Multiple Overseas Dispersal in Amphibians
Total Page:16
File Type:pdf, Size:1020Kb
Received 14March 2003 Accepted 9 July 2003 Publishedonline 2October 2003 Multipleoverseas dispersal in amphibians Miguel Vences 1* ,David R.Vieites 1,2, Frank Glaw3,HennerBrinkmann 1, Joachim Kosuch 4,Michael Veith 4 and AxelMeyer 1 1Departmentof Biology, University ofKonstanz, 78457 Konstanz, Germany 2Laboratoriode Anatom ´õ aAnimal,Departamento de Ecolox ´õ a e Biolox´õ aAnimal,Facultade de Ciencias Biolo´xicas, Universidade de Vigo,Buzo ´n137,36201 Vigo, Spain 3ZoologischeStaatssammlung, Mu ¨nchhausenstrasse21, 81247 Mu ¨nchen, Germany 4Institut fu¨rZoologie,Universita ¨tMainz, Saarstrasse21, 55099 Mainz, Germany Amphibians are thought tobe unable to disperse over oceanbarriers becausethey donot tolerate the osmoticstress of salt water.Their distribution patternshave thereforegenerally beenexplained byvicari- ancebiogeography. Here,we present compelling evidencefor overseasdispersal offrogs in theIndian Oceanregion basedon the discovery oftwo endemic species on Mayotte. This island belongsto the Comoroarchipelago, which is entirely volcanic andsurrounded by seadepths of more than 3500 m.This constitutesthe first observation ofendemic amphibians onoceanic islands that didnot have any past physical contactto other land masses.The twospecies of frogs had previously beenthought tobe non- endemicand introduced from Madagascar, butclearly representnew species based on their morphological andgenetic differentiation. They belong tothe genera Mantidactylus and Boophis in thefamily Mantellidae that is otherwiserestricted to Madagascar, andare distinguishedby morphology andmitochondrial and nuclearDNA sequencesfrom mantellid speciesoccurring in Madagascar. This discovery permits usto updateand test molecular clocksfor frogs distributedin this region. The newcalibrations are in agreement with previous rate estimatesand indicate two further Cenozoictransmarine dispersal eventsthat had previously beeninterpreted as vicariance: hyperoliid frogs from Africa toMadagascar ( Heterixalus ) and from Madagascar tothe Seychelles islands ( Tachycnemis ).Our resultsprovide thestrongest evidence so far that overseasdispersal ofamphibians existsand is norare exception,although vicariance certainly retains muchof its importance in explaining amphibian biogeography. Keywords: Amphibia; Mantellidae;Madagascar; Comoros;phylogeny; biogeography 1. INTRODUCTION andPacific regions.Some of theselandmasses were prob- ably belowsea-level in thepast, and their overseascoloniz- Amphibians are akey group in historical biogeography ation following emergencehas beenhypothesized (Hedges becausethey are oftenthought tobe unable to disperse et al. 1992; Hedges1999). However,all ofthese islands over saltwater barriers (Duellman &Trueb 1986; Meirte are made upat least partly by continentalfragments for 1999; Bossuyt& Milinkovitch 2001; Inger &Voris 2001; which past land connectionscannot be excluded (Heaney Brown& Guttman2002). They are well knownto be 1985; Whitmore 1987; Crother &Guyer1996). Geologi- extremely sensitiveto osmotic stress and do not survive in cal data are seldomsufficiently definitive toascertain the salt water,although somespecies of frogs tolerate or par- full submersionof a landmass,and small emerging tially inhabit brackish water (Balinsky 1981). Therefore, remains wouldbe sufficent to harbour relict amphibian amphibians are consideredto be excellent models for vica- populations. riance scenariosas explanation for general biogeographic Adifferentsituation is that ofoceanic islands that never patterns,and major biogeographic hypotheseshave been had physical contactto other landmasses.These are of influencedby theoccurrence of endemic amphibians on completevolcanic origin or werebuilt by coral reefs.They islandsor continents(Duellman &Trueb 1986; Rich- are surroundedby deepwaters that make land connec- ards& Moore1996; Worthy et al. 1999; Bossuyt& tionsthrough fluctuating sealevels impossible, andthey Milinkovitch 2001; Brown& Guttman2002). One are mostly tooyoung toassume vanished connections to important argument for suchinterpretations, ever since drifting continents.No endemicamphibian speciesare Darwin (1859), has beenthat heretoforth noendemic knownfrom truely oceanicislands. Some are populatedby amphibians wereknown from oceanicislands. By contrast, non-endemicfrogs orsalamanders,but these are knownor reptiles are presenton many islandsand some are known assumedto have beenintroduced. Such has beenthought tobe excellent over-water dispersers(Censky et al. 1998; tobe the case for Mayotte,an island belonging tothevol- Arnold 2000; Schoener et al. 2001). canicComoro archipelago in theIndian Ocean,located Amphibians are widespreadon many archipelagos, for betweenAfrica andMadagascar. Mayotteis separated instanceon the Philippines andin theSunda, Caribbean from Madagascar by ageographical distanceof 300 km andby seadepths of more than 3600 m,and its origin datesback nofurther than 10–15 Myr ago (Emerick & *Authorand present address for correspondence: Institutefor Biodiversity and Ecosystem Dynamics, Mauritskade 61,1092 AD Amsterdam, The Duncan1982; Nougier et al. 1986). The twofrog species Netherlands ([email protected]). knownfrom Mayotteare seenas conspecific with taxa Proc.R. Soc.Lond. B (2003) 270, 2435–2442 2435 Ó 2003 TheRoyal Society DOI10.1098/ rspb.2003.2516 2436M. Vences and others Multiple overseasdispersal in amphibians from Madagascar andof allochthonousorigin (Blommers- threemitochondrial (12S and 16SrRNA, tRNA Val) gene Schlo¨sser& Blanc 1991; Meirte1999). fragments fromrepresentatives of allmantellid genera, The evolutionand biogeography ofthe highly diverse, subgenera and speciesgroups to resolvethe relationships butstrongly endangeredanimal diversity ofMadagascar within the family;representatives of two other ranoidfam- andother islandsin thewestern Indian Oceanhave been ilies(Rhacophoridae: Polypedates ; Ranidae: Rana) were subjectto intense debates in recentyears (Krause et al. usedas the outgroup. 1997; Murphy &Collier 1997; Jansa et al. 1999; (iii)We chose members of majorclades of ranoidfrogs Bossuyt& Milinkovitch 2001; Farias et al. 2001; (Bossuyt &Milinkovitch2000), including previously Meegaskumbura et al. 2002; Raxworthy et al. 2002). unstudiedAfrican taxa, and of other familiesthat could Hypotheseson thetime ofthe origin ofthesefaunas must beinformative regarding biogeographic relationshipsin the largely rely onphylogenies ofextant taxa becauseno ter- IndianOcean region: to resolvethe relationshipsamong restrial or freshwaterfossils are knownfrom theTertiary these deepclades, we analyseda morecomprehensive period (65–2Myr ago) ofMadagascar (Krause et al. dataset includingtwo nuclear(rhodopsin, tyrosinase) and 1997). Deepvicariance has oftenbeen invoked to explain four mitochondrial(12S and 16SrRNA, tRNA Val, cyto- theorigin ofMadagascar ’sendemicvertebrates chrome b)genefragments. Asalamanderand representa- (Duellman &Trueb 1986; Richards &Moore1996; tivesof archaicfrogs (familiesDiscoglossidae and Pipidae) Murphy &Collier 1997; Bossuyt& Milinkovitch 2001; wereused as hierarchicaloutgroups. Farias et al. 2001): their ancestorssupposedly evolved in isolation after thebreakup ofthe southern supercontinent Gondwana.During this geological process,Madagascar (b) DNA extraction, amplification andsequencing had beenseparated from other landmassesin theJurassic DNA was extracted fromtissue samples preserved in ethanol andCretaceous (Briggs 2003). Recentphylogenies ofcha- and sequencedon ABI3100 and ABI377 automated meleonsand rodents ( Jansa et al. 1999; Raxworthy et al. sequencersafter directamplification using primersfrom pre- 2002), however,proposed area cladograms that are notin viousstudies (Palumbi et al. 1991;Bossuyt &Milinkovitch accordancewith thesuccession of plate tectonicalevents. 2000;Vences et al. 2003)or that weredeveloped for this work Dispersal scenariostherefore seem plausible for these (sequencesin a 5 9–39 directiongiven only for new primers;F, groups buthave notbeen considered for amphibians, forward primers;R, reverse primers). Cytochrome b (up to which in thewestern Indian Oceanregion are mostly rep- 1016bp): CBJ10933(F); Cytb-a (F);MVZ15L-mod (F) —AAC resentedby frogs.Caecilians occuron the Seychelles and TWA TGG CCCMCA CMA TMC GWA A; Cytb-c (R); continentalAfrica andAsia, whereas salamanders are CytbAR-H-mod (R) —TAW ARGGRT CYTCKA CTG GTT completely absent.Except for theenigmatic Seychellean G.Tyrosinase (exon1; 632 bp): Tyr-1b (F);Tyr-1d (F);Tyr- sooglossids,all anuransfrom theSeychelles and Madagas- 1a(F); Tyr-F40 (F) —AARGAR TGY TGY CCIGTI TGG; car are includedin thesuperfamily Ranoidea,a highly Tyr-Fx3 (F)—ACTGGC CCA YTG THT TYT ACAAC; diversegroup oflargely unsolvedsystematics (Duellman & Tyr-Fx4 (F)—YTG GCCYWY TGT NTT YTA YAAC; Tyr- Trueb 1986; Feller &Hedges1998; Vences& Glaw 1g (R);Tyr-1e (R);Tyr-SPA (R) —GAIGAG AAR AAR GAI 2001). GCTGGG CT. Rhodopsin (exon1; 334 bp): Rhod-ma (F) — Here,we report onour recent discovery that theCom- AACGGA ACA GAA GGY CC; Rhod-1a (F);Rhod-md (R) — oro frogs representpreviously undescribedspecies GTA GCGAAG AAR CCT TC; Rhod-1d (R);Rhod-1c (R). endemicto Mayotte. We use mitochondrial andnuclear 12SrRNA and tRNA Val (ca.700bp): 12SA-L (F);12SB-H (R); DNA sequencesto demonstrate the close phylogenetic 16SR3(R). 16S rRNA (5 9 fragment; ca.650bp): 16S-L3(F); relationships ofthese species to the endemic Malagasy 16SA-H(R); 16S rRNA (3 9 fragment; ca.550bp): 16SA-L (F); radiation ofmantellid frogs,thereby providing evidence 16SB-H(R). The moleculardataset was