ebook img

Distinctive Phylogenetic Patterns of Cone Snails from Two Ancient Oceanic Archipelagos PDF

17 Pages·2014·1.01 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Distinctive Phylogenetic Patterns of Cone Snails from Two Ancient Oceanic Archipelagos

Copyeditedby:PN MANUSCRIPTCATEGORY: Article Systematic Biology Advance Access published September 18, 2014 Syst.Biol.0(0):1–17,2014 ©TheAuthor(s)2014.PublishedbyOxfordUniversityPress,onbehalfoftheSocietyofSystematicBiologists.Allrightsreserved. ForPermissions,pleaseemail:[email protected] DOI:10.1093/sysbio/syu059 EvolutionataDifferentPace:DistinctivePhylogeneticPatternsofConeSnailsfromTwo AncientOceanicArchipelagos REGINAL.CUNHA1,∗,FERNANDOP.LIMA2,MANUELJ.TENORIO3,ANAA.RAMOS1,RITACASTILHO1,AND SUZANNET.WILLIAMS4 1CentreofMarineSciences-CCMAR,UniversidadedoAlgarve,CampusdeGambelas,8005-139Faro,Portugal,2CIBIO,CentrodeInvestigaçãoem BiodiversidadeeRecursosGenéticos,UniversidadedoPorto,CampusAgráriodeVairão4485-661Vairão,Portugal,3DepartmentCMIMyQ. Inorgánica-INBIO,FacultaddeCiencias,TorreNorte,1ªPlanta,UniversidaddeCadiz;11510PuertoReal;Cádiz,Spain;and4DepartmentofLifeSciences, TheNaturalHistoryMuseum,LondonSW75BD,UnitedKingdom ∗ Correspondencetobesentto:CCMAR–CentreofMarineSciences,CampusdeGambelas,UniversidadedoAlgarve,8005-139Faro,Portugal; E-mail:[email protected]. RitaCastilhoandSuzanneT.Williamscontributedequallytothisarticle. Received20August2013;reviewsreturned15July2014;accepted5August2014 AssociateEditor:AdrianPaterson Abstract.—Ancientoceanicarchipelagosofsimilargeologicalageareexpectedtoaccruecomparablenumbersofendemic D o lineageswithidenticallifehistorystrategies,especiallyiftheislandsexhibitanalogoushabitats.Wetestedthishypothesis w n usingmarinesnailsofthegenusConusfromtheAtlanticarchipelagosofCapeVerdeandCanaryIslands.Togetherwith lo a Azores and Madeira, these archipelagos comprise the Macaronesia biogeographic region and differ remarkably in the d e sdtiuvderys,iotynloyfttwhoisngornoeunpd.eMmoicresptheacines5,0inecnlduedminigcaCponuutastsivpeecsipeeschieasvceobmepenlexd,ewscerriebethdofurogmhtCtoaopcecuVrerindet,hweChearneaarsypIsriloanrdtos.tWhies d fro m combinedmolecularphylogeneticdataandgeometricmorphometricswithbathymetricandpaleoclimaticreconstructions h tounderstandthecontrastingdiversificationpatternsfoundintheseregions.Ourresultssuggestthatspeciesdiversityis ttp evenlowerthanpreviouslythoughtintheCanaryIslands,withtheputativespeciescomplexcorrespondingtoasingle ://s species, Conus guanche. One explanation for the enormous disparity in Conus diversity is that the amount of available y s habitatmaydiffer,ormayhavedifferedinthepastduetoeustatic(global)sealevelchanges.Historicalbathymetricdata, bio however,indicatedthatsealevelfluctuationssincetheMiocenehavehadasimilarimpactontheavailablehabitatareain .o x bothCapeVerdeandCanaryarchipelagosandthereforedonotexplainthisdisparity.Wesuggestthatrecurrentgeneflow fo betweentheCanaryIslandsandWestAfrica,habitatlossesduetointensevolcanicactivityincombinationwithunsuccessful rd colonizationofnewConusspeciesfrommorediverseregions,werealldeterminantinshapingdiversitypatternswithin jou theCanarianarchipelago.WorldwideConusspeciesdiversityfollowsthewell-establishedpatternoflatitudinalincreaseof rn a speciesrichnessfromthepolestowardsthetropics.However,theeasternAtlanticrevealedastrikingpatternwithtwomain ls .o peaksofConusspeciesrichnessinthesubtropicalareaanddecreasingdiversitiestowardthetropicalwesternAfricancoast. rg ARandomForestsmodelusing12oceanographicvariablessuggestedthatseasurfacetemperatureisthemaindeterminant b/ ofConusdiversityeitheratcontinentalscales(easternAtlanticcoast)orinabroadercontext(worldwide).Otherfactors y g suchasavailabilityofsuitablehabitatandreducedsalinityduetotheinfluxoflargeriversinthetropicalareaalsoplay ue s animportantroleinshapingConusdiversitypatternsinthewesterncoastofAfrica.[Conus;eustaticsealevelchanges; t o latitudinalgradientofspeciesdiversity;oceanicislands;RFmodels;speciesdiversity;SST.] n O c to b e On oceanic islands, the area of available shallow- numerous hypotheses have been proposed to explain r 8 marine habitat is strongly related to changes in the this pervasive pattern (Turner 2004; Jablonski et al. , 20 1 eustatic sea level throughout evolutionary time scales. 2006;.Anextensivebodyofliteraturehassuggestedsea 4 Episodes of falling sea level are often associated with surfacetemperature(SST),habitatarea,andchangesin extinction,whereassealevelriseisthoughttopromote primary productivity, among others, as major factors marine species radiation (Jablonski and Flessa 1986). shaping broad-scale richness gradients (Wright et al. This view has however been contradicted by several 1993; Hawkins et al. 2003). For instance, a strong studies that either found no relationship between sea- correlationbetweenspeciesdiversityandSSTwasfound level changes, shallow marine habitat availability and inthewesternAtlanticandeasternPacificoceans(coastal diversity (Raup 1976; Valentine and Jablonski 1991), marinegastropods[Royetal.1998]andinfiddlercrabs or showed that “sea level rise does not consistently [LevintonandMackie2013]). generate an increase in shelf area, […] suggesting that Two Atlantic oceanic archipelagos, Cape Verde thediversityresponsetosealevelchangewillbelargely and Canary Islands, provide an excellent model idiosyncratic” (Holland 2012). Other studies showed system for studying gradients of species diversity and thatperiodsofremarkablylowsealevelcoincidedwith phylogenetic patterns over evolutionary time scales, speciation or population expansions of coastal marine giventheirsimilargeologicalageandtheavailabilityof mollusksfromoceanicislands(Cunhaetal.2005;2011). paleoclimaticandbathymetricdataforboth.Thesetwo The latitudinal increase of species richness from archipelagos, along with Madeira and Azores, belong the poles toward the tropics has been recognized by totheMacaronesianregion(Fig.1).Thisbiogeographic biogeographerssincethe18thcentury(Hillebrand2004). regionisdefinedonthebasisofsharedfloraandfaunas Nevertheless, there is a lack of consensus regarding andhighlevelsofendemismespeciallyinthesouthern the mechanisms driving this diversity gradient and islands(Fernández-Palaciosetal.2011).CapeVerdehas 1 [17:5711/9/2014Sysbio-syu059.tex] Page:1 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 2 SYSTEMATICBIOLOGY 40º No volcanism Historical volcanism (last 500 yr) Lanzarote Quaternary volcanism Azores 15.5 35º La Palma Fuerteventura Madeira 1.77 Tenerife Gomera Gran Canaria 30º 20.6 1.12 12.0 11.6 Atlantic Ocean Hierro 14.5 Africa Canaries 25º Island Current sea level Africa 20º Historical sea level 50 km 50 km (200m lower) Historical 100 m depth Cape Verde D 15º ow n Buffer size loa d e -30º -25º -20º -15º -10º -5º d fro m FIGURE1. MapofthebiogeographicregionofMacaronesiashowingtheoceanicarchipelagosoftheAzores,Madeira,Canaries,andCapeVerde. h Thedashedlineshowsthemaximumareausedtoestimatetheshallow-marineareaaroundeacharchipelago.Theschematicrepresentation ttp ofthe50-kmperimeteraroundthedepthcontour,enoughtoencompassthe0–100moftheshallow-marinehabitatofeacharchipelagoeven ://s whentheglobalsealevelwas200mlowerthanitscurrentposition,isalsoshown.ThevolcanismofeachislandoftheCanarianarchipelagois y s representedbydifferentcolors(Quaternary-darkgray;inthelast500years—lightgray;novolcanism—white).Numbersrepresentthegeological bio age,inmillionyears,oftheCanaryIslands. .o x fo avolcanicoriginandnoneoftheislandshaseverbeen on the islets of Graciosa and Lobos. Conus guanche can rdjo u in contact with a continental landmass. The shortest also be found along the West coast of Africa (from rn a distance between the archipelago and the African Mauritania to south of Morocco, including all Western ls .o continent is about 450 km (Cap Blanc: Senegal, West Sahara). Based on variation in shell morphology, rg Africa). Geochronological data places the age of Cape Lauer (1993) described a subspecies (C. guanche nitens) b/ y VerdeIslandsbetween5.9(Brava)and25.6(Sal)million distributed across the islands of Lanzarote, Lobos, g u e years (Griffiths et al. 1975; Mitchell-Thomé 1976; Torres Graciosa, and Fuerteventura. Given the plasticity of s t o et al. 2002). The seven islands included in the Canary shell banding patterns in Conus (Röckel et al. 1995), n O archipelagowereformedinaneast-to-westprogression, independent sources of data (molecular and geometric c to as a result of volcanic activity (Coello et al. 1992). The morphometrics of radula and shell) are needed to test b e CanaryIslandsarelocated<100kmoffthecoastofWest whether these populations are in fact, different taxa. r 8 Africa. The oldest island, Fuerteventura, arose 20.7 Ma There are no aquarium observations regarding the , 20 1 (Carracedo1999)whereasthemostrecent,ElHierro,has larval development of C. guanche, but the paucispiral 4 anestimatedageof1.12myr(Guillouetal.2004). form(shellwithfewwhorls)ofitsprotoconchsuggests A major radiation of more than 50 endemic species that this species has non-planktonic larvae (E. Rolán, in Cape Verde was described for marine snails from personalcommunication),likeallConusspeciesendemic the genus Conus (Rolán 1992). Molecular analyses toCapeVerdeIslands.Benthicinvertebrateswhoselife of specimens from Cape Verde (Cunha et al. 2005; histories lack a planktonic larva often exhibit reduced Duda and Rolán 2005) showed the existence of two dispersalabilities(Scheltema1989).Therefore,lowlevels groups with distinct shell sizes (small and large shell of gene flow and differentiation might be predicted clades), and established a correspondence between amongC.guanchepopulations,supportingtheideathat cladogeneticeventsandperiodsoflowsealevel(Cunha crypticspeciesmayoccur. et al. 2005). The diversification patterns showed that Here, we used an integrative approach combining speciesclusteredintomonophyleticislandassemblages, molecular phylogenetics with geochronological and probablyasaconsequenceoftheirnon-planktoniclarvae paleoclimatic reconstructions to further understand (Cunhaetal.2005). speciation patterns in a group of marine snails from On the Canary Islands, this genus is represented two oceanic archipelagos (Cape Verde and Canary by two nonendemic species: Conus pulcher siamensis Islands) with similar geological ages but remarkably Hwass in Bruguière, 1792, which occurs throughout different Conus species richness. Our objectives were: the entire archipelago and also in Madeira, and Conus (i) to compare phylogenetic patterns within the genus guanche Lauer, 1993, which occurs on the islands of Conus between Cape Verde and Canary Islands; (ii) to Lanzarote, Tenerife, Gran Canaria, Fuerteventura, and investigate whether C. guanche is a single polymorphic [17:5711/9/2014Sysbio-syu059.tex] Page:2 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 2014 CUNHAETAL.—IDIOSYNCRATICCONUSPATTERNSONOCEANICISLANDS 3 species or represents a cryptic species complex using 1X PCR buffer, 0.2 mM of each dNTP, 0.2 μM of each molecular and geometric morphometric data; (iii) to primer, 1 μl of template DNA, and Taq Advantage®2 determine whether the number of Cape Verde Conus Polymerase mix DNA polymerase (1 U; CLONTECH- lineagesthroughtimeisconsistentwithaconstantrate Takara), using the following program: one cycle of 1 ofspeciationoriftherewereshiftsinthediversification min at 95 °C, 35 cycles of 30 s at 95 °C, 30 s at 56 °C rate that explain the striking species richness in this (NADH4), 50 °C (16S rRNA), 54 °C (ITS1), 50–90 s at archipelago; (iv) to evaluate the effect of available 68°C,andfinally,onecycleof5minat68°C. habitat area according to sea level fluctuations over PCRampliconswerepurifiedusingethanol/sodium geological time on Conus diversity in the main islands acetate precipitation and directly sequenced with of Macaronesia; (v) to test whether the irregularity of the corresponding PCR primers. Sequencing was thecoastlinepromotesspeciationinthisgroup;and(vi) performed in an automated sequencer (ABI PRISM given the similarity of habitats and geological ages, to 3700) using the BigDye Deoxy Terminator v3.1 Cycle propose an explanation for the observed disparity in SequencingKit(AppliedBiosystems),andfollowingthe Conusspeciesrichnessbetweenthetwoarchipelagos. manufacturer’s instructions. All new sequences were deposited in GenBank (accession numbers: 16S rRNA - KC748581 - KC748695; ITS1 - KC748696 - KC748824; NADH4-KC748825-KC748967;sequencesofC.guanche MATERIALANDMETHODS usedindatinganalyses:cytb-KF049928-KF049938;12S D o SpecimenCollection,DNAExtraction,Amplification, rRNA+tRNAVal+16SrRNA-KF049928-KF049949). wnlo a andSequencing d e d Atotalof201specimensofC.guanche(Supplementary MorphometricCharacterizationofShellandRadula from m10a.5t0er6i1a/ldrySa1d,.48asv5a3i)labwleerefrcoomllectehdttpf:r/o/mdx.fidvoei.orogu/t VariabilityinC.guanche http of six islands/islets of the Canarian archipelago (30 Ageometricmorphometricanalysisoftheshellshape ://s y sample locations on Gran Canaria, Fuerteventura, wasperformedon92specimensof C.guanchecollected sb io Graciosa, Tenerife, and Lanzarote) and also from from six locations (Fuerteventura, Lanzarote, Graciosa, .o x Tarfaya (Morocco) and Cap Blanc (Western Sahara/ Gran Canaria, Tenerife, and Western Sahara). To avoid fo Mauritania). All specimens were preserved in 70–96% the effects of allometry, only adult specimens with rdjo u ethanol, and total genomic DNA was isolated from lengthsbetween20and47mmandwell-preservedspires rn a foot muscle tissue using the cetyltrimethylammonium wereselected.Tophotographtheshellofeachspecimen, ls .o bromide (CTAB) protocol (Doyle and Doyle 1987). we followed the procedures described by Cunha et al. rg Specimens were gathered over several years from (2008). A total of 17 points were selected to capture b/ y private collections. As most of the samples had been shellshape(SupplementarymaterialS2andS3).Points g u e storedforlongperiodsinethanol70%,DNAextractions 1–11 were treated as landmarks, and the remaining as st o were frequently unsuccessful. Although the genus semilandmarks. These points cover most of the ratios n O Conus has been recently divided into many different usedforclassicalmorphometriccharacterizationofcone c to genera(TuckerandTenorio2009;2013),thetaxonomyof shellmorphometry(KohnandRiggs1975). b e thefamilyConidaeinthepresentworkfollows(Bouchet Radular tooth morphology was examined in a total r 8 andRocroi2005)forconvenienceandsimplicity. of42specimensfromsixlocations.Fromeachlocation, , 20 1 Two partial mitochondrial genes obtained through 6–8specimenswereselected.Landmarkpositioningwas 4 polymerase chain reaction (PCR) amplification were performedusingtwoteethperspecimen.Toobtainthe used in the phylogenetic analyses. One fragment radular teeth and the digital images of each tooth, we obtainedwiththeuniversalprimers16Sar-Land16Sbr- followed procedures described by Cunha et al. (2008). H (Palumbi 1996), amplified a 472 bp of the 16S All radular teeth included in this study were obtained rRNA gene in 115 specimens. The other fragment was from adult specimens, and all teeth were between 0.49 obtained using C. guanche-specific primers CGnad4 – and0.89mminlength.Atotalof84teethweredigitized F (GATACTTGCTAGTCAAAGATCGG) and CGnad4 – usingtpsDig2(Rohlf2005),and16pointswereselected R(GAGGAGCAGCTATATTAATAATGC)andamplified for capturing the shape of each tooth (Supplementary a 777-bp portion of the NADH dehydrogenase subunit material S2 and S3). These points cover most of the 4 (NADH4) mitochondrial genes in 143 specimens. ratios used for classical morphometric characterization Additionally, a partial fragment (505 bp) of the ofthevermivorousradulartooth(RolánandRaybaudi- internal transcribed spacer (ITS1) of nuclear-encoded Massilia 1994; Kohn et al. 1999). Points 1–4, 6–8, and ribosomal DNA was amplified in 129 specimens 13–16weretreatedaslandmarks,andtheremainingas using C. guanche-specific primers CGITS1 – F – semilandmarks. (GTGCGCAGTACAGCAGACGCTCG) and CGITS1 – Shellandradulartoothsemilandmarkswereallowed R–(GTGTCCTGCAATTCACATTAG). to slide (Zelditch et al. 2012) to the position that PCR amplifications of the two mitochondrial minimizestheProcrustesdistancebetweentheformof fragments (NADH4 and 16S rRNA) and the nuclear the shell (or the form of the radular tooth) of a given ITS1 were carried out in 25 μl reactions containing specimen, and the consensus reference form for all [17:5711/9/2014Sysbio-syu059.tex] Page:3 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 4 SYSTEMATICBIOLOGY analyzed shells (or radular teeth), using Semiland6 parameterswereestimatedindependentlyforeachofthe (Sheets 2003-2005). Landmarks were aligned by a datapartitionsusingtheunlinkcommandinMrBayes. standard generalized least squares (GLS) procedure. Analysesaccommodatedamong-partitionratevariation The within-population Procrustes distances were through use of the “prset applyto = (all) ratepr = calculated as the mean ± standard deviation (SD) of variable;” command in MrBayes. Metropolis-coupled the shell or radular tooth pairwise Procrustes distance Markov chain Monte Carlo analyses were run for matrices, respectively, as generated by CoordGen6h 20000000generationswithasamplefrequencyof1000. (Sheets 2003-2005). The pairwise Procrustes distances Lengthofburn-inwasdeterminedbyvisualexamination between locations based upon either shell or radular oftracesinTracerv.1.6(DrummondandRambaut2008). tooth morphometry were obtained with TwoGroup6h The first 20000 generations were discarded as burn- (Sheets 2003-2005) using resampling methods (500 in, and robustness of the inferred trees was evaluated bootstraps) in order to circumvent problems related usingBayesianposteriorprobabilities.Thedatasetwas to the adjustment of the correct number of degrees of analyzed in two independent runs and the final tree freedomwhensemilandmarksareused.Thesignificance was obtained from the combination of the accepted of the Goodall’s F-test was assessed by a bootstrapped treesfromeachrun.Convergencebetweenthetworuns F-test. The 95% probability interval and standard error wasassessedbyexaminingthepotentialscalereduction were also determined by bootstrapping. Partial warps factors (PSRF) and effective sample size (ESS). We also and their principal components (relative warps) were used Awty (Nylander et al. 2008) to confirm that the D o w computedusingPCAGen6p(Sheets2003-2005)included standarddeviationofsplitfrequenciesapproachedzero, n in the IMP software suite. This program was also used whichalsoindicatesconvergenceoftheresults. loa d for generating the two-dimension (2D) scatter plots To further analyze phylogenetic relationships within ed and the corresponding deformation grids and vectors. C. guanche, haplotype genealogies based on the fro m Canonical Variate Analysis (CVA) was performed concatenateddatasetandoneachgeneseparately(16S h using CVAGen6n (Sheets 2003-2005). Allometry and rRNA, NADH4, and ITS1) were reconstructed with ttp Multivariate Analysis of Variance (MANOVA) tests Network v4.6 ([Bandelt et al. 1999] available at fluxus- ://s y s were performed with Regress6k and MANOVA board, engineering.com). Median-joining networks (Bandelt b io respectively(Sheets2003-2005). etal.1999)thatcontainedallpossibleequallyshorttrees .o x were simplified by running the maximum parsimony fo rd calculation option to eliminate superfluous nodes and jo SequenceAnalysisandPhylogeneticReconstruction u links(PolzinandDaneschmand2003). rn a All DNA sequences were aligned using MAFFT ls .o version 6.0 (Multiple alignment using Fast Fourier rg Transform)(KatohandToh2010)usingtheautooption BayesianEvolutionaryAnalysisbySamplingTrees b/ y that automatically selects the appropriate strategy WeusedBayesianevolutionaryanalysisbysampling gu e accordingtodatasize.Thealignmentofmitochondrial trees (Beast) v.1.7.4 (Drummond et al. 2012) to obtain st o sequence data required no insertions and amino acid an ultrametric tree to be used in the estimation of the n O translations were checked using MacClade v. 408 number of Cape Verde Conus lineages through time. c to (Maddison and Maddison 2005). Several sequences for This analysis was based on published sequence data b e the nuclear ITS1 gene showed ambiguities that were from 51 species of Conus, including species from Cape r 8 coded with International Union of Pure and Applied Verde, Senegal, Portugal, Angola, and Canary Islands , 20 1 Chemistry(IUPAC)codes. of a mitochondrial fragment of 2044 bp. This fragment 4 (cid:3) To analyze phylogenetic patterns within C. guanche, includes the 3-end of the 12S rRNA, the complete (cid:3) weperformedmaximumlikelihood(ML)andBayesian tRNA-Valine (Val), the 5 portion of the 16S rRNA, and Inference (BI) analyses using PhyML v.3.0 (Guindon a partial sequence of the cytochrome b gene—cyt b and Gascuel 2003) and MrBayes v.3.1.2 (Huelsenbeck (GenBankaccessionnumbersretrievedfromCunhaetal. and Ronquist 2001), respectively. The data set used in [2005]).Weamplifiedthisfragmentin11newspecimens the ML analysis comprised partial sequences of the of C. guanche, which encompasses almost the entire mitochondrialNADH4(777bp)and16SrRNA(472bp) geographic distribution of this species, following the genes,andthenuclearITS1(508bp)from80specimensof methodsofCunhaetal.(2005). C.guanchethatwereconcatenatedintoasingledatasetof WeusedaYuletreeprior,whichassumesaconstant 1757bp.TheAkaikeinformationcriterion(Akaike1974) rateofspeciationamonglineages(Yule1925;Gernhard implemented in Modeltest (Posada and Crandall 1998) 2008). The Yule tree prior does not include a model selectedtheGTR+Iastheevolutionarymodelthatbest of coalescence and rate estimation for closely related fits the combined data set. Inferred model parameters sequences can be overestimated (Ho et al. 2005; wereusedintheMLanalysis. McCormack et al. 2011), but in this analysis we were The BI analysis was performed using the same data interested in dating the age of the crown group of set and considering three data partitions: NADH4, 16S C. guanche, therefore, we used a total of 13 specimens, rRNA,andITS1.Eachpartitionwasanalyzedaccording 11 from this study and sequences from two specimens to the best-fit model selected by Modeltest (NADH4: retrieved from Cunha et al. (2005). The analysis was HKY;16SrRNA:TVM;ITS1:TIM+I).Modelandmodel performedconsideringtwopartitions,oneincludingthe [17:5711/9/2014Sysbio-syu059.tex] Page:4 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 2014 CUNHAETAL.—IDIOSYNCRATICCONUSPATTERNSONOCEANICISLANDS 5 partialsequenceofcytbandtheother,thetworibosomal convex shape indicate early radiation events, whereas RNAs(12Sand16S)togetherwiththecompletesequence concaveshapessuggestrecentburstsofspeciation(Nee of the tRNA-Val. Sequence variation was partitioned et al. 1994). We also used the constant-rate (CR) test among partitions and specific models selected by the withthe(cid:3)statistic(PybusandHarvey2000)toevaluate AIC criterion implemented in Modeltest were used whethertheLTTplotswereconsistentwithaconstantnet (cytb:GTR+I+(cid:2);tRNA-Val+12S+16S:HKY+I+(cid:2)). rateofdiversificationthroughtime.Significantpositive We used two calibration points based on the (cid:3) values indicate an increase of the diversification rate paleontological record. One refers to Conus pulcher toward present (recent burst of cladogenesis), whereas Lightfoot,1786,anextantspeciesdatedfromtheLower negative values specify a slowdown in the rate of Pliocene (5.32–3.2 myr) of Cuenca de Siena (Italy) diversification through time (Pybus and Harvey 2000). (Spadini 1990) and Velerín (Estepona, Spain) (Muñiz Inphylogenieswithincompletetaxonsampling,suchas Solís 1999). The second calibration corresponded to the Conus phylogeny, the (cid:3) statistic has to be obtained Conus ventricosus Gmelin, 1791, an extant species that from simulations using the Monte Carlo constant-rate occurs in the Mediterranean and neighboring Atlantic (MCCR) test (Pybus 2000). We further examined rates includingAlgarve(SouthofPortugal)reportedfromthe of diversification across the reconstructed Bayesian MiddleMiocene(16.4–20.5myr)ofCuencadePiemonte chronogram using the relative cladogenesis statistic as (Italy) (Sacco 1893). Calibrations using the two fossils implementedintheRpackageLaserv.2.3thatidentifies weremodeledwithalognormaldistribution,where95% branchesinthetreethathavehigherratesthanexpected D o w ofthepriorweightfellwithinthegeologicalintervalin underaconstant-ratebirth–deathmodel. n which each fossil was discovered. For C. pulcher [5.32– loa d 3.2 myr], the parameters of the lognormal calibration ed prior were: hard minimum bound 3.2, mean in real Shallow-MarineHabitatAreaintheMacaronesiaVersus from tshpeacpea0ra.4m9eatnerdsSoDf1th.0e.lFoogrnCo.rmveanltrciacolisbursat[i2o0n.5p–1ri6o.4rwmeyrre]:, SeaLevelChanges http hard minimum bound 16.4, mean in real space 0.953 To test whether historical variations in habitat ://s y andSD1.0.Alognormaldistributionwasusedbecause availabilitycouldexplainthedisparityinthediversityof sb io evolutionary rates change along the branches instead ConusspeciesamongtheMacaronesiaarchipelagos,we .o x ofconvergingatthenodesasoccursintheexponential calculated the evolution of the relative shallow-marine fo distribution(Drummondetal.2006). habitatareaasafunctionoftheeustaticsealevelchanges rdjo u The starting tree for the Beast analysis was an inthelast25myr.Wedefinedtheshallow-marinehabitat rn a ultrametric tree obtained from a short run (1000000 astheseafloorareabetween0and100mdepth,whichis ls .o generations and a sample frequency of 100) using thedepthrangewhereConusspeciesaremostfrequently rg the same parameters as above but without fossil found(KohnandPerron1994). b/ y calibrations. Markov chain Monte Carlo (MCMC) Seafloor bathymetry data at 1 arc-min resolution gu e analyseswererunfourtimes(eachrunwith20000 000 were acquired from the ETOPO1 data set (Amante st o generationsandasamplefrequencyof1000)followinga and Eakins 2009) available at NOAA’s National n O discardedburn-inof2000000steps.Treeandlogoutput GeophysicalDataCenter(http://www.ngdc.noaa.gov/ c to filesfromthefourrunswerecombinedinLogCombiner mgg/global/global.html). Eustatic sea level curves for b e v.1.7.4 (Drummond et al. 2012), and the final tree was the last 25 myr, varying between ~−150 and 150 m, r 8 produced by TreeAnnotator v.1.7.4 (Drummond et al. wereobtainedfromMilleretal.(2005).Variabilitycould , 20 1 2012) using the “maximum clade creditability” option appear higher in periods where the sea level data had 4 andmeannodeheight.Theconvergencetothestationary higher resolution; hence, we used R (R Development distributionwasconfirmedbyinspectionoftheMCMC Core Team 2013) to aggregateMiller et al. (2005) data, samples and of effective sample sizes (ESS should be creating a regular time series with a regular frequency >200)usingTracerv1.6(DrummondandRambaut2008). (timeincrements)of0.1myr.FollowingHolland(2012), BIandMLanalyseswereperformedontheCCMAR we used GMT5 (generic mapping tools; Wessel and Computational Cluster Facility (http://gyra.ualg.pt) at Smith [1998]) to convert the binary bathymetry data theUniversityofAlgarve,andonthewebserverMobyle files into ASCII xyz format, which were subsequently (http://mobyle.pasteur.fr),respectively. processed in R 3.0.2 (R Development Core Team 2013). Then,foreacharchipelago,wecomputedthetotalarea between 0 and 100 m depth as a function of time. We limitedtheareasectiontoa50-kmperimeteraroundthe NumberofLineagesThroughTimeandDiversificationRates 300mdepthcontour,enoughtoencompassthe0–100m To determine whether the diversification of Cape habitatevenwhentheglobalsealevelwas200mlower Verde endemic Conus was constant through time, a than its current level (Fig. 1). This step was necessary lineage through time (LTT) plot was produced using R to exclude spurious data from areas not related to the 3.0.2(RDevelopmentCoreTeam2013)andthepackage main archipelago islands (e.g., the Desertas Islands or Laser v.2.3 (Rabosky 2006a; 2006b). Based on Tenorio theAfricancontinentalshelfwerenotconsideredwhen (2004) and Rolán (2005) we considered a total of 56 computing historical changes in the relative shallow- endemic Cape Verde Conus species. LTT plots with a marinehabitatareafortheCanaryIslands). [17:5711/9/2014Sysbio-syu059.tex] Page:5 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 6 SYSTEMATICBIOLOGY FractalDimensionofCapeVerdeandCanaryIslands genusConus(meanchlorophyllAconcentration,CHLA; Coastlines mean calcite concentration, CAL; mean percentage of cloud cover, CC; mean diffuse attenuation coefficient, To test the hypothesis that islands with more DA; mean dissolved oxygen concentration, DOX; mean convolutedcoastlines(withalargernumberofisolated nitrate concentration, NIT; photosynthetically available bays) would maximize the opportunities for speciation radiation, PAR; mean pH, PH, mean phosphate within Conus due to its reduced dispersal ability, concentration,PHOS;meansalinity,SAL;meansilicate we measured the island contour jaggedness using concentration,SIL;meanseasurfacetemperature,SST). fractal dimension analysis. The spatial dimension of A coastal diversity map for the East Atlantic was natural lines (e.g., mountain outlines) is called “fractal produced by digitizing the distribution ranges of 111 dimension” (Mandelbrot 1983). Fractal dimension has species of Conus based on (Monteiro et al. 2004) and been successfully applied to estimate the contour calculating latitudinal species richness by each 1 arc- of coastlines, and the “box-counting” dimension degreeoflatitude/longitudeusingthepackage“raster” (Hilborn 2000) is the most frequently used method (Hijmans and Etten 2013) on R (R Development Core for measurements due to its simplicity and automatic Team2013).Asimilarmapwasgeneratedfortheworld computability(Peitgenetal.2004). (longitudinalrange:180W–180E,latitudinalrange:41S– We used the software Fractal Dimension 48N)usingdatafromIUCN(2012).Environmentaldata Calculator (FDC) V1.01 (http://paulbourke.net/ were aggregated into 1 arc-degree spatial resolution D fractals/fracdim/). The method involves subdividing ow the image space into square boxes of equal size (s), tdoismtriabtuchtiosnpsecoiefsCdoivneurssidtyivdearstait.yThobettawinoedgeforgormaphbioctahl nloa and counting the number of boxes that contain at least d the eastern Atlantic and the global data sets were ed sopnleittpinixgeltohfethimeaigmeagsepa(cne).iTnhtoe psmroaclelsesr caonndtinsumeasllbeyr thenrelatedwiththeBio-ORACLEenvironmentaldata from boxes. When the series of log [n(s)] is plotted against uobsijnecgtiuvseedofRarnadnkoimngFotrheestse(nRvFir)omnmodeenltsa,lwvitahritahbelemsaiinn http log (1/s), the least squares regression line obtained functionoftheirimportanceforthedistributionofthis ://s is the fractal dimension of the object, which ranges ys genus.RFmodelsofferveryhighclassificationaccuracy b between 1 and 2, and because the fractal index is rep- io even when dealing with strong interactions among .o resented in a logarithmic scale, each increase of 0.1 in x variables(Cutleretal.2007;EvansandCushman2009), fo thefractaldimensionrepresentsatwo-foldincreaseon rd thestructuralcomplexity. show very low susceptibility to overfitting (Breiman jou Becauseimagetreatmentcanaffectthebox-counting 2001;Prasadetal.2006)andhavesuccessfullybeenused rna tomodelthedistributionofcoastalspecies(Hilbishetal. ls calculations, it is important to treat images exactly .o the same way to obtain comparable results. All 2tr0e1e2s).(PMarotdyeplsacwkeargeebfouriltR,u(sSintrgobcloentdaitli.o2n0a0l8i)n),fewrehnicche brg/ igslloabnadl csoealfs-tclionnessistewnetr,e hoiebrtaarinchedicalf,romhighG-rSeHsoHluGt—iona are specially recommended when variables might be y gue geography database (http://www.ngdc.noaa.gov/ highly correlated (Strobl et al. 2009). Our models were st o parameterized with 5000 trees per run and included n mgg/shorelines/gshhs.html) as PDF files, and O three randomly preselected predictor variables on c processed in ImageJ (Abràmoff et al. 2004), obtaining to eachsplit. b j1u6s0t0t×he1c6o0a0sptaixleclosn.tourwithabinaryimagefilesizeof er 8, 2 0 Two important aspects of fractal dimension 1 RESULTS 4 calculation are the choice of the box size and the positioning of the grid over an object. FDC offers a GeometricMorphometricsofShellandRadulaShapes maximumof12boxsizeswithinasinglerun.Weused ofC.guanche thelargestboxsize,automaticallydeterminedaccording Results of the one-way MANOVA tests for shell to an algorithm based on the size of the image, and and radular shape data of C. guanche using location the smallest box size equal to 1 pixel. FDC also allows (Fuerteventura, Lanzarote, Graciosa, Gran Canaria, the user to select the number of “offsets” for each box Tenerife, and Mauritania/Western Sahara) as the size to allow for placement of the grid at different grouping criterion are shown in Table 1. Differences startingpositions,andtocarryoutanexhaustivecount with every possible starting position. Because there is no preferred origin for the boxes that initiates the TABLE1. Resultsoftheone-wayMANOVAfromshellandradula positioningofthegrid,fractaldimensionmeasuresare shapeofC.Guancheusinglocalityasgroupingcriteria theaveragecomputedfrom100differentboxorigins. Groupingfactor: Explained Unexplained %Explained Fa Pa Locality SS SS ConusDiversityVersusEnvironmentalVariables Shell 0.0124 0.0412 23.17 5.19 0.002 Radula 0.0189 0.0595 24.13 4.96 0.002 We selected 12 environmental variables from the Bio-ORACLE database (Tyberghein et al. 2012) likely Note:SS,sumofsquares. to directly or indirectly influence the diversity of the aSignificanceisassessedbyabootstrappedF-test(500bootstraps). [17:5711/9/2014Sysbio-syu059.tex] Page:6 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 2014 CUNHAETAL.—IDIOSYNCRATICCONUSPATTERNSONOCEANICISLANDS 7 in mean shell length between locations (Fuerteventura, yielded the topology shown in Supplementary mate- 29.4 mm; Graciosa, 38.0 mm; Gran Canaria, 30.7 mm; rial S5. The few clades that received statistical support Lanzarote,34.1mm;WesternSahara,30.7mm;Tenerife showed no geographic structure (Supplementary 29.1 mm) were not significant (P >0.05), except for materialS5). the differences between the length of the shells from Haplotypegenealogiesbasedon16SrRNA,NADH4, Graciosa and the rest of the islands, but excluding and ITS1, separately, showed that the central and most Lanzarote. The low percentages of the explained commonhaplotypeissharedamongalllocations(Fig.2). variance(23%;Table1)suggestlowcorrelationbetween Inthecombineddataset,themostcommonhaplotypeis shell shape and location. The CVA of the shape alsosharedamongalllocationswiththesingleexception of 92 shells showed a large overlap among groups ofGraciosa.Mosthaplotypesdivergedfromthecentral (Supplementary material S3a). The CVA using locality haplotypebyonly1–4mutations. as grouping factor and the principal components 1–10 TheultrametrictreeobtainedwithBEASTshowedESS of the partial warp scores as variables (which account values>200inTRACERindicatingthatchainshadreached for the 92.3% of the variance in shell shape) correctly stationarity. Our age estimates for lineage splitting classified only 54 specimens out of 92 (58.7%) for events within Conus (Fig. 3a) differed by <1 myr from the location. Pairwise Procrustes distances between theresultsshowninCunhaetal.(2005). different groups were small (between 0.009 and 0.031; Supplementary material S4b). Procrustes distances D LTTPlotsandDiversificationRateswithin ow between and within groups defined by location and n based on shell morphometry are of the same order of CapeVerdeConus loa d magnitude(SupplementarymaterialS4a,b)suggesting According to the relative cladogenesis statistic, one ed that there is little variability of shell shape among instanceofsignificantincreaseinthediversificationrate fro m populations.Nevertheless,somevalueswerestatistically of Cape Verde Conus was identified within the BEAST h significant (for instance, between Fuerteventura and tree (represented by an asterisk in Fig. 3a) occurring ttp Graciosa;SupplementarymaterialS4b). withinthesmall-shelledclade.Theconcaveshapeofthe ://s y The large overlap among clusters corresponding to LTT plot (Fig. 3c) and the positive (cid:3) value (3.62), both sb io each island suggests little variability of the radular support an increase in the number of lineages toward .o x shape between specimens from different locations the present (recent radiation) ( Nee et al. 1994; Pybus fo rd (Supplementary material S3b). Differences in radular andHarvey2000).CR/MCCRtestsindicatedstatistically jo u tooth mean size between locations (Fuerteventura, significant changes in rates of diversification through rn 0.70 mm; Graciosa, 0.70 mm; Gran Canaria, 0.66 mm; time (species sampled = 42; estimated total number of als Lanzarote,0.68mm;WesternSahara,0.68mm;Tenerife endemicspecies=56;(cid:3)=3.62;criticalvalueof(cid:3)atP= .org 0.61 mm) were not significant (P >0.05), except for the 0.05,one-tailedtest,(cid:3)0.05=1.10). by/ differencesbetweenthemeansizesoftheradularteethof gu e specimensfromTenerifeandtherestoftheislands,but Shallow-MarineHabitatAreaandSeaLevelChange st o excluding Western Sahara. The low percentages of the n O explainedvariance(24%;Table1)suggestlowcorrelation Theavailablerelativeshallow-marinehabitatareafor c to between radular shape and location. The CVA of the allMacaronesianarchipelagosfromtheMiocenetothe b e radular shape of the entire data set (84 radular teeth) presentbasedonsealeveldatafrom(Milleretal.2005) r 8 isshowninSupplementarymaterialS3b.Locationwas isshowninFigure3b.Therearesubstantialdifferences , 20 1 used as grouping factor and the principal components in the relative shallow-marine area available among 4 1–10ofthepartialwarpscoresasvariables.Usingthese archipelagosandwithineacharchipelago,buttheeffect CVaxes,27outof84specimensweremisclassified(67.8% of the sea level variation (lower or higher sea stands) correct).PairwiseProcrustesdistancesbetweendifferent is identical on each archipelago. The relative shallow- groups were also small (between 0.011 and 0.045) but marineareainMadeiraandAzoresseemstobeaffected significantinseveralcases(e.g.,betweenLanzaroteand in a lesser extent by variations of the sea level. Cape Tenerife;SupplementarymaterialS4d). Verde and Canary Islands show identical variation of the relative shallow-marine area over time, and the area of each archipelago almost double the area of PhylogeneticReconstructions MadeiraorAzores.Inthelast5myr,allfourarchipelagos Potential scale reduction factors in the BI analysis experiencedareductionintherelativehabitatareawith wereabout1.00andESSvalueswereall>200,indicating Cape Verde and the Canary Islands showing higher convergence of runs. Assessment of convergence losses(Fig.3b). statisticsinAwtyindicatedthattheanalysesconverged given that the standard deviation of split frequencies approached zero. ML (−ln L = 3370.16) and Bayesian FractalDimensionoftheCoastline (−lnL=3431.00)analysisbasedontheconcatenateddata The average fractal dimension of the Cape Verde set (NADH4, 16S rRNA, and ITS1) arrived at a similar coastline is 1.432 ± 0.075 and the fractal dimension of topology showing that phylogenetic relationships the Canary Islands is 1.411 ± 0.080 (Fig. 4). Islands withinC.guancheweregenerallyunresolved.BIanalysis that showed more convoluted coastlines (and higher [17:5711/9/2014Sysbio-syu059.tex] Page:7 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 8 SYSTEMATICBIOLOGY D o w n lo a d e d fro m h ttp ://s y s b io .o FIGURE2. Median-joiningnetworkofConusguanchebasedontheconcatenateddatasetandoneachgeneseparately(16SrRNA,NADH4,and x ITS1)reconstructedwith‘Network’.Theareaofeachcircleisproportionaltothenumberofindividualssharingaparticularhaplotype.Colors ford refertosamplelocation,andthesizeofeachsliceisproportionaltothenumberofindividualswiththathaplotype.Haplotypesareconnected jo bybranchlengthsapproximatelyequaltotheinferredmutationalsteps.Numbersrepresentmutationalsteps. urn a ls .o fractal dimensions) were Gomera and Maio whereas PAR, SAL, and CC, all explained more than 10% of the brg/ Graciosa and Santa Luzia exhibited the most reduced model (Fig. 5b). Of all environmental variables tested, y g valuesoffractaldimension(Fig.4).Alinearregression SSTexplainedthemostvariationinConusdiversity.The ue s betweenthefractaldimensionofeachislandandthetotal relationship between SST and Conus diversity across t o n numberofspeciesrevealedanonsignificantrelationship the West African coast and the biogeographic region O c (R2=0.04,datanotshown). of Macaronesia is shown in Figure 6. The number of to Conus species is higher in the tropics (between ≈20ºS ber 8 and20ºN),withthemaximumdiversity(upto20species , 2 RelationshipbetweenEnvironmentalVariablesand per1ºlatitude/longitude)occurringoffSenegalandoff 01 4 ConusSpeciesDiversity theCapeVerdearchipelagowherethepolewardNorth Equatorial Counter Current meets the Canary Current We analyzed the relationship between 12 oceano- formingalargeoceanographicdiscontinuity(Longhurst graphic variables (CHLA; CAL; CC; DA; DOX; NIT; 1998), and where average SST is ~24 ºC (Fig. 6). In PAR; PH; PHOS; SAL; SIL; SST) and Conus diversity the worldwide RF model, SAL, NIT, DOX, and SST to explain the observed latitudinal gradient of species contributedmorethan10%tothemodelvariability,and richness along the West African coast, applying the RF SSTexplainedthemostvariationinConusdiversity. model (Cutler et al. 2007). We also used the RF model andthesame12parameterstoexplainworldwideConus diversity. Results from the RF model indicated that all DISCUSSION parameters showed a significant correlation with Itisareasonableexpectationthatspeciationpatterns diversity for both eastern Atlantic and worldwide of a given group of organisms be comparable on two analyses except CHLA and CAL (eastern Atlantic; ancient oceanic archipelagos sharing a similar setting. Fig.5a)andCHLA(worldwide;Fig.5b).TheRFmodel Although this premise is valid for several groups of explained83.2%ofthetotalvariancefortheworldwide terrestrial organisms, such as the emblematic case of data and 67.5% of the total variance for the eastern Anolis lizards from the Greater Antilles (Losos et al. Atlanticdata.Thecontributionofeachvariableforeach 1998), where islands with similar resources support RF model is shown in Figure 5, and in more detail in nearly identical communities (Case and Cody 1987), SupplementarymaterialS6.IntheeasternAtlantic,SST, its application in the marine environment is yet to be [17:5711/9/2014Sysbio-syu059.tex] Page:8 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 2014 CUNHAETAL.—IDIOSYNCRATICCONUSPATTERNSONOCEANICISLANDS 9 (a) 3.29 [1.89-9.25] 1 6.52 [4.32-9.25] 1 3.68 [2.28-5.45] 8.85 [6.04-12.13] 1* Conus small shells 0.8 1.85 [1.05-2.91] Cape Verde 1 9.76 [6.73-13.43] 1 14.79 [11.2-18] 0.76 6.5 [3.68-9.89] West Africa 1 3.99[2.06-6.42] 16.9 [16.42-180.7.555] 10.81[0.42-1.391] C. Cgaunaanrciehse South Portugal 3.03 [1.83-4.69] 4.65 [3.22-7.141] 0.99 Conus Claarpgee Vsheerdlles C. pulcher siamensis Canaries Miocene Pliocene Pleist. 23 5.3 2.6 0 million years D o w (b) m2) 5 nlo 1000 k 4 aded fro x m a ( h e are 3 ttp://s 00m) marin 2 CCaapnea rVieesrde ysbio.oxford w (0-1 1 Azores journa o ls hall 0 Madeira .org S 20 15 10 5 0 b/ Time (Myr) y g (c) s40 ue neage st on O ber of li20 ctober 8 Num 0 , 2014 (d) 50 m) el ( 0 v e a l −50 e S −100 20 18 16 14 12 10 8 6 4 2 0 Time (Myr) FIGURE3. a)BeastmaximumcladecredibilitychronogramshowingmaincladogeneticeventswithinConusbasedonthemitochondrial 2044bpdataset(12SrRNA,tRNA-VAL,16SrRNA,andcyt.)Numbersaboveandbelownodesrepresentnodeheights(ages)and95%highest posteriordensity(HPD)intervalsandBayesianposteriorprobabilities,respectively.TheshadedarearepresentsthecladeofspeciesfromSal; b)Variationoftheshallow-marinehabitatareaofthefourMacaronesianarchipelagos(Canaries,CapeVerde,Azores,andMadeira)according tosealevelchangessincetheMiocene(basedondatafromMilleretal.[2005]);c)LineagethroughtimeplotfortheendemicCapeVerdeConus showinganincreasednumberoflineagestowardthepresent.Thex-axiscorrespondstotime(inmyr);d)Eustaticsealevelchangesduringthe last20myrbasedondatafromMilleretal.(2005).TheverticaldashedlinerepresentstheupturninthenumberofCapeVerdeConuslineages thatstarted≈10Ma. [17:5711/9/2014Sysbio-syu059.tex] Page:9 1–17 Copyeditedby:PN MANUSCRIPTCATEGORY: Article 10 SYSTEMATICBIOLOGY Fractal dimension Number of species 1.6 20 n o N nsi 1.5 15 um e b m e al Di 1.4 10 r of s ct p a e r c F 1.3 5 ie s 1.2 0 GraciosaTenerifeFuerteventuGra.CanariaLanzaroteLa PalmaHierro Gomera Santa LuziaBrava Santo AntãoS. NicolauSantiagoBoavistaFogo Sal S. VicenteMaio Canaries Cape Verde D Average fractal dimension = 1.411 ± 0.080 Average fractal dimension = 1.432 ± 0.075 o w n FIGURE4. CoastlinefractaldimensionofCapeVerdeandCanarianarchipelagosusingFractalDimensionCalculatorV1.01.Barsrepresent loa thefractaldimensionofeachisland(lightgray)andthetotalnumberofspecies(darkgray),endemics,andnonendemics.Thecontourofthe de d coastlineofeachislandisshown. fro m (a) Eastern Atlantic (b) Worldwide h ttp SST SST ://sy s b PAR SAL io .o SAL NIT xfo rd CC DOX jo u DOX CC rna ls NIT PHOS .o rg PH SIL b/ y g DA PH u e s PHOS PAR t o n O SIL DA c to CHLA CAL be r 8 CAL CHLA , 2 0 1 4 0 10 20 30 40 50 0 5 10 15 20 25 30 variable contribution (%) variable contribution (%) FIGURE5. RelationshipbetweenConusdiversityand12environmentalvariables(meanCHLA;meanCAL;meanpercentageofCC;meanDA coefficient;meanDOXconcentration;meanNITconcentration;PAR;PH;meanPHOS;meanSAL;meanSILconcentration;meanSST).Predictors torightofdashedverticallinearesignificant.a)easternAtlanticConusdiversity;b)worldwideConusdiversity. determined. The remarkable marine radiation of the relative shallow-marine area during eustatic sea level genus Conus in the Cape Verde Islands (Rolán 2005) fluctuations observed since the Miocene. Hence, we hasnocorrespondenceintheCanaryarchipelagowhere explored alternative hypotheses that could explain the this genus is represented by only two, nonendemic distinctivediversitypatternsfoundinbotharchipelagos. species. The integrative approach presented here intends to elucidate patterns and processes underlying species diversity in comparable environments. Our Conusguanche:ACrypticSpeciesComplexorRecent findings suggest that species diversity in the Canarian Colonization? archipelago is even lower than initially anticipated, as no cryptic speciation was detected within C. guanche. Our results show that the genus Conus diversified Moreover,CapeVerdeandCanaryIslandsexperienced in very different ways on the Canary Islands and comparable changes in the amount of available in Cape Verde, despite similar geological ages and [17:5711/9/2014Sysbio-syu059.tex] Page:10 1–17

Description:
Biodiversidade e Recursos Genéticos, Universidade do Porto, Campus Agrário de Vairão 4485-661 Vairão, Portugal, 3Department CMIM y Q years (Griffiths et al. 1975; Mitchell-Thomé 1976; Torres et al. 2002). The seven islands included in the Canary archipelago were formed in an east-to-west
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.