Ecosystem-dependent adaptive radiations of picocyanobacteria inferred from 16S rRNA and ITS-1 sequence analysis

Anneliese Ernst1, Sven Becker2, Ute I. A. Wollenzien1 and Christine Postius2

1 NIOO-Centre for Estuarine and Marine Ecology, NL-4400 AC Yerseke, The Netherlands
2 Lehrstuhl für Physiologie und Biochemie der Pflanzen, Universität Konstanz, D-78457 Konstanz, Germany

Correspondence
Anneliese Ernst
a.ernst{at}nioo.knaw.nl


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Small, coccoid and rod-shaped Synechococcus-type cyanobacteria with either phycoerythrin or phycocyanin as major accessory pigments were isolated from several large, temperate-zone lakes and the brackish Baltic Sea. The picocyanobacteria had two ribosomal operons with a long internal transcribed spacer (ITS-1) separating the 16S rDNA and 23S rDNA. A 16S rRNA-based phylogenetic analysis assigned all isolates to the picophytoplankton clade [sensu Urbach, E., Scanlan, D. J., Distel, D. L., Waterbury, J. B. & Chisholm, S. W. (1998). J Mol Evol 46, 188–201], which also comprises marine Synechococcus spp. and Prochlorococcus spp. The strains assorted to five paraphyletic clusters each containing two or more strains with 99·4–100 % 16S rRNA sequence identity. Five corresponding strain clusters were deduced from analysis of ITS-1 sequences. Sequence divergence in ITS-1 varied between 23 % in the most divergent and 1 % in the phylogenetically most conserved cluster. Clustered strains with low sequence divergence in ITS-1 were frequently isolated from a single ecosystem or hydrographically comparable lakes in the same region. They represent physiologically distinct ecotypes of species which, among other phenotypic variations, frequently differed in their major accessory pigments, the phycobiliproteins. The reproduction of the various pigment traits in different lineages was not correlated with the phylogenetic divergence deduced from 16S rRNA or ITS-1 sequences but rather seemed to be related to characteristics of the ecosystem and habitat from which the strains were isolated. The occurrence of a comparable spectrum of phenotypes in different lineages and ecosystems indicates that different strain clusters developed similar ecotypes during independent adaptive radiations.

Abbreviations: CCA, complementary chromatic adaptation; ITS, internal transcribed spacer; PC, phycocyanin; PE, phycoerythrin

The GenBank accession numbers for sequences reported in this paper are AF317071AF317079, AF330246AF330254 and AF466990AF467000.


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Synechococcus is a botanical form-genus comprising coccoid and rod-shaped cyanobacteria with a diameter of 0·6–2·1 µm that divide in one plane. In pelagic environments, small Synechococcus-type cyanobacteria often thrive as solitary organisms or small colonies and form together with other non-bloom-forming small algae, the autotrophic picoplankton (Stockner et al., 2000). However, Synechococcus spp. are also important components of benthic cyanobacterial mats (Ward et al., 1998) and they can form compacted epiphytic or epilithic colonies (S. Becker, A. K. Singh, C. Postius, P. Böger & A. Ernst, unpublished results). Analyses of the DNA composition (mol% G+C; Herdman et al., 1979) and of 16S rRNA sequences showed that the form-genus Synechococcus is polyphyletic and representatives were located in five of eight deeply branching cyanobacterial lineages (Honda et al., 1999; Robertson et al., 2001). To make things worse, one of the most prominent phenotypic traits, the composition of photosynthetic pigments, determining the colour and autofluorescence of these organisms, varied within these lineages. In one of them, termed the picophytoplankton clade by Urbach et al. (1998), Synechococcus spp. utilizing the phycobiliproteins phycoerythrin (PE) or phycocyanin (PC) as major accessory pigments, and Prochlorococcus spp., which only use chlorophylls for light harvesting, are combined in a monophyletic group based on 16S rRNA and DNA-dependent RNA polymerase gene sequences (Palenik & Swift, 1996; Urbach et al., 1998). The instability of the pigment trait seemed surprising because many other traits, such as complementary chromatic adaptation (CCA), salt tolerance or diazotrophy (Rippka et al., 1979), motility (Toledo et al., 1999), thermotolerance (Miller & Castenholz, 2000) and extreme halotolerance (Garcia-Pichel et al., 1998) seemed to have a limited phylogenetic distribution and seemed to be characteristics useful in cyanobacterial taxonomy.

Genetic traits often reflect characteristics of a corresponding selective force. For example, light quality (i.e. light of limited range of light wavelengths) varies with depth of the water column, with distance from the coast and also with biotic factors. Light quality is also known to affect the synthesis of phycobiliproteins, which allows cyanobacteria capable of CCA to acclimate to the ambient light climate (for review see Grossman et al., 1993). However, most of the Synechococcus-type cyanobacteria of the autotrophic picoplankton produce either PE or PC as major light-harvesting pigments. Thus, they cannot acclimate to light by synthesis of complementary pigments. In this case, light quality is thought to influence the abundance and distribution of the different pigment types in the ecosystem. For example, PC-rich strains, which absorb mainly red light, occur frequently in shallow and eutrophic ecosystems, where mixing depth is limited either physically or by very stable stratification (Vörös et al., 1991). On the other hand, PE-rich strains, which can take advantage of blue light penetrating deeply in clear water columns, dominate in the autotrophic picoplankton of deep oligotrophic and mesotrophic lakes and in the open ocean (Pick, 1991; Wood et al., 1998).

The PE- and PC-rich picocyanobacteria described in this study have been isolated from several deep subalpine lakes (central Europe), Lake Biwa (Japan), Lake Balaton (Hungary) and the Baltic Sea, a large brackish water ecosystem. In all these ecosystems, both pigment types coexist and dominate the autotrophic picoplankton. Physiological studies showed that apart from the obvious difference in the major accessory pigments, PC and PE, the strains also differed in carotenoid content, light stress tolerance, response to nutrient limitation and surface structures (for review see Postius & Ernst, 1999). In addition, most freshwater isolates have previously been characterized as distinct genotypes by a genomic fingerprinting technique, RFLP of psbA genes (Ernst et al., 1995; Postius et al., 1996). However, the phylogenetic relationships of these strains have not been examined. Phylogenetic relationships are often inferred from comparison of coding and non-coding sequences in the ribosomal operon. In cyanobacteria, this operon includes genes encoding 16S rRNA, 23S rRNA and 5S rRNA and none, one or two tRNAs (Iteman et al., 2000). 16S rRNA has been frequently used to establish a gene-based phylogeny of cyanobacteria (Giovannoni et al., 1988; Wilmotte et al., 1992, 1993, 1994; Nelissen et al., 1996; Turner, 1997; Honda et al., 1999; Lyra et al., 2001; Robertson et al., 2001). However, 16S rRNA sequences often do not allow discrimination of bacterial strains at the subgeneric level (Fox et al., 1992). Therefore, the less conserved internal transcribed spacer (ITS) sequences, which are removed from the primary transcript during rRNA processing, have also been used for bacterial identification, typing and evolutionary studies (Gürtler & Stanisich, 1996). In several studies the amplified 16S–23S rDNA spacer (ITS-1) of cyanobacteria has been used to genetically characterize strains by PCR-RFLP (Lu et al., 1997; West & Adams, 1997; Scheldeman et al., 1999) or by sequence analyses (Wilmotte et al., 1994; Otsuka et al., 1999; Boyer et al., 2001; Rocap et al., 2002). These studies indicated the presence of multiple ribosomal operons with differences in ITS-1 in some strains. Genome analysis confirmed that cyanobacteria (including Prochlorococcus spp.) comprise one, two or four ribosomal operons [Kaneko et al., 1996; see also ongoing sequencing projects at the DOE Joint Genome Institute (http://www.jgi.doe.gov./JGI_microbial/html/index.html) and at the Kazusa DNA research institute (http://www.kazusa.or.jp/cyano/)]. As the number of ribosomal operons can vary among phylogenetically closely related cyanobacteria (i.e. among different Prochlorococcus spp.), this information may be an important criterion in strain characterization.

In this study, we analysed the 16S rRNA and ITS-1 of 19 picocyanobacteria that were isolated from different sources and that differed in genotype (RFLP of psbA genes) and/or phenotype. For 10 additional isolates from the same ecosystems only ITS-1 sequences were determined. The number of ribosomal operons per genome was examined by hybridization of genomic DNA with probes targeting 16S rDNA and ITS-1 in Southern blots. The analysis of the sequences and the deduced phylogenetic assignment of strains did not support our original distinction of isolates according to phenotype, i.e. pigmentation. Rather, the genetic and phenotypic diversification indicated that selection processes may have shaped local populations leading to similar phenotypes in different phylogenetic lineages. Therefore, we discuss the possibility that diversification among Synechococcus spp. is caused by ecosystem-specific adaptive radiations.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Source and maintenance of cyanobacteria.
Between 1988 and 2000, 16 different genotypes of Synechococcus spp. were isolated from the pelagic and the littoral zone of Lake Constance, a large (476 km2) and deep (254 m max. depth) subalpine lake in central Europe. Similar organisms were isolated from the pelagic zone of two other subalpine lakes, Lake Zurich (Switzerland; Chang, 1980) and Lake Maggiore (Italy; Callieri et al., 1996), and from two large lakes of non-glacial origin, Lake Balaton (Hungary; Callieri et al., 1996) and Lake Biwa (Japan; Maeda et al., 1992). In most cases, non-axenic liquid cultures were obtained from colonies growing on mineral medium BG11 solidified with washed agar (Difco) and maintained in liquid culture (for details see Ernst et al., 1992).

Picocyanobacteria from the Baltic Sea were obtained from 10 m water depth in August 1996 at sampling sites in the Bornholm Sea (surface salinity 9 g l-1; strains BS 4, BS 5, BS 7, BS 8) and in June 1999 in the Gotland Sea (salinity 7 g l-1; strain BS 20). Original samples contained untreated seawater (sampling 1996) or a fraction that had passed a 3 µm filter and remained on a 0·2 µm filter (sampling 1999). The original material was cultivated in a growth medium prepared from 1 part ASN III and 3 parts BG11. Successively, seven strains were isolated that were able to form colonies on ASN/BG11 solidified with 0·7 % agarose (Sigma).

Synechococcus rubescens SAG B 3.81, originating from Lake Zurich (Chang, 1980), and Synechococcus sp. SAG 1403-1 (synonyms Anacystis nidulans and Synechococcus PCC 6301) were obtained from Sammlung von Algenkulturen Göttingen (SAG, Universität Göttingen, Albrecht-von-Haller-Institut für Pflanzenwissenschaften, Untere Karspüle 2, 37073 Göttingen, Germany). Synechococcus sp. SAG 1403-1 will be addressed under its former name, Anacystis nidulans, to emphasize its distinct phylogenetic origin. Both strains were maintained in BG11.

DNA preparation and Southern analysis.
Two to three weeks after inoculation of 40 ml batches containing BG11, cultures were harvested by centrifugation (7 min, 7000 g), frozen in liquid nitrogen and stored at -20 °C. Genomic DNA was extracted according to Postius et al. (1996). For RFLP analysis, 5–10 µg genomic DNA was completely digested with different restriction endonucleases (Boehringer), separated by agarose gel electrophoresis and blotted on Hybond (Amersham) membranes (Ernst et al., 1995). Restriction fragments were probed with a 16S rDNA fragment labelled with UTP-digoxigenin in a PCR reaction with the primer pair P16S3p/P16S4m or the primer pair PITS2/PITS4 (see Table 1) using DNA of S. rubescens as template. Labelled fragments and DIG-labelled molecular size markers (Boehringer) on Southern blots were detected using chemiluminescent substrates of alkaline phosphatase.


View this table:
[in this window]
[in a new window]
 
Table 1. Primers used in this study

 
DNA amplification and restriction analysis.
Different segments of the rDNA were amplified by PCR (Mullis & Faloona, 1987). PCR primers (see Table 1) targeting coding regions of the ribosomal operon (16S rDNA, tRNAIle, tRNAAla and 23S rDNA) were deduced from the ribosomal sequences of A. nidulans (PCC 6301) obtained from GenBank. Primers PITS5–PITS8 were developed for pelagic Synechococcus spp. examined in this study. DNA amplification assays contained in 100 µl: 1x times; buffer from the enzyme supplier, supplemented with 0·5 mg BSA ml-1 if required by the manufacturer, 1·5 or 2·5 mM MgCl2, 1 µM of each primer, 200 µM of each deoxynucleoside triphosphate, 10–40 ng DNA and 0·5–2·5 units Taq polymerase from various sources (Boehringer, MBI Fermentas or Eurobio). The mixtures were heated to 95 °C for 2 min followed by 30 cycles of 45 °C, 0·6 min; 70 or 72 °C, 2 min; 95 °C, 0·6 min and a final extension at 72 °C for 5 min in a thermal cycler (Grant autogene II or PTI, MJ Research). The amplified DNA was purified by passage through ion exchange columns (from Qiagen or Boehringer) and resuspended in 50 µl H2O or Tris/EDTA (1/0·1 mM), pH 7·5. Aliquots were checked for homogeneity on 2 % agarose gels stained with ethidium bromide. Restriction enzymes for characterization of PCR products, AvaII and HaeII, were obtained from Boehringer and Amersham-Buchler. DNA size markers were from AGS.

DNA sequencing.
The PCR fragments were sequenced by a non-radioactive version of the dideoxynucleotide chain-termination method (Sanger et al., 1977), employing the GATC 1500 System for direct blotting electrophoresis as described by Pohl & Maier (1995). Non-radioactive labels were introduced in cyclic sequencing reactions either via DIG-labelled primers PITS1–PITS8, using the DIG- Sequencing Kit, including Taq polymerase (Boehringer) as described by Brass et al. (1996), or via Biotin-labelled dideoxynucleotides using ThermoSequenase (Amersham-Buchler; reaction kit provided by GATC) as described by Neuschaefer-Rube et al. (2000). Ribosomal sequences of strains isolated from the Baltic Sea and from the littoral zone of Lake Constance were obtained with an ABI sequencer (GATC).

Sequence alignment and analysis.
The sequences were assembled, analysed and aligned using programs of the software package PCGENE, version 6.7, from IntelliGenetics (Mountain View, CA, USA). For sequence comparison, sequences were searched in GenBank using the alignment tools provided by Altschul et al. (1990). Preliminary sequence data of several unicellular cyanobacteria were obtained from the DOE Joint Genome Institute (JGI) at http://www.jgi.doe.gov./JGI_microbial/html/index.html The sequences were aligned using CLUSTALX, originally developed by Higgins & Sharp (1988). 16S rRNA and ITS-1 sequences were analysed separately. For 16S rRNA analysis, the 20 nt encoded by the primer P16SAnf were excluded, resulting in analysis of 1456 nt (98 %) of 16S rRNA. Aligned sequences were corrected manually using the program GENEDOC (http://www.psc.edu/biomed/genedoc/). For analysis of ITS-1, sequences encoding 16S and 23S rRNA were excluded. Genetic distances were calculated and corrected for multiple base exchanges by the method of Jukes & Cantor (1969). The phylogenetic trees were reconstructed by the neighbour-joining method of Saitou & Nei (1987), including bootstrap analysis (Felsenstein, 1985) as provided by TREECON (Van de Peer & De Wachter, 1994). Secondary structures of segments of ITS-1 were analysed using the program RNADRAW V1.1 of Ole Mazura (ole{at}mango.mef.ki.se).


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Sequence polymorphism in the 16S rRNA and phylogenetic inference
Using different primer pairs (P16SAnf+P16Send3, P16SAnf+PITS5, P16SAnf+PITS2; see Fig. 1 and Table 1) DNA fragments that encode 16S rRNA and/or ITS-1 were amplified from DNA of picocyanobacteria grown in non-axenic cultures. DNA sequences were obtained by direct sequence analysis of PCR fragments. In all cases the ribosomal sequences overlapped unambiguously. For 9 PC-rich and 10 PE-rich strains the complete 16S rRNA sequence was obtained except for 27 nt extending from the 5' terminus to the 3' end of primer P16SAnf. The 16S rRNA of these 19 strains exhibited only 84 (5·8 %) variable positions, 54 of which located in five variable locations, representing less than 13 % of the gene. According to a secondary structure model of 16S rRNA, 30 mutations occurred in stem sections of eight hairpin structures identified by Wilmotte et al. (1993). In each case a complementary nucleotide exchange was established. Two exchanges of motifs, one in helix 12 and the other in helix 49, were observed. The variable internal loop in helix 12 contained one of the triplets TTC, ATT or GAC. The terminal loop of helix 49 was either formed by a pentanucleotide (CTTGT) or by tetranucleotides (GCAA or GTAA) with elongation of the stem by 1 bp (Fig. 2). Due to this elongation the analysed 16S rRNA in the strains LB P1 (Lake Biwa), BO 8805, BO 8807, BO 0014 (Lake Constance) and BS 4 and BS 5 (Bornholm Sea) was 1 nt longer than that of the other strains.



View larger version (20K):
[in this window]
[in a new window]
 
Fig. 1. Location of primers for PCR and sequencing reactions in 16S rRNA and the 16S–23S rRNA ITS-1 of Synechococcus-type cyanobacteria. The polymorphism of ITS-1 of two strains, A. nidulans (Synechococcus sp. PCC 6301) and S. rubescens SAG 3.81, containing 710 and 909 nt, respectively, is depicted schematically. The primer positions in the non-coding sections of ITS-1 of S. rubescens were conserved in all strains examined in this study. Genes are hatched and non-coding sequences are depicted with a white background; the 5'->3' orientation of primers is indicated by arrowheads.

 


View larger version (55K):
[in this window]
[in a new window]
 
Fig. 2. Alignment of the terminal part of helix 49 in the 16S rRNA of Synechococcus spp. Black bars on the right indicate phylogenetic assignment to 16S rRNA-inferred strain clusters depicted in Fig. 3. Deviant nucleotides in the stem are shaded in grey; the deviant motifs in the terminal loop are printed in white letters on a black background. S. rub., S. rubescens.

 
In a pairwise comparison of 16S rRNA, strains formed five clusters, each one comprising a lineage of strains with 99·4–100 % sequence identity. Seven strains, six PC-rich isolates from Lake Constance and Lake Balaton and one PE-rich isolate from the Baltic Sea, formed a cluster exhibiting only nine variable positions in 16S rRNA. Phylogenetic analysis based on 1405 aligned positions showed that these strains were closely related to Cyanobium gracile strain PCC 6307T, the type strain of the Cyanobium cluster sensu Rippka et al. (1979) (Fig. 3). A second cluster including S. rubescens SAG B3.81 comprised all pelagic PE-rich strains from three subalpine lakes, Lake Constance, Lake Zurich and Lake Maggiore. Strains of this subalpine cluster I exhibited eight variable positions in 16S rRNA, five of which related to the aforementioned motif exchange in helix 49 (Fig. 2). As two other mutations represented a complementary base exchange, we can assume that the topology of the 16S rRNA of this cluster varied in only three positions. Three other lineages were identified, each including at least on PE-rich and one PC-rich isolate. They were formed by two isolates from Lake Constance (subalpine cluster II), by two out of four strains isolated from Lake Biwa (Japan) and by the strains isolated from the Bornholm Sea. All five lineages were supported by high bootstrap values and exhibited unambiguous monophyletic origins in the picophytoplankton clade as defined by Urbach et al., 1998 (Fig. 3).



View larger version (50K):
[in this window]
[in a new window]
 
Fig. 3. Phylogenetic inference of the picophytoplankton clade sensu Urbach et al. (1998) comprising Synechococcus-type picocyanobacteria isolated from freshwater, brackish and marine ecosystems, and Prochlorococcus spp. Three phylogenetic clusters that correspond with the form-genera Cyanobium, Synechococcus cluster 5 (marine) and Prochloroccus described in Boone et al. (2001) are indicated on the right. The fourth cluster, comprising Antarctic strains, has not been taxonomically assigned. Within the form-genus Cyanobium, four lineages have been highlighted with a grey background to facilitate comparison with Fig. 4. The tree is based on comparison of 1405 nt of the 16S rRNA and is inferred by the distance matrix function and the neighbour-joining algorithm provided by the program TREECON. The distance between two organisms, in substitutions per nt, is obtained by summing up the horizontal branches connecting them. Numbers at nodes indicate the frequency with which the cluster descending from that node was found in 100 bootstrap trees. Strains are identified by strain numbers and the accession number under which the sequences are available in GenBank. Three species indicated, C. gracile, S. rubescens and P. marinus have been proposed as type strains for the corresponding strain clusters, while another strain, Microcystis elabens, is a mis-classified Synechococcus spp. (Otsuka et al., 1999). Sequences added in this study are printed in bold type and isolates from Lake Constance are indicated by an asterisk.

 
Sequence polymorphism in ITS-1 sequences
ITS-1 sequences were obtained by direct sequencing of PCR fragments using the primer pair PITS4+PITS2 (Fig. 1). The complete ITS-1 sequences, including 16 nt of the 23S rDNA, were obtained from 29 isolated strains characterized by a unique genetic fingerprint of psbA genes or by a unique phenotype. All ITS-1 sequences embraced genes encoding tRNAIle and tRNAAla. Each genotype defined by a unique pattern in RFLP analysis of psbA genes also exhibited unique mutations in the non-coding sections of ITS-1. Base ambiguities detected in both strands were observed in ITS-1 of strain BO 9403 and in two strains isolated from the Bornholm Sea (BS 5 and BS 6). On the other hand, strains with identical RFLP of psbA genes, such as the two PE-rich strains BO 8807 and BO 8810 or the three PC-rich strains BO 9301, BO 9302 and BO 9303 (Ernst et al., 1995; Postius et al., 1996), exhibited sequence identity in ITS-1 (data not shown).

Because of extended length and sequence divergence between the ITS-1 sequences of different 16S rRNA-defined lineages, sequences were initially aligned per lineage. Within these clusters the largest variability was observed among strains assigned to the C. gracile cluster (see above). The length of ITS-1 varied between 976 and 1033 nt and the maximum divergence in pairwise comparison was 22·9 %. In contrast, ITS-1 of stains of the subalpine cluster I, which exhibited a similar number of variable bases in 16S rRNA, contained 911–914 nt with only 5·0 % difference in pairwise comparison. The ITS-1 sequences of the two clustered strains from Lake Biwa, LB B3 and LB G2 (length 812/813 nt), and two strains of the subalpine cluster II, BO 0014/BO 8805 (length 847/851 nt), differed at 5·8 and 11·2 % of aligned positions, respectively. The least divergence was observed among ITS-1 sequences of six isolates from the Bornholm Sea. The PC- and PE-rich strains of this cluster exhibited only 12 variable positions in 887 nt of ITS-1 with a maximum of 1·0 % sequence divergence between two strains.

A low degree of sequence conservation and extended insertions/deletions in the ITS-1 complicated the alignment of ITS-1 sequences of the five lineages and the construction of a phylogenetic tree (Fig. 4). Unambiguous alignment of all strains, including a unique isolate with the shortest ITS-1 sequence in this study, strain LB P1 (703 nt), was possible only for the two tRNAs, for about 30 nt adjacent to the 16S rDNA and the target sites of primers PITS5/6 and PITS7/8 (Fig. 1). We tried to improve the alignment using structures in the non-coding sections of the sequence that were identified by Iteman et al. (2000). Domains D2 and D4, which form the base of highly complex stem–loop structures carrying the 16S rRNA and the 23S rRNA, were located in the target sequence of primers PITS5/6 and of primers PITS7/8, respectively. The latter also contained the antiterminator site, box B of the spacer. No other common structures were detected. The lack of conserved elements in this long spacer sequence led to deep branching in the phylogenetic inference with low statistical support of the arrangement of the lineages by bootstrap analysis (Fig. 4).



View larger version (52K):
[in this window]
[in a new window]
 
Fig. 4. Phylogenetic tree of picocyanobacteria isolated from different temperate zone lakes and the Baltic Sea based on sequence data from ITS-1. Phylogenetic relationships were inferred using the distance matrix function of the program TREECON. Sequences within clusters were aligned before the alignment of clusters. The two conserved genes located in ITS-1 and encoding tRNAIle and tRNAAla (see Fig. 1) were included. The final alignment was corrected manually. Numbers at nodes indicate the frequency (%) at which the cluster was found in 1000 bootstrap trees. The origin of the isolates is indicated at the long branches and when depicted with a black background refers to clusters with a high bootstrap support in the 16S rRNA inferred phylogeny (Fig. 3). PE- and PC-rich isolates are marked by a light and dark grey background, respectively. Sequences added by this study are printed in bold type and isolates from Lake Constance are indicated by an asterisk.

 
Number of ribosomal operons per genome and PCR-based strain typing
Southern analysis of chromosomal DNA of 20 clonal isolates from the pelagic zones of Lake Constance, Lake Zurich, Lake Maggiore and Lake Biwa was performed using DIG-labelled probes targeting either the 16S rDNA between P16S3p and P16S4m or the ITS-1 between PITS2 and PITS4 (compare Fig. 1). Southern blots of DNA digested with endonucleases that cut in non-conserved adjacent sequences of the ribosomal operon and that lacked restriction sites within the probed area (BamHI, BstEI) produced two long labelled fragments with both probes (data not shown). This is consistent with the assumption that the genome of each strain carries two ribosomal operons. However, frequently different genotypes within a phylogenetic lineage were indistinguishable by the genomic RFLP of the ribosomal operons (Fig. 5).



View larger version (118K):
[in this window]
[in a new window]
 
Fig. 5. Southern analysis of genomic DNA of PE-rich Synechococcus spp. assigned to subalpine cluster I. Cyanobacteria were isolated from Lake Constance (strains BO 8807, BO 8808, BO 8809, BO 9102, BO 9203, BO 9402, BO 9403), Lake Zurich (S. rubescens) and Lake Maggiore (LM 92, LM 94). DNA was digested with the restriction endonuclease BstEII. The blotted fragments were hybridized with a DIG-labelled 318 nt fragment of 16S rDNA. MII, DIG-labelled DNA marker II.

 
Southern analysis of genomic DNA requires large amounts of purified DNA from cultivated strains, while PCR can be performed with a small number of cells, derived from a single colony and furthermore, purification of DNA is often not required (Becker et al., 2002). This makes PCR-based strain typing with restriction enzymes much more convenient than genomic typing. Sequence analysis had shown that the ITS-1 has only few common elements, such as the genes encoding tRNAIle and tRNAAla. This property was used to develop an ITS-1-based typing applicable for all strains carrying these genes. The ITS-1 was amplified with primers targeting conserved sequences in the 3' terminus of 16S rDNA and the 5' terminus of 23S rDNA, and the amplified fragments were restricted with the endonucleases AvaII and HaeII, targeting sequences in the tRNAIle and tRNAAla, respectively. With this protocol a single fragment was amplified in all strains except LB P1 (Fig. 6a) and restriction resulted in characteristic patterns for each of the phylogenetic lineages (Fig. 6b). Notably, strains of the Cyanobium cluster, which exhibited a length polymorphism in ITS-1 (Fig. 6a) and hence, did not appear to be related by this criterion, exhibited a characteristic PCR-RFLP (Fig. 6b).



View larger version (126K):
[in this window]
[in a new window]
 
Fig. 6. PCR products covering ITS-1 (a) and restriction of these fragments with endonucleases AvaII and HaeII (b). The amplified fragments contain the complete ITS-1 plus short flanking sequences in 16S and 23S rDNA. Strain numbers refer to strains described in this study; bar I marks strains assigned to the C. gracile cluster, bar II marks strains assigned to the subalpine cluster I on the basis of 16S rRNA inferred phylogeny. Other bacteria: A. nidulans=Synechococcus sp. PCC 6301; E. coli, Escherchia coli K-12; A. ATCC 29413, Anabaena variabilis ATCC 29413; A. PCC 7120, Anabaena sp. PCC 7120; S. BO 8402, Synechocystis sp. BO 8402. In lanes 1–18 of (a), 3 µl of a PCR reaction or size markers was applied, while 9 µl was dispensed in lanes 19–22 and in all lanes of (b). Marker, DNA base ladders.

 

   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Phylogenetic assignment and the significance of different typing methods
The phylogenetic relationships of PE- and PC-rich picocyanobacteria isolated from five temperate zone lakes and the brackish Baltic Sea were inferred from almost complete 16S rRNA sequences and compared with Synechococcus spp. from other ecosystems using TREECON (van de Peer & de Wachter, 1994) (Fig. 3). All strains were unambiguously assigned to the picophytoplankton clade sensu Urbach et al. (1998), which corresponds to the ‘major lineage 6’ described by Honda et al. (1999) and lineages 6a+6b by Robertson et al. (2001). The clade is robust and can be reconstructed with outgroups consisting of heterotrophic bacteria as well as with the paraphyletic strains A. nidulans and Synechococcus spp. PCC 7942 of lineage 6c in the aforementioned analysis of Robertson et al. (2001). The genetic diversification of 16S rRNA appeared slightly distorted by the occurrence of numerous complementary base exchanges in hairpin structures and by motif exchanges, which are treated as independent events in phylogenetic programs, but which may have been introduced by recombination of small pieces of homologous DNA during horizontal gene transfer (Jain et al., 1999). In particular, the motif exchange in helix 49 (Fig. 2) may have unduly increased the phylogenetic distance of strain BO 8807 in subalpine cluster I (Fig. 3). However, phylogenetic analysis of ITS-1 not only confirmed all 16S rRNA-deduced clusters, but also the offside position of strain BO 8807 (Fig. 4). The clustering of sequences was statistically strongly supported in both phylogenetic inferences, but due to the small variance in 16S rRNA on one hand and a largely arbitrary alignment of the five ITS-1-inferrred clusters in regions of extended sequence divergence, the relationships between the paraphyletic lineages remained ambiguous (Figs 3 and 4).

Initially, most of the isolated strains were typed and identified using the monocistronic psbA genes, a gene family encoding a protein of photosystem II, as targets in Southern analyses (for review see Postius & Ernst, 1999). By selecting restriction enzymes that do not cut within the probed area we showed that each genome of the picocyanobacteria analysed contained two ribosomal operons (Fig. 5). PCR amplifications, restriction analysis of the ITS-1 and results of direct sequence analysis indicated that in most cases both ribosomal operons comprised identical sequences. This is consistent with results of genome analysis of so far three other unicellular cyanobacteria, Synechocystis sp. PCC 6803 (Kaneko et al., 1996), Synechococcus WH 8102 and Prochlorococcus marinus MIT 9313. However, it should be noted that the closely related P. marinus MED4 possesses only one ribosomal operon (see http://www.jgi.doe.gov./JGI_microbial/html/index.html). Even with two ribosomal operons present, genomic typing with targets in this polycistronic operon provided much less resolution (Fig. 5) than that achieved with psbA genes (cf. Postius & Ernst, 1999).

We also examined restriction typing of amplified fragments of the less conserved non-coding sequences of ITS-1. The length of ITS-1 of the Synechococcus isolates described in this study varied between 704 nt in strain LB P1 (Lake Biwa) and 1034 nt in strain BGS 171 (Lake Balaton). This range was larger than the length polymorphism observed among marine Synechococcus spp. (747–810 nt) and Prochlorococcus spp. (537–829 nt; Rocap et al., 2002). In total the length polymorphism of ITS-1 in this strictly monophyletic group of organism ranges from 537 to 1034 nt and hence, is larger than that of all other major cyanobacterial lineages, for which ITS-1 sequences between 283 and 648 nt have been reported (Itemann et al., 2000; Boyer et al., 2001). Restriction of this long intergenic region produced a characteristic pattern of restriction fragments for each of the lineages studied (Fig. 6b). Nevertheless, this method was in most cases not suitable to distinguish genotypes and ecotypes within these lineages.

Strain cluster descriptions and taxonomic considerations
Based on a sequence identity of >99 % in 16S rRNA, we assigned six PC-rich isolates, five from Lake Constance, one from Lake Balaton, and one PE-rich isolate from the brackish Baltic Sea, to a cluster with the type strain C. gracile PCC 6307T. Strains of this cluster are non-motile, obligately photoautotrophic, freshwater strains with a high DNA G+C content and peripheral thylakoids. Similar strains have been isolated from temperate zone lakes in Europe, America (Rippka et al., 1979) and Japan (Robertson et al., 2001). Phylogenetic inference from 16S rRNA and ITS-1 also assigned two strains from brackish ecosystems, Synechococcus sp. PCC 7009 (Rippka et al., 1979) and BS 20 (this study), and although the majority of isolated strains use the blue pigment PC as the major accessory pigment, there are also two red, PE-rich strains in this lineage (strains BS 20, this study; and PS-715, Robertson et al., 2001).

All PE-rich strains isolated from the pelagic zones of the subalpine Lake Constance, Lake Zurich and Lake Maggiore form a tight cluster in 16S rRNA as well as in ITS-1 inferred phylogenetic analyses: subalpine cluster I. S. rubescens SAG B3.81, which was the first PE-rich Synechococcus species isolated (Chang, 1980), is suggested as type strain of this cluster. All strains, except strain BO 8807, exhibit a coccoid morphology and lack particular surface structures. Strain BO 8807, which also confers a distinguished position in 16S rRNA and ITS-1 sequence analysis, exhibits elongated cells that are completely covered by an S-layer formed by a glycosylated protein (Ernst et al., 1992, 1996). This demonstrates that phylogenetically closely related strains can exhibit major morphological and biochemical variations apart from variations in accessory pigments.

Two clusters were formed by a single PE-rich and a single PC-rich isolate co-existing in the same ecosystem. The PE-rich strain Synechococcus sp BO 0014 was isolated from artificial substratum (tile) deposited for 6 weeks in the littoral zone of Lake Constance, while the phylogenetically closely related PC-rich strain BO 8805 was isolated from the pelagic zone of this lake. Accordingly, a PE-rich isolate LB B3 clustered with the PC-rich strain LB G2 from Lake Biwa, Japan. Both clusters are paraphyletic to the above-mentioned clusters, but their relative positions remained ambiguous in all analyses (Figs 3 and 4).

The Bornholm Sea cluster contained three PC-rich and three PE-rich isolates that differ in cell morphology and growth characteristics (B. E. M. Schaub, S. Vegos, U. I. A. Wollenzien, L. A. Stal & A. Ernst, unpublished results), but exhibit very few differences in ITS-1 sequences. The morphological differences observed in this cluster indicate that phenotypic diversification can occur with significantly higher frequency than fixation of mutations in the non-coding sections of the ribosomal operon. In the 16S rRNA-inferred phylogenetic analysis the Bornholm Sea cluster showed a weak affinity to two halotolerant PC-rich strains from coastal origins, Synechococcus spp. PCC 7001 and PCC 9005, originally assigned to the marine Synechococcus cluster B (Waterbury & Rippka, 1989). This taxonomic assignment was revised recently to a newly established form-genus Cyanobium Rippka & Cohen-Bazier 1983, characterized by a higher G+C content (66–71 mol%), but a more limited range of cell diameters (0·8–1·4 µm) than the cyanobacteria that remained provisionally assigned to the form-genus Synechococcus (Rippka et al., 2001).

The newly established form-genus Cyanobium comprises two physiologically distinct groups of strains: those not capable of sustained growth in a marine medium (cluster 1, type strains C. gracile PCC 6307T) and strains with a higher salt tolerance (cluster 2; Rippka et al., 2001). As far as morphological and ultrastructural data are available (Chang, 1980; Ernst et al., 1992, 2000) all picocyanobacteria described in this study conform to the criteria of Cyanobium. However, whether they exhibit a high DNA G+C content remains difficult to prove, because, with the notable exception of subspecies of C. gracile, many freshwater strains do not form isolated colonies on agarose amended with BG11 and, therefore, are difficult to obtain as axenic cultures. Phylogenetic analysis of the 16S rRNA concurred with the taxonomic definition, but received a low bootstrap support in our analysis (Fig. 3). Thus, the reassignment of all Synechococcus-type cyanobacteria from the autotrophic picoplankton of freshwater and brackish ecosystems to one genus, Cyanobium, remains debatable. As a putative genus, Cyanobium is paraphyletic to three other putative genera formed by the Antarctic Synechococcus strains (Vincent et al., 2000), the marine Synechococcus spp. and the marine Prochlorococcus spp. Together these four genera seem to dominate the autotrophic picoplankton of marine, brackish and freshwater ecosystems. Within these genera, the small divergence in 16S rRNA often limited further resolution (Fig. 3). Supplementary phylogenetic information obtained from ITS-1 provided convincing evidence for the existence of different lineages in this clade representing species and subspecies in a taxonomic context (Fig. 4)

Ecosystem-dependent adaptive radiations
The data presented in this study show that many closely related genotypes, classified as subspecies of Cyanobium spp. can coexist in a particular ecosystem. We emphasized that in four of the five paraphyletic strain clusters (i.e. species) the subspecies comprise two pigment variants, the PE-rich and the PC-rich strains (see Fig. 4). From this observation we conclude that adaptation in lineages unable to acclimate to different light quality by synthesis of complementary pigments (CCA) can occur by loss or gain of genes required to synthesize PE, the outermost pigment of the phycobilisome. This may be caused by point mutations, by extended deletions or by recombination with intact genetic information provided from a closely related organism by horizontal gene transfer. Evidence for the exchange of genetic material between closely related strains was demonstrated in the ribosomal operon (see Fig. 2) but evidence also comes from other cyanobacteria (Barker et al., 2001).

Pigmentation as well as cell-surface structures are characteristics that determine the ecophysiology of strains and changes are expected to create new niches for the organism. Mutation and selection can therefore lead to an adaptive radiation of strains within a particular ecosystem. Recent adaptive radiations are characterized by marginal differences in 16S rRNA sequences, limited divergence in ITS-1 and very high bootstrap support in distance-matrix-based phylogenies (Figs 3 and 4) as well as in maximum-likelihood trees (data not shown). Putative endemic radiations were demonstrated in the subalpine clusters I and II and in the Bornholm Sea cluster (Figs 3 and 4). A different genetic structure in ITS-1 was observed among strains assigned to the C. gracile cluster. Strains of this cluster were isolated all over the world and the sequence divergence in ITS-1 between isolates from Lake Constance was larger than between strains isolated from other ecosystems (Fig. 3). The lack of a geographic coherence in the genetic diversification of this lineage is not consistent with a hypothesis of endemic radiations. Interestingly, cluster-specific DGGE revealed that isolates of the subalpine cluster I are abundant in the autotrophic picoplankton of Lake Constance (Becker et al., 2002), while those assigned to the universal species C. gracile remained below the detection limit (S. Becker, A. K. Singh, C. Postius & A. Ernst, unpublished results). Thus, an adaptive radiation may have allowed the subalpine cluster I to gain dominance in the pelagic autotrophic picoplankton, but not C. gracile.

Recent adaptive radiations may also be seen in the genetic structure of marine Synechococcus spp. and Prochlorococcus spp. (Rocap et al., 2002). However, while some of these marine clusters can be associated with a particular phenotype, such as motility, chromatic adaptation or lack of phycourobilin, or differ in chlorophyll composition, phenotypic diversification among the most recently diversified genotypes, which is prerequisite for the hypothesis of an adaptive radiation, has not been reported. Nevertheless, the similarities in the genetic structure of different genera of the autotrophic picoplankton of marine, brackish and freshwater ecosystems are intriguing and inevitably lead to the question what has caused the sudden boosts in diversification in particular lineages, while others seemingly remained unaffected?


   ACKNOWLEDGEMENTS
 
Preliminary sequence data were obtained from The DOE Joint Genome Institute (JGI) at http://www.jgi.doe.gov./JGI_microbial/html/index.html. The support by P. Böger and Deutsche Forschungsgemeinschaft through its Sonderforschungsbereich 454 'Bodenseelitoral' is appreciated. This is publication 2880 from NIOO, Centre for Estuarine and Coastal Ecology, Yerseke, The Netherlands.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. (1990). Basic local alignment search tool. J Mol Biol 215, 403–410.[CrossRef][Medline]

Barker, G. L. A., Handley, B. A., Vacharapiyasophon, P., Stevens, J. R. & Hayes, P. K. (2001). Allele-specific PCR shows that genetic exchange occurs among genetically diverse Nodularia (cyanobacteria) filaments in the Baltic Sea. Microbiology 146, 2865–2875.

Becker, S., Fahrbach, M., Böger, P. & Ernst, A. (2002). Quantitative tracing, by Taq nuclease assays, of a Synechococcus ecotype in a highly diversified natural population. Appl Environ Microbiol 68, 4486–4494.[Abstract/Free Full Text]

Boone, R., Castenholz, R. W. & Garrity, G. M. (editors) (2001). Bergey's Manual of Systematic Bacteriology, 2nd edn, Vol. 1. New York: Springer.

Boyer, S. L., Flechtner, V. R. & Johansen, J. R. (2001). Is the 16S–23S rRNA internal transcribed spacer region a good tool for use in molecular systematics and population genetics? A case study in cyanobacteria. Mol Biol Evol 18, 1057–1069.[Abstract/Free Full Text]

Brass, S., Ernst, A. & Böger, P. (1996). An insertion element prevents phycobilisome synthesis in N2-fixing Synechocystis sp. strain BO 8402. Appl Environ Microbiol 62, 1964–1968.[Abstract]

Callieri, C., Amicucci, E., Bertoni, R. & Vörös, L. (1996). Fluorometric characterization of two picocyanobacteria strains from lakes of different underwater light quality. Int Rev Gesamten Hydrobiol 81, 13–23.

Chang, T.-P. (1980). Zwei neue Synechococcus-Arten aus dem Zürichsee. Schweiz Z Hydrol 42, 247–254.

Ernst, A., Sandmann, G., Postius, C., Brass, S., Kenter, U. & Böger, P. (1992). Cyanobacterial picoplankton from Lake Constance: II. classification of isolates by cell morphology and pigment composition. Bot Acta 105, 161–167.

Ernst, A., Marschall, P. &, Postius. C. (1995). Genetic diversity among Synechococcus spp. (cyanobacteria) isolated from the pelagial of Lake Constance. FEMS Microbiol Ecol 17, 197–204.[CrossRef]

Ernst, A., Postius, C. & Böger, P. (1996). Glycosylated surface proteins reflect genetic diversity among Synechococcus spp. of Lake Constance. Arch Hydrobiol 48, 1–6.

Ernst, A., Becker K. Hennes, S. & Postius, C. (2000). Is there a succession in the autotrophic picoplankton of temperate zone lakes? In Microbial Biosystems: New Frontiers. Proceedings of the 8th International Symposium on Microbial Ecology, pp. 623–629. Edited by C. R. Bell, M. Brylinski, P. Johnson-Green. Halifax, Canada: Atlantic Canada Society for Microbial Ecology.

Felsenstein, J. (1985). Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783–791.

Fox, G. E., Wisotzkey, J. D. & Jurtshuk, P., Jr (1992). How close is close: 16S rRNA sequence identity may not be sufficient to guarantee species identity. Int J Syst Bacteriol 42, 166–170.[Abstract]

Garcia-Pichel, F., Nübel, U. & Muyzer, G. (1998). The phylogeny of unicellular, extremely halotolerant cyanobacteria. Arch Microbiol 169, 469–482.[CrossRef][Medline]

Giovannoni, S. J., Turner, S., Olsen, G. J., Barns, S., Lane, D. J. & Pace, N. R. (1988). Evolutionary relationships among cyanobacteria and green chloroplasts. J Bacteriol 170, 3584–3592.[Medline]

Grossman, A. R., Schaefer, M. R., Chiang, G. G. & Collier, J. L. (1993). The phycobilisome, a light-harvesting complex responsive to environmental conditions. Microbiol Rev 57, 725–749.[Abstract]

Gürtler, V. & Stanisich, V. A. (1996). New approaches to typing and identification of bacteria using the 16S–23S rDNA spacer region. Microbiology 142, 3–16.[Medline]

Herdman, M., Janvier, J. B., Waterbury, J. B., Rippka, R., Stanier, R. Y. & Mandel, M. (1979). Deoxyribonucleic acid base composition of cyanobacteria. J Gen Microbiol 111, 63–71.

Higgins, D. G. & Sharp, P. M. (1988). CLUSTAL: a package for performing multiple sequence alignment on a microcomputer. Gene 73, 237–244.[CrossRef][Medline]

Honda, D., Yokota, A. & Sugiyama, J. (1999). Detection of seven major evolutionary lineages in cyanobacteria based on 16S rRNA gene sequence analysis with new sequences of five marine Synechococcus strains. J Mol Evol 48, 723–739.[Medline]

Iteman, I., Rippka, R., Tandeau de Marsac, N. & Herdman, M. (2000). Comparison of conserved structural and regulatory domains within divergent 16S rRNA–23S rRNA spacer sequences of cyanobacteria. Microbiology 146, 1275–1286.[Abstract/Free Full Text]

Jain, R., Rivera, M. C. & Lake, J. A. (1999). Horizontal gene transfer among genomes: the complexity hypothesis. Proc Natl Acad Sci U S A 96, 3801–3806.[Abstract/Free Full Text]

Jukes, T. H. & Cantor, C. R. (1969). Evolution of protein molecules. In Mammalian Protein Metabolism, Vol. III, pp. 21–132. Edited by H. N. Muntu. New York: Academic Press.

Kaneko, T., Sato, S., Kotani, H. & 21 other authors (1996). Sequence analysis of the genome of the unicellular cyanobacterium Synechocystis sp. strain PCC 6803. II. Sequence determination of the entire genome and assignment of potential protein-coding regions. DNA Res 3, 109–136.[Medline]

Lu, W., Evans, E. H., McColl, S. K. & Saunders, V. A. (1997). Identification of cyanobacteria by polymorphisms of PCR-amplified ribosomal DNA spacer region. FEMS Microbiol Lett 153, 141–149.[CrossRef]

Lyra, C., Suomalainen, S., Gugger, M., Vezie, C., Sundman, P., Paulin, L. & Sivonen, K. (2001). Molecular characterization of planktic cyanobacteria of Anabaena, Aphanizomenon, Microcystis and Planktothrix genera. Int J Syst Evol Microbiol 51, 513–526.[Abstract]

Maeda, H., Kawai, A. & Tilzer, M. M. (1992). The water bloom of cyanobacterial picoplankton in Lake Biwa, Japan. Hydrobiologia 24, 93–103.

Miller, S. R. & Castenholz, R. W. (2000). Evolution of thermotolerance in hot spring cyanobacteria of the genus Synechococcus. Appl Environ Microbiol 66, 4222–4229.[Abstract/Free Full Text]

Mullis, K. B. & Faloona, F. A. (1987). Specific synthesis of DNA in vitro via a polymerase catalyzed chain reaction. Methods Enzymol 155, 335–350.[Medline]

Nelissen, B., De Baere, R., Wilmotte, A. & De Wachter, R. (1996). Phylogenetic relationships of nonaxenic filamentous cyanobacterial strains based on 16S rRNA sequence analysis. J Mol Evol 42, 194–200.[Medline]

Neuschaefer-Rube, O., Westermann, M., Bluggel, M., Meyer, H. E. & Ernst, A. (2000). The blue-colored linker polypeptide L55 is a fusion protein of phycobiliproteins in the cyanobacterium Synechocystis sp. strain BO 8402. Eur J Biochem 267, 3623–3632.[Abstract/Free Full Text]

Otsuka, S., Suda, S., Li, R., Watanabe, M., Oyaizu, H., Matsumoto, S. & Watanabe, M. M. (1999). Phylogenetic relationships between toxic and non-toxic strains of the genus Microcystis based on 16S to 23S internal transcribed spacer sequence. FEMS Microbiol Lett 172, 15–21.[CrossRef][Medline]

Palenik, B. & Swift, H. (1996). Cyanobacterial evolution and prochlorophyte diversity as seen in DNA-dependent RNA polymerase gene sequences. J Phycol 32, 638–646.

Pick, F. R. (1991). The abundance and composition of freshwater picocyanobacteria in relation to light penetration. Limnol Oceanogr 36, 1457–1462.

Pohl, T. M. & Maier, E. (1995). Sequencing 500 kb of yeast DNA using a GATC1500 direct blotting electrophoresis system. BioTechniques 19, 482–486.[Medline]

Postius, C. & Ernst, A. (1999). Mechanisms of dominance: coexistence of picocyanobacterial genotypes in a freshwater ecosystem. Arch Microbiol 172, 69–75.[CrossRef][Medline]

Postius, C., Ernst, A., Kenter, U. & Böger, P. (1996). Persistence and genetic diversity among strains of phycoerythrin-rich cyanobacteria from the picoplankton of Lake Constance. J Plankton Res 18, 1159–1166.[Abstract]

Rippka, R., Deruelles, J., Waterbury, J. B., Herdman, M. & Stanier, R. Y. (1979). Generic assignments, strain histories and properties of pure cultures of cyanobacteria. J Gen Microbiol 111, 1–61.

Rippka, R., Castenholz, R. W. & Herdman, M. (2001). Form-genus IV. Cyanobium Rippka and Cohen-Bazire 1983. In Bergey's Manual of Systematic Bacteriology, 2nd edn, Vol. 1, pp. 498–499. Edited by D. R. Boone, R. W. Castenholz & G. M. Garrity. New York: Springer.

Robertson, B. R., Tezuka, N. & Watanabe, M. M. (2001). Phylogenetic analyses of Synechococcus strains (cyanobacteria) using sequences of 16S rDNA and part of the phycocyanin operon reveal multiple evolutionary lines and reflect phycobilin content. Int J Syst Evol Microbiol 51, 861–871.[Abstract]

Rocap, G., Distel, L., Waterbury, J. B. & Chrisholm, S. W. (2002). Resolution of Prochlorococcus and Synechococcus ecotypes by using 16S–23S ribosomal DNA internal transcribed spacer sequences. Appl Environ Microbiol 68, 1180–1191.[Abstract/Free Full Text]

Saitou, N. & Nei, M. (1987). The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4, 406–425.[Abstract]

Sanger, F., Nicklen, S. & Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sci USA 74, 5463–5467.[Abstract]

Scheldeman, P. B., Baurain, D., Bouhy, R., Scott, M., Muhling, M., Whitton, B. A., Belay, A. & Wilmotte, A. (1999). Arthrospira (‘Spirulina’) strains from four continents are resolved into only two clusters, based on amplified ribosomal DNA restriction analysis of the internally transcribed spacer. FEMS Microbiol Lett 172, 213–222.[CrossRef][Medline]

Stockner, J. G., Callieri, C. & Cronberg, C. (2000). Picoplankton and other non-bloom-forming cyanobacteria in lakes. In The Ecology of Cyanobacteria, pp. 195–231. Edited by B. A. Whillon & M. Potts. Dordrecht: Kluwer.

Toledo, G., Palenik, B. & Brahamsha, B. (1999). Swimming marine Synechococcus strains with widely different photosynthetic pigment ratios form a monophyletic group. Appl Environ Microbiol 65, 6247–5251.

Turner, S. (1997). Molecular systematics of oxygenic photosynthetic bacteria. Plant Syst Evol 11 suppl., 13–52.

Urbach, E., Scanlan, D. J., Distel, D. L., Waterbury, J. B. & Chisholm, S. W. (1998). Rapid diversification of marine picoplankton with dissimilar light harvesting structures inferred from sequences of Prochlorococcus and Synechococcus (cyanobacteria). J Mol Evol 46, 188–201.[Medline]

Van de Peer, Y. & De, Wachter, R. (1994). TREECON: a software package for the construction and drawing of evolutionary trees. Comput Appl Biosci 9, 177–182.

Vincent, W. F., Bowman, J. B., Powell, L. M. & McMeekin, T. A. (2000). Phylogenetic diversity of picocyanobacteria in Arctic and Antarctic ecosystems. In Microbial Biosystems: New Frontiers. Proceedings of the 8th International Symposium on Microbial Ecology, pp. 317–322. Edited by C. R. Bell, M. Brylinski, P. Johnson-Green. Halifax, Canada: Atlantic Canada Society for Microbial Ecology.

Vörös, L., Gulyas, P. & Nemeth, J. (1991). Occurrence, dynamics and production of picoplankton in Hungarian shallow lakes. Int Rev Gesamten Hydrobiol 76, 617–629.

Ward, D. M., Ferris, M. J., Nold, S. C. & Bateson, M. M. (1998). A natural view of microbial biodiversity within hot spring cyanobacterial mat communities. Microbiol Mol Biol Rev 62, 1353–1370.[Abstract/Free Full Text]

Waterbury, J. B. & Rippka, R. (1989). Subsection I. Order Chroococcales Wettstein 1924, emend. Rippka et al., 1979. In Bergey's Manual of Systematic Bacteriology, vol. 3, pp. 1728–1746. Edited by J. T. Staley, M. P. Bryant, N. Pfennig & J. G. Holt. Baltimore: Williams & Wilkins.

West, N. J. & Adams, D. G. (1997). Phenotypic and genotypic comparison of symbiotic and free-living cyanobacteria from a single field site. Appl Environ Micriobiol 63, 4479–4484.[Abstract]

Wilmotte, A., Turner, S., Van de Peer, Y. & Pace, N. R. (1992). Taxonomic study of marine Oscillatorian strains (cyanobacteria) with narrow trichomes. II. Nucleotide sequence analysis of the 16S ribosomal RNA. J Phycol 28, 828–838.

Wilmotte, A., Van der Auwera, G. & De Wachter, R. (1993). Structure of the 16S ribosomal RNA of the thermophilic cyanobacterium Chlorogloeopsis HTF (‘Mastigocladus laminosus HTF’) strain PCC 7518 and phylogenetic analysis. FEBS Lett 317, 96–100.[CrossRef][Medline]

Wilmotte, A., Neefs, J. M. & De Wachter, R. (1994). Evolutionary affiliation of the marine nitrogen-fixing cyanobacterium Trichodesmium sp. strain NIBB 1067, derived by 16S ribosomal RNA sequence analysis. Microbiology 140, 2159–2164.[Abstract]

Wood, A. M., Phinney, D. A. & Jentsch, C. S. (1998). Water column transparency and the distribution of spectrally distinct forms of phycoerythrin-containing organisms. Mar Ecol Prog Ser 162, 25–31.

Received 17 January 2002; revised 9 September 2002; accepted 16 October 2002.