Please download to get full document.

View again

of 13

Molecular taxonomy and phylogenetic affinities of two groundwater amphipods, Crangonyx islandicus and Crymostygius thingvallensis, endemic to Iceland

Molecular taxonomy and phylogenetic affinities of two groundwater amphipods, Crangonyx islandicus and Crymostygius thingvallensis, endemic to Iceland
0 views13 pages
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  Molecular taxonomy and phylogenetic affinities of two groundwater amphipods, Crangonyx islandicus  and  Crymostygius thingvallensis , endemic to Iceland Etienne Kornobis a, ⇑ , Snæbjörn Pálsson a , Dmitry A. Sidorov b , John R. Holsinger c , Bjarni K. Kristjánsson d a Department of Biology, University of Iceland, Askja, Sturlugata 7, 101 Reykjavik, Iceland b Institute of Biology and Soil Science, Far Eastern Branch of the Russian Academy of Sciences, Vladivostok 690022, Russia c Department of Biological Sciences, Old Dominion University, Norfolk, VA, USA d Department of Aquaculture and Fish Biology, Hólar University College, Háeyri 1, 550 Sauðárkrókur, Iceland a r t i c l e i n f o  Article history: Received 20 October 2010Revised 17 December 2010Accepted 20 December 2010Available online 30 December 2010 Keywords: rDNACrustaceaAmphipodaCrangonyctoideaMolecular phylogenyAlignment methods a b s t r a c t The amphipod superfamily Crangonyctoidea is distributed exclusively in freshwater habitats worldwideand is characteristic of subterranean habitats. Two members of the family,  Crangonyx islandicus  and Crymostygius thingvallensis , are endemic to Iceland and were recently discovered in groundwater under-neath lava fields.  Crangonyx islandicus  belongs to a well-known genus with representatives bothin NorthAmerica andin Eurasia.  Crymostygius thingvallensis  defines anewfamily, Crymostygidae. Considering theincongruences observed recently between molecular and morphological taxonomy within subterraneanspecies, we aim to assess the taxonomical status of the two species using molecular data. Additionally,thestudycontributestothephylogeneticrelationshipsamongseveralcrangonyctoideanspeciesandspe-cifically among species from four genera of the family Crangonyctidae. Given the available data we con-siderhowthetwoIcelandicspeciescouldhavecolonizedIceland,bycomparinggeographicalsrcinofthespecies with the phylogeny.Regions of two nuclear (18S and 28S rRNA) and two mitochondrial genes (16S rRNA and COI) for 20different species of three families of the Crangonyctoidea were sequenced. Four different methods wereused to align the RNA gene sequences and phylogenetic trees were constructed using bayesian and max-imum likelihood analysis. The Crangonyctidae monophyly is supported.  Crangonyx islandicus  appearedmore closely related to species from the Nearctic region.  Crymostygius thingvallensis  is clearly divergentfrom the other species of Crangonyctoidea.  Crangonyx  and  Synurella  genera are clearly polyphyletic andshowed a geographical association, being split into a Nearctic and a Palearctic group.ThisresearchconfirmsthatthestudiedspeciesofCrangonyctidaeshareacommonancestor,whichwasprobablywidespread intheNorthernhemisphere well beforethebreakupof Laurasia. TheIcelandicspe-cies are of particular interest since Iceland emerged after the separation of Eurasia and North America, isgeographicallyisolatedandhasrepeatedlybeencoveredbyglaciersduringtheIceAge. Thecloserelationbetween  Crangonyx islandicus  and North American species supports the hypothesis of the Trans-Atlanticland bridge between Greenland and Iceland which might have persisted until 6 million years ago. Thestatus of the family Crymostygidae is supported, whereas  Crangonyx islandicus  might represent a newgenus. As commonly observed in subterranean animals, molecular and morphological taxonomy led todifferent conclusions, probably due to convergent evolution of morphological traits. Our molecular anal-ysis suggests that the family Crangonyctidae needs taxonomic revisions.   2010 Elsevier Inc. All rights reserved. 1. Introduction Two endemic species of the exclusively freshwater superfamilyCrangonyctoidea, characteristic of subterranean habitats, were re-cently discovered in Iceland and described using morphologicaland meristic criteria. The two species are  Crangonyx islandicus , anew species within a known genus in the family Crangonyctidae(Svavarsson and Kristjánsson, 2006), and  Crymostygius thingvallen-sis , a new monotypic family (Crymostygidae) (Kristjánsson andSvavarsson, 2004). According to morphological taxonomy,  Crang-onyx islandicus  belongs to the genus  Crangonyx , distributed inNorthAmericaandEurasia.Thediscoveryofthesetwonewspeciesin Iceland, which was repeatedly covered by glaciers during thecold periods of the Ice Age (Pleistocene, from 2.59Myr to12,000years ago), has raised questions about when and how they 1055-7903/$ - see front matter    2010 Elsevier Inc. All rights reserved.doi:10.1016/j.ympev.2010.12.010 ⇑ Corresponding author. Fax: +354 525 4632. E-mail addresses: (E. Kornobis), (S. Pálsson), sidorov (D.A. Sidorov), (J.R. Holsinger), (B.K. Kristjánsson).Molecular Phylogenetics and Evolution 58 (2011) 527–539 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage:  colonized Iceland. A recent study has found strong evidence forsurvival of   Crangonyx islandicus  in sub-glacial refugia in Iceland.Using mtDNA divergence among monophyletic and geographicallyisolated populations, Kornobis et al. (2010) showed that the ob-served divergence occurred within Iceland, during the Ice Ageand even before its onset.The existence of Iceland has been traced to a geological hotspotwhichmigratedsoutheastwardabout40MyafromtheEastGreen-land coast to its current location at the boundary of the Atlanticridge and the Greenland–Scotland Transverse Ridge (Lawver andMüller, 1994; Lundin and Doré, 2002). Geological and biologicalevidence support that the Greenland–Scotland Ridge, submergedat great depthsat present, was above sea level fromearlyCenozoicto late Miocene, first as a continuous land bridge and later as achain of islands (McKenna, 1983; Eldholm et al., 1994; Grímssonet al., 2007; Poore, 2008; Denk et al., 2010). Fossil records showthat plants migrated along this land bridge between Scotland andproto-Iceland until 24Mya and until 15Mya or even 6Mya be-tween Greenland and proto-Iceland (Grímsson et al., 2007; Denket al., 2010). Because Crangonyx islandicus  and Crymostygius thingv-allensis  belong to a superfamily of exclusively freshwater amphi-pods, it is possible that their ancestors colonized Iceland throughthis land connection rather than by marine ancestors.The taxonomy of subterranean organisms based on morpholog-ical information faces several problems. The morphological traitsare generally characterized both by progressive (e.g. elongationof the trunk and/or sensory appendages) and regressive evolution(loss of eyes and pigmentation) (Porter, 2007; Väinölä et al.,2008). Thesetroglomorphictraits areapparentlyshapedbysimilarselective pressures (absence of light, low nutrients, low oxygen)encountered in subterranean systems worldwide, which have ledto convergent evolution of morphological traits among differentspecies. Theabundance of suchtraits mayhinder theidentificationof morphologically informative characters for phylogenetic recon-struction (Englisch et al., 2003; Wiens et al., 2003). Taxonomicrevision using molecular markers is thus especially important forsubterranean species (Fišer et al., 2008). Molecular taxonomy of groundwater amphipods has been widely used in the past decade(Englisch et al., 2003; Fišer et al., 2008) and has indicated thatmorphological markers are to a great extent homoplasic. In addi-tion, molecular analysis has recently revealed cryptic diversity(Lefébureetal.,2006a;Kornobisetal., 2010)andtaxonomicincon-sistenciesamongvariousgroundwateramphipodspecies(Lefébureet al., 2006a).The aim of this study is to evaluate the taxonomic status of thetwo endemic species  Crymostygius thingvallensis  and  Crangonyxislandicus  and to contribute to the phylogenic classification of thecrangonyctoidean species, with a special focus on the family Cran-gonyctidae. In addition we consider the phylogenetic relationshipswith respect to the species distribution (Palearctic or Nearctic) toinfer putative colonization routes, focusing on the colonization of Iceland. The study is based on DNA sequence variation amongthree families (Crymostygidae, Crangonyctidae, Pseud-ocrangonyctidae) of the superfamily Crangonyctoidea and amongfive genera ( Crangonyx ,  Amurocrangonyx ,  Bactrurus ,  Synurella  and Stygobromus ) in the family Crangonyctidae for two nuclear genes(18S and 28S ribosomal RNA genes), and two mitochondrial genes(Cytochrome oxydase subunit I COI and 16S ribosomal RNA).Thesubunits, 18Sand28S, of ribosomalgenes havebeenexten-sively used as markers for phylogenies at the family level, e.g.within amphipods (Englisch et al., 2003; MacDonald et al., 2005;Lefébure et al., 2006a) and even for deeper phylogenies withincrustaceans(MallattandGiribet, 2004; Pageet al., 2008). Althoughthe use of RNA genes for phylogenetic purposes has been success-ful, some methodological problems exist (Edgar and Batzoglou,2006). Like the tree building methods (Morrison, 2008), alignment methods can have profound effects on phylogenies and are of spe-cial concern for the ribosomal genes where insertions and/or dele-tions are commonly observed. Numerous sequence alignmentmethods have been developed recently (Edgar and Batzoglou,2006; Novák et al., 2008; Bradley et al., 2009). We apply differentalignment techniques and weighting schemes to evaluate the im-pact of the alignment and variation along the sequence. 2. Material and methods  2.1. Samples Twelve amphipod species were collected from 15 locations inNorth America, Europe and Asia (Table 1) and were preserved in96% ethanol. For the construction of phylogenetic trees, sequencesof 18S RNA (14 species) and 28S RNA (7 sp.) genes together withthe mtDNA 16S (4 sp.) and COI (5 sp.) genes (Table A1) from 17species in total belonging to the suborder Gammaridea were ob-tained from Genbank (,13 of them belonging to the superfamily Crangonyctoidea. Thetotal sample resulted in 20 different species of the crangonyctoidsin five genera. The number of species sequenced for each genevaried (Table A1). To evaluate the divergence among species andgenera of the Crangonyctidae, comparisons were made with corre-sponding divergence between 18S sequences from the amphipodsuperfamilies Gammaroidea (33 species), Lysianassoidea (3),Talitroidea (10) and Eusiroidea (7), and 28S sequences fromGammaroidea (54), Lysianassoidea (18) and Talitroidea (4),obtained from Genbank.  2.2. Molecular work DNA was extracted using 6% Chelex 100 (Biorad) from wholespecimens or only a pereopod frombigger specimens. Two regionsofthenucleargenome,the18SRNAand28SRNAgenes,andtwoof themitochondrialgenome,COIand16SRNAgenes,wereamplifiedusingprimerssummarizedinTable2.PCRamplificationswereper-formed in 10 l l, containing 0.15mM dNTPs, 0.1% Tween, 0.5 l g/ l lof BSA, 1X Taq buffer, 0.35 l M primers, 0.5 unit of Taq (New Eng-land Biolabs) and 10–100ng of template. Amplifications were ob-tained with the PCR conditions: 94  C 4min, followed by 39cycles of 94  C 30s, annealing temperature ranging from 45 to55  C (depending on species and genes) for 45s, 72  C for 1.5min(1min for both mt genes), with a final elongation step at 72  Cfor 6min. The mtDNA fragments were purified and sequenced di-rectly as in Kornobis et al. (2010). The 18S and 28S PCR productswere cut out of the gel and purified using the Nucleospin ExtractII Kit (Macherey–Nagel), ligated to TOPO vector (TOPO TA CloningKit, Invitrogen) and cloned in chemically competent cells DH5 a ™-T1 R  . Plasmids purified using the Nucleospin Plasmid Kit (Mache-rey–Nagel) and Exosap purified PCR products were sequenced inboth directions and with internal primers (Table 2) using ABI Big-Dye Terminator v3.1 (Applied Biosystems). Three to five clones perindividual of the nuclear genes were sequenced and run on an ABIPRISM™3100 Genetic Analyzer. Raw sequences were checked andeditedusingBioEdit7.0.9.0.(Hall,1999).Sequenceshavebeensub-mitted to GenBank under accession numbers from HQ286000 toHQ286020 and from HQ286022 to HQ286037.  2.3. Alignment  Four different methods were used to align the RNA sequences:(1) ClustalW2 with default parameters (Larkin et al., 2007). Clu-stalW has been extensively used, but newly developed algorithmsare considered to be more accurate (Morrison, 2009). We chose 528  E. Kornobis et al./Molecular Phylogenetics and Evolution 58 (2011) 527–539  this method for comparison with other alignment techniques. (2)MAFFT 6 (Katoh et al., 2009), withQ-INS-i strategy in order to takeinto account the secondary structure of the RNA. This method waschosen for the recent implementation of secondary structuresearch(KatohandToh,2008)anditsgoodratioaccuracy/computa-tion time (Edgar and Batzoglou, 2006). (3) FSA 1.15.2, a statisticalalignment software with a limited runtime and which avoids effi-ciently false/positive alignments (Bradley et al., 2009), was usedwith a gap factor of 1 and the Tamura Nei model. (4) RNAsalsa0.8.1(Stocsitsetal., 2009), arecentsoftwarewhichusessecondarystructure information for adjusting and refining the sequencealignment, was used with default parameters and the consensussecondary structure of preliminary alignments of the RNA se-quencesasastructural constraint. TheCOI sequenceswerealignedby eye using BioEdit.The mean sequence identity was estimated for each alignmentmethod and for each RNA gene, using the APE package (Paradiset al., 2004) in R (R Development Core Team, 2010). According to Hall (2008) and following the results of  Kumar and Filipski(2007), an alignment of non-coding DNA that presents a mean se-quence identity of 66% ensures about 50% of alignment accuracy.Variable alignment accuracies higher than 50% are considered tohave little effect on the phylogenetic reconstruction, both forBayesian and Maximum likelihood methods (Ogden and Rosen-berg, 2006).The optimal model of evolution for each dataset was chosenaccordingtothemaximumlikelihoodtreewithrespecttosubstitu-tions, transition/transversion ratio, proportion of invariant sitesand the gamma distribution parameters selected with PhyML (Guindon and Gascuel, 2003) and implemented in the R-package  Table 1 Species included in the phylogenetic analyses with geographic srcin and sampling date. References are given for specimens obtained from genbank. Species from theGammaridae and Megaluropidae families were used as outgroups. ‘‘NA’’ stands for not available. Family Species Locality Latitude Longitude Reference or year of samplingCrangonyctidae  Amurocrangonyx arsenjevi  Khabarovsk, Russia 47.902 135.340 This study (2005)Crangonyctidae  Bactrurus brachycaudus  St. Louis Co., Missouri, USA NA NA Englisch and Koenemann (2001)Crangonyctidae  Bactrurus mucronatus  Saline Co., Illinois, USA 37.680   88.420 Englisch and Koenemann (2001)Crangonyctidae  Bactruruspseudo mucronatus  Oregon Co., Missouri, USA 36.815   91.181 Englisch and Koenemann (2001)Crangonyctidae  Crangonyx chlebnikovi  Perm, Russia 57.446 57.017 This study (2005)Crangonyctidae  Crangonyx chlebnikovi  Perm, Russia 57.061 57.528 This study (2003)Crangonyctidae  Crangonyx floridanus  Gainsville, Florida, USA   81.657 30.275 Slothouber Galbreath et al. (2009)Crangonyctidae  Crangonyx forbesi  St. Louis Co., Missouri, USA 38.616   90.701 Englisch AND Koenemann (2001)Crangonyctidae  Crangonyx islandicus  Thingvallavatn, Iceland 64.241   21.053 This study (2007)Crangonyctidae  Crangonyx islandicus  Svartarvatn, Iceland 65.469   17.233 This study (2007)Crangonyctidae  Crangonyx islandicus  Klapparos Kopasker, Iceland 66.361   16.400 This study (2008)Crangonyctidae  Crangonyx pseudogracilis  Lake Charles, Louisiana, USA 30.262   93.221 Slothouber Galbreath et al. (2009)Crangonyctidae  Crangonyx serratus  Virginia, USA NA NA MacDonald et al. (2005)Crangonyctidae  Crangonyx  sp. Barnishee Slough, Tenessee, USA 35.579   89.963 Slothouber Galbreath et al. (2009)Crangonyctidae  Crangonyx subterraneus  Biberach, Schwarzwald, Germany 48.326 8.126 Fišer et al. (2008)Crangonyctidae  Stygobromus gracilipes  (H3970) West Virginia, USA 39.302   77.850 This study (2002)Crangonyctidae  Stygobromus gracilipes  (H4070) Virginia, USA 39.014   78.276 This study (2000)Crangonyctidae  Stygobromus mackini  USA NA NA Englisch and Koenemann (2001)Crangonyctidae  Stygobromus stegerorum  Virginia, USA 38.301   79.255 This study (2000)Crangonyctidae  Synurella ambulans  Ljubljana, Slovenia 46.050 14.500 This study (2010)Crangonyctidae  Synurella dentata  Preble Co., Ohio, USA 39.772   84.722 Englisch et al. (2003)Crangonyctidae  Synurella  sp. Lake Charles, Louisiana, USA 30.262   93.221 Slothouber Galbreath et al. (2010)Crymostygiidae  Crymostygius thingvallensis  Thingvallavatn, Iceland 64.241   21.053 This study (2007)Gammaridae  Gammarus abstrusus  Lushan, Sichuan, China 30.280 102.970 Hou et al. (2007)Megaluropidae  Megaluropus longimerus  Curacao, Karibische See NA NA Englisch et al. (2003)Niphargidae  Niphargus fontanus  Grundwasser am Ruhrufer, Germany NA NA Englisch and Koenemann (2001)Niphargidae  Niphargus kochianus  Grundwasser am Ruhrufer, Germany NA NA Englisch et al. (2003)Pseudocrangonyctidae  Procrangonyx primoryensis  Primory, Russia 47.185 138.743 This study (2003)Pseudocrangonyctidae  Procrangonyx primoryensis  Primory, Russia 47.256 138.800 This study (2002)Pseudocrangonyctidae  Pseudocrangonyx korkishkoorum  Primory, Russia 43.100 131.547 This study (2006)  Table 2 Oligonucleotides used for PCR and sequencing of the four different genes. Genes Primer name Primer sequence (5 0 –3 0 ) References18S 18SF CCTAYCTGGTTGATCCTGCCAGT Englisch et al. (2003)18S 700R CGCGGCTGCTGGCACCAGAC Englisch et al. (2003)18S 1500R CATCTAGGGCATCACAGACC Englisch et al. (2003)18S R TAATGATCCTTCCGCAGGTT Englisch et al. (2003)18S F700+ AATTCCAGCTTCAGCAGCAT This study28S 28SF TTAGTAGGGGCGACCGAACAGGGAT Hou et al. (2007)28S700F AAGACGCGATAACCAGCCCACCA Hou et al. (2007)28S1000R GACCGATGGGCTTGGACTTTACACC Hou et al. (2007)28SR GTCTTTCGCCCCTATGCCCAACTG Hou et al. (2007)28Sa TTGGCGACCCGCAATTTAAGCAT Cristescu and Hebert (2005)28Sb CCTGAGGGAAACTTCGGAGGGAAC Cristescu and Hebert (2005)COI LCO1490 GGTCAACAAATCATAAAGATATTGG Folmer et al. (1994)HCO2198 TAAACTTCAGGGTGACCAAAAAATCA Folmer et al. (1994)16S 16Stf GGTAWHYTRACYGTGCTAAG Macdonald et al. (2005)16Sbr CCGGTTTGAACTCAGATCATGT Palumbi et al. (1991) E. Kornobis et al./Molecular Phylogenetics and Evolution 58 (2011) 527–539  529  APE. The Akaike Information Criterion (AIC) (Akaike, 1974) wasused to detect the model which best fitted the data, as recom-mended by Posada and Buckley (2004).Nuclear datasets were partitioned in stem and loop regionsaccording to the consensus structure obtained in RNAalifold(Bernhart et al., 2008) for each alignment method. Variation,including the number of phylogenetically informative sites andgaps at different regions, outside hairpins, in stems and at internaland terminal loops, was scored for the four different methods.Mountain plots were used to compare the secondary structuresobtained by RNAalifold.  2.4. Phylogenetic analysis Phylogenetic trees were constructed separately for each genefragment as the available number of taxa for the different genesvaried, and also based on combined datasets. The analysis basedondifferentgenesmayprovideastrongersupportfortheobservedclusters of species, or indicate some gene-specific evolutionarydivergence.Search for optimal trees was carried out with MrBayes v3.1.2(Ronquist and Huelsenbeck, 2003) and PhyML (Guindon and Gascuel, 2003) using the pre-specified evolutionary model (withthe lowest AIC value). MrBayes is particularly suited for our analy-sis since it allows partitioning of the data according to secondarystructure and implementation of indels as morphological charac-ters. In MrBayes, the parameters of the selected model were opti-mized during searches as recommended by Ronquist andHuelsenbeck (2003), running two independent MCMC with onecold and three hot chain searches during 2  10 6 generations forthe16Sand4  10 6 generationsfortheothergenes,sampledevery100 generations. In most cases a 10% ‘‘burn-in’’ was estimated suf-ficient after checking graphically for convergence to stable   ln L scoreswithTracer(RambautandDrummond,2003).Wecomparedposterior probabilities of the splits betweenruns, as well as duringruns using default parameters in the program AWTY (Nylanderet al., 2008). High correlation was observed between posteriorprobabilities of the splits between runs (Pearson correlation test: r   >0.98,  p  <2.2  10  16 ) and no particular trend diagnosing lackof convergence was observed (see Supplementary materialFig. A1). The secondary structure and the indels were taken intoaccount for the nuclear genes with commands implemented inMrBayes. The doublet model of evolutionwas applied to the stemsin order to take into account compensatory mutations. Two differ-ent partitions were used in order to test the sensitivity of the treetopologytothepartitioning:runswereperformedconsideringfourdifferent partitions (outside hairpins, stems, internal and terminalloops) and for just two partitions (stems and loops). Indels werecoded as binary characters and included as a morphological data-set. PhyML does not provide these options. Branch supports wereassessed by using the approximate likelihood ratio test (aLRT) inPhyML, applying the non-parametric method based on a Shimoda-ira–Hasegawa-like procedure (Anisimova and Gascuel, 2006). BothMrBayes and PhyML trees were rooted by the most closely relatedtaxa available in Genbank outside the superfamily: the 18S with Megaluropus longimerus  and the 28S with  Gammarus abstrusus (see Table 1) (Englisch et al., 2003). The secondarystructureof the 16S fragment was not takenintoaccount in the phylogenetic reconstruction, due to the short se-quences. A codon model of evolution was applied to the completeCOI dataset in MrBayes, which allows for different rates for synon-ymous and non-synonymous substitutions as well as on 0-folddegenerated codon positions in order to avoid phylogenetic signalsaturation. The COI and 16S phylogenies were rooted by  G. abstru-sus  (see Table 1), as used in previous amphipod phylogenies (Eng- lischetal.,2003).Inadditiontothephylogeniescomputedforeachgene, we reconstructed trees following the same procedure as de-scribed above, separately with two combined datasets: one withthe nuclear genes (18S and 28S), and another with the mitochon-drial genes (COI and 16S).The results from the different alignment methods were com-pared by considering the variation in number of trees obtainedwith MrBayes (Ronquist and Huelsenbeck, 2003) and their likeli-hoodsusingTracer(RambautandDrummond,2003),thensumma-rized with the mean values and high posterior density interval(HPD). As the overall likelihood value depends on the number of bases compared, the best alignment method was selected afteradjusting the log likelihoods of the trees using ANCOVA, with thenumber of bases in the alignment as a covariate.  2.5. Variation within Crangonyctoidea compared with other amphipodsuperfamilies Pairwise evolutionary distances between 18S and 28S se-quences, aligned with ClustalW2, were calculated using the APEpackage with the evolutionary model selected from the PhyML analysis. The distances were grouped at different taxonomic levels(withinandbetweengenera,withinandbetweenfamilies)inorderto assess both the taxonomic status of the Icelandic species and toevaluate the classification within the family Crangonyctidae. Forcomparing divergence within this family with divergence withinother amphipod families, we computed distances at the same tax-onomic levels for four other amphipod superfamilies available forthe 18S in Genbank (Gammaroidea, Lysianassoidea, Talitroidea,Eusiroidea) and three superfamilies for the 28S (Gammaroidea,Lysianassoidea and Talitroidea). 3. Results  3.1. Sequences Sequences of the 18S gene ranged in length from 2310bp to2440bp (Genebank accession numbers: HQ286012–HQ286018)while the 28S sequences (HQ286019–HQ286020 and HQ286022–HQ286024) ranged from 850 to 1440bp. The length of the se-quences retrieved from Genbank varied from 1150 to 2500bp forthe 18S and from 820 to 1290bp for the 28S sequences.  Crymosty- gius thingvallensis  had the longest fragments among the Crang-onyctoidea sequences for both 18S and 28S genes, reflecting itsunique phylogenetic status within the superfamily. The variationin length is due to several indels, ranging from 1 to 53 bases. Thegenetic distances observed between clones from the same individ-ual never exceeded0.9%, irrespective of the alignment methodandthenucleargene.Wethereforechoseasingleclonefromeachindi-vidual as representativeof eachspecies. The 16S sequences rangedfrom 362 to 411bp (Genebank accession numbers: HQ286000–HQ286011). No indels and no stop codons were observed in the582bp long COI alignment (HQ286025–HQ286037).  3.2. Alignment and secondary structure Differentalignment methodsresultedindifferentalignmentsof sequences as reflected by the overall lengths and the number of phylogeneticinformativesitesforeachRNAgenedataset(Table3).The number of phylogenetically informative sites and sites wheregaps were introduced varied among the alignment methods. Thenumber of gaps introduced was smallest with ClustalW2 and larg-est with FSA, for 18S: 532 (ClustalW2), 724 (MAFFT), 1277 (FSA)and 664 (RNAsalsa), and for 28S: 446 (ClustalW2), 474 (MAFFT),1094 (FSA) and 554 (RNAsalsa). Variation among species dependson the regions of secondary structure. Stems were generally least 530  E. Kornobis et al./Molecular Phylogenetics and Evolution 58 (2011) 527–539  variable: in 18S, the proportion of variable sites ranged from0.055to 0.072, and the largest differences were found among the loops(0.116–0.175). The difference in number of informative sites be-tween four regions (outside hairpins, stems, internal and terminalloops) was significant, independent of the alignment (Fisher exacttests:  P   <3.44  10  6 ), for both 18S and 28S datasets.The mean sequence identity for all nuclear gene alignmentswas high, ranging from 71% to 95%, and is substantially higherthan the percentage needed to ensure the alignment accuracyrequired for phylogenetic reconstruction (66%, see Hall, 2008;Kumar and Filipski, 2007). The 16S mean sequence identity val-ues are more marginal and ranged from 60% to 70%; the max-imum value was obtained with the FSA alignment. Based on theAIC, the best-fit model, independently chosen for all genes, wasthe GTR+I+G model (Lanave et al., 1984). The TN93 model(Tamura and Nei, 1993), the most similar substitution modelto GTR and implemented in R, was used for pairwise distancecalculations.Similar consensus secondary structures were obtained with thedifferentalignmentmethodsofthe18Sand28Sgenesasshownbythemountainplots (Fig. 1), exceptforthefirst part of theRNAsalsaalignmentofthe18Sgenewhichexhibitsmuchshorterstemstruc-tures than those observed with other alignment methods. The FSAalignments induced a consensus secondary structure with rela-tively longer loops for both genes, along with shorter stem struc-tures for the first part of the 18S. No difference was observedbetweenphylogenetictreesconstructedconsideringthefourparti-tions based on the secondary structure (outside hairpins, stems,internal and terminal loops) or two partitions (stems and loops)in MrBayes.The sample of most probable trees after ‘‘burn-in’’ in Bayesiananalysis varied considerably in size between the datasets alignedwith different methods. Samples of trees based on FSA and Clu-stalW2 were dominated by few, highly probable, different topol-ogies whereas MAFFT-aligned datasets produced a sample of numerous different topologies with low posterior probabilities.RNAsalsa showed intermediate numbers of most probable trees.The log likelihood of the trees based on the different alignmentsshowed clear patterns which depended on the length of the se-quence alignments (ANCOVA, adjusted  R 2 =0.998) (Fig. 2). TheFSA alignments produced significantly the most likely trees inall cases (Table 3 and Fig. 2). FSA is a conservative method, avoid- ing stringently false homologies, and consequently shows align-ments with less informative sites (Table 3). For these reasonswe chose to present the trees based on FSA alignment techniquesand discuss the differences obtained with the other alignmentmethods.The COI trees reconstructed with the different datasets weresimilar in topology. To avoid a misleading phylogenetic signaldue to saturation, we choose the conservative strategy to presentthe tree based on 0-fold degenerated sites.  3.3. Phylogenetic relationships among species of Crangonyctoidea,based on nuclear genes Species belonging to the Crangonyctidae forma well-supportedmonophyletic group for both 18S and 28S genes (Fig. 3), indepen-dently of the alignment method (posterior probability:  pp  >0.95). Crymostygius thingvallensis  is clearly differentiated from both Nip-hargidae and Crangonyctidae based on both the 18S and 28S phy-logenies (Fig. 3). The 28S dataset aligned with ClustalW2 and the18SdatasetalignedwithRNAsalsaaretheonlydatasetssupportingan early divergence of the Crymostygidae from Niphargidae andCrangonyctidae, whichcluster together(Table4). All other phylog-enies clustered  Crymostygius thingvallensis  together with the Cran-gonyctidae. The family Crangonyctidae is composed of twomonophyletic groups (Fig. 3 and Table 4). One group is formed by the genera  Bactrurus  and  Stygobromus  clustering together withthe European  Synurella ambulans  and species of   Crangonyx  fromEurasia (i.e.  Crangonyx chlebnikovi  for the 18S and both  C. chlebnik-ovi  and  Crangonyx subterraneus  for the 28S). The other group iscomposed of species of   Crangonyx  fromNorth America and Iceland(in both phylogenies) and the species of   Synurella  from NorthAmerica (18S phylogeny only). Independent from the alignmentmethod and the gene studied, all phylogenies support that  Crang-onyx  is clearly polyphyletic (Fig. 3 and Table 4). For example, spe- cies of   Crangonyx  from Eurasia ( C. chlebnikovi  and  C. subterraneus) are more closely related to  Stygobromus  and  Bactrurus  (  pp  >0.95)than to species of   Crangonyx  from North America and Iceland(Fig. 3a). Similarly,  S. ambulans  fromEuropeis more closelyrelatedto species of   Stygobromus  and  Bacturus  (  pp  >0.95) than to speciesof  Synurella speciesfromNorthAmerica( Synurella sp.and Synurelladentata ), based on the 18S dataset. This dataset also supports agroupformedby  Amurocrangonyx arsenjevi , Bactrurus , Stygobromus ,and the species of   Crangonyx  and  Synurella  fromEurasia (  pp  >0.95,see Fig. 3a).The 18S phylogenies also show an early divergence of   Crangonyxislandicus  from other  Crangonyx  from North America, which aremorecloselyrelatedto Synurella sp.and S. dentata  (Fig.3a).Thispat-tern is observed, though less frequently, for 28S phylogenies andcombined datasets, depending on the alignment and the phyloge-netic methods used (Table 4). Moreover, the divergence observedbetween  Crangonyx islandicus  and other  Crangonyx  greatly exceedthe one observed between species of the genera  Stygobromus  and Bactrurus  or even between those species and  A. arsenjevi  (Fig. 3a).Conversely,phylogeniesbasedonthe28Sdataset,notencompassingspecies of   Synurella  from North America, support the monophyly of the group formed by species of   Crangonyx  from North America and Crangonyx islandicus  (  pp >0.95,seeFig.3b).Thespeciesof  Crangonyx from North America ( Crangonyx forbesi ,  Crangonyx  sp.,  Crangonyx pseudogracilis  and  Crangonyx floridanus ) only clustered together inthe phylogeny based on MAFFT alignment for the 18S gene. None-theless, the early divergence of   C. forbesi  from the other species isonly supported by informative indels. All 18S phylogenies con-structed without indels as morphological characters strongly sup-ported the monophyly of the North American species of   Crangonyx (  pp >0.95, data not shown).The different alignment methods result in slight differences intopology, which are mostly in the external nodes of the trees,grouping together species of   Stygobromus ,  S. ambulans ,  C. chlebnik-ovi  and  C. subterraneus . These changes appeared even for nodeswhich were highly supported in the phylogeny based on FSAalignment. The monophyly of the genus  Bactrurus , based on 18S,  Table 3 Comparison of the likelihoods between genes and alignment methods. The tablepresents the results from the Bayesian analysis for the four alignment methods. ln  L :the log likelihoods of the Bayesian trees, HPD: high posterior density interval of the  log likelihoods, bp: number of base pair in the alignment, IS: number of phylogenetically informative sites. Methods Genes -lnL 95%HPD bp IS18S 12020 12010-12040 2629 439ClustalW 28S 9859 9849-9870 1561 51816S 4592 4582-4602 424 26218S 12140 12130-12150 3224 352FSA 28S 9984 9974-9995 2115 42716S 4459 4440-4471 689 24918S 11350 11340-11360 2840 426MAFFT 28S 9588 9578-9598 1660 48116S 4553 4543-4564 431 25518S 11910 11900-11920 2663 345RNAsalsa 28S 10040 10030-10050 1595 46616S 4299 4290-4309 455 260 E. Kornobis et al./Molecular Phylogenetics and Evolution 58 (2011) 527–539  531
View more
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!