail the contact and / or hybrid zones of the spe- Bn/goomantis found by us in western Madagas- cies pairs. car were assignable to M. idcewsns or to unde- scribed species. The record of Mantidactylns In- Acknowledgements tens (Blommers-Schlösser & Blanc 1991, appar- ently based on MNHN 1975.327) is due to a We are grateful to F. Andreone (Torino), G. Garcia mistake with M. corviis or M. pseudoasper (pers. (Jersey), V. H. Raherisoa (Antananarivo), J. Randri- obs.). The record of M. granidatns from the west- anirina (Antananarivo), A. Rasoamamonjinirina ern locality Antsingy by Blommers-Schlösser & (Antananarivo), K. Schmidt (München) and D. R. is Blanc (1991) unconfirmed as we were unable Vieites (Vigo), who assisted in the field. J. Kosuch to locate vouchers for this locality in the MNHN and C. Schäfer (Mainz) contributed some of the and ZMA collections on which the distribution DNA sequences. The field work was made possible maps of these authors were almost entirely by a Cooperation accord between the Zoological Institute at the University of Antananarivo (UAD- based. The Scaphiophryne mannorata population BA) and the Zoologisches Forschungsinstitut und from the Tsingy de Bemaraha shows different Museum A. Koenig (ZFMK). Financial Support was colour patterns to those from the east, and may granted both to the field work of F. Glaw and the correspond to a distinct species. And at least in work of M. Vences in the MNHN by the "Deutscher Mantella betsileo, genetic evidence (pers. obs.) Akademischer Austauschdienst" (DAAD). indicates that the western populations are taxo- nomically distinct from those in the east and References northwest. No data have so far been published on the western populations of Mantidactylns Angel, F. 1934. Description d'un viperide nouveau, wittei and Boophis Intens, but further studies du Congo Beige, et de deux batraciens de may confirm that the anuran endemicity rates Madagascar. - Bull. Soc. Zool. France 59: 169- of western Madagascar are much higher than 172 previously thought. Avise, J. C. 2000. Phylogeography. The history and On the other hand, at least four pairs of formation of species. - Harvard University sister species show a vicariant east- Press, Cambridge, 447 p. west distribution: Boophis albilabris (east) and Battistini, R. 1996. Paleogeographie et variete des ß. occidentalis (west), Mantidactylus pseudoasper milieux naturels ä Madagascar det dans les iles

83 ©Zoologische Staatssammlung München;download: http://www.biodiversitylibrary.org/; www.biologiezentrum.at

voisines: quelques donnees de base pour l'etu- Nussbaum, R. A. & C. J. Raxworthy 1998. Revision de biogeographique de la "region malgache". of the genus Ebenavia Boettger (Reptilia: Squa-

Pp. 1-17 in Louren^o, W. R. (ed.): Actes du mata: Gekkonidae). - Herpetologica 54 (1): Colloque International Biogeographie de Ma- 18-34 -- dagascar. - Societe de Biogeographie - Mu- -- , & O. Pronk 1998. The ghost geckos of seum, Paris Madagascar: A further revision of the Mala- (Reptilia, Beentje, H. J. 1998. J. M. Hildebrandt (1847-1881): gasy leaf-toed geckos Squamata, Notes on his travels and plant collections. - Gekkonidae). - Mise. Publ. Mus. Zool. Univ. Kew Bulletin 53: 835-856 Michigan 186: 1-26 Blommers-Schlösser, R. M. A. 1978. Cytotaxonomy Swofford, D. L. 1998. PAUP*. Phylogenetic Analy- of the Ranidae, Rhacophoridae, sis Using Parsimony (* and other methods), (Anura) from Madagascar with a note on the Version 4. - Sinauer Associates, Sunderland, karyotype of two of the Seychel- Massachusets

les. - Genetica 48: 23-40 Vences, M., F. Glaw & W. Böhme 1999. A review of — 1979. Biosystematics of the Malagasy . IL the genus Mantella (Anura, Ranidae, Mantelli- The genus Boophis (Rhacophoridae). - Bijd. nae): , distribution and conservation Dierk. 49: 261-312 of Malagasy poison frogs. - Alytes 17: 3-72

— & C. P. Blanc 1991. Amphibiens (premiere par- — , —, R. Jesu & G. Schimmenti 2000a. A new

tie). - Faune de Madagascar 75 (1): 1-379 species of Heterixalus (Amphibia: Hyperolii- Emanueli, L. & R. Jesu 1995. The herpetofauna of dae) from western Madagascar. - Afr. Zool. 35: the World Heritage Site "Tsingy de Bemaraha" 269-276 - - (Western Madagascar). Pp. 341-348 in Lloren- , J. Kosuch, S. Lötters, A. Widmer, K.-H. Jung- - te, G. et al. (eds.): Scientia Herpetologica. fer, J. Köhler, & M. Veith 2000b. Phylogeny and Societas Europaea Herpetologica, Barcelona Classification of poison frogs (Amphibia: Den- mitochondrial Felsenstein, J. 1985. Confidence limits on phyloge- drobatidae), based on 16S nies: an approach using the bootstrap. - Evolu- and 12S ribosomal RNA gene sequences. - tion 39: 783-791 Mol. Phyl. Evol. 14: 34-40 Glaw, F. & M. Vences 1994. A fieldguide to the amphibians and reptiles of Madagascar. 2nd edition, including mammals and freshwater Appendix fish. - Vences & Glaw Verlag, Köln, 480 p. — & - - 1997. New species of the Boophis tephrae- The taxon Rhacophorus doulioti, as described by An- omystax group (Anura: Ranidae: Rhacophor- gel (1934), is based on two specimens housed in inae) from arid western Madagascar. - Copeia the Museum National d'Histoire Naturelle, Paris 1997 572-578 (3): (MNHN): the lectotype (MNHN 1891.356) and the — , — & W. Böhme 1998. Systematic revision of paralectotype (MNHN 1891.357) (Glaw & Vences the genus Aglyptodactylus Boulenger, 1919 1997). Measurements of the lectotype (paralecto- (Amphibia: Ranidae), and analysis of its phylo- type in parentheses), taken with a caliper to the genetic relationships to other Madagascan ra- nearest 0.1 mm, are: snout-vent length 37.2 (31.3); nid genera (Tomopterna, Boophis, Mantidadylus, head width 12.5 (11.5); head length 14.1 (12.3); hor- and ManteUa).-]ouTn. Zool. Syst. Evol. Res. 36: izontal eye diameter 3.6 (3.5); eye-nostril distance 17-37 3.1 (2.8); nostril-snout tip distance 2.4 (2.1); nostril- Hawkins A. F. A., 1994. Isalo Faunal Inventory. nostril distance 3.4 (3.1); horizontal tympanum di- Final Report to the ANGAP. - Unpublished ameter 2.6 (2.3); hand length 10.8 (8.8); forelimb Report, Landeil Mills Limited length 20.6 (17.1); hindlimb length 56.4 (50.5); foot Lees, D. C. 1996. The Perinet effect? Diversity gradi- length 15.4 (13.8); foot length including tarsus 24.3 ents in an adaptive radiation of Madagascan (22.0); length of inner metatarsal tubercle 1.4 (1.4). butterflies (Satyrinae: Mycalesina) contrasted When the hindlimb is adpressed along the body, the with other species-rich taxa. Pp. 479- tibiotarsal articulation reaches between eye and 490 in Louren^o, W. R. (ed.): Actes du Col- nostril in both specimens. Webbing formula (ac- loque International Biogeographie de Mada- cording to Blommers-Schlösser & Blanc 1991) is 1(0), gascar. Societe de Biogeographie - Museum, 2i(l), 2e(0), 3i(l), 3e(0), 41(1.25), 4e(l), 5(0) in the Paris lectotype and 1(0.25), 21(1.25), 2e(0), 31(1.5), 3e(0), Lockhart, P. M. A. Steel, M. D. Hendy & D. Penny J., 41(1.5), 4e(1.25), 5(0.25) in the paralectotype. 1994. Recovering evolutionary trees under a more realistic model of sequence evolution. - Mol. Biol. Evol. 11: 605-612

84