3Present address: Department of Organismic and Evolutionary Biology, 16 Divinity Ave., Harvard University, Cambridge, MA 02138.

Running head: Molecular characterization of Microcoleus.

Thirty-one strains of Microcoleus were isolated from diverse arid regions in the western United States, including the Mojave Desert, Chihuahuan Desert, Great Basin Desert, and Colorado Plateau. Although all of these taxa fit the broad definition of Microcoleus vaginatus in common usage by current soil algal phycologists, sequence data for the 16S rRNA gene and associated 16S-23S ITS region indicated that more than one species was represented in the set of isolations. Combined sequence and morphological data revealed the presence of two morphologically similar taxa, Microcoleus vaginatus (Vaucher) Gomont and Microcoleus steenstrupii Boye-Petersen, in our collections. The rRNA operons of these taxa are sufficiently dissimilar that we suspect the two taxa belong in separate genera. 16S rRNA sequences from our isolates of M. vaginatus are highly similar to the sequences for M. vaginatus generated recently by other workers. The combined M. vaginatus clade is most similar to published sequences from Trichodesmium and Arthrospira. When 16S sequences from the isolates we identified as M. steenstrupii are compared to published sequences, our strains group with M. chthonoplastes, and may have closest relatives among several genera in the Phormidiaceae. Organization within the 16S-23S ITS regions is variable between the two taxa. M. vaginatus has either two tRNA genes (tRNA^Ile and tRNA^Ala) or a fragment of the tRNA^Ile gene in its ITS regions, whereas M. steenstrupii has rRNA operons with either the tRNA^Ile gene or no tRNA genes in its ITS regions. We were unable to detect any geographic signal in our 16S data set. M. vaginatus showed no subspecific variation within the combined morphological and molecular characterizations, with 16S similarities ranging from 97.1-99.9%. M. steenstrupii did show considerable genetic variability, with 16S similarities ranging from 91.5-99.4%. In phylogenetic analyses, we found that this variability was not congruent with geography, and we suspect that our M. steenstrupii strains represent several cryptic species.

Key index words: cyanobacteria; ITS; microbiotic crusts; Microcoleus; phylogeny; rRNA sequence data, 16S; rRNA, operon variation.

Abbreviations: ITS, internal transcribed spacer; PCR, polymerase chain reaction.

Molecular approaches to cyanobacterial systematics have been embraced by phycologists eager for large, unambiguous character data sets. Molecular studies have focused primarily on higher level phylogenies as reconstructed from analysis of the 16S (small subunit) rRNA sequences (e.g. Turner 1997, Wilmotte et al. 1992, Nübel et al. 1996). Some researchers have examined within-genus variability in the 16S (Otsuka et al. 1998), but this gene is generally thought to be too highly conserved for use in estimating relationships below the genus level (Fox et al. 1992). The nifH gene has also been used in phylogenetic analysis of higher cyanobacterial taxa (Zehr et al. 1997) and among species (Ben-Porath and Carpenter, 1993).

Population genetic studies are few in cyanobacteria, and often involve the 16S-23S internal transcribed spacer (ITS) region. In most instances, RFLP of the PCR-amplified ITS has been used to analyze population-level and species-level relationships (e.g. Lu et al. 1997, West and Adams 1997, Scheldeman et al. 1999). Thus far, Otsuka et al. (1999) are the only researchers to have used direct sequencing of the ITS in a population-level study. The ribosomal rRNA operon is normally present in several copies in cyanobacteria (Boyer et al. 2001, Iteman et al. 2000), and the 16S-23S ITS may vary considerably in sequence, length, secondary structure, and presence/absence of tRNA genes among the multiple operons in a single strain. This means that ITS sequences cannot be compared among strains unless orthology (rather than paralogy) is established among those sequences. Therefore, the ITS region is problematic for use in population genetic and molecular systematic studies, and should be used with great caution.

There is currently disagreement as to how many species of cyanobacteria exist. Some researchers take a conservative approach and recommend few species be recognized (Drouet 1968), whereas others are involved in actively describing new taxa based on slight differences in morphology and ecophysiology (Komárek and Anagnostidis 1999). Regardless of taxonomic practice, most researchers acknowledge that morphologically similar strains can differ in their biogeography and physiological characteristics.

Our research group is testing the feasibility of using cyanobacterial amendments for reclamation of disturbed arid and semi-arid lands. As part of this study, we wished to ascertain the extent of the genetic variability within Microcoleus vaginatus (Vaucher) Gomont, the most cosmopolitan cyanobacterial taxon in these environments. During the course of our work, our molecular data suggested that some isolates originally classified as M. vaginatus might actually represent a different species. This paper reports the results of a study of the 16S rRNA in numerous strains of edaphic Microcoleus isolated from diverse regions in the western United States and demonstrates how a polyphasic approach that combines morphological and molecular data can be used to solve questions of cyanobacterial phylogenetics.

Microcoleus strains used in this study were all isolated from arid soils in the western United States and identified by workers in our research group. Microbiotic soil crusts were collected from four arid biomes: Great Basin Desert, Colorado Plateau, Chihuahuan Desert, and Mojave Desert (Table 1). Sites within desert regions were generally chosen such that each site was at least 1 km from other sites. Dry soil samples were crushed, subsampled, and dilution plated as described in Flechtner (1999). In most instances, only a single isolate was used to represent a site.

In our initial studies, single filaments of Microcoleus were isolated from rough cultures into unialgal culture and grown on sterile sand plates. Strains were kept in dim light (<50 µE·cm^-2·s^-1 illuminance) at 7°C on a 12:12 hr light:dark cycle. These strains all fit the rather broad description of Microcoleus vaginatus found in Geitler (1932) and in common usage in many ecological and floristic studies of crusts (Anderson & Rushforth 1976, Ashley et al. 1985, Johansen et al. 1981, others). Molecular characterization of the 16S rRNA gene in these strains demonstrated that two distinct clades were represented. We resolved our taxonomic confusion in subsequent studies through morphological characterization and molecular analysis of single filaments isolated from wetted soil samples. The presence of M. steenstrupii only became clear to us after these latter studies, which demonstrated complete congruence between morphological characters and 16S rRNA sequence data. All isolates were examined using Olympus photomicroscopes with Nomarski DIC optics. In characterization of the single filaments, all were scored as to cell and trichome dimensions, motility, presence or absence of calyptra, type of cell division (Anagnostidis and Komárek 1988), and apical cell characteristics.

Strains were assigned code designations with the site code first (TAA1), followed by the isolate number from that site (TAA1-4). Single strand sequences were coded by site code (FI6), followed by an isolate number preceded by "MC" (FI6MC1). Two exceptions to the MC designation are EM3J1 and EM1B1C4, which were named before standardization of the single filament codes. For all sequences from these strains, a code identifying the PCR reaction (number) and TA clone (letter) is given last (e.g. TAA1-4-1A, TAA1-4-1B, and FI6MC1-2A, FI6MC1-2B).

DNA was extracted from 20 mg of fresh unialgal tissue using the Cullings (1992) modification of the Doyle and Doyle extraction (Doyle and Doyle 1987). The resultant DNA was suspended in 50 µL TE and stored at –20ºC.

Primers were modified from Wilmotte et al. (1993) and Nübel et al. (1997). They were designated:

Primer 1 5’ CTC TGT GTG CCT AGG TAT CC 3’ (after Wilmotte et al. 1993)

Primer 2 5’ GGG GAA TTT TCC GCA ATG GG 3’ (after Nübel et al., 1997)

Primer 5 5’ TGT ACA CAC CGG CCC GTC 3’ (after Wilmotte et al. 1993)

Primer 6 5’ GAC GGG CCG GTG TGT ACA 3’ (after Wilmotte et al. 1993)

These primers are complimentary to conserved stretches of the cyanobacterial 16S RNA gene and the 23S RNA gene (Fig.1). Primer 6 is the reverse complement of Primer 5. Primers were ordered from the Midland Certified Reagent Company in concentrations that we brought to 100 µM. For use in PCR, a mix of 1.2 µL each of two primers and 7.6 µL sterile water was made.

Each DNA sample was amplified using primers 1 and 2. This resulted in a chain approximately 1600 bp long that was then used as a template for a reamplification using primer pairs 1 & 5 and 2 & 6.

Each 100 µL reaction contained 86 µL sterile water, 10µL 10x buffer (Promega), 0.5 µL of each dNTP (G, A, T, C) at 10 mM, 0.5 µL of the primer mixture described above, 0.5 uL Taq polymerase (Promega), and, typically, 1.0 µL template DNA (either genomic DNA or PCR product).

The most commonly used protocol for the initial 16S + ITS PCR reaction using primers 1 and 2 was 94ºC for 1 min, 57ºC for 1 min, 72ºC for 4 min (35 cycles), followed by a 10 minute extension at 72ºC. For PCR reamplifications, the most commonly used protocol was 94ºC for 1 min, 56ºC for 45 s, 72ºC for 2 min (20 cycles). Less commonly used protocols had annealing temperatures that varied by one or two degrees from the common protocols. Reactions were carried out using Thermolyne’s Amplitron and Temptronic thermocyclers. The presence of PCR products was detected by standard agarose gel electrophoresis and ethydium bromide staining.

Microcoleus filaments in moistened soils were examined at 1250X magnification in order to characterize the strain and guarantee the absence of contaminants.

If free of contaminants, a filament was photographed and then transferred to sterile 0.5 mL eppendorf tubes. Eighty-seven micro-liters of sterile water and 10 µL of Taq polymerase buffer (Promega) were added, and the mixture was frozen at -80°C. Later, it was thawed, refrozen, and thawed again to help shear the cells open, and then the rest of the PCR reagents were added. Procedures from this point were as above. Cultures for strains processed in this way do not exist.

PCR products were cloned into plasmids containing the sites for the universal primers M13 forward and reverse on either side of the cloning site using Invitrogen’s TOPO TA Cloning Kit for Sequencing, Version A. Plasmid DNA was obtained from, generally, 3 of the resultant clones using Qiagen’s QiaPrep Spin Kit. In the case of clones containing PCR product generated with primers 1 and 5 (ITS primers), 9 minipreps were digested with EcoRI enzyme and run on a long gel (1h+) to visualize the size of inserts. The three clones chosen for sequencing included as many different insert lengths as were distinguishable.

Automatic sequencing with the universal primers M13 forward and reverse was performed by Cleveland Genomics.
Data Analysis

Forward and reverse primer sequences were checked against each other by generating the reverse complement of the "reverse" sequence with Oxford Molecular Group’s Omiga™ and aligning it with the "forward" sequence with CLUSTAL W Multiple Sequence Alignment Program, version 1.7 (Thompson et al. 1994) via the Baylor College of Medicine’s Search Launcher (Smith et al. 1996) at http://dot.imgen.bcm.tmc.edu:9331/. This resulted in the longest possible read of the sequence, in addition to acting as a check on the sequencing. Where forward and reverse sequences did not agree with each other or with published sequences, the sequence data were carefully checked against the sequence chromatograms. Sequences from amplifications using the three different primer pairs were put together using alignments of overlaps between the sequences. All sequences were BLAST searched (Altschul et al. 1997) to ensure that they were cyanobacterial in origin. All sequences were submitted to GenBank.

Sequences from different clones and isolates were aligned using CLUSTAL W. These alignments were checked by eye, and again, when sequences did not agree with each other or published sequences, the sequence data were carefully checked against the sequence chromatograms. Analysis of secondary structure also informed sequence proofreading. The variability among sequences from different clones was preserved in the analysis through the inclusion of multiple non-identical sequences.

Similarity among 16S rRNA sequences in the two taxa, M. vaginatus and M. steenstrupii, was determined through construction of a pair-wise matrix of distance values in MEGA (Kumar et al. 1993). Minimum, maximum, and mean percent similarity for the partial sequences we obtained were calculated for each species.

Phylogenetic trees based on 16S sequence data were constructed by a variety of methods using PAUP 4.0b4a (Swofford 1998). Methods used include: neighbor-joining using LogDet and Jukes-Cantor distance metrics assuming equal substitution rates; HKY85 distance metric assuming rates followed a continuous gamma distribution (a = 0.5); maximum parsimony (equal weights); and maximum likelihood using the HKY85 model and assuming equal substitution rates. Initially, a heuristic search was run with 100 replicates and the TBR branch-swapping option. Boostraps were performed with 1000 replicates on all trees, but the NNI branch-swapping option was used to permit analysis within a reasonable timeframe. Following construction of the large trees, a second set of analyses was performed just on the Microcoleus vaginatus sequences together with sequences from the species’ apparent sister group (Trichodesmium + Oscillatoria sancta).

The following sequences from GenBank (accession numbers in parentheses) were used as outgroups in various analyses: Arthrospira fusiformis Ethi-B2 (AF260510), Arthrospira maxima Ethi-A2 (AF260509), Arthrospira sp. PCC8005 (X70769.1), Geitlerinema PCC7105 (AF132780), Halospirulina tapeticola CCC Baja-95 C1.2 (Y18791.1), Halospirulina sp. CCC Baja-95 C1.3 (Y18790.1), Leptolyngbya foveolarum Komarek 1964/112 (X84808.1), Leptolyngbya schmidlei (AF355398), Microcoleus chthonoplastes PCC7420 (X70770.1), Oscillatoria sancta PCC7515 (AF132933), Oscillatoria sp. 2 (AJ133185), Oscillatoria sp. (AJ133106), Phormidium ambiguum M-71 (AB003167), Phormidium mucicola M-221 (AB003165), Planktothrix sp. FP1 (AF212922), Plectonema boryanum UTEX485 (AF132793), Pseudanabaena PCC6903 (AF132778), Pseudanabaena PCC7402 (AF132787), Spirulina sp. MPI-S4 (Y18792.1), Synechococcus sp. PS721 (AF216954), Trichodesmium contortum (AF013028), Trichodesmium erythraeum (AF013030), and Trichodesmium hildebrandtii (AF091322).

Secondary structure of all stems and loops in the 16S rRNA gene between bp 379-1333 was determined using Mfold version 3.0 at http://mfold2.wustl.edu/~mfold/rna/form1.cgi (based on Zuker et al. 1999, Mathews et al. 1999). These secondary structures were used to assess the significance of observed differences in 16S rRNA sequence data. Designation of the variable regions and stem structures are based on terminology used in Wilmotte et al. (1993). Secondary structure for major domains of the 16S-23S ITS region was also determined to facilitate alignment and interpret variability. Terminology for the domains in the ITS region are based on Iteman et al. (2000).

All sequence data were deposited with GenBank. Accession numbers for M. vaginatus are AF355388-AF355374 (16S) and AF363924-AF363940 (ITS). Accession numbers for M. steenstrupii are AF355379-AF355397 (16S) and AF363941-AF363950 (ITS).

The phycological community is certainly not in agreement as to what constitutes a species. Among cyanobacteria, which are studied by microbiologists, botanists, and ecologists, species concepts are even more contested. We cover the application of various modern species concepts to cyanobacteria in detail in another paper (Flechtner et al. 2002). For the purposes of clarity, we here define how we use species concepts in this paper.

A cyanobacterial morphospecies is a species recognized based on morphological characters alone, and may be a fairly artificial construct. A morphospecies can include strains or uncultivated populations that are morphologically inseparable, but have very different ecophysiology and do not share a recent common ancestry (as revealed by DNA-DNA hybridization studies or DNA sequence analysis). All species of cyanobacteria described before the advent of the electron microscope and molecular methods are morphospecies by definition, although many early workers considered habitat to also have taxonomic significance. Because of plasticity in morphology, it is possible that some morphospecies may have multiple morphotypes, but a single genotype. This was the central premise that led to the drastic combination of all 1300+ cyanobacterial species into 64 species by Drouet.

A genospecies is a species recognized based on its genetic information alone. Wayne et al. (1987) recommended that two strains of bacteria which have less than 70% DNA-DNA hybridization are genetically distinct enough to be considered, by definition, different species (i.e. genospecies). When no phenotypic characters exist that can be used to separate the different genospecies, Wayne et al. (1987) recommend that the genospecies not be taxonomically recognized. Stackebrandt & Goebel (1994) found a relationship between DNA-DNA hybridization and 16S rRNA gene sequence data that suggested that two strains with less than 97.5% sequence similarity always had less than 70% DNA-DNA hybridization, and so could likewise be considered genospecies. We use the term genospecies to mean strains or populations that are demonstrated to be genetically different by one of these two standards, but which have not yet been formally recognized as named species.

We feel the best species concept for cyanobacteria is the monophyletic phylogenetic species concept of Mischler and Theriot (2000). Under this definition a species is defined by an autapomorphic character state that separates it from other similar species. Thus, species are defined by their differences in derived character states (a cladistic approach), not by their similarities (a phenetic approach). Once the species is described, other collections may be assigned to that species as long as they do not possess autapomorphic characters that require they be recognized as separate species. A polyphasic approach to identifying cyanobacteria uses a combination of different character sets (morphology, ultrastructure, biochemistry, physiology, DNA sequence data) to better characterize strains and uncultured populations. We feel that when cyanobacteria are characterized more thoroughly, any of these character sets can provide the autapomorphic characters necessary for species recognition. Species so defined have both ecological and evolutionary relevance.

Species which are difficult to separate based on their morphology alone are sometimes referred to as cryptic species. Morphologically similar genospecies are thus cryptic species. With further and more careful polyphasic phenotypic characterization, cryptic species may eventually be taxonomically recognized, although subsequent observations of these taxa will likely continue to be difficult to assign to the proper species due to the continued need for polyphasic characterization.

Filaments with many trichomes per sheath. Sheath colorless, unlamellated. Trichomes highly motile, bright blue-green, tapering towards the ends, unconstricted at the crosswalls, with oscillatoriacean cell division (new wall formation begins before previous wall formation is complete), (3.8)-4.5-5.5 µm wide. End cells bluntly rounded following trichome breakage, conical when mature, shorter than wide, with a capitate calyptra. Cells usually but not always granular at the crosswalls, often grainy to rarely granular throughout the cell, with no thylakoid structure evident, 2-5-(6.7) µm long.

The morphological variability in this species was fairly small, and primarily limited to the presence or absence of granules at the crosswalls and throughout the cell (Fig. 2b, d). Trichome diameter was fairly limited within strains (not more than 0.5 µm range), and was most commonly 4.5-5 µm. Cells were most commonly 2-5 µm long. One strain (FB1MC1) was distinct in the presence of olive brown trichomes and long cells (5-8 µm). Calyptra were not always evident in the single filaments, but our sense of this species is that cultures and field material usually have at least some capitate cells with calyptra in some of the trichomes present. The shortness of the end cell was an especially helpful feature in separating this taxon from the forms we here consider M. steenstrupii, as length of end cell was fairly consistent in both taxa (Fig. 2a-h). The type of cell division was also a critical feature (Fig. 2d), but not always discernable.

Microcoleus steenstrupii Boye-Petersen (1923). Figs. 2f-h.

Filaments with few to many loosely arranged trichomes per sheath. Sheath colorless, unlamellated to faintly lamellated. Trichomes usually immotile, infrequently weakly motile, bright blue-green, tapering towards the ends, unconstricted to indistinctly constricted at the crosswalls, with phormidiacean cell division (cells grow to full size before new cell walls begin to form), (3.8)-4-5 µm wide. End cells bluntly rounded following trichome breakage, when mature conical, elongated, never capitate, without a calyptra, up 7.5-12 µm long. Cells grainy to granular throughout, but never granular at the crosswalls, with radial thylakoid structure evident in some cells, 3.5-9 µm long.

This species was most reliably identified by the long, conical, non-capitate end cells (Fig. 2f-h). The presence of slightly to indistinctly constricted crosswalls was sufficient to identify it (Fig. 2f), but this feature was often not evident. M. steenstrupii was described as having slightly constricted crosswalls, and it may be that many of the strains in our soils do not really belong in this taxon. Phormidiacean cell division, when discernable, was also a key diagnostic feature, but was often unclear due to faint septation (Fig. 2h). The cells were generally longer than in M. vaginatus. The absence or reduction of motility is likely an ecologically important characteristic.

This species is almost identical in morphology to M. chthonoplastes Thuret ex Gomont, but that taxon was described from marine benthic habitats and has an ecophysiology certainly very different from that of our desert soil isolates. M. steenstrupii was described from warm springs in Iceland. This habitat is more similar to desert soils, but is still fairly distinct. Our isolates could well be genotypically and physiologically separate from the source material for the original description. We chose to use the morphospecies M. steenstrupii instead of M. chthonoplastes because it is a better match with regards to habitat, and to avoid confusion with a fairly well described, ecologically distinct taxon such as M. chthonoplastes.

Direct PCR amplification of DNA obtained from single filaments demonstrated excellent congruence between morphology and 16S rRNA sequence data. Each filament was carefully characterized under the microscope, and then sacrificed for DNA amplification. A total of 17 strains were studied in this manner. Sequences from the cultured strains could all be clearly assigned to either M. vaginatus or M. steenstrupii based upon the sequences collected as part of the single filament work.

The most notable synapomorphy separating M. vaginatus from M. steenstrupii was an 11 bp insert in variable loop six (V6) on the 16S rRNA gene (bp 423-433) in M. vaginatus. This insert was recently discovered by Garcia-Pichel et al. (2001) in M. vaginatus strains isolated from various soil localities, and thus far has been seen in no other cyanobacterial taxon. Genetic similarity of our M. vaginatus clade was high, 97.1-99.9% similar, with mean similarity of 98.9%. Most of the variability occurred in V3 and V8, although the secondary structure of these loops was generally conserved, indicating the mutations were probably neutral. The morphology of all of these strains is very similar, and matches the narrow original description of M. vaginatus (Vaucher) Gomont.

A BLAST search of GenBank revealed that the M. vaginutus clade was most similar in sequence to the genera Arthrospira and Trichodesmium, at about 91-92% similarity. A complete sequence of M. vaginatus became available on GenBank (PCC9802, AF284803) after data analysis and manuscript preparation for this paper. This strain was 100% identical in sequence to our strain TAA1-4-1B. Other partial sequences for the species (Garcia-Pichel et al. 2001) had little sequence overlap with our partial sequences, and could not be matched to single strains in our study with confidence.

The group of strains without the 11 bp insert that we assigned to the morphospecies M. steenstrupii showed considerably more variability among strains. Similarity ranged from 91.5-99.4%, with a mean similarity of 94.5%. Variability was evident not only in the variable loops (V3, V6, V7, V8, V9), but also in many of the loops traditionally not considered variable. Even some of the stem regions, which are usually very conservative, showed variability (Stem 23 and Stem 36 in particular).

We suspect that there are likely a number of cryptic species within our M. steenstrupii clade, given that the low similarity of the sequences indicates the presence of a number of genospecies. Further phenotypic characterization of these strains is thus warranted. In the remainder of the paper we will continue to call this clade M. steenstrupii, recognizing that it is actually a species cluster. Indeed, given the genetic distance between M. vaginatus and M. steenstrupii, it is likely that M. steenstrupii will eventually be transferred out of Microcoleus into an existing or new genus of cyanobacteria. A BLAST search using GenBank revealed that this clade was most similar in sequence to Microcoleus chthonoplastes, Leptolyngbya, and some Oscillatoria species.

In the phylogenetic tree generated from our aligned sequence data, both M. vaginatus and the M. steenstrupii species cluster formed monophyletic clades using distance, parsimony, and maximum likelihood, indicating that our molecular data were congruent with our morphological characterization (Fig. 3). However, bootstrap analyses often did not support the inclusive nodes for each species when the full data set was used in analysis. M. vaginatus sequences were united by a unique synapomorphy (the 11 bp insert), which was left out of our distance and maximum likelihood analyses due to the fact that gaps are treated as missing data. Because all sequences in each of the two Microcoleus clades were more closely related to each other than to anything else, we can be confident that all of these sequences came from the Microcoleus taxa of interest. The Microcoleus vaginatus strains were analyzed separately with a limited outgroup (Fig. 4) in order to achieve better resolution of this group, but internal branch support was not high in this tree.

One unexpected finding is that although there was some structure within each of the two Microcoleus clades, it did not appear to be congruent with geography in either case. In the M. steenstrupii clade, sequences generally clustered with one another according to strain (Fig. 3); for example, the two sequences from MOA1MC4 were most closely related to one another. The M. vaginatus clade had poor bootstrap support within internal nodes, but there did not appear to be any clustering of strains based on geographical origin. It was notable that in the M. vaginatus clade, multiple sequences from single isolates, including multiple sequences from single filaments, were present in different parts of the clade (Fig. 4). For instance, the sequence WD1MC2-4A was most closely related to CSV2MC1-7A, whereas the sequence WD1MC2-4B was most closely related to three sequences from OTA3-2. This may in part be due to the fact that sequences were similar enough across strains that they were grouped according to single shared nucleotides.

Within the Microcoleus vaginatus sequences, the variability in the 16S gene was limited when compared with that found within the Microcoleus steenstrupii species cluster. When the data were examined closely, it was evident that identical V3 regions were present in WD1MC2-4A and SEV2MC1-1B. These two sequences were different from WD1MC2-4B and SEV2MC1-1A, which in turn were identical to one another. This suggests that the differences between sequences from single isolates may not be due merely to PCR "errors" but rather to the presence of multiple non-identical rRNA operons within single isolates. The presence of multiple non-identical rRNA operons is well-documented in non-photosynthetic eubacteria. Further study of multiple PCR clones in the 16S rRNA genes of cyanobacteria is needed before we can accurately conclude whether differences in sequences within strains are due to PCR error or real differences in sequences among operons.

The presence of Microcoleus chthonoplastes within our M. steenstrupii clade indicates that our M. steenstrupii species cluster is closely related to other described Microcoleus species. However, this cluster of Microcoleus species is so genetically and morphologically distinct from M. vaginatus that the two clades likely represent separate genera. M. vaginatus is the type species of Microcoleus Desmazières ex Gomont (Drouet 1968). Drouet (1968) did not explain his typification of the genus, but it was likely based on the fact that the earliest Microcoleus taxon described, M. terrestris Desmazières (1823), was combined with the earlier species Oscillatoria vaginata Vaucher (1803) to create the epithet M. vaginatus (Vaucher) Gomont ex Gomont (1892). Drouet’s typification is valid according to Article 10 of the International Code of Botanical Nomenclature (Greuter et al. 2000). Thus it is M. steenstrupii and M. chthonoplastes that will need to be transferred to a different new or existing genus. The sister taxa of M. vaginatus are Trichodesmium and Oscillatoria sancta (according to parsimony analysis). The three published Oscillatoria sequences used in our analysis appear in very different parts of our phylogenetic tree: one is closest to Trichodesmium, one is closest to Arthrospira, and one is closest to the Plectonema-Planktothrix-Leptolyngbya subclade. This suggests that Oscillatoria is an invalid, polyphyletic taxon in need of revision. Some of these revisions have indeed occurred (Anagnostidis & Komárek 1988), but are not reflected in the names used in GenBank for these particular strains or in the recently constructed phylogenies using GenBank sequences (Garcia-Pichel et al. 2001, Wilmotte & Herdman 2001).

The section between conserved ITS regions D1 and D1’ had a secondary structure that was conserved in all strains of M. vaginatus, regardless of ITS configuration or sequence differences (Fig. 5a). This same region for M. steenstrupii was more variable, and revealed a variety of secondary structures, all of which were different than that seen in M. vaginatus (Fig. 5a, secondary structures not shown). It is evident from these data that a number of distinct rRNA operons are present in M. steenstrupii. The Box B spacer, on the other hand, was more variable in M. vaginatus than M. steenstrupii, particularly in the base pair composition in and below the tip of the loop (Fig. 5b). The Box A spacer and D4 region were the most conserved part of the ITS, not varying by a single base pair (Fig. 6). The ITS V3 stem-loop structure was fairly consistent in sequence and secondary structure among the M. vaginatus strains, but quite variable in the M. steenstrupii cluster (Fig. 6).

The configurations of the ITS region were also different for the two species (Fig. 7). In M. vaginatus, we found ITS regions with either both tRNA genes or a fragment of the tRNA^Ile gene. In M. steenstrupii, we found ITS regions with the tRNA^Ilegene only or with no tRNA genes. We saw much greater consistency in ISR (intergenic spacer) sequence lengths within M. vaginatus than within M. steenstrupii (Fig. 7). These ITS data further support our conclusion that we have two unrelated species in this study, likely in separate genera. They also support our conclusions from the analysis of the 16S rRNA gene that M. steenstrupii is more variable than M. vaginatus, and likely includes multiple genospecies.

Our 16S rRNA data showed no population structure congruent with geography within strains of either M. vaginatus or M. steenstrupii (Fig. 3, Fig. 4). This is likely due in part to the nature of the gene, which is thought to be too highly conserved to detect population-level differences (Fox et al. 1992). The combined genetic, morphological, and ecological uniformity of M. vaginatus indicates that our strains (together with those of Garcia-Pichel et al. 2001) form a well-defined species. Significant synapomorphies of the species include the formation of calyptra, the shortened end cell, and the 11 bp insert in the 16S rRNA gene. Thus this species comforms to both the phenetic species concept of "cluster of similar forms" of Castenholz (1992) and the monophyletic phylogenetic species concept of Mischler and Theriot (2000). On the other hand, the genetic diversity in the morphospecies M. steenstrupii indicates that this clade represents several genospecies. M. chthonoplastes has been shown to be a well-defined cosmopolitan taxon by DGGE analysis of its 16S gene (Garcia-Pichel et al. 1996); therefore its presence within our M. steenstrupii clade further supports our contention that M. steenstrupii represents more than one cryptic species.

The 16S-23S ITS region may eventually help us in understanding population structure in cyanobacteria, but presently our inability to determine homology among sequences precludes the use of this region for any phylogenetic analyses. Currently, the primary utility of the ITS region seems to be the study of sequence configuration (tRNA gene presence and intergenic spacer region (ISR) lengths) for the purpose of identifying species. For example, in this study we found ITS regions with either two tRNA gene sequences (tRNA^Ile and tRNA^Ala) or a fragment of the tRNA^Ile gene in M. vaginatus, while in M. steenstrupii we found ITS regions with either one tRNA (tRNA^Ile) gene or no tRNA gene sequences. These patterns of configuration may also be useful for studies of higher level phylogeny (Boyer et al. 2001).

Researchers have long considered the dominant species of microbiotic soil crusts to be M. vaginatus (Metting 1991). It has been reported from the Great Basin Desert (Anderson and Rushforth 1976, Johansen and St. Clair 1986, Johansen et al. 1982, 1984, St. Clair et al. 1986), Uintah Basin Desert (Ashley et al. 1985, Johansen and Rushforth 1981), Colorado Plateau (Anderson and Rushforth 1976, Belnap and Gardner 1993, Campbell et al. 1989, Grondin and Johansen 1993, Johansen et al. 1981), Columbia Basin shrub-steppe (Johansen et al. 1993), the Sonoran Desert (Cameron 1964), Baja California (Flechtner et al. 1998), the Sahara Desert (Novichkova-Ivanova 1980), the Mideast (Novichkova-Ivanova 1980), and the steppes of Russia and Tibet (Novichkova-Ivanova 1980). Only a very few workers have recognized more than one species of Microcoleus in arid soils (Cameron 1964, Novichkova-Ivanova 1980). These species include M. chthonoplastes Thuret ex Gomont, M. lacustris (Rabenh.) Farlow, M. paludosus (Kütz.) Gomont ex Gomont, M. sociatus West et West, and M. subtorulosus (Bréb.) Gomont. It is possible that Schizothrix friesii (Ag.) Gomont ex Gomont is also in this clade, as it bears many morphological and physiological similarities to the group. To our knowledge, no one reported M. steenstrupii until very recently (Flechtner et al. 1998).

Garcia-Pichel et al. (2001) found that his Colorado Plateau isolates of M. vaginatus had the 11 bp insert in the 16S rRNA gene that we observed in our strains. Furthermore, he has seen this same insert in M. vaginatus isolated from Israel, Spain, and even polar areas. Of the strains of constricted Microcoleus taxa reported from soils, only M. chthonoplastes has had its sequence published on GenBank. This taxon was described from marine habitats, and is cosmopolitan in shallow marine and hypersaline waters. It is extremely doubtful that the marine form also occurs in desert soils, and the constricted forms reported from terrestrial habitats are more likely M. steenstrupii, a taxon originally described from soil. The greatest genetic similarity between any of our M. steenstrupii strains and the single published M. chthonoplastes strain is 95%, indicating that all our strains are not conspecific with M. chthonoplastes, despite the striking morphological similarity.

There did appear to be some ecological differentiation between M. vaginatus and M. steenstrupii. Although both taxa often occurred together, M. vaginatus appeared to be common in the soils of the Great Basin and Colorado Plateau, but scarce in the soils of the northern Great Basin and Columbia Basin. M. steenstrupii had a broader distribution, occurring commonly in the Mojave and Chihuahuan Deserts to the south and the Columbia Basin to the north. The two species likely have very different ecological strategies. M. vaginatus is highly motile, and has a protective calyptra that allows it to push through soil particles. It is likely a more aggressive colonizer. M. steenstrupii has limited motility and no protective calyptra. These ecological features have ramifications for crust reclamation technologies (Buttars et al. 1998). M. vaginatus, with its higher motility and ability to better penetrate soils, would likely be more successful moving from cyanobacterial amendments into the soil than its less motile counterpart M. steenstrupii. M. vaginatus has been reported to be an early colonizer (Belnap 1993), although the author of that finding likely did not separate the two species.

M. paludosus, M. sociatus, M. subtorulosus, and M. lacustris have slight to distinct constrictions at the crosswalls, indicating a Phormidium-like cell division, while M. vaginatus is not so constricted and demonstrates an Oscillatoria-like cell division. The abundance of potential cryptic species in what we have called M. steenstrupii may include some of these other morphospecies, which are at present poorly delineated and difficult to separate from one another. Additional undescribed taxa are likely to be discovered with further study. The genus Microcoleus needs revision, as the cell division differences are indicative of genetic separation at the family level. M. vaginatus is the type species of the genus, and will likely retain the generic epithet. The constricted Microcoleus taxa will require transfer to one or more other genera, but their morphological, ultrastructural, and molecular characters need to be more fully characterized before these taxonomic revisions are undertaken.

The authors thank Ferran Garcia-Pichel for his protocol for PCR from single filaments and for very helpful comments during the early stages of this project. Jim Lissemore provided continued support in the lab and Alison Dillon helped establish molecular protocols. We thank Michael Payne for providing his Leptolyngbya schmidlei sequence. This work was supported by funds from the U.S. Army Construction Engineering Research Laboratory (Contract DACA88-97-D-0001) as well as from the Biotic Surveys and Inventories Program of the National Science Foundation (Proposal 9870201). Some of the supplies and sequencing costs were paid for through funds from John Carroll University, which also provided release time for research for Johansen and Flechtner.

Altschul, S. F., Madden, T. L., Schäffer, A. A., Zhang, J., Zhang, Z., Miller, W. and Lipman, D. J. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25:3389-402.

Anagnostidis, K. and Komárek, J. 1988. Modern approach to the classification system of cyanophytes. 3. Oscillatoriales. Arch. Hydrobiol. Suppl. 80, Algological Studies 50-53:327-472.

Anderson, D. C. and Rushforth, S. R. 1976. The cryptogamic flora of desert soil crusts in southern Utah. Nova Hedwigia 28:691-729.

Ashley, J., S. R. Rushforth, and J. R. Johansen. 1985. Soil algae of cryptogamic crusts from the Uintah Basin, Utah, U. S. A. Great Basin Nat. 45:432-442.

Belnap, J. 1993. Recovery rates of cryptobiotic crusts: inoculant use and assessment methods. Great Basin Nat. 53:89-95.

Belnap, J. and Gardner, J. S. 1993. Soil microstructure in soils of the Colorado Plateau: the role of the cyanobacterium Microcoleus vaginatus. Great Basin Nat. 53:40-7.

Ben-Porath, J. and Carpenter, E. J. 1993. Genotypic relationships in Trichodesmium (Cyanophyceae) based on nifH sequence comparisons. J. Phycol. 29:806-10.

Boyer, S. L., Flechtner, V. R. and Johansen. J. R. 2001. Is the 16S-23S rRNA internal transcribed spacer (ITS) region a good tool for use in molecular systematics and population genetics? A case study in cyanobacteria. Mol. Biol. Evol. 18:1057-69.

Buttars, S. M., St. Clair, L. L., Johansen, J. R., Sray, J. C., Payne, M. C., Webb, B. L., Terry, R. E., Pendleton, B. K. and Warren, S. D. 1998. Pelletized cyanobacterial soil amendments: laboratory testing for survival, escapability, and nitrogen fixation. Arid Soil Res. Rehab. 12:165-78.

Cameron, R. E. 1964. Terrestrial algae of southern Arizona. Trans. Am. Microsc. Soc. 133:212-8.

Campbell, S. E., Seeler, J. and Golubic, S. 1989. Desert crust formation and soil stabilization. In Skujinš, J. [Ed.] Soil microbiology and organic matter in desert rehabilitation. Arid Soil Res. Rehab. 3:217-28.

Castenholz R.W. 1992. Species usage, concept, and evolution in the cyanobacteria (blue-green algae). J. Phycol. 28:737-745.

Cullings, K. W. 1992. Design and testing of a plant-specific PCR primer for ecological and evolutionary studies. Mol. Ecol. 1:233-40.

Desmazières. 1823. Catalogue des plantes omises dans la botanographie belgique. Plantes Cryptogames de France. Edition 1, no. 55.

Doyle, J. J. and Doyle, J. L. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19:11-5.

Drouet, F. 1968. Revision of the classification of the Oscillatoriaceae. Monogr. Acad. Nat. Sci. Philad. 15:1-370.

Flechtner, V. R. 1999. Enigmatic desert soil algae. In Seckbach, J. [Ed.] Enigmatic Microorganisms and Life in Extreme Environments. Kluwer Academic Publishers, Amsterdam, pp. 231-41.

Flechtner, V. R., Johansen, J.R. and Clark, W. H. 1998. Algal composition of microbiotic crusts from the central desert of Baja California, Mexico. Great Basin Nat. 58:295-311.

Fox, G. E., Wisotzkey, J. D. and Jurtshuk, P., Jr. 1992. How close is close: 16S rRNA sequence identity may not be sufficient to guarantee species identity. Int. J. Syst. Bacteriol. 42:166-70.

Garcia-Pichel, F., López-Cortés, A. and Nübel, U. 2001. Phylogenetic and morphological diversity of cyanobacteria in soil desert crusts from the Colorado Plateau. Appl. Environ. Microbiol. 67:1902-10.

Garcia-Pichel, F., Prufert-Bebout, L. and Muyzer, G. 1996. Phenotypic and phylogenetic analyses show Microcoleus chthonoplastes to be a cosmopolitan cyanobacterium. Appl. Environ. Microbiol. 62:3284-91.

Geitler, L. 1932. Cyanophyceae. In Rabenhorst, L. [Ed.] Kryptogamen-Flora von Deutschland, Österreich und der Schweiz. Reprinted 1985, Koeltz Scientific Books, Koenigstein, Germany, pp. 1-1196.

Gomont, M.M. 1892. Monographie des Oscillariées (Nostocacées homocystées). Ann. Sci. Nat. Bot. Ser. 7 15:263-368, 16:91-264.

Greuter, W., McNeill, J., Barrie, F.R., Burdet, H.M., Demoulin, V., Filgueiras, T.S., Nicolson, D.H., Silva, P.C., Skog, J.E., Trehane, P., Turland, N.J., and Hawksworth, D.L. 2000. International Code of Botanical Nomencalture (Saint Louis Code). Koeltz Scientific Books, Königstein, Germany, 474 pp.

Grondin, A. E. and Johansen, J. R. 1993. Microbial spatial heterogeneity in microbiotic crusts in Colorado National Monument. I. Algae. Great Basin Nat. 53:24-30.

Iteman, I., Rippka, R., Tandeau de Marsac, N. and Herdman, M. 2000. Comparison of conserved structural and regulatory domains within divergent 16S rRRNA-23S rRNA spacer sequences of cyanobacteria. Microbiology 146:1275-86.

Johansen, J. R. and Rushforth, S. R. 1981. Algae of surface waters and soils of areas of eastern Utah. Nova Hedwigia 34:333-9.

Johansen, J. R., Ashley, J. and W. R. Rayburn. 1993. Effects of rangefire on soil algal crusts in semiarid shrub-steppe of the Lower Columbia Basin and their subsequent recovery. Great Basin Nat. 53:73-88.

Johansen, J. R. and St. Clair, L. L. 1986. Cryptogamic soil crusts: Recovery from grazing near Camp Floyd State Park, Utah, U. S. A. Great Basin Nat. 46:632-40.

Johansen, J. R., Rushforth, S. R. and Brotherson, J. D. 1981. Subaerial algae of Navajo National Monument, Arizona. Great Basin Nat. 41:433-9.

Johansen, J. R., Javakul, A. and Rushforth, S. R. 1982. The effects of burning on the algal communities of a high desert soil near Wallsburg, Utah, U. S. A. J. Range Manage. 35:598-600.

Johansen, J. R., St. Clair, L. L., Webb, B. L. and Nebeker, G. T. 1984. Recovery patterns of cryptogamic soil crusts in desert rangelands following fire disturbance. Bryologist 87:238-43.

Komárek, J and Anagnostidis, K. 1999. Cyanoprokaryota. 1. Chroococcales. Gustav Fischer, Jena, Germany. 548 pp.

Kumar, S., Tamura, K. and Nei, M. 1993. MEGA: Molecular Evolutionary Genetics Analysis, version 1.01. The Pennsylvania State University, University Park, PA 16802.

Lu, W., Evans, E. H., McColl, S. K. and Saunders, V. A. 1997. Identification of cyanobacteria by polymorphisms of PCR-amplified ribosomal DNA spacer region. FEMS Microbiol. Let. 153:141-9.

Mathews, D.H., Sabina, J., Zuker, M. and Turner, D. H. 1999. Expanded Sequence Dependence of Thermodynamic Parameters Improves Prediction of RNA Secondary Structure. J. Mol. Biol. 288:911-40.

Metting, B. 1991. Biological surface features of semiarid lands and deserts. In Skujinš, J. [Ed.] Semiarid Lands and Deserts: Soil Resource and Reclamation. Marcel Dekker, Inc., New York, pp. 257-93.

Mishler, B.D. and Theriot, E.C.. 2000. The phylogenetic species concept (sensu Mishler and Theriot): monophyly, apomorphy, and phylogenetic species concepts. In: Wheeler, Q.D. & Meier, R. [Eds.], Species concepts and phylogenetic theory, a debate, Columbia University Press, New York, pp. 44-54.

Novichkova-Ivanova, L.N. 1980. Soil algae of Sahara-Gobi Desert region. Nauka Publishing House, Leningrad (In Russian). 255 pp.

Nübel, U., De Baere, R., Wilmotte, A. and De Wachter, R. 1996. Phylogenetic relationships of nonaxenic filamentuos cyanobacterial strains based on 16S rRNA sequence analysis. J. Mol. Evol. 42:194-200.

Nübel, U., Garcia-Pichel, F. and Muyzer, G. 1997. PCR primers to amplify 16S rRNA genes from cyanobacteria. Appl. Env. Microbiol. 63:3327-32.

Otsuka, S., Suda, S., Li, R., Watanabe, M., Oyaizu, H., Matsumoto, S. and Watanabe, M. M. 1998. 16S rDNA sequences and phylogenetic analyses of Microcystis strains with and without phycoerytherin. FEMS Microbiol. Let. 163:119-24.

Otsuka, S., Suda, S., Li, R., Watanabe, M., Oyaizu, H., Matsumoto, S. and 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. Let. 172:15-21.

Petersen, J.B. 1923.

Scheldeman, P., Baurain, D., Bouhy, R., Scott, M., Mühling, M., Whitton, B. A., Belay, A. and 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. Let. 172:213-22.

Smith, R. F., Wiese, B. A., Wojzynski, M. K., Davison, D. B. and Worley, K. C. 1996. BCM Search Launder—An Integrated Interface to Molecular Biology Data Base Search and Analysis Services Available on the World Wide Web. Genome Res. 6:454-62.

Stackebrandt, E. and Goebel, B.M. 1994. Taxonomic note: a place for DNA-DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. Int. J. Syst. Bacteriol. 44:846-849.

St. Clair, L. L., Johansen, J. R. and Webb, B. L. 1986. Rapid stabilization of fire-disturbed sites using a soil crust slurry: Inoculation studies. Reclamation Reveg. Res. 4:261-9.

Swofford, D.L. 1998. PAUP^* . Phylogentic analysis of parsimony (^*and other methods). Version 4 - Sinauer Associates, Sunderland, MA.

Thompson, J. D., Higgins, D. G. and Gibson, T. J. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-80.

Turner, S. 1997. Molecular systematics of oxygenic photosynthetic bacteria. Plant Syst. Evol., Suppl. 11:13-52.

Wayne, L.G., Brenner, D.J., Colwell, R.R., Grimont, P.A.D., Kandler, O., Krichevsky, M.I., Moore, L.H., Moore, W.E.C., Murray, R.G.E., Stackebrandt, E., Starr, M.P. and Trüper, H.G.. 1987. Report of the Ad Hoc committee on reconciliation of approaches to bacterial systematics. Int. J. Syst. Bacteriol. 37:463-464.

West, N. J. and Adams, D. G. 1997. Phenotypic and genotypic comparison of symbiotic and free-living cyanobacteria from a single field site. Appl. Environ. Microbiol. 63: 4479-84.

Wilmotte, A and Herdman, M. 2001. Phylogenetic relationships among the cyanobacteria based on 16S rRNA sequences. In: Boone, D.R. & Castenholz, R.W. [Eds.] Bergey’s manual of systematic bacteriology, 2^nd ed. vol. 1, Springer Verlag, New York, pp 487-493.

Wilmotte, A., Turner, S., Van der Peer, Y. and 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. 28:828-38.

Wilmotte, A., Van der Auwera, C. and De Wachter, R. 1993. Structure of the 16S ribosomal RNA of the thermophilic cyanobacteria Chlorogloeopsis HTF (Mastigocladus laminosus HTF) strain PCC 7518 and phylogenetic analysis. FEBS Let. 317:96-100.

Zehr, J. P., Mellon, M. T. and Hiorns, W. D. 1997. Phylogeny of cyanobacterial nifHgenes: evolutionary implications and potential applications to natural assemblages. Microbiology 143:1443-50.

Zuker, M., Mathews, D. H. and Turner, D. H. 1999. Algorithms and Thermodynamics for RNA Secondary Structure Prediction: A Practical Guide. In Barciszewski, J. and. Clark, B. F. C [Eds.] RNA Biochemistry and Biotechnology, NATO ASI Series, Kluwer Academic Publishers, pp. 11-43.

Table 1. Site data for all strains. All sites are at least 1.0 km apart except for DG6 and DAC, and DSD and DCSD, each pair representing separate samples from closely situated sites. All strain designations are prefixed by their site code.