oner

ionar

tmos

Article history:Received 5 January 2012Revised 12 April 2012Accepted 20 April 2012Available online 3 May 2012

Keywords:

In marine Synechococcus there is evidence for the adaptive evolution of spectrally distinct forms of the

tive agent expected to promote evolution of phenotypes adapted todifferent spectral environments (Wood, 1985; Stomp et al., 2004).This is supported by data showing that the photosynthetic perfor-mance or growth of cultured strains is highest in light elds thatcomplement the spectral phenotype of the culture (Wood, 1985;

used for light harvesting (Grossman et al., 1993; Six et al., 2007).The dominant phycobiliprotein pigments in the PBS confer any ofa wide range of colors on cells ranging from blue-green to orange,pink and brick-red. Phycobiliproteins are pigments composed ofcolorless apoproteins that bind several light-absorbing phycobilinchromophores. Within a PBS, these phycobiliproteins can be foundas antennae (light harvesting) and core (energy transfer to photo-synthetic reaction center II) pigments (for reviews on PBS structureand energy transfer see Gantt, 1981; Glazer, 1989; MacColl, 1998).

Synechococcus is a form-genus with marine representativesbelonging to three genetically distinct subclusters collectively

Abbreviations: AU, approximately unbiased; CA, chromatic adaptation; HGT,horizontal gene transfer; ML, maximum likelihood; MP, maximum parsimony; PBS,phycobilisome; PE, phycoerythrin; PEB, phycoerythrobilin; PUB, phycourobilin. Corresponding author. Fax: +81 45 508 7363.

Molecular Phylogenetics and Evolution 64 (2012) 381392

Contents lists available at

Molecular Phylogene

.e lE-mail address: rcraig@riken.jp (R.C. Everroad).1. Introduction

Marine Synechococcus are a globally important group of photo-synthetic prokaryotes found in a wide range of habitats where thequantity and quality of light available for photosynthesis can varydramatically over relatively short time-scales (Waterbury et al.,1986; Li and Wood, 1988; Olson et al., 1990; Li, 1994; Partenskyet al., 1999; Crosbie et al., 2003; Scanlan et al., 2009). These photo-synthetic organisms rely on light capture for growth, thus thechanging spectral quality of their environment represents a selec-

Glover et al., 1987; Stomp et al., 2004, 2007). Additionally, eldstudies indicate that while different spectral phenotypes oftenco-exist, there is an optical biogeography for marine Synechococcussuch that the most spectrally complementary pigment types pre-dominate in natural waters of different colors (Olson et al., 1988,1990; Wood et al., 1998; Coble et al., 2004; Katano and Nakano,2006; Haverkamp et al., 2008).

The molecular mechanisms that underlie the spectral pheno-type of a given cell are due to the presence of complex water-soluble macromolecular structures called phycobilisomes (PBS)Adaptive evolutionApproximately unbiased testCyanobacteriaHorizontal gene transferPhycoerythrinPhylogenetics1055-7903/$ - see front matter 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.ympev.2012.04.013major light harvesting pigment phycoerythrin (PE). Recent research has suggested that these spectralforms of PE have a different evolutionary history than the core genome. However, a lack of explicit sta-tistical testing of alternative hypotheses or for selection on these genes has made it difcult to evaluatethe evolutionary relationships between spectral forms of PE or the role horizontal gene transfer (HGT)may have had in the adaptive phenotypic evolution of the pigment system in marine Synechococcus. Inthis work, PE phylogenies of picocyanobacteria with known spectral phenotypes, including newly co-iso-lated strains of marine Synechococcus from the Gulf of Mexico, were constructed to explore the diversi-cation of spectral phenotype and PE evolution in this group more completely. For the rst time,statistical evaluation of competing evolutionary hypotheses and tests for positive selection on the PElocus in picocyanobacteria were performed. Genes for PEs associated with specic PE spectral phenotypesformed strongly supported monophyletic clades within the PE tree with positive directional selectiondriving evolution towards higher phycourobilin (PUB) content. The presence of the PUB-lacking pheno-type in PE-containing marine picocyanobacteria from cyanobacterial lineages identied as Cyanobiumis best explained by HGT into this group from marine Synechococcus. Taken together, these data providestrong examples of adaptive evolution of a single phenotypic trait in bacteria via mutation, positivedirectional selection and horizontal gene transfer.

2012 Elsevier Inc. All rights reserved.a r t i c l e i n f o a b s t r a c tPhycoerythrin evolution and diversicatiSynechococcus and related picocyanobact

R. Craig Everroad a,b,, A. Michelle Wood a,ca Institute for Ecology and Evolutionary Biology, formerly Center for Ecology and EvolutbRIKEN Advanced Science Institute, 2-1 Hirosawa, Wako, Saitama 351-0198, JapancAtlantic Oceanographic and Meteorological Laboratory, National Oceanographic and A

journal homepage: wwwll rights reserved.of spectral phenotype in marineia

y Biology, 5289 University of Oregon, Eugene, OR 97403, USA

pheric Administration, Miami, FL 33149, USA

SciVerse ScienceDirect

tics and Evolution

sevier .com/ locate /ympev

known as cluster 5; these likely are not a monophyletic groupingand each subcluster may represent a genus-level group (Herdmanet al., 2001; Scanlan et al., 2009). In two of these subclusters, 5.1and 5.3, the antennae phycobiliprotein phycoerythrin (PE) is foundin two distinct forms that enhance light absorption in the green orblue regions of the light spectrum (Six et al., 2007). Always presentare Class I PEs (PEI), which always contain the green-absorbingchromophore phycoerythrobilin (PEB, Absmax 550 nm). Addition-ally Class II PEs (PEII) can be found; these always contain both PEBand the blue/bluegreen absorbing chromophore phycourobilin(PUB, Absmax 495 nm; Ong and Glazer, 1991; Six et al., 2007).These PE apoproteins are coded for by either cpeBA which formsthe basis of PEI, or mpeBA which forms the basis of PEII (Ong andGlazer, 1991; Wilbanks et al., 1991). When a strain synthesizesboth PEI and PEII, the PEI apoprotein may also bind the PUB chro-mophore, but PUB chromophores are unknown in the PEs of iso-lated marine Synechococcus that lack PEII (Ong and Glazer, 1991;Six et al., 2007).

Phenotypically, the ratio of PUB and PEB chromophores in theassembled PE molecule produce characteristic peaks in thein vivo uorescence excitation and absorption spectra of wholecells. Thus, the ExPUB/ExPEB ratio can be used as a quantitativedescriptor of spectral phenotype within marine Synechococcus(Fig. 1). These phenotypes are classied into four categories based

Synechococcus genomes have revealed PE and PBS genes follow apattern of evolution incongruent with that of the core genome; this

382 R.C. Everroad, A.M. Wood /Molecular Phylogeon ExPUB/ExPEB ratios measured in whole cells or on pigment ex-tracts, with higher ExPUB/ExPEB ratios indicating stronger absorp-tion of blue light relative to green light (Wood, 1985; Woodet al., 1998; Everroad and Wood, 2006; Six et al., 2007). These phe-notypes are: PEB-only or PUB-lacking, where the PUB chromo-phore is absent; low-PUB (ExPUB/ExPEB < 0.6); mid-PUB (ExPUB/ExPEB = 0.61.0) and high PUB (ExPUB/ExPEB > 1.0). Physiologicaladapters that regulate PUB:PEB ratios by a process called type IVchromatic adaptation (CA) may be thought as a fth spectral phe-notype (Palenik, 2001; Everroad et al., 2006).

Apt and colleagues (1995) suggested that PBS and PE evolutionhas proceeded towards increased absorption of shorter wave-lengths, but this hypothesis has never been formally tested. Anevolutionary analysis by Zhao and Qin (2006) found elevated dN/dS ratios and possible positive selection between different

0

5

10

15

20

25

30

35

40

450 475 500 525 550 575

M16.3

M12.1

M11.2

Wavelength (nm)

Rel

ativ

e Fl

uore

scen

ce

Fig. 1. In vivo excitation spectra for PE uorescence emission at 588 nm for thestudy strains M12.1, M16.3, and M11.2. Phycoerythrobilin (PEB) absorbs maximallyaround 550 nm, while phycourobilin (PUB) absorbs maximally around 495 nm.Characteristic peaks or shoulders in the spectra at these wavelengths are due tothese chromophores. Strain M12.1 represents the PUB-lacking phenotype, M16.3represents the low-PUB phenotype, and M11.2 represents the high-PUB phenotype.The low-PUB M16B.1 is omitted for clarity, but it possesses a near identical

excitation spectrum to M16.3. Likewise spectra for the chromatic adapters M11.1and M16.17 are omitted. Spectra for these adapters under blue and white lightconditions can be found in Everroad et al. (2006).pattern has been interpreted as due to unspecied horizontal genetransfer (HGT) events (Everroad, 2007; Six et al., 2007). Based onthe incongruence between phycocyanin (PC; cpcBA-IGS) and 16SrRNA gene trees for several lineages of freshwater picocyanobacte-ria, and the by-phenotype clustering of PE- and PC-rich isolates inthe cpcBA-IGS tree, Jasser et al. (2011) proposed HGT as a likelymechanism explaining the distribution of the PE phenotype inthese groups.

Marine Synechococcus are the most widely recognized PE-richpicophytoplankton in the marine environment, yet they are oftenaccompanied by representatives from other lineages of PE-contain-ing picocyanobacteria. In particular, PE-rich strains closely relatedto Cyanobium spp. are observed frequently in coastal and brackishwaters where the predominant wavelengths of available light aregenerally longer than in the open ocean (Wingard et al., 2002;Ernst et al., 2003; Chen et al., 2004; Haverkamp et al., 2009). Cyan-obium is traditionally described as a fresh- and brackish-waterlineage lacking PE that in most cyanobacterial phylogenies is foundsister to the radiation that gives rise to both marine Synechococcuscluster 5 and Prochlorococcus along with several somewhat-relatedbut unique lineages of picocyanobacteria (Urbach et al., 1998;Herdman et al., 2001; Rippka et al., 2001; Robertson et al., 2001;Ernst et al., 2003). In an earlier study, Everroad and Wood (2006)noted that PE genes from Cyanobium-like isolates shared a com-mon ancestry with PEs from marine Synechococcus 5.1, but couldnot determine if PE genes in the Cyanobium-like isolates evolvedby direct descent from a common PE-containing ancestor or byHGT. Similarly, Synechococcus 5.3 appears basal to the marine Syn-echococcus 5.1 and Prochlorococcus lineages but contains a PBS con-taining both PEI and PEII and a spectral phenotype most similar tothose of PUB-containing Synechococcus 5.1 strains (Six et al., 2007;Dufresne et al., 2008).

The aim of the present research was to test hypotheses aboutthe evolution of spectral diversity within marine picocyanobacte-ria and the evolutionary relationships between Cyanobium andSynechococcus subclusters 5.1 and 5.3. The spectral and geneticdiversity of several strains of marine Synechococcus were exploredas part of a more phylogenetically comprehensive group of taxathan has been examined to date. The deduced phylogenetic rela-tionships between these and other picocyanobacteria were usedto explicitly test the evolutionary patterns inferred for PE genesassociated with different spectral phenotypes and the whole cellphylogenies derived from 16S rRNA gene sequences.

2. Materials and methods

2.1. Isolation, selection, and maintenance of strains used in this study

The four newly reported Gulf of Mexico strains, along with thepreviously-described chromatic adapters M11.1 and M16.17(Everroad et al., 2006) used in this study are a subset of a largerculture isolation project. Isolation details have been previously de-phycobiliprotein lineages; among other places, these elevated ra-tios were observed in the chromophore-binding regions of the phy-cobiliproteins examined. PE phenotypes do not map onphylogenies based on housekeeping and/or ribosomal loci (Supple-mentary Fig. S1; Toledo et al., 1999; Fuller et al., 2003; Everroad,2007; Six et al., 2007) or other molecular data (Wood and Town-send, 1990). Recent analyses of PE genes from many types of cya-nobacteria, red algae and of the PBS gene cluster from 11 marine

netics and Evolution 64 (2012) 381392scribed (Everroad et al., 2006). Briey, these strains were isolatedfrom water collected from a 10-L Niskin bottle on a conductiv-itytemperaturedepth (CTD) cast on February 9th, 2003 from a

logedepth of 275 m at a site immediately above the Brine Pool, a meth-ane seep located in the Gulf of Mexico offshore Louisiana, USA(27430N, 91150W; MacDonald et al., 1990). This water was en-riched with nutrients equivalent to f/50 Si culture medium (Guil-lard and Ryther, 1962), and once growth was visible, cells wereslowly transitioned into sterile seawater with f/2 Si nutrient lev-els. Clonal, but not axenic isolates were obtained from theseenrichments using the three successive pour plate steps as de-scribed in Brahamsha (1996). Based on microscopic observationsof morphology, size and cell division pattern, each clonal strainwas determined to match the description of the form genus Syn-echococcus (Herdman et al., 2001).

2.2. Determination of spectral phenotype

Spectral phenotype for each strain was determined using uo-rescence criteria described in Wood et al. (1998) and Everroadand Wood (2006). Fluorescence excitation spectra were obtainedon an Aminco Bowman series 2 luminescence spectrometer withthe following settings: scan range 450580 nm, variable excitationmonochromator, emission monochromator at 588 nm, 4 nm band-pass for both monochromators and a scan speed of 4 nm s1. PEexcitation spectra were collected and the ratio of excitation at495 nm to excitation at 550 nm was used to determine ExPUB/ExPEBvalues for classication of strains as high, medium, or low PUBstrains, or as lacking PUB. All strains were screened for CA by com-paring the ExPUB/ExPEB ratios of acclimated cultures grown in blueand white light. For these experiments cultures were maintained inexponential phase as described in Wood et al. (2005).

2.3. DNA isolation, amplication, cloning and sequencing

Sequencing for subsequent identity and phylogenetic analyseswere performed for two loci: the cotranscribed PE apoproteingenes cpeBA (PEI) and mpeBA (PEII), and a portion of the 16S ribo-somal RNA gene. DNA extraction, PCR and sequencing procedureswere largely as performed in Everroad and Wood (2006). For eachstrain, culture media was harvested by centrifugation at 27,000gfor 15 min. Cell pellets were vortexed and genomic DNA was iso-lated to sterile water using the Chelex 100 method (de Lamball-erie et al., 1992). All PCR amplications utilized reagent types andconcentrations as previously described (Everroad and Wood,2006). A fragment of the 16S rRNA gene was amplied using thecyanobacteria-specic primers CYA106F (Nbel et al., 1997) andPLG2.3 (Miller and Castenholz, 2000).

For amplication and sequencing of the PE genes, the primerpairs SynB1F/SynA3R and SynB3F/SynA2R were used with condi-tions as previously reported (Everroad and Wood, 2006). BothcpeBA and mpeBA are amplied by these primers; consequentlyPCR products were cloned using the pGEM-T vector (Promega,Madison, WI). For each clone, the PE sequence was re-ampliedand subsequently sequenced in both directions with the originaland internal primers on a Beckman Coulter CEQ capillary sequen-cer using dye terminator chemistry at the Genomics Core Facility,University of Oregon (Eugene, OR, USA).

2.4. Nucleotide sequence editing, alignment, and phylogeneticreconstruction

Sequences were manually edited and assembled using the Bio-Edit Sequence Alignment Editor v. 5.0.9 (Hall, 1999). For eachstrain, consensus PE sequences were created from the overlappingfragments. Nucleotide alignments of the 16S rRNA gene and amino

R.C. Everroad, A.M. Wood /Molecular Phyacid alignments of the translated cpeBA and mpeBA nucleotide se-quence data were obtained using CLUSTALX (Higgins and Sharp,1988, 1989; Thompson et al., 1997). Default settings were usedfor gap penalties. It is known that the a and b subunits of cpeand mpe form distinct evolutionary groups in phycobiliproteinphylogenies, but they are co-transcribed (i.e. cpeBA and mpeBA)and possess similar divergence patterns. Previous analyses haveshown these subunits are suitable for analysis together (Newmanet al., 1994; Apt et al., 1995; Everroad and Wood, 2006; Everroad,2007). To conrm this with the added sequence data, an incongru-ence-length difference test (Farris et al., 1995) was performed onthe translated amino acid PE data using the partition homogeneitytest in PAUP 4.0 b v1 (Sinauer Associates, Sunderland, MA, USA;Swofford, 2003).

16S rRNA gene and PE alignments were made with 46 and 38sequences, respectively. The ribosomal gene alignment was com-posed with genes frommembers of the marine and freshwater pic-ocyanobacterial clade, namely Synechococcus (subclusters 5.1, 5.2and 5.3), Prochlorococcus, Cyanobium and Cyanobium-like strains.The freshwater Synechococcus PCC 7942 (cluster 1) and Synechococ-cus PCC 7002 (cluster 3) were used as the outgroup taxa for thisdataset. The PE alignment was composed of representatives fromall spectral phenotypes for CpeBA and MpeBA in Synechococcus5.1, the mid-PUB Synechococcus 5.3, as well as CpeBA from non-picocyanobacterial taxa, Prochlorococcus, and red algae(Rhodophyta).

Hierarchical likelihood ratio tests were performed on the 16Sand PE data sets using the BIC criterion in jModelTest version0.1.1 and ProtTest version 2.4, respectively (Posada, 2008; Abascalet al., 2005). Phylogenetic analyses for both datasets were per-formed using maximum likelihood (ML), maximum parsimony(MP) and Bayesian methods. Selected models with model parame-ters estimated were used in subsequent parametric phylogeneticanalyses. Gapped positions were excluded from all phylogeneticanalyses.

Maximum likelihood was implemented with PhyML online(Guindon and Gascuel, 2003; Guindon et al., 2010) using the bestof SPR and NNI tree rearrangements of 5 random starting BIONJtrees (Gascuel, 1997). ML analyses were bootstrap replicated 100times. Maximum parsimony (MP) was performed in PAUP usingthe heuristic search option and the tree-bisection-reconnectionbranch-swapping algorithm. Starting trees were obtained by step-wise addition with 100 replications of random sequence addition.MP trees were bootstrapped 100 times. Bayesian analyses were runusing MrBayes v3.1.2 (Huelsenbeck and Ronquist, 2001). For theseanalyses, two runs of four chains (one cold) were run until thedeviation of split frequencies between the runs was

logeclades of marine Synechococcus based on 16S rRNA gene sequencedata (clades IIX, XVI and clade X/Synechococcus subcluster 5.3;Rocap et al., 2002; Fuller et al., 2003; Dufresne et al., 2008) wereincluded in the analysis. Clades with multiple representatives inFig. 2 were recovered. Strains M16B.1, M16.3 and M12.1 afliatedwith clade II strain CC9605, indicating considerable phenotypicand genetic diversity within this clade. Previous work has demon-strated variation of spectral phenotype and nitrogen utilizationcapabilities within clade II (Fuller et al., 2003; Ahlgren and Rocap,2006). M16.17, a member of clade XVI, afliated most closely withthe clade IX chromatic adapter RS9916. Strains M11.1 and M11.2clustered together sister to clade III members WH 8102, WH8103 and Max42.

This analysis also included several freshwater, brackish andmarine picocyanobacteria (including the previously reported PE-containing Arabian Sea strains from Wingard et al., 2002 andEverroad and Wood, 2006) provisionally belonging to the formgenus Cyanobium (Rippka et al., 2001; Crosbie et al., 2003; Ernstet al., 2003). The Synechococcus 5.3 strains RCC307 and Minos11were found deep in the tree, sister to the picocyanobacterial radi-ation, as discussed below.using a maximum parsimony approach. To identify positive (direc-tional and non-directional) selection, synonymous and non-synon-ymous substitutions are classied as invariant (change is retainedin all descendant taxa) or variant (position is subject to additionalchanges) to create four classes: replacement invariant (RI), replace-ment variant (RV), silent invariant (SI) and silent variant (SV). Bycomparing the ratios of RI:RV with SI:SV using the G-test, the nullhypothesis of neutral selection (RI:RV will be similar to SI:SV) canbe tested against the hypotheses of positive selection (RI:RV signif-icantly different from SI:SV). Further, either directional or non-directional positive selection can be identied if the rejection ofneutrality is due to a high number of RI or RV substitutions, respec-tively (Creevey and McInerney, 2002).

2.6. Sequence data

Sequences have been deposited in Genbank under the accessionnumbers JN566227JN566230 for 16S rRNA gene sequences andJN566231JN566237 for cpeBA and mpeBA. The 16S accessionnumbers are also available in Fig. 2.

3. Results and discussion

3.1. Phenotypic characterization of new strains

The study strains from the Gulf of Mexico represent all but oneof the basic categories of spectral phenotype described in the intro-duction. Fluorescence excitation spectra for representative strainsare given in Fig. 1. Strain M12.1 possesses the PUB-lacking pheno-type, strains M16.3 and M16B.1 possess the low-PUB phenotype,strain M11.2 possesses the high-PUB phenotype, and the previ-ously reported strains M11.1 and M16.17 possess the CA pheno-type (Everroad et al., 2006).

3.2. Evolutionary relationships between Cyanobium and marineSynechococcus

The inferred ML tree for the 16S rRNA gene is shown in Fig. 2.The phylogenetic position of all the strains from the Gulf of Mexicoindicates they belong to marine Synechococcus cluster 5.1. Threestrains of Prochlorococcus as well as most previously described

384 R.C. Everroad, A.M. Wood /Molecular PhyThe evolutionary relationships between genera within the pico-cyanobacterial lineage are generally agreed to follow a patternrecovered in Fig. 2 (Everroad and Wood, 2006; Scanlan et al.,2009). However, these relationships are not fully resolved, withSynechococcus 5.1 sometimes reported as paraphyletic with respectto Prochlorococcus (Urbach et al., 1998; Herdman et al., 2001).Additionally, ambiguities exist within Synechococcus 5.1. Recentphylogenies (including the one presented here in Fig. 2) place thePE-lacking Synechococcus 5.2 strain WH 8101 within clade VIII ofSynechococcus 5.1 (Scanlan et al., 2009). Likewise, Synechococcus5.1 clade X, consisting of strains RCC307 and Minos11, was re-cently described as sister to Synechococcus 5.1 and Prochlorococcusand proposed to be renamed Synechococcus 5.3 (Dufresne et al.,2008).

The remaining strains of picocyanobacteria not included amongthe marine Synechococcus or Prochlorococcus lineages appear clo-sely related to presently dened Cyanobium strains and togetherform amonophyletic lineage of Cyanobium-like strains (sensu Ernstet al., 2003) that includes isolates from fresh, brackish and marinewaters. However, this monophyletic group is not well-supported inthis work (Fig. 2) or elsewhere (e.g. Crosbie et al., 2003; Ernst et al.,2003; Jasser et al., 2011). Several distinct lineages and naming sys-tems exist for these Cyanobium-like organisms (Fig. 2), and therelationships of these lineages to one another remain ambiguous(Crosbie et al., 2003; Ernst et al., 2003). Although Cyanobium isformally dened as lacking PE in Bergeys Manual of SystematicBacteriology (Rippka et al., 2001), many strains from these Cyanobi-um-like lineages have since been identied that possess PE, includ-ing the previously reported marine Cyanobium strains from theArabian Sea (Wingard et al., 2002; Everroad andWood, 2006). Thuslike marine Synechococcus in subcluster 5.1, which are dened ashaving PE (Herdman et al., 2001 but occasionally lack PE, strainsin the Cyanobium clade do not always conform to the pigmentcomplement expected from the formal description of the lineage,i.e. sometimes containing PE even though the taxon is dened aslacking PE. In other words, both Synechococcus 5.1 and Cyanobiumappear to include members both with and without a PE phenotype.Based on the results presented in Fig. 2, it is clear that future effortsshould be made to revise the taxonomic status, formal descrip-tions, and naming conventions for the Synechococcus 5 cluster aswell as the Cyanobium and Cyanobium-like lineages of picocyano-bacteria, as has been called for previously (Herdman et al., 2001;Robertson et al., 2001; Dufresne et al., 2008).

Given the somewhat unclear relationships between the individ-ual picocyanobacterial groups, and the dispersal of similar pigmenttypes throughout the entire radiation, a more robust analysis of the16S rRNA gene tree was performed. The AU test was used to deter-mine if an alternative phylogeny that placed either the PE-contain-ing Cyanobium from the Arabian Sea or the mid-PUB Synechococcus5.3 as sister to the remaining members of the Synechococcus 5.1radiation could be rejected. The results, shown in Table 1, clearlyreject a phylogeny with these marine Cyanobium as members ofSynechococcus 5.1.

Conversely, a phylogeny that places the proposed Synechococcus5.3 as a sister group to Synechococcus 5.1 clade cannot be rejectedas signicantly worse than the best tree, although it could be con-sidered only marginally non-signicant (p = 0.063, SE 0.004;Table 1). As discussed briey by Dufresne et al. (2008), phyloge-nomic analyses lacked sufcient taxon sampling for rm assign-ment of Synechococcus 5.3 to an existing clade (Six et al., 2007;Dufresne et al., 2008), but other evidence from analyses of perox-iredoxins and glnB suggest that Synechococcus 5.3 is more similarto Synechococcus 5.2 than 5.1, incongruent with a placement ofSynechococcus 5.3 as a member of the 5.1 clade (Scanlan et al.,2009).

netics and Evolution 64 (2012) 3813923.2.1. Evolution of picocyanobacterial PEsThe PE genes of the new PUB-lacking, low-PUB and high-PUB

Synechococcus 5.1 strains were sequenced to build a more complete

62329)00028)0012905)80

7)4)43528300)

logeM16B.1 (JN56M16.3 (JN5662

71/-/-

CC9605 (II; CP53/57/-

M12.1 (JN566285/60/0.93

CC9902 (IV; CP0

71/-/0.89

WH 8103 (AF31Max42 (AY1728WH 8102 (BX54

*

M11.2 (JN56622M11.1 (DQ2242068/60/0.76

75/-/0.88

CC9311 (CP000WH 8020 (AY17Almo3 (AY1728

100/87/0.99

AB

No PE (PC-rich)PUB-lackingLow PUBMid PUBHigh PUBVariable PUB/PEBdiv-Chl a/b

R.C. Everroad, A.M. Wood /Molecular PhyPE phylogeny for marine Synechococcus. The partition homogeneitytest did not detect incongruence between the a- and b-subunits ofthe PE genes, so they were analyzed together in subsequent phylo-genetic analyses. The best-t model of evolution for the deducedPE apoprotein sequence alignment conformed to the LG substitu-tion model with gamma distribution and empirical amino acid fre-quencies (LG + G + F). This model was coded into MrBayes using amanually checked template substitution matrix from the GARLIproject (Zwickl, 2006). The inferred ML tree, which is unrooted(Fig. 3), shows very strong support for the existence of ve majorclades of PE, as reported previously (Everroad and Wood, 2006).These include the PE genes of red algae, Prochlorococcus, cyanobac-teria not from the picocyanobacterial radiation (i.e. other cyano-bacteria), and the CpeBA and MpeBA clades from Synechococcus

0.05

A) 94/88/0.91B) 86/82/-C) 71/60/0.90D) 68/69/0.94

Eum14 (AY172804)WH 7805 (AAOK010WH 7803 (CT97158369/63/0.8

84/75/0.74

RS9916 (IX; AYM16.17 (XVI; DQ22WH 8101 (VIII; AF0

52/57/-

CCMP1375 (AE0PAC1 (AF001471

MIT 9313 (AF0533982/95 0.96P211 (AF0983MW 100C3 (AY

*

*

WH 5701 (AY1BO 8805 (AF31

78/87/-

PCC 6307 (AF00PCC 7009 (AF

*

MW 33B4 (AMW 25B5 (AY

*

MW 4C3 (AY15BO 8807 (AF317

66/81 -

77/53 0.93

BS5 (AF330253) BG6.1 (DQ24800

G11 (DQ248008

E) 73/51/0.83F) 84/91/0.85G) 65/54/0.99H) 50/-/0.99

G4.1 (DQ248003G10.1 (DQ24

PCC 7001 (AB01505PS717 (AF216953)

RCC307 (NC_009482)*

*

PCC 7942 (CP000100)

-/75/0.98

-/62/-

HG

FE

DC

Oli31 (AY172810 )

ML/MP/PP

Minos 11 / RCC61 (AY17

77/95 1.0

Fig. 2. Maximum likelihood (ML) phylogram of the marine cyanobacterial clade (Prochlor16S rRNA gene. Clade names for recovered clades are given. For Synechococcus 5.1; whparentheses with the strain name. The freshwater Synechococcus PCC 7942 (cluster 1) andis shown with MP/ML bootstrap values >50 above the node. Bayesian posterior probabilitindicated at >90 for both ML and MP and a Bayesian posterior probability >0.97. DasheStrain pigment information derived from Herdman et al. (2001), Rippka et al. (2001), REverroad et al. (2006), Everroad and Wood (2006), Six et al. (2007), and Haverkamp et a0)

110)

097)3)

20)

)5)

Synechococcus 5.1Clade I

Clade III

netics and Evolution 64 (2012) 381392 385subcluster 5.1. None of the deeper relationships between thesegroups were well supported, although all methods suggest thered algal CpeBA lineage is sister to the Synechococcus 5.1 CpeBAclade. PE sequences from Prochlorococcus appear the most evolu-tionarily distant, which may in part be due to differential selectionon PE in these taxa (Ting et al., 2001). The placement of the prim-itive cyanobacteria Gloeobacter spp. PCC 7421 sister to the red algallineage is another interesting feature of the overall tree (Fig. 3).

Within the CpeBA lineage frommarine Synechococcus and Cyan-obium, there are well-supported subclades in which the sequencesfrom strains with xed PUB-lacking, low-PUB and high-PUB phe-notypes occur together (Fig. 3). Specically, all CpeBA sequencesobtained from strains with a PUB-lacking phenotype, includingfrom both marine Synechococcus and Cyanobium, are found

00001))172826)4203)01480)17126))9)73)151249)

ABRAXAS (AF098372)ACE (AF098370)

72832)7073)1477)

216945)Y151237)151233)1238)

074)ornholm sea

5)))8007)8) Cyanobium 2 Group E (Lake Biwa)

PCC 7002 (CP000951)

Prochlorococcus

Cyanobium 1 (Group A)

Group I

Antarctic

Group B (subalpine I)

Group H

Subalpine II / Synechococcus 5.2

Arabian Sea

Synechococcus 5.3 (5.1 clade X)

Non-marine picocyanobacteria Cyanobium

Clades V/VI/VII

2807)

ococcus/marine Synechococcus/Cyanobium) inferred from a 1414-bp alignment of theere only one representative of a clade is present, the clade number is included inSynechococcus PCC 7002 (cluster 3) were used as outgroups. Support for each node

ies >0.7 are shown below the node. Asterisks indicate bootstrap support for the nodes indicate bootstrap or posterior probability support below 50 or 0.7, respectively.obertson et al. (2001), Crosbie et al. (2003), Ernst et al. (2003), Fuller et al. (2003),l. (2008). The PUB content of the Antarctic strains is unreported.

cocll oer C

ccuschoc

or M

etai

ancechoc

logeexclusively together as a monophyletic group sister to the remain-ing CpeBA sequences. Within this sister clade, all sequences fromstrains with a low-PUB phenotype form a second monophyleticgroup sister to the remaining mid, high, and variable-PUB se-quences. For this latter grouping of CpeBA sequences from strainswith mid, high and variable-PUB (or CA) strains, two subcladesare observed. One clade, not well supported, places the mid-PUB

Table 1Approximately unbiased likelihood tests of alternative tree topologies.

Rank Tree topoogy

Phycoerythrin spectral clades1 Best treec

2 MpeBA (ML)Hd

3 CpeBA (HM)(LN)4 CpeBA ((HM)N)L

Phycoerythrin taxonomic clades1 Best treee

2 PUB-lacking Cyanobium sister to no-PUB Synecho3 PUB-lacking Synechococus 5.1 as sister clade to a4 PUB-lacking Cyanobium as sister clade to all oth

5 RCC307 as sister to remaining CpeBAs

6 RCC307 as sister to remaining MpeBAs

16S rRNA locus1 Best treeg

2 Synechococcus 5.3 sister to remaining Synechoco3 Arabian Sea Cyanobium sister to remaining Syne

a Log likelihood difference.b p-Value for the approximately unbiased test.c Tree in Fig. 3 except RS9916 CpeBA is excluded. For CpeBA ((High Mid)Low)No; F

RCC307 in mid-PUB Synechococcus 5.1.d N, L, M and H refer to the No, Low, Mid and High clades of the PE tree tested. De Tree shown in Fig. 3.f Underlined values indicate topologies of interest that were rejected at a signicg Tree shown in Fig. 2. Arabian Sea Cyanobium sister to Cyanobium cluster 2, Syne

386 R.C. Everroad, A.M. Wood /Molecular PhyRCC307 sister to a group of chromatic adapters. The second cladeputs the origin of the branch leading to the chromatic adapterRS9916 basal to a well-supported monophyletic group of xedhigh-PUB strains. Consequently, this analysis makes the CA pheno-type paraphyletic with respect to both a monophyletic high-PUBcluster and the mid-PUB RCC307. However these placements arenot well supported in Fig. 3, and for RS9916 are incongruent withthe evolutionary relationships found in the MpeBA sequences.Likewise, the placement of the RCC307 CpeBA and MpeBA se-quences are not well supported; until further mid-PUB strain PEsequences are available a true mid-PUB cluster cannot be identi-ed. As RCC307 appears to be the only member of Synechococcus5.3 examined that does not have the CA phenotype it may alsobe simply that this particular strain has lost its ability to chromat-ically adapt (Six et al., 2007).

The pattern found in the MpeBA sequences is similar to that ofCpeBA, but with better resolution between high- and mid-PUBclades. The xed low-PUB group is placed sister to two stronglysupported clades that contain either the mid-PUB RCC307 withchromatic adapters or the xed high-PUB strains. The nearly com-plete congruence of the phylogenies for MpeBA and CpeBA instrains that contain both is consistent with a hypothesis that thetwo sets of genes represent paralogs evolving in tandem after agene duplication event that facilitated accommodation of thePUB chromophore. Earlier work by Apt et al. (1995) emphasizedthe role of gene duplication in the evolution of phycobiliproteinsand this appears to have been an important mechanism for acqui-sition of a PUB phenotype by marine Synechococcus.

The clustering of the Synechococcus PEs by spectral type and thebranching patterns of these clades suggest successive evolution to-wards higher PUB content, particularly because PE sequences fromstrains with lower or PUB-lacking phenotypes share more basalancestors compared to strains with mid- or high PUB phenotypes.To further explore this idea, two sets of tests were performed. TheAU test was used to test the evolutionary branching order of PEspectral clades within the Synechococcus 5.1 CpeBA and MpeBA lin-eages as shown in Fig. 3, while the relative rate ratio test of Creeveyand McInerney (2002) was used to reconstruct ancestral sequences

D lnLa AUb

2.3 0.7962.3 0.4603.6 0.4595.0 0.112

6.3 0.875cus 5.1 6.3 0.18ther CpeBAs 10.5 0.085peBAs 10.7 0.025f

20.2 0.03631.3 0.001

6.1 0.9755.1 6.1 0.063occus 5.1 23.1 0.005

peBA (High Mid)Low; no-PUB Cyanobium paraphyletc with no-PUB Synechococcus;

ls in main text.

level of 0.05.occus 5.3 basal to all non-outgroup taxa.

netics and Evolution 64 (2012) 381392and test for positive selection on a simpler 11 taxon CpeBA phylog-eny containing all phenotypic groups.

3.2.2. Approximately unbiased testsFor the PE AU tests, using the nomenclature H = high-PUB,

M = mid-PUB, L = low-PUB and N = PUB-lacking, all possible evolu-tionary relationships of the H, M, L and N clades within CpeBA andMpeBA were tested. These clades were dened as follows: ForMpeBA, the low-PUB group L included all sequences from low-PUB strains, the mid-PUB clade M included RCC307 and severalstrains capable of chromatic adaptation, while the high-PUB cladeH contained all the high-PUB strains. For CpeBA test, the no- andlow-PUB groupings were clear (N, and L, respectively). Based onthe best tree however, assignment of high and mid-PUB clades wassomewhat less clear. The clade containing Synechococcus spp.CC9311, CC9902 and WH 8020 and the true mid-PUB RCC307 arereferred to as M. When acclimated to white light, these CA strainspossess PEs with a mid-PUB phenotype and were historically con-sidered mid-PUB strains (Six et al., 2007). Due to the questionableplacement of the chromatically adapting RS9916 within the Hclade in the CpeBA phylogeny as discussed above, a preliminaryAU test was performed to asses whether an alternative tree withit as a member of the M clade with the remaining chromaticadapters could be rejected. This test could not reject such a place-ment (p > 0.205), so RS9916 CpeBA was removed from the analysis.This result, combined with the phenotype of RS9916 and the pat-tern of the MpeBA locus makes it seem possible that the placementof RS9916 with the CpeBA H clade in the ML and Bayesian trees isthe result of some phylogenetic artifact in the analysis.

For PE spectral clades, these tests effectively reduced to sepa-rate four and three taxon questions. Trees could be considered

ApeB

eBAA

AA

loge98/98/1.0

75/87/1.0

M11.2 CpeBWH 8102 C

CC9605 CpRS9916 CpeB

CC9311 CpeBWH 8020 CpeB

73/71/0.98

CC9902 CpeBA

100/100/1.0

74/-/0.98

A

B

R.C. Everroad, A.M. Wood /Molecular Phyrooted since all Synechococcus/Cyanobium PEs share a most recentcommon ancestor (Everroad and Wood, 2006). Thus the CpeBA testincluded 14 alternative trees (rooted four-taxon problem), whilethe MpeBA test included 2 alternative trees (rooted three-taxonproblem), for a total of 17 trees (16 alternatives plus the best tree).

Five additional tests were performed to assess the evolutionaryrelationships between the PE-sequences from the Arabian SeaCyanobium, Synechococcus 5.3, and Synechococcus 5.1 strains. Spe-cically, these tests aimed to demonstrate whether the genes forPEs from the Cyanobium and Synechococcus 5.3 strain groups weretruly members of the same PE clade as the CpeBA and MpeBA

0.1

97/100/1.0

72/55/ 0.91

98/100/1.0

99/100/1.067/56/0.75

100/100/1.0

86/78/1.0

69/53/1.0

89/94/ 1.0

86/92/1.0

-/99/1.0

100/ 1.

100/100/1.0

67/53/0.7100/100/1.0

-/81/ 1.0

100/100/1.0

RCC307 CpeBA (SM16B.1 CpeBAM16.3 CpeBAWH 7803 CpeBA

90/87/1.0

WH 7805 CpeBAM12.1 CpeBA

D

G5.1 CpeBA

E

G4.1 CpeBAG11 CpeBA

PorpPorp

CeCoralli

Gloeobacter

F

PseudFremyella diplo

Synechocy

100/100/1.0

Prochlorococcus MIT 9313 CpeBProchlorococcus CCMP 1375 CpeBAProchlorococcus PAC1 CpeBA

A) 87/93/1.0B) 59/56/0.9C) 51/52/0.94D) 99/92/1.0E) 70/88/1.0F) 92/77/0.99G) 76/94/1.0H) 71/84/1.0

ML/MP/PP

C

H

G

L

P

Fig. 3. Unrooted ML phylogram inferred from a 321 amino acid alignment of the a-cyanobacteria and red algae. ML/MP bootstrap >50 and Bayesian posterior probabilities >of the CpeBA and CpeBA genes of marine Synechococcus, note that MpeBA is absent fromsupport values given to the left of the tree.A

Synechococcus 5.1

Variable PUB/PEB

High PUB H

netics and Evolution 64 (2012) 381392 387genes from all other PE-containing strains from the Synechococcus5.1 group, and secondarily if the sequences from PUB-lackingCyanobium and Synechococcus strains in the 5.1 group could formmutually monophyletic groupings (and a clearer demonstrationof HGT) as had been shown previously (Everroad, 2007). For thesetests trees were constructed by alternatively placing PEs fromCyanobium as a sister group to PUB-lacking Synechococcus 5.1 oras a sister clade to all marine Synechococcus CpeBA. Likewise, thealternative placement of the CpeBA and MpeBA sequences fromstrain RCC307 as a sister clade to the respective Synechococcus5.1 PE clades was tested. Finally, the PUB-lacking Synechococcus

94/-/-100/0

yn 5.3; Mid PUB M)

hyra yezoensis CpeBAhyra purpurea CpeBAramium boydenii CpeBAna officinalis CpeBAPCC 7421 CpeBA

WH 8020 MpeBACC9902 MpeBA

RS9916 MpeBA

RCC307 MpeBA (Syn 5.3; M)

M11.2 MpeBACC9605 MpeBA

WH 8102 MpeBA

M16B.1 MpeBAM16.3 MpeBA

WH 7803 MpeBanabaena PCC 7409 CpeBAsiphon CpeBA

stis PCC 9413 CpeBAA

CC9311 MpeBA

Synechococcus 5.1

Rhodophyta

Arabian Sea (Cyanobium)

ow PUB L

UB-lacking N

VariablePUB/PEB

High PUB H

Low PUB L

and b-subunits of phycoerythrin (PE) I and PEII (CpeBA and MpeBA) for selected0.7 are given for each node. Study strains are in bold. When comparing the topologystrains with a phycourobilin (PUB)-lacking phenotype. Letters indicate nodes with

loge(A) MpeBA

(B) CpeBA

ML best tree p= 0.46

p= 0.112

p = 0.459

High Mid Low No High Mid Low No

High Mid LowNo

ML best tree

HighHigh Mid Low Mid Low

Fig. 4. Schematic representation of possible phycoerythrin (PE) evolutionarypatterns for spectral phenotype. (A) The most likely tree for MpeBA, followed byone alternative tree not rejectable using the approximately unbiased (AU) test. (B)

388 R.C. Everroad, A.M. Wood /Molecular Phy5.1 strains were placed sister to the PUB-lacking Cyanobium strainsand remaining Synechococcus 5.1 strains (making them the mostancestral-like Synechococcus 5.1 CpeBA).

Selected results of the AU tests are shown in Table 1 and Fig. 4.For MpeBA the best tree was (HM)L with the mid- and high-PUBPEs nested to the exclusion of the low-PUB PEs (Fig. 3, Fig. 4A).However, the AU test could not reject an alternative possibility,namely (ML)H where the mid- and low-PUB PEs grouped togetherto the exclusion of the high-PUB PEs (Fig. 4A, Table 1). This wouldindicate successive adaptation to lower PUB content, counter towhat was observed in the most likely tree. In either case, evolutionappeared directional, but the direction is unknown based on theseresults alone.

In the CpeBA phylogeny however, evidence of directional adap-tation towards higher PUB content could be seen clearly. The bestCpeBA tree is ((HM)L)N, with high- and mid-PUB PEs nested to-gether to the exclusion of low-PUB PEs (just as in the best MpeBAtree), followed by PUB-lacking PEs sister to all PUB-containing PEs(Fig. 4B). Like the best MpeBA tree, this evolutionary pattern sug-gests strongly that successive adaptation towards higher PUB con-tent has occurred in PE. Alternative statistically supportedphylogenies are also consistent with evolution towards higherPUB content and include (HM)(LN), with the high- and mid-PUBPEs sister and sharing a common ancestor with a clade that con-tains sister low- and PUB-lacking PEs (Fig. 4B). The second alterna-tive tree of ((HM)N)L again places high- and mid-PUB PEs as sister,but here, the PUB-lacking PEs are found at the next deepest bifur-cation, followed by a basal low-PUB PE clade. All 12 remainingtopologies, including three consistent with evolution toward lowerPUB content were rejected at a signicance level of 0.05 (Table 1).Identical AU tests were performed that included RS9916 CpeBA asa member of the H clade (as in the ML tree presented in Fig. 3);

The most likely tree for CpeBA, followed by two non-rejectable trees. p-Values arefor the AU test. Small bars in alternative CpeBA phylogenies represent presumedlosses of phycourobilin (PUB) and MpeBA.these tests resulted in identical patterns of signicance for theclades (not shown).

The two alternative CpeBA patterns of (HM)(LN) and ((HM)N)Lboth retain high-PUB, mid-PUB and CA PEs together to the exclu-sion of the low- and PUB-lacking PEs, but must invoke a secondaryloss of PUB and MpeBA in the PUB-lacking clades (shown inFig. 4B), with the relationships between low- mid- and high-PUBPEs still following directional evolution toward higher PUB con-tent. Thus for both CpeBA and MpeBA, the best tree and all butone non-rejectable topology agree on the direction of evolution.Only the MpeBA phylogeny of (ML)H, shown in Fig. 4A, disagreeswith the other potentially acceptable trees but it is incongruentwith all CpeBA phylogenies that could not be rejected. Converselythe best MpeBA phylogeny is congruent with all three possibleCpeBA phylogenies. The actual evolutionary history for these co-transcribed and co-assembled loci should match even if the actualsignal is somewhat ambiguous and, if this reasonable assumptionis made, then the best tree from MpeBA can be accepted as correct.Given that the branching patterns are congruent between the Mpe-BA and CpeBA loci, the best tree fully supports the hypothesis thatevolution of the apoprotein genes for PEs associated with differentphycobilin composition and different spectral phenotypes at thewhole organism level proceeded towards higher PUB content andadaptation of PE to bluer water. Whether or not the ancestral formsof marine Synechococcus had the genetic architecture now associ-ated with a PUB-lacking phenotype or rather contained the genesfor both PEI and PEII apoproteins remains an unanswered question.If the latter, then the PUB-lacking strains present in the lineage to-day would have to have evolved by gene loss since they do not re-tain any residual sign of the apoprotiens for MpeBA, an apparentrequirement for PUB-containing phenotypes.

Next, the AU test was used to determine if PEs grouped by theirtaxonomic afliation based on the 16S rRNA gene (Fig. 5). The testcould not reject a monophyletic Cyanobium clade sister to a mono-phyletic PUB-lacking Synechococcus 5.1 clade, nor could it reject atree that has PUB-lacking Synechococcus 5.1 sister to a CpeBA cladethat includes the remaining Synechococcus 5.1 and Cyanobium se-quences (Fig. 5A and Table 1). These two possible topologies, likethe ML topology in Fig. 3, make Synechococcus 5.1 CpeBA ancestralto Cyanobium CpeBA, incongruent with the relationship betweenthese groups observed in the 16S rRNA gene phylogeny (Fig. 2).However, the results did reject both CpeBA and MpeBA fromRCC307 as sister to the respective Synechococcus 5.1 PE clades (Ta-ble 1, Fig. 5B). Likewise the hypothesis that CpeBA sequences froma PE-containing Cyanobium are sister to the Synechococcus 5.1CpeBA clade, consistent with the 16S rRNA gene phylogeny, anda hypothesis that Cyanobium CpeBA is the ancestral form, can berejected.

3.2.3. Relative rates ratio testTo further examine the evolution of the picocyanobacterial PE

genes, a smaller 11-taxon tree with all phenotypes representedwas made and tested for evidence of positive selection using therelative rates ratio test of Creevey and McInerney (2002; Fig. 6).This method is more sensitive than traditional approaches suchas pairwise dN/dS ratios, which are calculated for the whole se-quence and can often overlook positive selection at one or a fewpositions. For all pairwise sequence comparisons in Fig. 6, the nullhypothesis of neutrality could be rejected using a codon-based Z-test, and all had dN/dS ratios

UB

UB

lack

UB

UB

ack

ies

logeHigh-P

Mid-P

PUB-

Mid-P

Low-P

PUB-l

(A) Non-rejected CpeBA phylogen

R.C. Everroad, A.M. Wood /Molecular PhyConversely, positive selection was not detected on the branch lead-ing from the ancestor of the (HM)L clade to the ancestor of the Lsequences, nor on the branch leading to the N sequences from theancestor of all the Synechococcus 5.1 CpeBAs (Fig 6).

3.3. Hypervariable region of CpeB

Everroad and Wood (2006) reported a hypervariable region ofapproximately 35 amino acids long near the C-terminus of theCpeB and MpeB sequences. In this previous study, they proposedthis region as a suitable taxonomic marker for PE, as well as for dis-tinguishing CpeB from MpeB. However, this previous analysis didnot contain PUB-lacking PEs from Synechococcus 5.1. With theaddition of several new CpeB and MpeB sequences, including CpeB

ML best tree p

PUB-lack

PUB-lacking clades sister p = 0.18

Synechococcus 5.3 sister to remaining CpeBA sequences p = 0.036

(B) Rejected CpeBA phylogenies

S

High-PUB (Syn 5.1)

Mid-PUB (Syn 5.3)

PUB-lacking (Syn 5.1)

Mid-PUB (Syn 5.1)

Low-PUB (Syn 5.1)

PUB-lacking (Cbm)

PUB-lacking (Cbm)

High-PUB (Syn 5.1)

Mid-PUB (Syn 5.3)

PUB-lacking (Syn 5.1)

Mid-PUB (Syn 5.1)Low-PUB (Syn 5.1)

PUB-lacking (Cbm)

Fig. 5. Schematic representation of possible CpeBA evolutionary patterns in context of tderived. (A) The most likely tree for CpeBA with Cyanobium (Cmb) shown twice to emphaphylogenies placing phycourobilin (PUB)-lacking CpeBA from Cyanobium as sister toSynechococcus 5.1 as sister to all other CpeBA sequences (right) using the approximatelyeither Synechococcus 5.3 (left) or Cyanobium (right) sister to all CpeBA sequences from SyCmb = Cyanobium (Arabian Sea strains). p-Values are for the AU test. (Syn 5.1)

(Syn 5.3)

ing (Syn 5.1)

(Syn 5.1)

(Syn 5.1)

ing (Cbm)

netics and Evolution 64 (2012) 381392 389from PUB-lacking Synechococcus, the present study reassessed thishypervariable area. Supplementary Fig. S2 reveals this larger align-ment and the utility of this region. As previously shown, it can dif-ferentiate between PEs from Prochlorococcus, Rhodophyta, non-picocyanobacterial PEs, and CpeB and MpeB from Synechococcus5.1. It can also distinguish low-PUB MpeB sequences from otherMpeB sequences, and it can differentiate between PUB-lacking,high-PUB and other PUB-containing CpeB sequences. However, itis not truly a taxonomic marker as previously proposed; it cannotdifferentiate between PUB-lacking PEs from Cyanobium and Syn-echococcus 5.1, nor can it differentiate between PEs from Synecho-coccus 5.3 and 5.1. However, since this hypervariable region of thegene was excluded from the phylogenetic analysis shown in Fig. 3,the similarities that make it a poorer taxonomic marker than

= 0.875

ing (Cbm)

Cyanobium sister to remaining CpeBA sequences p = 0.025

ynechococcus 5.1 PUB-lacking sister toremaining CpeBA sequences p = 0.085

High-PUB (Syn 5.1)

Mid-PUB (Syn 5.3)

PUB-lacking (Syn 5.1)

Mid-PUB (Syn 5.1)Low-PUB (Syn 5.1)

PUB-lacking (Cbm)

High-PUB (Syn 5.1)

Mid-PUB (Syn 5.3)

PUB-lacking (Syn 5.1)

Mid-PUB (Syn 5.1)

Low-PUB (Syn 5.1)

PUB-lacking (Cbm)

he 16S rRNA gene phylogeny of the strains from which the CpeBA sequences weresize its paraphyly with respect to Synechococcus (top), followed by two non-rejectedPUB-lacking CpeBA from Synechococcus 5.1 (left) or PUB-lacking CpeBA from

unbiased (AU) test. (B) Two rejected phylogenies that place CpeBA sequences fromnechococcus 5.1. Syn 5.1 = Synechococcus 5.1, Syn 5.3 = Synechococcus, 5.3 (RCC307),

distance (Whitaker and Baneld, 2006). Here it has been shown

logeoriginally proposed do provide additional evidence that the PEsfrom the Arabian Sea Cyanobium and Synechococcus 5.3 are evolu-tionarily truly Synechococcus 5.1 PEs.

3.4. Concluding remarks

These data give an interesting picture of the evolution of PE inmarine picocyanobacteria. As a group, these organisms differ frommany cyanobacteria and red algae in their widespread distribution

WH 8020

CC9311

CC9902

WH 8102

CC9605

M16.3

M16B.1

WH 7803

M12.1

WH 7805

PCC 7421

High-PUB H

Variable-PUB M

Low-PUB L

PUB-lacking N

24:1766:137

51:3791:288

156:99190:462

75:53141:353

18:1066:137

Fig. 6. Relative rate ratio test for CpeBA. The number of replacement invariant andreplacement variant, silent invariant and silent variant substitutions (RI:RV andSI:SV, respectively) are shown above the branches where positive directionalselection was detected (all branches at p < 0.001).

390 R.C. Everroad, A.M. Wood /Molecular Phyover a vast, uid environment in which the spectral composition ofthe natural light eld varies dramatically. Three clades of marinepicocyanobacteria were detailed in this study: marine Synechococ-cus 5.1, marine Synechococcus 5.3, and Cyanobium (for CpeBA). Allthree share the same unique ancestry of CpeBA and MpeBA apo-protein genes when considered in the context of other PE-contain-ing cyanobacteria and red algae, indicating that the apoprotein oftheir PEs evolved independently with respect to these other PEs.The inferred phylogenies for both these genes clearly indicate thatPE has evolved directionally towards higher PUB content.

These PE results, in the context of the 16S rDNA gene tree, re-veal that while both Synechococcus 5.3 and the Arabian Sea Cyano-bium clearly possess Synechococcus 5.1-like PEs, these members ofCyanobium are distinct from Synechococcus 5.1 based on their 16SrRNA gene sequences. This strongly supported taxonomic separa-tion between strains associated with the Cyanobium lineage andthe 5.1 lineage, combined with the data on the evolution of thePE apoprotein genes from members of the PE-containing Cyanobi-um provides strong support for the acquisition of PE by these mar-ine Cyanobium strains by HGT. Specically, the PEs from the marineCyanobium strains and Synechococcus 5.1 are mutually paraphyleticin the best CpeBA PE tree (Fig. 3 and Fig. 5A) and PE fromCyanobium is nested within Synechococcus 5.1 in the two alterna-tive spectral phylogenies for CpeBA not rejected by the AU test(Fig. 5A). If these results are combined with rejection of a tree withCyanobium CpeBA sister to all Synechococcus CpeBAs (Fig. 5B andTable 1), the data strongly support a conclusion that the studyCyanobium strains have acquired PE from marine Synechococcusby HGT. This line of evidence is supported independently by theshared identity of the length and sequences of the hypervariableclearly with strong statistical support that this model holds forthe evolution of PE genes in marine picocyanobacteria even thoughthey have a different evolutionary history than that of housekeep-ing loci like the 16S rRNA gene. For the groups studied here, thephylogenetic history of PE apoprotein genes is the same for se-quences that underlie any single spectral phenotype. Additionally,the data show that gene duplication appears to have preceded evo-lution of mpeBA genes, and the data indicate that this event wasalso required for inclusion of a PUB chromophore on any picocy-anobacterial PE. More revealing in terms of evolutionary mecha-nism is the fact that the evolution of PE clades associated withdifferent phenotypes is ordered in a sequence that suggests an evo-lutionary adaptation from a pigment system adapted to use greenwavelengths that predominate in waters with high chlorophyllcontent towards one with a greater ability to use the blue wave-lengths that predominate in the waters of the open ocean. This evi-dence for molecular evolution in response to natural selectioninvolves an apparent progression from simpler to more complexpigment systems after an episode of gene duplication. Togetherthe data presented here show marine picocyanobacteria to be anexceptional model system for demonstrating the molecular mech-anisms of adaptive phenotypic evolution.

Acknowledgments

The authors wish to thank Joe Thornton for numerous insightfuland helpful discussions. We also thank Scott Miller for critical com-ments andKevinEmerson for technical assistance. This researchwasfunded by the Ofce of Naval Research (ONR Grant 0149910177 toAMW), and NSF Grant OCE-0527139 (to AMW, CM Young, R Emlet,and W Jaeckles). RCE acknowledges the support of a RIKEN FPRfellowship. RCE was additionally supported by an NSF IntegratedTraining Grant (IGERT) in Evolution, Development, and Genomics(to the University of Oregon). Fieldwork in the Gulf of Mexico forRCE was also supported by NSF Grant OCE-0118733 and NOAA/NURP UNCW Grant NA96RU0260 (both to CM Young).

Appendix A. Supplementary material

Supplementary data associated with this article can be found,in the online version, at http://dx.doi.org/10.1016/j.ympev.2012.04.013.

References

Abascal, F., Zardoya, R., Posada, D., 2005. ProtTest: selection of best-t models ofprotein evolution. Bioinformatics 21, 21042105.

Ahlgren, N.A., Rocap, G., 2006. Culture isolation and culture-independent clonelibraries reveal new marine Synechococcus ecotypes with distinctive light and Nphysiologies. Appl. Environ. Microbiol. 72, 71937204.

Apt, K.E., Collier, J.L., Grossman, A.R., 1995. Evolution of the phycobiliproteins. J.region in the C-terminus of CpeBA as discussed above. The studyof additional PE genes from other Cyanobium and Cyanobium-likelineages will likely reveal more clearly the extent and nature ofsuch transfer events and the relationship between the distinct evo-lutionary lineages within the PE-containing picocyanobacteria.

The periodic selection model for bacterial populations predictsthat in clonal lineages, selective sweeps of strongly favored allelesresults in purges of less favored genotypes from the population(Atwood et al., 1951; Cohan, 2002; Whitaker and Baneld, 2006).For individual genes this results in an evolutionary pattern charac-terized by several niche-specic clusters of closely related se-quences separated from each other by considerable evolutionary

netics and Evolution 64 (2012) 381392Mol. Biol. 248, 7996.Atwood, K.C., Schneider, L.K., Ryan, F.J., 1951. Periodic selection in Escherichia coli.

Proc. Natl. Acad. Sci. USA 37, 146155.

logeBrahamsha, B., 1996. A genetic manipulation system for oceanic cyanobacteria ofthe genus Synechococcus. Appl. Environ. Microbiol. 62, 17471751.

Chen, F., Wang, K., Kan, J., Bachoon, D.S., Lu, J., Lau, S., Campbell, L., 2004.Phylogenetic diversity of Synechococcus in the Chesapeake Bay revealed byRibulose-1,5-bisphosphate carboxylase-oxygenase (RuBisCO) large subunitgene (rbcL) sequences. Aquat. Microb. Ecol. 36, 153164.

Coble, P., Hu, C., Gould, R.W., Chang, G., Wood, A.M., 2004. Colored dissolved organicmatter in the coastal ocean: an optical tool for coastal zone environmentalassessment and management. Oceanography 17, 5059.

Cohan, F.M., 2002. What are bacterial species? Annu. Rev. Microbiol. 56, 457487.Creevey, C.J., McInerney, J.O., 2002. An algorithm for detecting directional and non-

directional positive selection in protein coding DNA sequences. Gene 300, 4351.

Creevey, C.J., McInerney, J.O., 2003. CRANN: detecting adaptive evolution in protein-coding DNA sequences. Bioinformatics 19, 1726.

Crosbie, N.D., Pckl, M., Weisse, T., 2003. Dispersal and phylogenetic diversity ofnonmarine picocyanobacteria, inferred from 16S rRNA gene and cpcBA-intergenic spacer sequence analyses. Appl. Environ. Microbiol. 69, 57165721.

de Lamballerie, X., Zandotti, C., Vignoli, C., Bollet, C., de Micco, P., 1992. A one-stepmicrobial DNA extraction method using Chelex 100 suitable for geneamplication. Res. Microbiol. 143, 785790.

Dufresne, A., Ostrowski, M., Scanlan, D.J., Garczarek, L., Mazard, S., Palenik, B.P.,Paulsen, I.T., de Marsac, N.T., Wincker, P., Dossat, C., Ferriera, S., Johnson, J., Post,A.F., Hess, W.R., Partensky, F., 2008. Unraveling the genomic mosaic of aubiquitous genus of marine cyanobacteria. Genome Biol. 9, R90.

Ernst, A., Becker, S., Wollenzien, U.I.A., Postius, C., 2003. Ecosystem-dependentadaptive radiations of picocyanobacteria inferred from 16S rRNA and ITS-1sequence analysis. Microbiology 149, 217228.

Everroad, R.C., 2007. Diversication of Marine Picocyanobacteria: the Ecology andEvolution of Spectral Phenotype and Phycoerythrin. The University of Oregon,Eugene, OR.

Everroad, R.C., Wood, A.M., 2006. Comparative molecular evolution of newlydiscovered picocyanobacterial strains reveals a phylogeneticaly informativevariable region of b-phycoerythrin. J. Phycol. 42, 13001311.

Everroad, C., Six, C., Partensky, F., Thomas, J.C., Holtzendorff, J., Wood, A.M., 2006.Biochemical bases of type IV chromatic adaptation in marine Synechococcus spp.J. Bact. 188, 33453356.

Farris, J.S., Kllersj, M., Kluge, A.G., Bult, C., 1995. Testing signicance ofincongruence. Cladistics 10, 315319.

Fuller, N.J., Marie, D., Partensky, F., Vaulot, D., Post, A.F., Scanlan, D.J., 2003. Clade-specic 16S ribosomal DNA oligonucleotides reveal the predominance of asingle marine Synechococcus clade throughout a stratied water column in theRed Sea. Appl. Environ. Microbiol. 69, 24302443.

Gantt, E., 1981. Phycobilisomes. Annu. Rev. Plant Physiol. 32, 327347.Gascuel, O., 1997. BIONJ: an improved version of the NJ algorithm based on a simple

model of sequence data. Mol. Biol. Evol. 14, 685695.Glazer, A.N., 1989. Light guides. Directional energy transfer in a photosynthetic

antenna. J. Biol. Chem. 264, 14.Glover, H.E., Keller, M.D., Spinrad, R.W., 1987. The effects of light quality and

intensity on photosynthesis and growth of marine eukaryotic and prokaryoticphytoplankton clones. J. Exp. Mar. Biol. Ecol. 105, 137159.

Grossman, A.R., Schaefer, M.R., Chiang, G.G., Collier, J.L., 1993. The phycobilisome, alight-harvesting complex responsive to environmental conditions. Microbiol.Mol. Biol. Rev. 57, 725749.

Guillard, R.R.L., Ryther, J.H., 1962. Studies of marine planktonic diatoms. I. Cyclotellanana Hustedt and Detonula confervacea (Cleve) Gran. Can. J. Microbiol. 8, 229239.

Guindon, S., Gascuel, O., 2003. A simple, fast, and accurate algorithm to estimatelarge phylogenies by maximum likelihood. Syst. Biol. 52, 696704.

Guindon, S., Dufayard, J.F., Lefort, V., Anisimova, M., Hordijk, W., Gascuel, O., 2010.New algorithms and methods to estimate maximum-likelihood phylogenies:assessing the performance of PhyML 3.0. Syst. Biol. 59, 307321.

Hall, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor andanalysis program for windows 95/98/NT. Nucleic Acids Symp. Ser. 41, 9598.

Haverkamp, T., Acinas, S.G., Doeleman, M., Stomp, M., Huisman, J., Stal, L.J., 2008.Diversity and phylogeny of Baltic Sea picocyanobacteria inferred from their ITSand phycobiliprotein operons. Environ. Microbiol. 10, 174188.

Haverkamp, T.H.A., Schouten, D., Doeleman, M., Wollenzien, U., Huisman, J., Stal, L.J.,2009. Colorful microdiversity of Synechococcus strains (picocyanobacteria)isolated from the Baltic Sea. ISME J. 3, 397408.

Herdman, M., Castenholz, R.W., Waterbury, J.B., Rippka, R., 2001. Form-genus XIII.Synechococcus. In: Boone, D.R., Castenholz, R.W. (Eds.), Bergeys Manual ofSystematic Bacteriology, second ed., vol. 1. Springer-Verlag, NewYork, pp. 508-512.

Higgins, D.G., Sharp, P.M., 1988. CLUSTAL: a package for performing multiplesequence alignment on a microcomputer. Gene 73, 237244.

Higgins, D.G., Sharp, P.M., 1989. Fast and sensitive multiple sequence alignments ona microcomputer. Bioinformatics 5, 151153.

Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetictrees. Bioinformatics 17, 754755.

Jasser, I., Krlicka, A., Karnkowska-Ishikawa, A., 2011. A novel phylogenetic clade ofpicocyanobacteria from the Mazurian lakes (Poland) reects the early ontogenyof glacial lakes. FEMS Microbiol. Ecol. 75, 8998.

Katano, T., Nakano, S.I., 2006. Growth rates of Synechococcus types with different

R.C. Everroad, A.M. Wood /Molecular Phyphycoerythrin composition estimated by dual-laser ow cytometry inrelationship to the light environment in the Uwa Sea. J. Sea Res. 55, 182190.Li, W.K.W., 1994. Primary production of prochlorophytes, cyanobacteria, andeucaryotic ultraphytoplankton: measurements from ow cytometric sorting.Limnol. Oceanogr. 39, 169175.

Li, W.K.W., Wood, A.M., 1988. Vertical distribution of North Atlanticultraphytoplankton: analysis by ow cytometry and epiuorescencemicroscopy. Deep-Sea Res. A 35, 16151638.

MacColl, R., 1998. Cyanobacterial phycobilisomes. J. Struct. Biol. 24, 311334.MacDonald, I.R., Reilly, J.F., Guinasso, N.L., Brooks, J.M., Carney, R.S., Bryant, W.A.,

Bright, T.J., 1990. Chemosynthetic mussels at a brine-lled pockmark in thenorthern Gulf of Mexico. Science 248, 10961099.

Miller, S.R., Castenholz, R.W., 2000. Evolution of thermotolerance in hot springcyanobacteria of the genus Synechococcus. Appl. Environ. Microbiol. 66, 42224229.

Nei, M., Gojobori, T., 1986. Simple methods for estimating the numbers ofsynonymous and nonsynonymous nucleotide substitutions. Mol. Biol. Evol. 3,418426.

Newman, J., Mann, N.H., Carr, N.G., 1994. Organization and transcription of the classI phycoerythrin genes of the marine cyanobacterium Synechococcus sp.WH7803. Plant Mol. Biol. 24, 679683.

Nbel, U., Garcia-Pichel, F., Muyzer, G., 1997. PCR primers to amplify 16S rRNAgenes from cyanobacteria. Appl. Environ. Microbiol. 63, 33273332.

Olson, R.J., Chisholm, S.W., Zettler, E.R., Armbrust, E.V., 1988. Analysis ofSynechococcus pigment types in the sea using single and dual beam owcytometry. Deep-Sea Res. A 35, 425440.

Olson, R.J., Chisholm, S.W., Zettler, E.R., Armbrust, E.V., 1990. Pigment, size anddistribution of Synechococcus in the North Atlantic and Pacic oceans. Limnol.Oceanogr. 35, 4558.

Ong, L.J., Glazer, A.N., 1991. Phycoerythrins of marine unicellular cyanobacteria. I.Bilin types and locations and energy transfer pathways in Synechococcus spp.phycoerythrins. J. Biol. Chem. 266, 95159527.

Page, R.D.M., 1996. TREEVIEW: an application to display phylogenetic trees onpersonal computers. Comput. Appl. Biosci. 12, 357358.

Palenik, B., 2001. Chromatic adaptation in marine Synechococcus strains. Appl.Environ. Microbiol. 67, 991994.

Partensky, F., Blanchot, J., Vaulot, D., 1999. Differential distribution and ecology ofProchlorococcus and Synechococcus in oceanic waters: a review. Bull. Inst.Oceanogr. 19, 457475.

Posada, D., 2008. JModelTest: phylogenetic model averaging. Mol. Biol. Evol. 25,12531256.

Rippka, R., Castenholz, R.W., Herdman, M., 2001. Form-genus IV. Cyanobium. In:Boone, D.R., Castenholz, R.W. (Eds.), Bergeys Manual of SystematicBacteriology, second ed., vol. 1. Springer-Verlag, New York, pp. 498-499.

Robertson, B.R., Tezuka, N., Watanabe, M.M., 2001. Phylogenetic analyses ofSynechococcus strains (cyanobacteria) using sequences of 16S rDNA and partof the phycocyanin operon reveal multiple evolutionary lines and reectphycobilin content. Int. J. Syst. Evol. Microbiol. 51, 861871.

Rocap, G., Distel, D.L., Waterbury, J.B., Chisholm, S.W., 2002. Resolution ofProchlorococcus and Synechococcus ecotypes by using 16S23S ribosomal DNAinternal transcribed spacer sequences. Appl. Environ. Microbiol. 68, 11801191.

Scanlan, D.J., Ostrowski, M., Mazard, S., Dufresne, A., Garczarek, L., Hess, W.R., Post,A.F., Hagemann, M., Paulsen, I., Partensky, F., 2009. Ecological genomics ofmarine picocyanobacteria. Microbiol. Mol. Biol. Rev. 73, 249299.

Shimodaira, H., 2002. An approximately unbiased test of phylogenetic treeselection. Syst. Biol. 51, 492508.

Shimodaira, H., Hasegawa, M., 2001. CONSEL: for assessing the condence ofphylogenetic tree selection. Bioinformatics 17, 12461247.

Six, C., Thomas, J.C., Garczarek, L., Ostrowski, M., Dufresne, A., Blot, N., Scanlan, D.J.,Partensky, F., 2007. Diversity and evolution of phycobilisomes in marineSynechococcus spp.: a comparative genomics study. Genome Biol. 8, R259.

Stomp, M., Huisman, J., de Jongh, F., Veraart, A.J., Gerla, D., Rijkeboer, M., Ibelings,B.W., Wollenzien, U.I.A., Stal, L.J., 2004. Adaptive divergence in pigmentcomposition promotes phytoplankton biodiversity. Nature 432, 104107.

Stomp, M., Huisman, J., Vrs, L., Pick, F.R., Laamanen, M., Haverkamp, T., Stal, L.J.,2007. Colourful coexistence of red and green picocyanobacteria in lakes andseas. Ecol. Lett. 10, 290298.

Swofford, D.L., 2003. PAUP. Phylogenetic Analysis Using Parsimony ( and OtherMethods). Version 4. Sinauer Associates, Sunderland, Massachusetts.

Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. TheclustalX windows interface: exible strategies for multiple sequence alignmentaided by quality analysis tools. Nucleic Acids Res. 24, 48764882.

Ting, C.S., Rocap, G., King, J., Chisholm, S.W., 2001. Phycobiliprotein genes of themarine photosynthetic prokaryote Prochlorococcus: evidence for rapid evolutionof genetic heterogeneity. Microbiology 211, 31713182.

Toledo, G., Palenik, B., Brahamsha, B., 1999. Swimming marine Synechococcus strainswith widely different photosynthetic pigment ratios form a monophyleticgroup. Appl. Environ. Microbiol. 65, 52475251.

Urbach, E., Scanlan, D.J., Distel, D.L., Waterbury, J.B., Chisholm, S.W., 1998. Rapiddiversication of marine picophytoplankton with dissimilar light-harvestingstructures inferred from sequences of Prochlorococcus and Synechococcus(Cyanobacteria). J. Mol. Evol. 46, 188201.

Waterbury, J.B., Watson, S.W., Valois, F.W., Franks, D.G., 1986. Biological andecological characterization of the marine unicellular cyanobacteriumSynechococcus. Can. Bull. Fish Aquat. Sci. 214, 71120.

netics and Evolution 64 (2012) 381392 391Whitaker, R.J., Baneld, J.F., 2006. Population genomics in natural microbialcommunities. Trend. Ecol. Evol. 21, 508516.

Wilbanks, S.M., de Lorimer, R., Glazer, A.N., 1991. Phycoerythrins of marineunicellular cyanobacteria. III. Sequence of a class II phycoerythrin. J. Biol.Chem. 266, 95359539.

Wingard, L.L., Miller, S.R., Sellker, J.M.L., Stenn, E., Allen, M.M., Wood, A.M., 2002.Cyanophycin production in a phycoerythrin-containing marine Synechococcusstrain of unusual phylogenetic afnity. Appl. Environ. Microbiol. 68, 17721777.

Wood, A.M., 1985. Adaptation of photosynthetic apparatus of marineultraphytoplankton to natural light elds. Nature 316, 253255.

Wood, A.M., Townsend, D., 1990. DNA Polymorphism within the WH7803serogroup of marine Synechococcus spp. J. Phycol. 26, 576585.

Wood, A.M., Phinney, D.A., Yentsch, C.S., 1998. Water column transparency and thedistribution of spectrally distinct forms of phycoerythrin-containing organisms.Mar. Ecol. Prog. Ser. 162, 2531.

Wood, A.M., Everroad, R.C., Wingard, L.M., 2005. Measuring growth rates inmicroalgal cultures. In: Anderson, R.A. (Ed.), Algal Culturing Techniques.Elsevier Academic Press, Burlington, MA, pp. 269285.

Zhao, F., Qin, S., 2006. Evolutionary analysis of phycobiliproteins: implications fortheir structural and functional relationships. J. Mol. Evol. 63, 330340.

Zwickl, D.J., 2006. Genetic Algorithm Approaches for the Phylogenetic Analysis ofLarge Biological Sequence Datasets under the Maximum Likelihood Criterion.The University of Texas, Austin, TX.

392 R.C. Everroad, A.M. Wood /Molecular Phylogenetics and Evolution 64 (2012) 381392

Phycoerythrin evolution and diversification of spectral phenotype in marine Synechococcus and related picocyanobacteria1 Introduction2 Materials and methods2.1 Isolation, selection, and maintenance of strains used in this study2.2 Determination of spectral phenotype2.3 DNA isolation, amplification, cloning and sequencing2.4 Nucleotide sequence editing, alignment, and phylogenetic reconstruction2.5 Tests of evolution2.6 Sequence data

3 Results and discussion3.1 Phenotypic characterization of new strains3.2 Evolutionary relationships between Cyanobium and marine Synechococcus3.2.1 Evolution of picocyanobacterial PEs3.2.2 Approximately unbiased tests3.2.3 Relative rates ratio test

3.3 Hypervariable region of CpeB3.4 Concluding remarks

AcknowledgmentsAppendix A Supplementary materialReferences