Odonatologica37(3):235-245 September1.2008 TheFijian Nesobasis: a furtherexamination of species diversity andabundance (Zygoptera: Coenagrionidae) H.Van Gossum¹’*,C.D. Beatty²,M.Tokota’a³andT.N.Sherrat4t 1EvolutionaryEcologyGroup,University ofAntwerp,Groenenborgerlaan171, B-2020Antwerp,Belgium 2GrupodeEcologiaEvolutiva,DepartaraentodeEcologiayBiologiaAnimal, Universidad deVigo,EUETForestal,CampusUniversitario,ES-36005 Pontevedra,Galicia,Spain 3International Conservation,South Pacific Program,11Ma’afuStreet,Suva,Fiji 4DepartmentofBiology,Carleton University, 1125 Colonel ByDrive,Ottawa, Ontario,K1S5B6, Canada ReceivedJune11,2007/ReviewedandAcceptedOctober21, 2007 Recently,anoverview ofthediversity,abundance,distribution andmorphological characteristics ofspp. ofthegenusNesobasis,endemictoFiji,waspresentedforspp. occurringonthe2largestislands ofthearchipelago:VitiLevuandVanua Levu.Here, thisknowledgeisextendedbyprovidingmoreextensivediversityandabundance data fortheisland ofVanua Levu,aswellasfor4smallerislands inFiji: Taveuni, Koro, Ovalauand Kadavu. Previous research indicated that the Nesobasis spp.inhabiting VitiLevu andVanuaLevu areunique,withthese islands havingnospeciesin com- mon.Thenewdataconfirmthisproposalandalso show thatsmallerislandsinprox- imity tothese 2largerislands usuallycontainasubset ofthelargeisland’s Nesobasis fauna. The island ofKoro, however,isunusual in that,while itsNesobasis spp.are predominantlythose foundonVanuaLevu,it alsoharbours N.rufostigma,asp.oc- curringonVitiLevu.Further, N.recavaisendemic toKadavu and isnotfoundon VitiLevu,the nearestlargeisland. Speciesrichness ishigheronlargethan smallis- landswhile meanspeciesabundances wereconsistentlyhigheronlargeislandscom- paredtosmallislands. Thepatternofdistribution andspeciationinthisgenusisquite complex, andisthesubject ofongoingresearch. INTRODUCTION The study of oceanic islands offers an excellentopportunity to explore the process andpattern ofselectionas wellas thenature of speciation (DARWIN, *Correspondingauthor:[email protected] 236 H. vanGossum, C.D.Beatty,M.Tokota'a&T.N.Sherratt 1859; MacARTHUR & WILSON, 1967). Sadly, in many island groups, par- ticularly thosein the Southwest Pacific, our knowledge ofbasic biodiversity is limited. Moreover,thereis afundamentallack ofinformationon thebehaviour, lifehistoryandecology ofendemicislandfauna(WHITTAKER, 1998). Inthis context BEATTYetal.(2007) reported informationonthe diversity,abundance andbasic morphology ofspecies of thegenusNesobasis(Odonata: Zygoptera). Notonly is thegenusNesohasisendemictoFiji, itisalso species-rich, withspe- cies distributedoverseveral islandswithinFiji (DONNELLY, 1990). Moreover, for someofthe species withinthe genus,malesare seldomobservedat oviposi- tionsites(VAN GOSSUM etah, 2007). Thismalerarity hasnot causedsex role reversal as previously speculated (DONNELLY, 1990, 1994)butresearch is on- going to explore whatfactorsmayexplain malerarity in somespecies ofNeso- basisbut not inothers (seealso VANGOSSUM etal.,2007). Inour previous report, based on research conductedin 2005 (BEATTY et al.,2007), wefocussed onNesohasis observedon Viti Levuand partsofVanua Levu. Duringourresearch in2006wewereableto exploreotherregions onVan- ua Levuandalso to include informationonspecies diversityandabundancefor four additionalsmaller islands: Kadavu, Koro,Ovalau and Taveuni.For these fiveislands weprovide quantitative estimatesof Nesohasis species diversityand abundance.Furthermore,as BEATTY etal.(2007) didforotherspecies, wepro- videbasic morphological characteristicsforfieldidentificationforone described species (N. recava) andfora newspecies weencountered(Uds3). We also make note ofany reproductive behaviour(tandem,copulation wheel,andoviposition) observed forthesespecies. In thispaperwe characterisetheinter-island distributionofspeciesof Nesoha- sis basedprimarily onour ownobservational data,butalso onearlierpublished work (DONNELLY, 1990;BEATTY etal„ 2007). Theislands ofTaveuniand Koroare locatedclosertoVanuaLevu(the secondlargest islandinthearchipel- ago) than Viti Levu (the largest island inthe archipelago), while theislands of Ovalau andKadavu are located nearerto Viti Levu (Fig. 1). Ifsmallerislands sharespecies with nearby largerislands, weexpected the Nesohasis community on Taveuniand Koro tobea subsetofspecies occurring onVanuaLevu, while OvalauandKadavu species were expected to occuralso on Viti Levu. Many ecological factors can influencewhethera disperserfromthemainland establishes onanisland. However, abundantspecies are morelikely to produce higherabsolutenumbersofdispersers thanrarespecies(MacARTHUR &WIL- SON, 1967;GASTONetal.,2006). Therefore,if dispersal played an important role in determining establishmentsuccess, onemight expect that thosespecies foundonsmallerislandswouldcomprise asubset ofthemoreabundantspecies onthelarge island. Speciesdiversity and abundance inNesohasis 237 METHODS ThedamselflygenusNesobasisisabundant andwidespreadintheFijianislands.Theyarethedomi- nantZygopterainmoststreamhabitats,particularlyinmountainstreams,buttheyarenotfound in largeriversorponds(DONNELLY, 1990,1994).We usedtopographicmaps(1:50,000)in combina- tion withourpriorexperiencewith thegenus(BEATTYetal.,2007)toselect regionsonislandsthat werelikelytocontain appropriateNesobasis habitat. Inaddition,T.W. Donnellyand D.Polhemus communicated theirearlierexperienceswithNesobasis onthedifferent islands. TheFijiIslandsarelocated intheeasternMelanesian regionoftheSouth-westernPacificandhave awarm,humidtropicalmaritimeclimatewithmeanmonthlytemperaturesfrom23°Cin Julyto27°C in January(RYAN,2000).Theislands aresubequatorialandexposedtoheavierrainfall betweenNo- vemberandApril,especiallyonthelowislandsandtheleewardsidesofthelargeislands(EVENHUIS &BICKEL,2005).OurstudywasconductedduringthedryseasoninFiji fromAugust6toOctober4, 2006,atimeofyearsimilartothetimingofresearchsummarisedinBEATTYet al.(2007).Fijiconsists ofseveralhundred islands andoccupiesatotalareaofapproximately650,000km2 ofwhichtheland areaislessthan 3percent.There aretwolargeislands in thearchipelago(Fig.1):Viti Levu(10,388 km2)and Vanua Levu(5,535km2) (seealso EVENFIU1S& BICKEL,2005).Viti Levuand Vanua Levuarerugged,withlandformsincludingvolcanicplugs,erodedcalderas,deepgorges,andravines carvedbymountain streams,flat-bottomedvalleyswithextensivefood plains,andmangrovedominat- ed deltas(EVENHUIS& BICKEL,2005).Otherislandsinthearchipelagoareconsiderablysmaller, manyhavinglesspronouncedtopographyandlackingstreams.However,several islandsaresizable and predominantlyvolcanic in origin,havingsignificantel- evations and most impor- tantly mountain streams that could hold species of Nesobasis.Ofthesesmaller islandswesampledKadavu (408km2),Koro (109km2), Ovalau(102km2)andTave- uni (434km2). Study sites that com- pliedtoourselectioncrite- ria (see further) forVAN- UA LEVU(seeFig. 1)in- clude: KR-01 (I6°30.79’S, 179°32.70’ E, alt.265m),a largestreamonthe interi- orroad above Korotesere, crossingapproximately 1.2 kmsouthofSeavaciRd,we sampledupstreamfromthe road; KR-02 (16°28.55’S, 179°33.32’ E, alt. 170m), second stream above Ko- Fig.I. Mapofsixislands ofFiji forwhich dataarepresented,showing rotesere, we sampled on samplelocations. ForVANUA LEVU: (1)KR-01;- (2)KR-02; - thedownstreamside ofthe (3)NagaciCreek;— (4)VunigautulailaiCreek;— (5)WR-01.ForTA- road crossing;NagaciCre- VEUN1:(6)Somosomo;—(7)TV-02.ForKORO:(8)BuretiniCreek; ek(16°28.36’S,179°35.32’ - (9)K-01; - (10)K-02.ForOVALAU:(11)Lovoni.ForKADAVU: E, alt. 130m),smallstream (12)VuniseaEast.Sitedescriptionsandlatitude/longitudecoordinates crossing road near small areprovidedinthetext. 238 H.vanGossum, C.D.Beatty,MTokota'a&T.N.Sherratt settlement,wesampleddownstream oftheroad; VunigautulailaiCreek (16°30.06’S, 179°37.20’ E, alt. 110m), streamatbridgeonroad aboveVanuavou,wesampleddownstreamoftheroad;WR-01 (16°41.12’S, 178°55.29’E,alt.335m),riverathighelevation inwesternVanuaLevu.For TAVEUNI (seeFig. 1)wesampled;Somosomo(16°46.40’S, 179°58.22’W, alt. 30m),alargeriveratthesouth- ernedgeofSomosomovillage,wesampledupstream ofthebridge;TV-02 (16°44.12’S, I79°54.57’ W, alt.395m),small streamabove Somosomoinagriculturalarea.Wealso sampledatTavoroFalls, in BoumaNational HeritagePark,asite thatwasindicated tobe verygood Nesobasis habitat (D. Polhemus,pers. com.). Despitesearchinganhour with two persons wewere unable tolocate any Nesobasis. ForKORO(seeFig. l)wesampledBuretini Creek(I7°15.82’S,179°22,08’E,alt. 150m), astreamwith verylowflow, andpredominantlyexposedbedrock withboulders and numeroussmall pools,wesampledimmediatelyabovealargewaterfall,workingupstreamtotheroad crossing;K-01 (I7°15.82’S, 179°22.08’ E,alt. 40m),astreamalongtheroad crossingthe interior oftheisland in thenorth;K-02 (17°19.82’S, 179°25.92’E,alt. 50m), alargestreamontheeastside ofKoro,which crossesthemain road southoftheintersectionwith theinterior roadacrossthenorth sideoftheis- land.ForOVALAU(seeFig. I)wesampledLovoni (17°41.20’S,I78°47.38’ E,alt. 85m),arivernear Lovoni village,high inthe caldera of theisland’s volcano. For KADAVU (seeFig. 1)wesampled VuniseaEast(19°01.93’S, 178°10.79’ E,alt. 140m),alargestreamonthemain roadeast ofVunisea, samplingupstream ofthebridgeinaregionwithmuchexposedbedrock,boulders andwaterfalls. Oursamplingmethodconsisted ofnettingallobserved damselfliesatasite,identifyingeveryindi- vidualbelongingtothegenusNesobasis.Wedecidedonlytoincludesitesforwhichsamplinglasteda minimum ofonehourwithaminimum of2peoplesearching.Whileaone-hour sampleperiodmay notbelongenoughtodetectallspeciespresentatasite(forexample,those that areveryrareorthat may bepresentatthewateratdifferent timesofthedaythan thosesampled), this level ofsampling effortallowed ustocharacterize thecommonspeciesataparticularsiteeffectively.Wehave alsocho- sentoexclude sitesforwhich wesampledforonehouror greater,but whereoursampleeffortswere focussed onthecaptureofparticularspeciesforourresearch; in these casessamplingeffortwould not provideanunbiased sampleofspecies abundances atasite. For eachofthe sitesthat metour criteria werecorded date,latitude and longitude,streamwidth and lengthofthe streamsampled, and theapproximatesamplingduration (for2searchers). This information allowed ustocalculate coarseabundance estimates; streamwidth and lengthwereused tocalculate areasampled(number ofindividuals per unitofarea),while samplingtimeprovidedanalternativemeasureforcalculating species abundances (numberofindividuals perunitoftime)(seealso BEATTYetal.,2007). Analysis ofspecies distributions amongislands utilized aMann-WhitneyU Test(SPSS 13.0)to determinewhether speciesoccurring bothonthelargeisland andoneormoreofthe smallerislands differed inmeanabundances fromthose onlyoccurring onthelargeisland. SPECIES DESCRIPTIONS Thefollowing basicdescriptionsprovide characteristicsweusedinthefieldfor species identification.In some instancesidentificationofspecies required mag- nifying lenses(20X magnification) or a field-microscope, as a close lookat the exactstructureofgenitaliaand/ormesostigmallaminaewas required(seeDON- NELLY, 1990). Moredetaileddescriptions canbefoundin TILLYARD(1924), DONNELLY(1990) and will appearin Donnelly unpubl. ms. With respect to thelatterandas agreed uponwithT.W.Donnelly, inwhatfollowsweuse thefirst letter(s) of thespecies namesDonnelly will be using inhis forthcoming manu- script toallowconvenientcomparison (seealso BEATTYetal., 2007). Wefound onespecies newto science during our samplingin 2006,whichwerefertoas un- Species diversityand abundance in Nesobasis 239 describedspecies 3(Uds 3).Thisallowsfordistinctionbetweenthis species and two species {Uds 1 and Uds2) discussedinBEATTY etal.(2007). N. recava(DONNELLY, 1990):Kadavu.Thisis aslender, medium-sizeddam- selfly,endemicto theisland. Thethoraxofmalesisbright bluelaterally,dorsal- ly black, while femalesare morepale blueorbrownonthe lateralthorax. Fur- ther,males are predominantly blueonthe dorsumofabdominalsegmentsnine andten withbluecolourextending laterally, whilefemaleshavesomepaleblue/ brownonthedorsumofabdominalsegmentten and limited colorationon seg- TableI Speciesdiversity andabundance fornewlocationsvisitedonVanuaLevuin2006.Totalareasampled andduration ofsamplingaregiveninitalics LLooccaalliittyyaannddddaattee SSppeecciieess NNuummbbeerr ooff DDeennssiittyy DDeennssiittyy iinnddiivviidduuaallss ((iinnddiivviidduuaallss//110000mm22)) ((iinnddiivviidduuaallss//hh)) KK..RR--0011 ((2277//0099)) 22225500mm22 110055mmiinn NA.taall 2299 11..33 1166..66 NN..aauu 11 ~~00 00,,66 NN.. bbrraacchhyycceerrccaa 7788 33..55 4444..66 KN.ff 22 00..11 11..11 NN.. I1 2255 11..11 1144..33 NN..rr 55 00..22 22..99 UUddss33 33 00..11 11..77 KKRR--0022((2277//0099)) 11440000mm22 6600mmiinn NN.aal! 4422 33 4422 NN..bbrraacchhyycceerrccaa 4422 33 4422 NN. cc I1I1 00..88 1111 NN.. I1 7755 55..44 7755 NNaaggaacciiCCrreeeekk((1177//0099)) 9900mm22 6600mmiinn NN..aall 22 22..22 22 NN..aauu 22 22..22 22 NN..bbrraacchhyycceerrccaa 1100 I11I.I1 1100 NN..ff 22 22..22 22 NN..l1 44 44..44 44 NN.. vv 11 11..11 11 VVaanniiggaauuttuullaaiillaaiiCCrrkk(f1177//0099)) 770000mm22 6600mmiinn NN..l1 33 00..44 33 WWRR--0011((2266//0099)) 11550000mm22 l1O0S5mmiinn NN..aall 2255 11..77 1144..33 NN..aauu 33 22 11..77 NN..bbrraacchhyycceerrccaa 110055 77 6600 NN..ff 11 00.,11 00..66 NN. I/ 5599 33..99 3333..77 NN.rr 11 00,.11 00..66 240 H.vanGossum,CD. Beatty,M.Tokota'a&T.N.Sherratt ment nine.Thepterostigma ofbothsexes is darkbrown. N. recava is somewhat similarto N. selysionVitiLevu, though showing onaveragemoreextensiveblue colorationon thoraxandabdomenthan inN. selysi(BEATTYetal„ 2007). We didnot observeany matingassociationsorovipositions. Uds 3:Vanua Leva. Thisis a very large damselfly, withextensive blue-green colorationonthoraxandabdomen.Themale iseasily distinguished by thepter- ostigma onitsforewings, whichareenlarged andhaveapearlescent pinkcolora- tion.Femalesare generally blue-greenincolour, somewhatsimilarto femalesof N. brachycerca, butlarger. Maleswere observed displaying to one another, but notandemsoregg-laying wereobserved. RESULTSAND DISCUSSION In totalwesampled 27sites onfiveislands, collecting 886specimens ofNeso- basis, with 12 sites meeting our criteria for inclusion(see Methods): 5 sites for VanuaLevu(Tab. I), 2 forTaveuni(Tab. II),3 forKoro (Tab. Ill), 1 forOvalau (Tab. IV)and 1 for Kadavu(Tab. V).Sitesthatdidnot meet ourselection criteria containednospecies thatdidnot occurinnearby(withinisland) sites.Therefore, our interpretation ofspecies distributionandabundancewouldnot bechanged ifthesesites were included. Fig. 2.VenndiagramofNesohasisspecies distributionsamongtheislands ofVitiLevu,VanuaLevu, Ovalau,Kadavu, Koro and Taveuni. Species distributions arebased onBEATTYet at,2007and present data,except forthefollowing;Nesobasisaurantiaca, N. caerulescens, N.flavostigma,N. ingens, N. leveriand N.pedataonViti Levu,N. campioniandN. monticolaonOvalau(DONNELLY, 1990), and N. tonTaveuni (J.H. Skevington,pers. com.). Speciesdiversity andabundance inNesobasis 241 Weencountereda totalof9species fromthe 'VanuaLevu' group,whichwere distributedoverVanuaLevu,Taveuniand Koro(see Fig. 2 forarepresentation ofknown species distributions). All species foundon Taveuniwere also found onVanuaLevu. ForKoro, thiswas notthecase: whereasmostspecies foundon Koroarealso found onVanuaLevu,we additionally encounteredN. rufostigma, aspecies occurring onViti Levu, butnotVanuaLevu(Tab. III).We didnot find N. rufostigma to beabundanton Koro (only two specimens werecollectedat a single location). ThefactthatKorois locatedata middledistance betweenViti LevuandVanuaLevumayhelp toexplain thepresenceofspecies frombothViti LevuandVanuaLevuthere.As notedearlier, VitiLevuandVanuaLevu donot share species of Nesobasis(BEATTY etal., 2007andresults here included); at present Korois theonly islandwhereNesohasisspecies of thetwo large islands areknownto co-exist. We encountered6species from the Viti Levu' group,foundon Ovalauand Kadavu, (as wellas onKoro) (Fig. 2).All species encounteredonOvalau were also foundonViti Levu.Species werecordedonKadavuwerealso foundonViti Levu, except forN. recava., whichisendemictoKadavu (seealso DONNELLY, 1990). Our findings for Kadavu are similar to DONNELLY (1990), where the samethree species were found.For Ovalau threeof thefivespecies also found by DONNELLY(1990) were includedin our sample, whilewe didnot observe N. campionior N. monticola.Flowever, we didfind Uds I onOvalau, whichhad not previously been described for this island(Tab. IV). For Taveuni and Koro, nopreviously published species accounts exist forcomparison. SpeciesrichnessatsiteswithinTaveunivariedbetween2and3species, forKoro itvariedbetween3and4species, forOvalau4species andforKadavu 3species. Forboth VitiLevuandVanuaLevuthemostspeciose locationsharboured8spe- ciesofNesobasis(BEATTY etal.,2007). Maximalspecies richness persitethus appearshigher onthelarge islandscompared to thesmallislands. Table II Species diversity and abundance forlocations visited onTaveuni in 2006. Total areasampledand durationofsamplingaregiven initalics Localityanddate Species Number of Density Density individuals (individuals/!00m2) (individuals/h) Somosomo(3/10) 5500m2 60min N. au 1 -0 1 N. 1 17 0.3 17 TV-02(3/10) 600m! 60min N. hrachycerca 4 0.7 4 N. 1 1 0.2 1 N. v 3 0.5 3 242 H.vanGossum, C.D.Beatty,M.Tokota'a&T.N.Sherratt Table III Species diversity and abundance for locations visited onKoro 2006. Total area sampledand duration ofsamplingaregivenin italics LLooccaalliittyyaannddddaattee SSppeecciieess NNuummbbeerr ooff DDeennssiittyy DDeennssiittyy iinnddiivviidduuaallss ((iinnddiivviidduuaallss//I1O0O0mm22)) ((iinnddiivviidduuaallss//hh)) BBuurreettiinnii CCrreeeekk((77//0099)) i1m00)0mm22 6600mmiinn NA..bbrraacchhyycceerrccaa 44 00..44 44 NA.tf/ I1 00..11 11 NA.trruuffoossttiiggmmaa 22 00..22 22 NA..vv 44 00..44 44 KK.--0011 ((77//0099)) 220000mm22 6600mmiinn NA.tbbrraacchhyycceerrccaa 66 33 66 NAt./f 66 33 66 NA.vv 88 44 88 KK--0022 ((88//0099)) 44000000mm22 II4400mmiinn NA..bbrraacchhyycceerrccaa 11 ~~00 00..44 NAt./f 1133 00..33 55..66 NA.vv 11 ~~00 00..44 TableIV Species diversityand abundance forlocations visited onOvalau in 2006. Total area sampledand duration ofsamplingaregiveninitalics LLooccaalliittyy aannddddaattee SSppeecciieess NNuummbbeerr ooff DDeennssiittyy DDeennssiittyy iinnddiivviidduuaallss ((iinnddiivviidduuaallss//I1O0O0mm22)) ((iinnddiivviidduuaallss//hh)) LLoovvoonnii ((1155//0088)) 1111000000mm-2’ I13355mminin NA. hheetleerroonneeuurraa 4422 00..44 1188..77 UUddss /I 2 ~-00 00..99 NA. rruuffoosstliiggmmaa 2233 00..22 1100..22 NN.. sseellyyssii 88 00..11 33..66 TableV Species diversityand abundance forlocations visited onKadavu in 2006. Total areasampledand duration ofsamplingaregivenin italics LLooccaalliittyyaannddddaattee SSppeecciieess NNuummbbeerr ooff DDeennssiittyy DDeennssiittyy iinnddiivviidduuaallss ((iinnddiivviidduuaallss//I1O0O0mm22)) ((iinnddiivviidduuaallss//hh)) VVuunniisseeaaEEaasstt((2222//0099)) 33550000mm11 I10055mmimn NN.. llounnggiisstiyyllaa 44 00..11 22..33 NN..rreeccaavma 6655 11..99 3377..11 NN..rruuffoosattiiggmmaa 66 00..22 33..44 Speciesdiversityandabundance in Nesobasis 243 Fig. 3.Species/arearelationship(log/log)forspeciesofNesobasisontheislands ofViti Levu,Vanua Levu,Ovalau,Kadavu,Taveuniand Koro. An overallpositive correlation between islandsize and speciesrichnessisobserved. Comparing totalknownspecies diversityperisland(Fig. 2),thetwolarge islands, Viti LevuandVanuaLevacurrently have20 and11 knownspecies, respectively. Thetotalspecies diversityofthe foursmall islandsis: 4forKoro, 5 forTaveuni, 6forOvalauand3 forKadavu (DONNELLY, 1990;BEATTYetal., 2007; J.H. Skevington, pers.com.anddataherein).Considering theirsmallersizeandrela- tive isolationfrom VitiLevu andVanuaLevu(see Fig. 1) the lowerspecies di- versity of thesefourislands isnot surprising (e.g. MacARTHUR& WILSON, 1967).Aspecies/area curve (Fig. 3)comparing islandsize versus species number (log/log transformations) shows an overall positive correlationbetween island sizeandspecies richness (R =0.833,df= 4, P=0.039). Itshouldbe notedthat Kadavu, withits greatdistancefromViti Levucompared to Ovalau, isthe only smaller islandhavinganendemic, N.recava. Species relationships andthepattern ofspeciation withinandbetweenislandsare thesubjectsof current phylogenetic research, Apositive relationship between species abundanceandspecies range size/oc- cupancyhasbeenidentifiedinarangeoffloraandfauna(GASTON etal., 1997; GASTON etal.,2000). Thisrelationship hasalsobeenfoundto holdforspecies assemblages distributedon andamong islands(GASTON etal.,2006). On the large islandsin Fiji,Viti LevuandVanuaLevu,this general pattern alsoappears to holdfor species of Nesobasis, withrelativelyabundantspecies also being en- counteredat alarger numberofsites (BEATTY etal., 2007 and data herein). Betweenislands, thispatternis lessclear. Wecompared themeanabundancesof large islandspecies foundonadjoining small islandsto large islandspecies not foundonsmallislands. Whenconsidering combineddatafrom Viti Levu versus Ovalauand Kadavu, and VanuaLevu versus Taveuniand Koro(all abundance datainBEATTYetal.2007(tabs 1&2,pages22-23,25)andallabundancedata 244 H.vanGossum, C.D.Beatty,M.Tokota'a&T.N.Sherratt fromVanuaLevu(Tab, I) herein), therewas littleevidencethatspecies foundon oneofthelargeislandsandatleastonesmallislandhadhighermedianabundance thanthosespecies foundonly on thelarge islands.(Mann-Whitney U = 62.0,N - 25, P=0.406) (Fig.4).Whilequestions couldberaisedaboutthepowerofthis test, itsupportsthe general pattern seen in our species distributiondata;while some“common”species onthelarge islandsare likely tobefoundonsmalleris- lands, species such as N. monticola, rare onViti Levubut foundonOvalau, and N.f,rare onVanuaLevubutlocally abundanton Koro, suggest thatthis is not ahardandfastruleforNesobasis. Thedatapresented here greatly expand our knowledge on Nesohasis species distributionandabundance, particularly forTaveuni, Koro, Ovalau and Kada- vu, fourislandsforwhichdistributionand abundancedatawerenot previously published. As indicatedin BEATTYet al.(2007) the damselfly faunaofFiji re- mainsrelatively unexplored. Indeed,in total, threespecies previously unknown to scienceresultedfrom our sampling(BEATTY etal.,2007anddataherein). It Fig. 4. Box and whisker plot ofmeanspecies abundances (individuals/lOOm2).Values arefor (a) populationsofspeciesfound onlyononeofthe twolargeislands ofVitiLevu orVanua Levu(large island only), (b)largeisland populationsofspeciesfoundonthelargeislands andatleastonesmall island,(c) smallislandpopulations.Thedarklinewithin thebox represents the median,whiletheup- per and lowerboundaries ofthe boxrepresentthe 75lhand 25"'percentiles,respectively. Theextent ofthe upperwhisker representsthe90,hpercentile, while theextent ofthe lowerwhisker represents the 10thpercentile.Dotsrepresent anyobservations outsideofthese ranges. Anon-significanttrend isobserved wherein speciesfoundon the largeislands and atleast oneadjoiningsmallisland have higherabundances inlarge-islandpopulationsthanthosefoundonlyonthelargeisland.Smallisland populationshaveconsistentlylowerabundances than largerisland populations.