Your SlideShare is downloading. ×

Gutell 097.jphy.2006.42.0655


Published on

Alverson A.J., Cannone J.J., Gutell R.R., and Theriot E.C. (2006). …

Alverson A.J., Cannone J.J., Gutell R.R., and Theriot E.C. (2006).
The Evolution of Elongate Shape in Diatoms.
Journal of Phycology, 42(3):655-668.

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total Views
On Slideshare
From Embeds
Number of Embeds
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

No notes for slide


  • 1. THE EVOLUTION OF ELONGATE SHAPE IN DIATOMS1Andrew J. Alverson2Section of Integrative Biology and Texas Memorial Museum, The University of Texas at Austin, 1 University Station, Austin, Texas78712, USAJamie J. Cannone, Robin R. GutellSection of Integrative Biology and Institute for Cellular and Molecular Biology, The University of Texas at Austin, 1 UniversityStation, Austin, Texas 78712, USAandEdward C. TheriotSection of Integrative Biology and Texas Memorial Museum, The University of Texas at Austin, 1 University Station, Austin, Texas78712, USADiatoms have been classified historically as ei-ther centric or pennate based on a number of fea-tures, cell outline foremost among them. Theconsensus among nearly every estimate of the dia-tom phylogeny is that the traditional pennate dia-toms (Pennales) constitute a well-supported clade,whereas centric diatoms do not. The problem withthe centric–pennate classification was highlightedby some recent analyses concerning the phyloge-netic position of Toxarium, whereby it was conclud-ed that this ‘‘centric’’ diatom independently evolvedseveral pennate-like characters including an elon-gate, pennate-like cell outline. We performed sev-eral phylogenetic analyses to test the hypothesisthat Toxarium evolved its elongate shape indepen-dently from Pennales. First, we reanalyzed the orig-inal data set used to infer the phylogenetic positionof Toxarium and found that a more thorough heu-ristic search was necessary to find the optimal tree.Second, we aligned 181 diatom and eight outgroupSSU rDNA sequences to maximize the juxtaposi-tioning of similar primary and secondary structureof the 18S rRNA molecule over a much broadersampling of diatoms. We then performed a numberof phylogenetic analyses purposely based on dispa-rate sets of assumptions and found that none ofthese analyses supported the conclusion thatToxarium acquired its pennate-like outline inde-pendently from Pennales. Our results suggest thatelongate outline is congruent with SSU rDNA dataand may be synapomorphic for a larger, more in-clusive clade than the traditional Pennales.Key index words: 18S rDNA; Bacillariophyceae;centric; diatoms; Pennales; pennate; secondarystructure; small subunit rDNA; ToxariumAbbreviations: BPP, Bayesian posterior probabili-ty; GTR, General Time Reversible model of se-quence evolution; I, proportion of invariable sites;ML, maximum likelihood; MP, maximum parsimo-ny; NJ, neighbor joining; TBR, tree bisection re-connection; C, gamma distributionInterest in the classification of diatoms dates back toat least 1896 when diatoms with a round cell outline(centrics) were distinguished from those with a longand narrow cell outline (pennates) (Schu¨tt 1896). Be-yond cell outline, centric diatoms generally are ooga-mous, and pennate diatoms generally are isogamousor anisogamous (Edlund and Stoermer 1997, Chepur-nov et al. 2004). These sexual characteristics have re-inforced the traditional centric–pennate split, and insome cases, have taken precedence over cell outline inthe classification of taxa that are not unambiguouslycentric or pennate (Hasle et al. 1983). The distinctionbetween centrics and pennates persists largely out ofconvenience, and despite the fact that early specula-tions about the diatom phylogeny suggested that cen-trics were not a natural evolutionary lineage (Simonsen1979, Kociolek et al. 1989). Molecular phylogeneticanalyses consistently have shown that centric diatomsgrade into pennate diatoms, so only the ‘‘true’’ penn-ate diatoms (Pennales) are a monophyletic group(Medlin et al. 1993, 1996a, b, Medlin and Kaczmarska2004, Sorhannus 2004, see Alverson and Theriot 2005for a review).Because of this grade-like nature of relationships,some diatoms have a combination of ancestral(plesiomorphic) ‘‘centric’’ characters and derived1Received 31 August 2005. Accepted 8 March 2006.2Author for correspondence: e-mail Phycol. 42, 655–668 (2006)r 2006 by the Phycological Society of AmericaDOI: 10.1111/j.1529-8817.2006.00228.x
  • 2. (apomorphic) ‘‘pennate’’ characters. This perceivedcharacter conflict has obscured the higher level classi-fication of a number of elongate centric diatoms. Forexample, in establishing Cymatosiraceae, Hasle andSyvertsen (1983) struggled as to whether the familyshould be considered centric or pennate. They exam-ined numerous characters and ultimately concludedthat a majority of them affiliated Cymatosiraceae withother centrics. Two features traditionally associatedwith centric diatoms, flagellated gametes and develop-ment from an annulus, were particularly important totheir decision.Toxarium undulatum Bailey is another diatom withfeatures of both centrics and pennates. Toxarium is amonotypic genus with a distinctly elongate cell shape,though it lacks many of the other features traditionallyused to circumscribe Pennales. For example, poroidsare scattered on the valve face rather than being or-ganized around a longitudinal sternum and associatedtransapical ribs (Round et al. 1990, Kooistra et al.2003a). Together, these two structures impose the or-ganization of poroids into striae aligned perpendicularto the longitudinal axis of the cell in Pennales (Roundet al. 1990, Kooistra et al. 2003a). Sexual characteris-tics of Toxarium have not been observed. Kooistra et al.(2003a) sought to resolve the classification of Toxarium(i.e. whether it is centric or pennate) through phylo-genetic analysis of SSU rDNA sequences. They per-formed a number of phylogenetic analyses andconcluded that Toxarium represented independent evo-lution of several ‘‘true’’ pennate-like characteristics,including valve outline, which was described as ‘‘elon-gate,’’ ‘‘pennate-like,’’ and ‘‘pennate’’ (Kooistra et al.2003a, p. 186). We will use this convention, likewiserecognizing that some members of phylogeneticallyisolated lineages have independently evolved elongateshape (e.g. Stephanodiscus rhombus Mahood). All refer-ences to ‘‘pennate shape,’’ ‘‘pennate outline,’’ or ‘‘elon-gate outline’’ in this article describe an elongate,biangular, and/or oval valve outline (Round et al. 1990).In this article, we test the hypothesis that Toxariumevolved its elongate cell outline independently fromPennales. First, we reanalyzed the 40-taxon data set ofKooistra et al. (2003a). Second, because a growing lit-erature emphasizes the critical importance of taxonsampling on the accuracy of phylogenetic inferences(Hillis 1998, Pollock et al. 2002, Zwickl and Hillis 2002,Hillis et al. 2003), we also analyzed 189 SSU rDNA se-quences (181 diatom, eight outgroup) aligned by max-imizing the proper juxtaposition of homologousprimary and secondary structure. We analyzed thisalignment using two methods based on distinctly dif-ferent assumptions as a rough control over robustnessof our results to analytical details. We first used a Bay-esian method that accounts for nonindependence ofthe nucleotides that are base paired in the secondarystructure of the SSU rRNA molecule, and then weused equally weighted maximum parsimony, ignoringall information about nucleotide pairings in the SSUrRNA molecule. In the end, we found little evidence tosupport the hypothesis that Toxarium acquired its elon-gate outline independently from Pennales.ANALYSISSummary of Kooistra et al. (2003a) analysis. Thealignment combined 98 diatom SSU rDNA sequencesfrom Toxarium and other unspecified diatom taxa.Bolidomonas mediterranea and B. pacifica were used asoutgroup taxa. This 100-taxon matrix was alignedmanually, and ModelTest 3.0 (Posada and Crandall1998) was used to determine that the GTR þ G þ Imodel provided the best fit to their data set. A non-parametric bootstrap analysis (1000 pseudoreplicates)on the 100-taxon matrix was performed. For eachpseudoreplicate of the bootstrap analysis, the optimaltree was built with the neighbor joining (NJ) algorithmon distances corrected with the GTR þ G þ I model.The resulting bootstrap consensus tree was used as thebasis for an unspecified number of Kishino–Hasegawatests (Kishino and Hasegawa 1989), with the goal ofidentifying taxa that could be deleted without ‘‘im-pairing recovery of the phylogenetic position ofToxarium’’ (Kooistra et al. 2003a, p. 191). TheseKishino–Hasegawa tests were used to justify deletionof most radial centrics and pennates from the 100-taxon matrix because tree topologies with constraintsof Toxarium þ radial centrics and Toxarium þ Pennalesgave likelihood scores significantly worse than that ofthe initial bootstrap consensus tree. All radial and mul-tipolar centrics separated by pairwise distances greaterthan 0.2 were also removed (Kooistra et al. 2003a,p. 191). In the end, these two criteria were used todelete 60 diatom taxa from the original 100-taxonmatrix. ModelTest 3.0 again was used to determinethat the GTR þ G þ I model provided the best fit to the40-taxon matrix. They performed a ML analysis andfixed the values of their model parameters to thoseoutput by ModelTest. No description of their heuristicsearch was provided. They stated that a bootstrapanalysis on their initial 100-taxon matrix using fullheuristic searches in ML ‘‘would take years’’ and citedthis constraint as the primary reason for deleting 60taxa from the matrix. Ultimately, however, NJ wasused to find the optimal tree in each pseudoreplicateof the bootstrap analysis, rather than heuristic searcheswith ML.METHODSMultiple sequence alignment. All SSU rDNA sequences fromdiatoms available before April 19, 2004 were obtained fromGenBank for alignment (Table 1). Additional informationabout the sequences, sequence alignment, and secondarystructure analysis are available at: SSU rDNA sequences were aligned manually with thealignment editor ‘‘AE2’’ (developed by T. Macke, Scripps Re-search Institute, San Diego, CA—Larsen et al. 1993), which wasdeveloped for Sun Microsystems’ (Santa Clara, CA, USA) work-stations running the Solaris operating system. The manualalignment process involves first aligning positionally homolo-gous nucleotides (i.e. those that map to the same locations in thesecondary and tertiary structure models) into columns in theA. J. ALVERSON ET AL.656
  • 3. TABLE 1. List of GenBank annotations and accession numbers for small subunit (18S) rDNA sequences used in this study.Taxon (GenBank annotation)OutlinecodingGenBankAccession IDMelosira varians Agardh 0 X85402Melosira varians Agardh 0 AJ243065Stephanopyxis cf. broschii 0 M87330Aulacoseira baicalensis (K. Meyer) Simonsen 0 AY121821Aulacoseira islandica (O. Mu¨ller) Simonsen 0 AY121820Aulacoseira ambigua (Grunow) Simonsen 0 X85404Aulacoseira nyassensis (O. Mu¨ller) Simonsen 0 AJ535187Aulacoseira nyassensis (O. Mu¨ller) Simonsen 0 AY121819Aulacoseira distans (Ehrenburg) Simonsen 0 X85403Aulacoseira baicalensis (K. Meyer) Simonsen 0 AJ535186Aulacoseira baicalensis (K. Meyer) Simonsen 0 AJ535185Aulacoseira islandica (O. Mu¨ller) Simonsen 0 AJ535183Aulacoseira skvortzowii Edlund, Stoermer, and Taylor 0 AJ535184Aulacoseira skvortzowii Edlund, Stoermer, and Taylor 0 AY121822Aulacoseira subarctica (O. Mu¨ller) Haworth 0 AY121818Actinocyclus curvatulus Janisch 0 X85401Actinoptychus seniarius (Ehrenberg) He´ribaud 0 AJ535182Coscinodiscus radiatus Ehrenberg 0 X77705Rhizosolenia setigera Brightwell 0 M87329Guinardia flaccida (Castracane) H. Peragallo 0 AJ535191Guinardia delicatula (Cleve) Hasle 0 AJ535192Corethron hystrix Hensen 0 AJ535179Corethron criophilum Castracane 0 X85400Corethron inerme Karsten 0 AJ535180Paralia sol (Ehrenberg) Crawford 0 AJ535174Proboscia alata Brightwell (Sundstro¨m) 0 AJ535181Leptocylindrus danicus Cleve 0 AJ535175Leptocylindrus minimus Gran 0 AJ535176Thalassiosira oceanica Hasle 0 AF374479Thalassiosira weissflogii (Grunow) Fryxell & Hasle 0 AF374477Thalassiosira rotula Muenier 0 AF462058Thalassiosira rotula Muenier 0 AF462059Thalassiosira rotula Muenier 0 AF374480Rhizosolenia imbricate Brightwell 0 AJ535178Rhizosolenia similoides Cleve-Euler 0 AJ535177Papiliocellulus elegans Hasle, von Stosch et Syvertsen 2 X85388Ditylum brightwellii (West) Grunow in van Heurck 2/3/4 X85386Ditylum brightwellii (West) Grunow in van Heurck 2/3/4 AY188181Ditylum brightwellii (West) Grunow in van Heurck 2/3/4 AY188182Bellerochea malleus (Brightwell) van Heurck 2/3/4 AF525671Lithodesmium undulatum Ehrenberg 3/4 Y10569Helicotheca tamesis (Schrubsole) Ricard 2 X85385Lauderia borealis Cleve 0 X85399Porosira pseudodenticulata Hustedt (Jouse´) 0 X85398Detonula confervacea (Cleve) Gran 0 AF525672Thalassiosira guillardii Hasle 0 AF374478Thalassiosira weissflogii (Grunow) Fryxell & Hasle 0 AJ535170Skeletonema menzellii Guillard, Carpenter et Reim 0 AJ536450Skeletonema menzellii Guillard, Carpenter et Reim 0 AJ535168Skeletonema pseudocostatum Medlin 0 AF462060Skeletonema sp. 0 AJ535165Skeletonema subsalsum (Cleve-Euler) Bethge 0 AJ535166Skeletonema costatum (Grev.) Cleve 0 X85395Skeletonema costatum (Grev.) Cleve 0 X52006Skeletonema pseudocostatum Medlin 0 X85393Skeletonema pseudocostatum Medlin 0 X85394Planktoniella sol (Wallich) Schu¨tt 0 AJ535173Thalassiosira eccentrica (Ehrenb.) Cleve 0 X85396Thalassiosira pseudonana Hasle & Heimdal 0 AJ535169Thalassiosira sp. 0 AJ535171Thalassiosira pseudonana Hasle & Heimdal 0 AF374481Thalassiosira rotula Muenier 0 X85397Cyclotella meneghiniana Ku¨tzing 0 AJ535172Cyclotella meneghiniana Ku¨tzing 0 AY496206Cyclotella meneghiniana Ku¨tzing 0 AY496207Cyclotella meneghiniana Ku¨tzing 0 AY496210Cyclotella meneghiniana Ku¨tzing 0 AY496211Cyclotella meneghiniana Ku¨tzing 0 AY496212Cyclotella meneghiniana Ku¨tzing 0 AY496213EVOLUTION OF ELONGATE SHAPE IN DIATOMS 657
  • 4. TABLE 1 (Continued)Taxon (GenBank annotation)OutlinecodingGenBankAccession IDCyclotella cf. scaldensis 0 AY496208Cyclotella cf. scaldensis 0 AY496209Chaetoceros sp. 2 X85390Chaetoceros didymus Ehrenberg 2 X85392Chaetoceros debilis Cleve 2 AY229896Chaetoceros gracilis Schu¨tt 2 AY229897Chaetoceros curvisetus Cleve 2 AY229895Biddulphiopsis titiana (Grunow) von Stosch et Simonsen 2 AF525669Lampriscus kittonii Schmidt 0/3/4 AF525667Eucampia antarctica (Castracane) Mangin 2 X85389Chaetoceros rostratus Lauder 2 X85391Chaetoceros sp. 2 AJ535167Pleurosira cf. laevis 0 AJ535188Cymatosira belgica Grunow 2 X85387Pleurosira laevis (Ehrenberg) Compe`re 0 AF525670Odontella sinensis (Greville) Grunow 2 Y10570Chaetoceros sp. 2 AF145226Eunotia pectinalis (Dillwyn) Rabenhorst 2 AB085832Eunotia monodon var. asiatica Skvortzow 2 AB085831Eunotia formica var. sumatrana Hustedt 2 AB085830Eunotia sp. 2 AJ535145Eunotia cf. pectinalis f. minor 2 AJ535146uncultured diatom ? AY180015Navicula cryptocephala var. veneta (Ku¨tzing) Grunow 2 AJ297724Pseudogomphonema sp. 2 AF525663Pleurosigma sp. 2 AF525664Fragilariopsis sublineata Hasle 2 AF525665Thalassiosira antarctica Combera? AF374482Nitzschia apiculata (Gregory) Grunow 2 M87334Undatella sp. 2 AJ535163Rossia sp. 2 AJ535144Amphora cf. capitellata 2 AJ535158Amphora montana Krasske 2 AJ243061Planothidium lanceolatum (Bre´bisson ex Ku¨tzing) F. E. Round & L. Bukhtiyarova 2 AJ535189Lyrella atlantica (Schmidt) D. G. Mann 2 AJ544659Cymbella cymbiformis C. Agardh 2 AJ535156Surirella fastuosa var. cuneata (A. Schmidt) H. Peragallo & M. Peragallo 2 AJ535161Campylodiscus ralfsii Gregory 2 AJ535162Anomoeoneis sp. haerophora 2 AJ535153Gomphonema pseudoaugur Lange-Bertalot 2 AB085833Gomphonema parvulum Ku¨tzing 2 AJ243062Entomoneis cf. alata 2 AJ535160Lyrella sp. 2 AJ535149Eolimna subminuscula (Manguin) Gerd Moser 2 AJ243064Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544645Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544651Sellaphora pupula f. capitata (Skvortsov & K. I. Mey.) Poulin in Poulin, Hamilton & Proulx 2 AJ535155Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544646Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544647Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544648Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544649Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544650Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544652Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544653Sellaphora pupula (Ku¨tzing) Mereschkovsky 2 AJ544654Sellaphora laevissima (Ku¨tzing) D. G. Mann 2 AJ544655Sellaphora laevissima (Ku¨tzing) D. G. Mann 2 AJ544656Pinnularia cf. interrupta 2 AJ544658Pinnularia sp. 2 AJ535154Navicula pelliculosa (Bre´bisson ex Ku¨tzing) Hilse 2 AJ544657Amphora cf. proteus 2 AJ535147Eolimna minima (Grunow) Lange-Bertalot 2 AJ243063Cylindrotheca closterium (Ehrenberg) Reimann et Lewin 2 M87326Bacillaria paxillifer (O. F. Mu¨ller) Hendey 2 M87325Nitzschia frustulum (Ku¨tzing) Grunow 2 AJ535164Navicula diserta Hustedt 2 AJ535159Pseudogomphonema sp. 2 AJ535152Phaeodactylum tricornutum Bohlin 2 AJ269501Achnanthes bongrainii (M. Peragallo) A. Mann 2 AJ535150Achnanthes sp. 2 AJ535151A. J. ALVERSON ET AL.658
  • 5. alignment, maximizing their sequence and structure similarity.For regions with high similarity between sequences, the nucleo-tide sequence is sufficient to align sequences with confidence.For more variable regions in closely related sequences or whenaligning more distantly related sequences, however, a high-qual-ity alignment only can be produced when additional informa-tion (here, secondary and/or tertiary structure data) is included.The underlying SSU rRNA secondary structure model in-itially was predicted with covariation analysis (Gutell et al.1985, 1992). Approximately 98% of the predicted model basepairs were present in the high-resolution crystal structure fromthe 30S ribosomal subunit (Gutell et al. 2002). This model(based on the bacterium Escherichia coli) has been extended tothe eukaryotic SSU rRNA (Cannone et al. 2002), further usingcovariation analysis to assess eukaryote-specific features. Theadditional constraints of the eukaryotic model were used torefine the alignment of the diatom sequences iteratively untilpositional homology was established confidently for the entireTABLE 1 (Continued)Taxon (GenBank annotation)OutlinecodingGenBankAccession IDPseudo-nitzschia multiseries (Hasle) Hasle 2 U18241Pseudo-nitzschia pungens (Grunow ex Cleve) Hasle 2 U18240Cocconeis cf. molesta 2 AJ535148Encyonema triangulatum Ku¨tzing 2 AJ535157Uncultured eukaryote ? AY082977Uncultured eukaryote ? AY082992Uncultured diatom ? AY180016Uncultured diatom ? AY180014Peridinium foliaceum endosymbiont ? Y10567Peridinium balticum endosymbiont ? Y10566Uncultured diatom ? AY180017Asterionellopsis glacialis (Castracane) F. E. Round 2 X77701Talaroneis posidoniae Kooistra & De Stefano 2 AY216905Asterionellopsis glacialis (Castracane) Round 2 AY216904Asterionellopsis kariana (Grunow) F.E. Round 2 Y10568Convoluta convoluta diatom endosymbiont ? AY345013Cyclophora tenuis Castracane 2 AJ535142Diatoma tenue Agardh 2 AJ535143Diatoma hyemalis (Roth) Heiberg 2 AB085829Grammatophora oceanica Ehrenberg 2 AF525655Grammatophora gibberula Ku¨tzing 2 AF525656Grammatophora marina (Lyngbe) Ku¨tzing 2 AY216906Licmophora juergensii Agardh 2 AF525661Rhabdonema sp. 2 AF525660Asterionella formosa Hassall 2 AF525657Rhaphoneis belgica (Grunow in van Heurck) Grunow in van Heurck 2 X77703Staurosira construens Ehrenberg 2 AF525659Nanofrustulum shiloi (Lee, Reimer & McEmery) Round, Hallsteinsen et Paasche 2 AF525658Thalassionema sp. 2 AJ535140Striatella unipunctata (Lyngbye) Agardh 2 AF525666Hyalosira delicatula Ku¨tzing 2 AF525654Fragilaria sp. 2 AJ535141Fragilaria cf. islandica 2 AJ535190Fragilaria striatula Lyngbye 2 X77704Tabularia tabulata (Agardh) D.M. Williams & Round 2 AY216907Fragilariforma virescens (Ralfs) D.M. Williams & Round 2 AJ535137Synedra ulna (Nitzsch) Ehrenberg 2 AJ535139Fragilaria crotonensis Kitton 2 AF525662Thalassionema nitzshcioides (Grunow) Hustedt 2 X77702Synedra sp. 2 AJ535138Toxarium undulatum Bailey 2 AF525668uncultured marine diatom ? AF290085uncultured diatom ? AY180020Bolidomonas mediterranea Guillou et Chrete´innot-Dinet NA AF123596Bolidomonas pacifica Guillou et Chrete´innot-Dinet NA AF167153Bolidomonas pacifica Guillou et Chrete´innot-Dinet NA AF167154Bolidomonas pacifica Guillou et Chrete´innot-Dinet NA AF167155Bolidomonas pacifica Guillou et Chrete´innot-Dinet NA AF167156Bolidomonas pacifica Guillou et Chrete´innot-Dinet NA AF123595Bolidomonas pacifica Guillou et Chrete´innot-Dinet NA AF167157The outline of the valve was coded so this character could be mapped onto phylogenetic trees: 0, circular/subcircular; 2, elongate/biangular; 3, triangular; 4, quadrangular. The outline codings were based on generic descriptions from Round et al. (1990). Un-identified sequences were coded with ‘‘?,’’ and outgroup taxa were coded ‘‘NA.’’aThis sequence is presumably misannotated because it is consistently placed within the pennate diatoms. Small subunit rDNAsequences from cultures confirmed to be T. antarctica are placed, as expected, with other species of Thalassiosirales (A. J. A., un-published data).EVOLUTION OF ELONGATE SHAPE IN DIATOMS 659
  • 6. data matrix. The final SSU rDNA alignment contained 189sequences, with a final length of 2034 columns, after discardingdistant outgroup and partial sequences.Secondary structure model diagrams. Three older diatom sec-ondary structure model diagrams were updated to reflect thecurrent version of the diatom SSU rRNA structure model.From these diagrams, 32 new secondary structure model dia-grams were generated for complete (or nearly complete)sequences representing the major lineages of the diatomphylogeny and the Bolidomonas outgroup. The secondarystructure diagrams were drawn with the interactive second-ary structure program XRNA (developed in C for SunMicrosystems’ workstations running the Solaris operatingsystem by B. Weiser and H. Noller, University of California,Santa Cruz). All structure diagrams are available at: analyses. 40-taxon analyses. We attempted toreplicate the phylogenetic analysis of Kooistra et al. (2003a)with the same 40-taxon alignment (provided by W. Kooistra).We analyzed this alignment with ML, fixing the parametersof the GTR þ G þ I model to those set by Kooistra et al.2003a, p. 192). Details of their search algorithm were notavailable, so for our analysis of their data set, the ML tree wasfound using a heuristic search with 100 random addition se-quence replicates and TBR branch swapping. This analysisdid not result in the tree topology reported by Kooistra et al.(2003a), so we then experimented with different combina-tions of optimality criteria, model parameters, and heuristicsearch settings until we were able to reproduce their tree to-pology. All ML analyses were done with PAUP* (ver. 4.0b10,Sinauer Associates Inc., Sunderland, MA, USA).189-taxon analyses. The final structurally aligned matrixincluded 181 diatom sequences, with seven Bolidomonas se-quences and one chrysophyte (misannotated in GenBank as‘‘uncultured diatom’’) comprising the outgroup.For Bayesian analyses, the posterior probability distributionwas estimated using Metropolis-Coupled Markov Chain MonteCarlo (MCMCMC) as implemented in MrBayes (ver. 3.0b4—Ronquist and Huelsenbeck 2003). From an analytical perspec-tive, it is important to distinguish between base-paired nucleo-tides in rRNA helices and unpaired nucleotides in rRNA loopsand bulges. Nucleotide pairings in helices are maintained des-pite evolutionary changes in the nucleotide sequences. Sub-stitutions in one of the two base-paired nucleotides are oftencoordinated with a substitution at the second paired nucleotideto maintain canonical pairing (G:C, A:U, and G:U), which pre-serves the integrity of the helix. Phylogenetic methods assumethat all characters have evolved independently, an assumptionthat is violated for the positions that are base paired in rRNAmolecules. One strategy that overcomes this type of noninde-pendence is use of a model of sequence evolution that consid-ers pairs of sites rather than considering them singly (Savillet al. 2001). Numerous ‘‘doublet’’ models have been proposed(Savill et al. 2001), one of which considers all 16 possible com-binations of nucleotide pairs. This variant is implemented inMrBayes and was used for the paired-sites partition in all anal-yses; the GTR model was specified for the unpaired-sites par-tition. For each partition, rate heterogeneity among sites wasmodeled with its own gamma distribution (‘‘unlinked’’ in MrB-ayes), estimated with four discrete rate categories; a parameterfor the proportion of invariant sites also was included in thepaired-sites partition. We used the default priors from MrB-ayes. Each run used four chains, one cold and three heated,with temperature set to 0.1. We performed two independentruns for 20,000,000 generations per run, sampling every 100generations. Proposal parameters were adjusted to facilitatemixing of the chains. We assessed convergence, determinedthe burn-in, and combined samples from the two runs usingcriteria similar to those of Nylander et al. (2004). Samples fromthe first 4,000,000 generations were discarded as the burn-infor each run, and final results were based on pooled samplesfrom stationary phase of the two runs. A 50% majority-ruleconsensus tree was calculated with the ‘‘sumt’’ command inMrBayes. We used PAUP* to calculate bipartition posteriorprobabilities by computing a 50% majority rule consensus fromthe pooled distribution of trees from stationary phase of thetwo runs.In a purposeful departure from the detailed model andunderlying assumptions of the Bayesian analysis, we also analy-zed the 189-taxon matrix with equally weighted maximumparsimony, ignoring all information about nucleotide pairingsin the SSU rRNA molecule. All parsimony analyses were runwith Winclada–Nona (W/N) and TNT (Goloboff et al. 2004).The ratchet in W/N and the full suite of TNToptions (sectorialsearch, ratchet, drift, and tree fusion) were run with defaultsettings to get an estimate of run times. Maximum parsimonycladograms from those runs were saved. Previous experiencewith the ratchet in W/N (Goertzen and Theriot 2003) on asimilarly sized SSU rDNA data set indicated that at least oneshortest tree could be found in fewer than 20 replicates andthat effectiveness of the search (in terms of changes in the strictconsensus tree) began to plateau at approximately 500 repli-cates, with 20% of the characters perturbed, so we ran an ad-ditional 500 ratchet replicates with 20% of the charactersperturbed and saved all equally most parsimonious trees.With TNT, at least one most parsimonious tree was foundwithin seconds using default settings. We increased all cycles,rounds, and repetitions an order of magnitude beyond defaultvalues, and performed 500 random addition sequences. Ineach case, we calculated the number of nodes collapsed (themore nodes collapsed among the equally most parsimonioustrees, the greater the diversity of topologies) in collections oftrees from separate runs and in the pooled collection of equallymost parsimonious trees. Diversity in the pooled collection oftrees was not increased over the maximum obtained in anyseparate W/N or TNT run by pooling trees from the separateW/N ratchet runs (default and additional settings) and the sep-arate TNT runs (default and additional settings).RESULTSStructural alignment and secondary structure mod-els. The secondary structure model for the eukaryo-tic SSU rRNA molecule was used to guide thealignment of 189 diatom and outgroup sequences,with a maximum individual sequence length of1814 nt. The final alignment contained 2034 col-umns, accounting for insertions and deletions in theset of sequences (a 12% increase with respect to thelongest sequence). Of the 2034 columns in the orig-inal alignment, 40 columns (2%) that were outsidethe 50and 30boundaries of the structure model wereexcluded, and 100 columns (5%) containing six var-iable helices and loops were excluded, leaving 1894columns (93%) available for phylogenetic analyses. Atotal of 1028 columns contained unpaired nucleo-tides in loops and bulges, whereas the remaining1006 columns (503 pairs) were base-paired in helices.The secondary structure model for T. undulatum in-cludes all of the major structural elements that arecharacteristic of the eukaryotic SSU rRNA molecule(Fig. 1). GenBank entry AF525668 is a partial sequence(1730 nt), so N’s were added to the 50(22 N’s) and 30(28 N’s) ends to represent a complete sequence(1780 nt). Nucleotides involved in long-range interac-A. J. ALVERSON ET AL.660
  • 7. tions are connected with lines. Using this secondarystructure model diagram as the reference sequence,the diatom alignment was summarized in a conserva-tion secondary structure diagram (Fig. 2). The T. un-dulatum reference sequence contains 1780 nucleotides(including the added N’s described above); all percent-- Canonical base pair (A-U, G-C)- G-A base pair- G-U base pair- Non-canonical base pairFIG. 1. Small subunit rRNA secondary structure model for the Toxarium undulatum GenBank accession number AF525668. Canonicalbase-pairs (G:C, A:U) are shown with tick marks, wobble (G:U) base-pairs are marked with small closed circles, A:G base pairs areindicated with large open circles, and all other non-canonical base pairs are shown with large closed circles.EVOLUTION OF ELONGATE SHAPE IN DIATOMS 661
  • 8. ACGUacgu- 98+ % conserved- 90-98% conserved- 80-90% conserved- less than 80% conservedFIG. 2. Conservation secondary structure diagram for Bacillariophyceae SSU rRNA, using the Toxarium undulatum SSU rRNA sec-ondary structure model (Fig. 1) as the reference sequence. The conservation diagram summarizes the alignment of 181 diatom se-quences. Symbols are present for positions that contain a nucleotide in at least 95% of the sequences in the alignment: red capital letters,the given nucleotide is conserved at 98%–100% at the position; red lower-case letters, 90%–98% conservation; black closed circles, 80%–90%; black open circles, less than 80% conserved. Other positions (not containing a nucleotide in 95% of the sequences) are shown byarcs, which are labeled with the minimum and maximum numbers of nucleotides known to exist in the region. The blue tags indicateinsertions relative to the reference sequence that are either 1–4nt in length in at least 10% of the sequences or at least 5 nt in length in atleast one sequence. The label format is (maximum length of insertion:percentage of sequences having any length insertion).A. J. ALVERSON ET AL.662
  • 9. ages in this section are calculated against this number.The 1742 (98%) positions with a nucleotide in at least95% of the sequences are shown with symbols. A totalof 1111 positions (62%) have a single nucleotide con-served in at least 98% of our diatom sequences (indi-cated in Fig. 2 with red uppercase letters). Many ofthese positions are highly conserved in an all-eukar-yote conservation diagram (available at the Compara-tive RNA Web Site: additional 313 positions (18%) were conserved at90%–98% (red lowercase letters), 188 positions (11%)at 80%–90% (closed circles), and the remaining 130positions (7%) were less than 80% conserved (opencircles). The remaining 38 (2%) positions (representedin the diagram with arcs) are the sites with insertions ordeletions in the alignment and tend to occur at theperiphery of the three-dimensional structure. Conse-quently, these regions are the most variable and hard-est to align with confidence and were excluded fromphylogenetic analyses.40-taxon analyses. We analyzed the Kooistra et al.(2003a) data set using the same optimality criterion,model (GTR þG þ I), and parameter values set byKooistra et al. (2003a) (rAC 50.9394, rAG 5 2.4904,0.01 substitutions/siteBolidomonas mediterraneaBolidomonas pacificaStephanopyxis nipponicaRhizosolenia setigeraDetonula confervaceaThalassiosira eccentricaThalassiosira rotulaDitylum brightwelliiHelicotheca thamesisLithodesmium undulatumBellerochea maleusPleurosira laevisOdontella sinensisPapiliocellulus elegansCymatosira belgicaBiddulphiopsis titianaLampriscus kittoniiToxarium undulatumEucampia antarcticaChaetoceros rostratusAsterionellopsis glacialisRhaphoneis belgicaeStriatella unipunctataPhaeodactylum tricornutumPseudogomphonema sp.Pleurosigma sp.Bacillaria paxiliferTryblionella apiculataFragilariopsis sublineataCylindrotheca closteriumAsterionella formosaNanofrustulum shiloiStaurosira construensRhabdonema sp.Hyalosira delicatulaGrammatophora oceanicaGrammatophora gibberulaLicmophora juergensiiFragilaria crotonensisThalassionema nitzschioideselongate/pennatecircular/subcircularnot applicableequivocalFIG. 3. Phylogenetic tree from maximum likelihood analysis of this alignment provided by Kooistra et al. (2003a). Parameter valuesof the GTR þ G þ I model were fixed to those used by Kooistra et al. 2003a. This tree has a higher likelihood score than the tree fromKooistra et al. 2003a and suggests that Toxarium did not evolve its elongate valve shape independently from Pennales. The tree searchused 100 random addition sequence replicates and TBR branch swapping. Each diatom name is followed by at least one generalized linedrawing, based on figures and generic descriptions from Round et al. (1990). For taxa with multiple line drawings, the drawing im-mediately following the scientific name represents the most common outline for that genus, based on Round et al. (1990).EVOLUTION OF ELONGATE SHAPE IN DIATOMS 663
  • 10. rAT 5 1.1368, rCG 5 1.1279, rCT 5 3.7238;A 5 0.2576, C 5 0.1784, G 5 0.2413; a 5 0.5716;I 5 0.3360) (Kooistra et al. 2003a, p. 192). Kooistraet al. (2003a) did not specify their tree-search algo-rithm, so we used 100 random addition replicates andTBR branch swapping. The resulting tree (Fig. 3) hada different topology and a better log-likelihood score( À ln L 5 14,015.74660) than their tree. In our tree,Toxarium þ Lampriscus was part of a clade includingPennales and various ‘‘centric’’ taxa with a distinctlyelongate outline (Biddulphiopsis, Cymatosira, and Pap-illiocellulus) (Fig. 3). Thus, the Kooistra et al. (2003a)alignment, analyzed with their parameter values, pro-duced a tree on which the elongate cell outline of To-xarium is unambiguously symplesiomorphic, nothomoplasic, with Pennales (Fig. 3).To understand these discrepancies, we first cal-culated the ML score of the Kooistra et al. (2003a)tree topology using their alignment and model param-eters. We obtained a different, lower log-likelihoodscore (Àln L 5 14,017.95809) than they originallyreported (Àln L 5 13,985.63853). We then explored0.01 substitutions/siteBolidomonas mediterraneaBolidomonas pacificaStephanopyxis nipponicaRhizosolenia setigeraLampriscus kittoniiToxarium undulatumPleurosira laevisOdontella sinensisDetonula confervaceaThalassiosira eccentricaThalassiosira rotulaDitylum brightwelliiLithodesmium undulatumHelicotheca thamesisBellerochea maleusBiddulphiopsis titianaPapiliocellulus elegansCymatosira belgicaEucampia antarcticaChaetoceros rostratusAsterionellopsis glacialisRhaphoneis belgicaeStriatella unipunctataPhaeodactylum tricornutumPseudogomphonema sp.Pleurosigma sp.Bacillaria paxiliferTryblionella apiculataFragilariopsis sublineataCylindrotheca closteriumAsterionella formosaNanofrustulum shiloiStaurosira construensRhabdonema sp.Hyalosira delicatulaGrammatophora oceanicaGrammatophora gibberulaLicmophora juergensiiFragilaria crotonensisThalassionema nitzschioideselongate/pennatecircular/subcircularnot applicableFIG. 4. Phylogenetic tree with near identical topology to that found by Kooistra et al. 2003a, based on maximum likelihood analysis ofthe 40-taxon alignment used by Kooistra et al. (2003a). This tree suggests that Toxarium evolved its elongate, pennate-like valve shapeindependently from Pennales. Parameter values of the GTR þ G þ I model were fixed to those set by Kooistra et al. 2003a, except thatempirical base frequencies were used. The tree search used ‘‘as-is’’ addition of taxa and TBR branch swapping. Each diatom name isfollowed by at least one generalized line drawing, based on figures and generic descriptions from Round et al. (1990). For taxa withmultiple line drawings, the drawing immediately following the scientific name represents the most common outline for that genus, basedon Round et al. (1990).A. J. ALVERSON ET AL.664
  • 11. Y10570 Odontella sinensisAF525669 Biddulphiopsis titianaAF525667 Lampriscus kittoniiAF525668 Toxarium undulatumAJ535188 Pleurosira cf. laevisAF525670 Pleurosira laevisX85388 Papiliocellulus elegansX85387 Cymatosira belgicaX85389 Eucampia antarcticaX85392 Chaetoceros didymusX85391 Chaetoceros rostratusAJ535167 Chaetoceros sp.X85390 Chaetoceros sp.AF145226 Chaetoceros sp.AY229896 Chaetoceros debilisAY229897 Chaetoceros gracilisAY229895 Chaetoceros curvisetusAY216905 Talaroneis posidoniaeX77703 Rhaphoneis belgicaeY10568 Asterionellopsis karianaX77701 Asterionellopsis glacialisAY216904 Asterionellopsis glacialisAF525660 Rhabdonema sp.AF525666 Striatella unipunctataAJ535143 Diatoma tenueAB085829 Diatoma hyemalisAF525657 Asterionella formosaAF525654 Hyalosira delicatulaAF525655 Grammatophora oceanicaAF525656 Grammatophora gibberulaAY216906 Grammatophora marinaAF525659 Staurosira construensAF525658 Nanofrustulum shiloiAY345013 C. convoluta endosymbiontAF525661 Licmophora juergensiiAJ535142 Cyclophora tenuisAY216907 Tabularia tabulataAJ535138 Synedra sp.AJ535140 Thalassionema sp.X77702 Thalassionema nitzschioidesAJ535141 Fragilaria sp.AJ535190 Fragilaria cf. islandicaX77704 Fragilaria striatulaAJ535139 Synedra ulnaAJ535137 Fragilariforma virescensAF525662 Fragilaria crotonensisEunotia (5)raphid pennates (59)AY180017 (Chrysophyte)Bolidomonas (7)AJ535174 Paralia solThalassiosirales (34)Hemiaulales, Lithodesmiales (6)Coscinodiscophyceae (20)Coscinodiscophyceae (10)1.000.981. applicableequivocalFIG. 5. Consensus tree from Bayesian analysis of structurally aligned SSU rDNA sequences for 181 diatoms and eight outgroup taxa.A 50% majority-rule consensus tree was calculated from the pooled posterior distributions of two independent MCMCMC runs. Bay-esian posterior probability values greater than 0.5 are shown below nodes. Terminal taxa are identified by GenBank accession numberfollowed by scientific name. For simplicity, several clades were collapsed to triangles, with the number of taxa per clade noted to the right.Two clades (‘‘A’’ and ‘‘B’’) were highlighted to facilitate discussion in the text.EVOLUTION OF ELONGATE SHAPE IN DIATOMS 665
  • 12. alternative strategies for tree estimation and wereunable to reproduce their tree topology using arange of settings under ML, MP, and NJ. We eventu-ally found two tree topologies from an analysisthat used ‘‘as-is’’ taxon addition and empirical basefrequencies (the latter is the default setting in PAUP).One of these was nearly identical to their topol-ogy, including the trichotomy of Biddulphiopsis,Pennales and ((Chaetoceros þ Eucampia) þ Cymatosir-ales) (Fig. 4). The second tree (not shown) resolvedthis trichotomy, with Biddulphiopsis basal to the othertwo clades.AY180017 (Chrysophyte)AJ535174 Paralia solAJ535175 Leptocylindrus danicusAJ535176 Leptocylindrus minimusX85389 Eucampia antarcticaX85392 Chaetoceros didymusX85391 Chaetoceros rostratusAJ535167 Chaetoceros sp.X85390 Chaetoceros sp.AF145226 Chaetoceros sp.AY229896 Chaetoceros debilisAY229897 Chaetoceros gracilisAY229895 Chaetoceros curvisetusX85388 Papiliocellulus elegansX85387 Cymatosira belgicaAF525669 Biddulphiopsis titianaAF525667 Lampriscus kittoniiAF525668 Toxarium undulatumY10570 Odontella sinensisAJ535188 Pleurosira cf. laevisAF525670 Pleurosira laevisBolidomonas (7)Hemiaulales, Lithodesmiales (6)AY216905 Talaroneis posidoniaeX77703 Rhaphoneis belgicaeY10568 Asterionellopsis karianaX77701 Asterionellopsis glacialisAY216904 Asterionellopsis glacialisAJ535142 Cyclophora tenuisAF525660 Rhabdonema sp.AF525666 Striatella unipunctataAF525654 Hyalosira delicatulaAF525655 Grammatophora oceanicaAF525656 Grammatophora gibberulaAY216906 Grammatophora marinaAY345013 Convoluta convoluta endosymbiontAF525661 Licmophora juergensiiAJ535143 Diatoma tenueAB085829 Diatoma hyemalisAF525657 Asterionella formosaAF525659 Staurosira construensAF525658 Nanofrustulum shiloiAJ535137 Fragilariforma virescensAF525662 Fragilaria crotonensisAJ535141 Fragilaria sp.AJ535190 Fragilaria cf. islandicaX77704 Fragilaria striatulaAY216907 Tabularia tabulataAJ535140 Thalassionema sp.X77702 Thalassionema nitzschioidesAJ535139 Synedra ulnaAJ535138 Synedra sp.AJ535163 Undatella sp.AJ535150 Achnanthes bongraniiAJ535151 Achnanthes sp.Eunotia (5)raphid pennates (12)raphid pennates (44)Thalassiosirales (34)Coscinodiscophyceae (19)Coscinodiscophyceae (9)100969710010095941009993596210010081759868100100100100100100100100100elongatecircular/subcircularnot applicableequivocalFIG. 6. Strict consensus of 106 most parsimonious trees based on 752 parsimony-informative characters; tree length 5 7151, con-sistency index (excluding uninformative characters)5 0.2646; retention index 5 0.7040; rescaled consistency index 5 0.1863. Nonpar-ametric bootstrap values are shown below nodes. For simplicity, several clades were collapsed to triangles, with the number of taxa perclade noted to the right.A. J. ALVERSON ET AL.666
  • 13. 189-taxon analyses. The Bayesian analysis placedToxarium within a clade containing Odontella, Biddul-phiopsis, and Lampriscus (Fig. 5). This clade of elon-gate centrics was nested within a larger clade (Fig. 5,Clade B) containing most other centrics with an elon-gate outline (Pleurosira was the only taxon in Clade Bconsidered circular by Round et al. 1990—see Figs. 3and 4 for diagrams of cell outlines). Clade B was in anunresolved trichotomy with Pennales and ((Litho-desmiales þ Hemiaulales) þ Thalassiosirales) (latter 5Clade A, Fig. 5). In one resolution of the trichotomy(Clade A þ (Pennales þ Clade B)), elongate outline isunambiguously optimized as a synapomorphy forPennales þ Clade B, as it was in Fig. 3. In the othertwo resolutions, resemblance in elongate outlinebetween Clade B (including Toxarium) and Pennalesis ambiguously optimized as either plesiomorphic orhomoplasic.The parsimony strict consensus tree was somewhatless resolved than the Bayesian tree (Fig. 6). One dif-ference important to character optimization, however,is that Chaetoceros þ Eucampia is sister to the clade in-cluding Pennales, Pleurosira, Thalassiosirales, Hemiaul-ales þ Lithodesmiales, and various subclades ofelongate centrics including Toxarium. We mappedshape onto each most parsimonious tree found, andelongate shape mapped, either ambiguously or unam-biguously, as plesiomorphic resemblance between To-xarium and Pennales in every topology.DISCUSSIONDiatoms traditionally have been classified as eithercentric or pennate based on a number of features, celloutline foremost among them. This classification is notsupported by phylogenetic analyses of DNA sequencedata (Alverson and Theriot 2005), leading some toconclude that morphological data are misleading.There is, however, broad agreement between morpho-logical characters and phylogenetic hypotheses basedon DNA sequence data, but diatomists have resistedincorporating phylogenetic principles in diatom classi-fication (Round et al. 1990) and continue to recognizenonmonophyletic groups (Medlin and Kaczmarska2004). Illustrative of this, Simonsen (1972, 1979) pro-duced evolutionary scenarios based on an eclectic mixof phylogeny, ecology, phenetics, and geologic age, inwhich centrics gave rise to araphid pennates which inturn gave rise to raphid pennates, yet Simonsen (1972)explicitly eschewed phylogeny as the sole principle inclassification.Nearly every phylogenetic analysis of diatoms hasshown that Pennales constitutes a well-supported cladeand that centric diatoms do not (Medlin et al. 1993,1996a, b, 2000, Ehara et al. 2000, Medlin and Kacz-marska 2004, Sorhannus 2004, see Alverson and The-riot 2005 for a review). This relationship iscorroborated by a suite of morphological charactersthat are synapomorphic for Pennales, whereas centricdiatoms are defined simply by the fact that they arenon-pennate—centric diatoms are ‘‘united’’ bysymplesiomorphic characters and lack of pennateapomorphies. As a result, many of the apparent char-acter conflicts observed in derived centrics (or, alter-natively, basal pennates) vanish when viewed from aphylogenetic perspective. Still, there are real cases ofcharacter conflict in which the distribution of one char-acter implies homoplasy in another. For example,some ‘‘true’’ pennates are circular, not elongate, invalve view (e.g. Campylodiscus). Citing this sort of char-acter distribution, Round et al. (1990) recognized thatoutline was not the primary distinction between centricand pennate diatoms.Recent studies on Toxarium renewed interest inquestions about the evolution of shape in diatoms (Ko-oistra et al. 2003a, b). For example, the elongate valveoutline (a ‘‘pennate’’ character) was thought to conflictwith the orientation of areolar rows, which are morenearly radial from an annulus—albeit an elongate one(a ‘‘centric’’ character)—than perpendicular to a cen-tral line or sternum (a ‘‘pennate’’ character). The an-nulus is an ancestral condition and the sternum aderived one when mapped onto any of a number ofSSU rDNA-based trees (Figs. 3, 5, and 6; Kooistra et al.2003b, Sorhannus 2004). Toxarium, though elongate inshape, has simply retained the ancestral pattern center(an annulus), so it presents no character conflict. Noneof these trees provides compelling evidence that theelongate shape in Toxarium evolved independentlyfrom Pennales.In summary, we performed phylogenetic analyseson two data sets that differed greatly in taxonomiccomposition and method of alignment. We employedthree different optimality criteria and made disparateassumptions about evolution of the SSU rRNA gene.Notwithstanding these differences, results from ouranalyses were in broad agreement and similar to stud-ies using yet other sets of taxa, optimality criteria, andapproaches to alignment (e.g. Kooistra et al. 2003b,Sorhannus 2004). Together, these results suggest thatelongate outline is congruent with SSU rDNA data andmay be synapomorphic for a larger, more inclusiveclade than the traditional Pennales.We thank Mark Edlund, Bob Jansen, Michael Moore, Eliza-beth Ruck, and six anonymous reviewers for critical commentsand suggestions. We thank Gwen Gage for help with figures.Andrew Alverson and Edward Theriot were supported byNSF PEET grant DEB 0118883. Robin Gutell and Jamie Can-none were supported by an NIH grant GM067317.Alverson, A. J. & Theriot, E. C. 2005. Comments on recentprogress toward reconstructing the diatom phylogeny.J. Nanosci. Nanotechnol. 5:57–62.Cannone, J. J., Subramanian, S., Schnare, M. N., Collett, J. R.,D’Souza, L. M., Du, Y., Feng, B., Lin, N., Madabusi, L. V.,Muller, K. M., Pande, N., Shang, Z., Yu, N. & Gutell, R. R.2002. The Comparative RNA Web (CRW) Site: an online da-tabase of comparative sequence and structure information forribosomal, intron, and other RNAs. BMC Bioinformatics 3:2.EVOLUTION OF ELONGATE SHAPE IN DIATOMS 667
  • 14. Chepurnov, V. A., Mann, D. G., Sabbe, K. & Vyverman, W. 2004.Experimental studies on sexual reproduction in diatoms. Int.Rev. Cytol. 237:91–154.Edlund, M. B. & Stoermer, E. F. 1997. Ecological, evolutionary, andsystematic significance of diatom life histories. J. Phycol.33:897–918.Ehara, M., Inagaki, Y., Watanabe, K. I. & Ohama, T. 2000. Phylo-genetic analysis of diatom coxI genes and implications of afluctuating GC content on mitochondrial genetic code evolu-tion. Curr. Genet. 37:29–33.Goertzen, L. R. & Theriot, E. C. 2003. Effect of taxon sampling,character weighting, and combined data on the interpretationof relationships among the heterokont algae. J. Phycol. 39:1–22.Goloboff, P. A., Farris, J. S. & Nixon, K. C. 2004. TNT. Cladistics20:84.Gutell, R. R., Lee, J. C. & Cannone, J. J. 2002. The accuracy ofribosomal RNA comparative structure models. Curr. Opin.Struct. Biol. 12:301–10.Gutell, R. R., Power, A., Hertz, G. Z., Putz, E. J. & Stormo, G. D.1992. Identifying constraints on the higher-order structure ofRNA: continued development and application of comparativesequence analysis methods. Nucleic Acids Res. 20:5785–95.Gutell, R. R., Weiser, B., Woese, C. R. & Noller, H. F. 1985. Com-parative anatomy of 16S-like ribosomal RNA. Prog. Nucleic AcidRes. Mol. Biol. 32:155–216.Hasle, G. R., von Stosch, H. A. & Syvertsen, E. E. 1983. Cymato-siraceae, a new diatom family. Bacillaria 6:9–156.Hillis, D. M. 1998. Taxonomic sampling, phylogenetic accuracy,and investigator bias. Syst. Biol. 47:3–8.Hillis, D. M., Pollock, D. D., McGuire, J. A. & Zwickl, D. J. 2003. Issparse taxon sampling a problem for phylogenetic inference?Syst. Biol. 52:124–6.Kishino, H. & Hasegawa, M. 1989. Evaluation of the maximumlikelihood estimate of the evolutionary tree topologies fromDNA sequence data, and the branching order in Hominoidea.J. Mol. Evol. 29:170–9.Kociolek, J. P., Theriot, E. C. & Willliams, D. M. 1989. Inferring di-atom phylogeny: a cladistic perspective. Diatom. Res. 4:289–300.Kooistra, W., De Stefano, M., Mann, D. G., Salma, N. & Medlin, L. K.2003a. Phylogenetic position of Toxarium, a pennate-like lineagewithin centric diatoms (Bacillariophyceae). J. Phycol. 39:185–97.Kooistra, W. H. C. F., De Stefano, M., Mann, D. G. & Medlin, L. K.2003b. The phylogeny of diatoms. In Mu¨ller, W. E. G. [Ed.]Silicon Biomineralization. Springer, Berlin, pp. 59–97.Larsen, N., Olsen, G. J., Maidak, B. L., McCaughey, M. J., Over-beek, R., Macke, T. J., Marsh, T. L. & Woese, C. R. 1993. Theribosomal database project. Nucleic Acids Res. 21:3021–3.Medlin, L. K. & Kaczmarska, I. 2004. Evolution of the diatoms V:morphological and cytological support for the major cladesand a taxonomic revision. Phycologia 43:245–70.Medlin, L. K., Kooistra, W. H. C. F., Gersonde, R. & Wellbrock, U.1996a. Evolution of the diatoms (Bacillariophyta): II. Nuclear-encoded small-subunit rRNA sequence comparisons confirm aparaphyletic origin for the centric diatoms. Mol. Biol. Evol.13:67–75.Medlin, L. K., Kooistra, W. H. C. F., Gersonde, R. & Wellbrock, U.1996b. Evolution of the diatoms (Bacillariophyta): III. Molec-ular evidence for the origin of the Thalassiosirales. NovaHedwigia Beih. 112:221–34.Medlin, L. K., Kooistra, W. H. C. F. & Schmid, A.-M. M. 2000.A review of the evolution of the diatoms—a total approachusing molecules, morphology and geology. In Witkowski, A.& Sieminska, J. [Eds.] The Origin and Early Evolution of theDiatoms: Fossil, Molecular and Biogeographical Approaches. SzaferInstitute of Botany, Polish Academy of Sciences, Krako´w,pp. 13–36.Medlin, L. K., Williams, D. M. & Sims, P. A. 1993. The evolution ofthe diatoms (Bacillariophyta). I. Origin of the group and as-sessment of the monophyly of its major divisions. Eur. J. Phycol.28:261–75.Nylander, J. A. A., Ronquist, F., Huelsenbeck, J. P. & Nieves-Aldrey,J. L. 2004. Bayesian phylogenetic analysis of combined data.Syst. Biol. 53:47–67.Pollock, D. D., Zwickl, D. J., McGuire, J. A. & Hillis, D. M. 2002.Increased taxon sampling is advantageous for phylogeneticinference. Syst. Biol. 51:664–71.Posada, D. & Crandall, K. A. 1998. MODELTEST: testing themodel of DNA substitution. Bioinformatics 14:817–8.Ronquist, F. & Huelsenbeck, J. P. 2003. MrBayes 3: Bayesianphylogenetic inference under mixed models. Bioinformatics19:1572–4.Round, F. E., Crawford, R. M. & Mann, D. G. 1990. The Diatoms:Biology & Morphology of the Genera. Cambridge UniversityPress, Cambridge, 747 pp.Savill, N. J., Hoyle, D. C. & Higgs, P. G. 2001. RNA sequence ev-olution with secondary structure constraints: comparison ofsubstitution rate models using maximum-likelihood methods.Genetics 157:399–411.Schu¨tt, F. 1896. Bacillariales (Diatomeae). In Engler, A. & Prantl, K.[Eds.] Die Natu¨rlichen Pflanzenfamilien. Verlag von Wilhelm En-gelmann, Leipzig, pp. 31–153.Simonsen, R. 1972. Ideas for a more natural system of the centricdiatoms. Nova Hedwigia Beih. 39:37–54.Simonsen, R. 1979. The diatom system: ideas on phylogeny. Baci-llaria 2:9–71.Sorhannus, U. 2004. Diatom phylogenetics inferred based on di-rect optimization of nuclear-encoded SSU rRNA sequences.Cladistics 20:487–97.Zwickl, D. J. & Hillis, D. M. 2002. Increased taxon sampling greatlyreduces phylogenetic error. Syst. Biol. 51:588–98.A. J. ALVERSON ET AL.668