ARTICLE IN PRESS

0043-1354/$ - se

doi:10.1016/j.w

�CorrespondAgricultura, Un

Caixa Postal 96

Tel.: +5519 34

E-mail addr

Water Research 39 (2005) 5017–5026

www.elsevier.com/locate/watres

Characterization of nitrogen-fixing cyanobacteria in theBrazilian Amazon floodplain

Marli de Fatima Fiorea,b,�, Brett A. Neilanc, Janine N. Coppc,Jorge L.M. Rodriguesa, Siu M. Tsaia, Hung Leeb, Jack T. Trevorsb

aCentro de Energia Nuclear na Agricultura, Universidade de Sao Paulo, Av. Centenario 303, Caixa Postal 96,

CEP 13400-970, Piracicaba-SP, BrazilbDepartment of Environmental Biology, University of Guelph, Guelph, Ont., Canada N1G 2W1

cSchool of Microbiology and Immunology, The University of New South Wales, Sydney 2052, Australia

Received 1 September 2003; received in revised form 21 April 2005; accepted 5 October 2005

Available online 10 November 2005

Abstract

The diversity of the free-living nitrogen-fixing cyanobacterial community in the floodplain sediments along the

Solimoes and Amazon Rivers and some of their tributaries (Japura, Negro and Madeira) was investigated. Five

cyanobacterial genera were morphologically identified, four of which (Nostoc, Calothrix, Cylindrospermum and

Fischerella) have not previously been isolated from the Brazilian Amazon floodplain. Nostoc strains were the most

commonly found heterocyst-forming cyanobacteria. Five strains (N. muscorum CENA18 and CENA61, N. piscinale

CENA21, Cylindrospermum sp. CENA33 and Fischerella sp. CENA19) were selected for growth measurement, ability

to fix N2 and phylogenetic analysis, based on their widespread distribution and morphological distinction. Molecular

analyses employing 16S rRNA sequences indicated that some of the isolates may represent novel cyanobacterial species.

Dinitrogen fixed by these strains was measured indirectly as acetylene reduction activity and ranged from 11.5 to

22.2 nmolC2H4 mgChl a�1 h�1. These results provide evidence of widespread and importance of nitrogen-fixing

cyanobacteria as a source of N inputs in the Amazonian ecosystem.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Amazon floodplain; Cyanobacteria; 16S rRNA; N2 fixation

1. Introduction

The confluence of the Solimoes and Negro Rivers (near

the city of Manaus) forms the Amazon River, which is

the world’s largest river constituting roughly 20% of the

global freshwater discharge to the oceans (Devol et al.,

e front matter r 2005 Elsevier Ltd. All rights reserve

atres.2005.10.002

ing author at: Centro de Energia Nuclear na

iversidade de Sao Paulo, Av. Centenario 303,

, CEP 13400-970, Piracicaba, SP, Brazil.

29 4657; fax: +55 19 3429 4610.

ess: fiore@cena.usp.br (M.F. Fiore).

1995). The Amazon River and its large tributaries present

a floodplain region of about 300,000km2 (Iron et al.,

1997), which are wetlands periodically inundated by the

lateral overflow of the rivers. The floodplain regions from

the Amazon are called ‘varzea’ and ‘igapo’. The term

‘varzea’ is used for floodplains along white-water rivers,

which are rich in nutrients and suspended matter, and

‘igapo’ for those along black-water and clear-water rivers,

which are poor in both (Iron et al., 1997). The ‘varzea’ is

one of the most productive regions within the Amazon

Basin, known for its fertile soils and nutrient-rich waters

(Furch, 1984; Junk, 1984; Sioli, 1975). The excess of

d.

ARTICLE IN PRESSM.F. Fiore et al. / Water Research 39 (2005) 5017–50265018

nitrate (30%) in a nitrogen balance study conducted in

the Amazon ‘varzea’ region was suggested to be supplied

by the high rate of atmospheric N2 fixation by legumes

(and probably by Paspalum grasses) together with other

N2-fixing organisms in the floodplain system (Martinelli

et al., 1992).

Cyanobacteria are responsible for a significant pro-

portion of the biological fixation of nitrogen on Earth

(Haselkorn and Buikema, 1992). N2 fixation is catalyzed

by the nitrogenase enzyme complex which is sensitive to

oxygen (Fay, 1992). Some filamentous cyanobacteria

have the ability to produce differentiated cells called

heterocysts, which provide an anaerobic environment

for the oxygen-sensitive nitrogenase (Wolk et al., 1994).

N2 fixation is carried out by all heterocsyt-forming

cyanobacteria, while only a few species of unicellular

and non-heterocyst filamentous types can fix nitrogen

aerobically (Bergman et al., 1997). Recently, cyanobac-

terial nitrogen fixation has gained recognition as an

important source of fixed nitrogen in the Arctic region

(Solheim et al., 2002) and as a major component of

oceanic primary production (Capone et al., 1997).

An important source of nitrogen input in the Amazon

floodplain might be the free-living heterocyst-forming,

diazotrophic cyanobacteria. They are found in the

Amazon floodplain during inundated periods as plank-

tonic or attached to the underwater matrix of stems and

roots of floating macrophytes (periphyton). Cyanobac-

teria also colonize exposed sediments with or without

perennial vegetation during partial drying of wet

(previously inundated) areas. Several cyanobacterial

species were identified in the Amazon region (Rodrigues,

1994; Uherkovich, 1976; Uherkovich and Franken,

1980; Uherkovich and Schmidt, 1974), with the majority

being planktonic and non-heterocyst-forming. Cyano-

bacteria may also be important as a food (protein)

source for zooplankton and benthic organisms since

primary production occurs almost exclusively in and

along the floodplain area and not in the main river

channels (Junk, 1970; Sioli, 1975; Wissmar et al., 1981).

In the present study, we isolated free-living hetero-

cyst-forming species of cyanobacteria from floodplain

sediment samples of the Brazilian Amazon region and

evaluated their diversity and importance as a source of

nitrogen to this ecosystem. Five strains were selected,

based on their widespread distribution and morpholo-

gical distinction, for growth measurement, ability to fix

N2 and phylogenetic characterization.

Fig. 1. Solimoes/Amazon Rivers and major tributaries. Num-

bers indicate mainstream and tributary floodplain sampled (K)

as identified in the Table (redrawn from Forsberg et al., 1988;

Victoria et al., 1989).

2. Materials and methods

2.1. Cyanobacteria isolation and purification

Samples of ‘varzea’ sediments were collected at 14

locations along the mainstream Solimoes/Amazon River

channel and one location on each of three tributaries

(Japura and Madeira ‘varzea’ sediments and Negro

River ‘igapo’ sediment) (Fig. 1). They were taken during

the drying of wet (previously inundated) areas (June–-

July). Sediment samples were collected with a coring

device from recent alluvial deposits within 100m of the

river edge. The pH of the samples was determined after

equilibration of 10 g sediment with 25mL distilled water.

Sediment samples were homogenized using a sterile

spatula and dispensed (1 g of dry weight) into sterile test

tubes containing 9mL of BG-11 liquid medium (Allen,

1968) without nitrogen source. After mixing for 15min

at 200 rpm, 10-fold serial dilutions were used

to inoculate test tubes containing the same medium.

The tubes were incubated in a growth chamber for

30 days at 28 1C, under white fluorescent illumination

(50mmol photonsm�2 s�1) with a 12 h light/dark cycle.

Cyanobacterial cell numbers were calculated by

the most probable number (MPN) technique, using

probability tables (Postgate, 1969). Aliquots (0.1mL)

of appropriate dilutions were spread onto 1.2% (w/v)

BG-11 Noble agar (Difco Laboratories, Detroit, MI)

ARTICLE IN PRESSM.F. Fiore et al. / Water Research 39 (2005) 5017–5026 5019

plates and incubated under the same conditions as the

test tubes.

To obtain axenic clones of cyanobacterial strains,

filaments were fragmented in an ultrasonic bath until two

to four cells per filament were produced as observed

microscopically at 400� magnification. The fragmented

culture was rinsed three times with sterile AA liquid

medium (Allen and Arnon, 1955) diluted fourfold (AA/4)

with a nitrogen source omitted, and centrifuged for 5min

at 10,000g at 20 1C. Resuspended cells were collected on a

sterile membrane filter (2.7mm pore size) and washed

several times with fresh sterile AA/4 medium. Cells were

repeatedly streaked onto solid AA/4 medium until axenic

cultures were established. Following isolation of uni-

cyanobacterial cultures, strains were identified based on

morphotype description using the identification keys of

Anagnostidis and Komarek (1990) and Komarek

and Anagnostidis (1989). For routine cultivation, cells

were grown under constant, white fluorescent light

(50mmol photonsm�2 s�1) at 28 1C in AA/4 liquid

medium without a nitrogen source.

2.2. Plating efficiency

Cyanobacterial strains grown for 20 days in flasks

containing 40mL AA/4 medium were sonicated in an

ultrasonic cleaning bath (Branson, Smithkline Co.

Shelton, CT), to yield filaments of two to four cells.

The average number of cells per fragment was deter-

mined by assays of 200 fragments using serial dilutions.

The number of fragments per mL was determined with a

hemocytometer under light microscopy at 400� mag-

nification and the cell counts were recorded.

Cell suspensions, with an average of 200 fragments,

were plated in duplicate on AA medium supplemented

with 2.5mM NH4Cl and buffered with 5mM MOPS

(Sigma Chemical, St. Louis, MO) and incubated under

the conditions described above. Plating efficiency was

calculated by dividing the number of colonies appearing

on the plate by the number of fragments plated (Wolk

and Wojciuch, 1973).

2.3. Cell growth

Stock 15–20 day cultures were used as inocula for

growth measurements. Ten mL of the cultures were

collected by centrifugation at 1000g at 4 1C for 15min

and resuspended in 2mL sterile AA/4 medium. A 200mLaliquot was inoculated in flasks (50mL) containing

20mL liquid AA/4 medium and incubated under the

same conditions described above with orbital shaking at

110 rpm. Growth was monitored at 48 h intervals by

chlorophyll a (Chl a) measurements. Briefly, triplicate

subsamples (1mL of cell suspension) were harvested at

12,000g for 10min and washed twice with sterile water.

Cell pellets were resuspended in 1mL 90% (v/v)

methanol and incubated at 25 1C for 15min in the dark.

The methanol extract was determined at 663 nm using a

spectrophotometer (model DU-60, Beckman). The Chl a

content was estimated using the equation: C (mg/mL) ¼ OD663 nm� 12.7 (Meeks and Castenholz, 1971).

2.4. Nitrogenase activity

Nitrogenase activity was evaluated in 12-day-old

cultures using the acetylene reduction technique (Hardy

et al., 1973). A 10mL aliquot of each culture was

transferred to a flask with a 50mL total volume, which

was sealed with a suba-seal gas-tight stopper. A 5%

volume of air was replaced with pure acetylene using a

gas-tight syringe. Flasks were incubated for 2 h under

the same conditions described above. Ethylene quanti-

fication was analyzed in a FID gas chromatograph

equipped with a Porapak N column. Nitrogenase

activity was expressed as nmoles of ethylene produced

per mg Chl a.

2.5. DNA extraction, amplification, sequencing and

analysis

Genomic DNA was extracted from eight cyanobac-

terial strains using a previously published method

(Fiore et al., 2000). The strains included the five Amazon

isolates (Nostoc muscorum CENA18 and CENA61,

N. piscinale CENA21, Cylindrospermum sp. CENA33

and Fischerella sp. CENA19) and strains obtained from

the University of Toronto Culture Collection

(Phormidium autumnale UTEX1580, Nostoc commune

UTEX584 and Anabaena flos-aquae UTCC64). Cultures

were maintained in AA/4 or BG-11 (P. autumnale

UTEX1580) media at 28 1C with light intensity of

50mmol photonsm�2 s�1 until the mid-exponential to

late exponential growth phase. PCR amplifications of

the 16S rRNA gene were performed by using the primers

27F1 and 1494R together with PCR reagents as

previously described (Neilan et al., 1997). PCR was

performed in a PE2400 apparatus (Perkin-Elmer Cetus

Corporation, CA) with initial denaturation at 94 1C for

2min, followed by 35 cycles of 93 1C for 10 s, primer

annealing at 50 1C for 20 s and extension at 72 1C for

90 s, followed by a final extension time of 4min at 72 1C.

N. muscorum CENA18 and CENA61, A. flos-aquae

UTCC64, N. commune UTEX584 and P. autumnale

UTEX1580 were amplified with an annealing tempera-

ture of 58 1C. Genomic DNA of Fischerella CENA19

required an annealing temperature of 52 1C to produce

PCR amplicons of the expected size. Approximately,

50 ng of genomic DNA was used per 25mL reaction in

addition to 10 pmol of each of the bacterial 16S rRNA

primers. The remaining amplification reaction compo-

nents were as previously described (Neilan et al., 1997).

A positive control sample (cyanobacterial DNA from

ARTICLE IN PRESS

Table 1

Sample locations, number of strains isolated, most probable

number (MPN) of heterocyst-forming cyanobacteria and pH

from floodplain sediment samples collected in the main

Amazon channel and tributaries

Sample location Numbers

of isolates

MPN

(Cyanobacteria/g

dry weight

sediment)

pHa

Vargem Grande 30 41.4� 105 7.2

Santo Antonio do Ic-a 62 41.4� 105 6.5

Xibeco 33 41.4� 105 6.9

Tupe 42 41.4� 105 6.9

Mari 42 41.4� 105 6.1

Panamim 24 1.1� 105 6.6

Itapeua 21 41.4� 105 6.6

M.F. Fiore et al. / Water Research 39 (2005) 5017–50265020

Microcystis aeruginosa) was also used. Amplified DNA

was purified from surplus reaction components employ-

ing an 80% ethanol precipitation step and sequenced by

standard automated fluorescence techniques (Neilan

et al., 1997). Sequence analysis of PCR products was

performed using the computer programs NCBI-Blast

(Altschul et al., 1997). The sequences were then aligned

against the most similar sequences in the ARB small

subunit rRNA database using the alignment algorithm

in the ARB software package (Ludwig et al., 2004) and

the alignment was adjusted manually based upon

elements of primary sequence and secondary structure.

Aligned sequences were used to generate a phylogenetic

tree using parsimony analysis. The program PAUP* was

used for bootstrap analysis of the same data (Swofford,

2000).

Anori 25 41.4� 105 6.5

Manacapuru 32 41.4� 105 6.1

Sao Jose do Amatari 20 41.4� 105 6.4

Santa Luzia 19 2.0� 104 6.4

Paura 33 41.4� 105 5.7

Caldeirao 26 41.4� 105 5.6

Obidos (left margin) 23 41.4� 105 6.2

Obidos (right margin) 12 1.1� 105 6.2

Japura River 7 7.0� 102 5.3

Negro River 22 1.4� 104 5.1

Madeira River 9 9.5� 103 5.1

apH determined during the period when the samples were

taken in early falling water period.

3. Results

3.1. Cyanobacteria isolation and purification

Cyanobacterial strains (482) (Table 1) were isolated

through successive plating from sediment samples at 17

locations along the main channel and tributaries of the

Amazon River (Fig. 1). After several platings, micro-

scopic observations revealed that some of the isolates,

initially seen to exhibit differences in morphology, were

shown to be in fact the same. Sixty-one isolates were

obtained after careful selection using taxonomic de-

scriptors. A total of 20 cyanobacterial strains with

higher growth yields were selected for axenic cultures

and these were successfully purified (Table 2). Using our

protocol, axenic cultures were obtained for the majority

of the isolates (90%) in less than 30 days and we

observed blue–green microcolonies on agar plates even

after the third day for some of the isolates.

The MPN of heterocyst-forming cyanobacteria from

the sediment samples (ranging from 7� 102 to 1.4� 105

cells/g dry weight) showed that the main channel

sediments support larger cyanobacterial populations

than the three tributary sediments (Table 1). In general,

lower MPN values and numbers of isolates were

observed in sediments with lower pH values.

3.2. Plating efficiency

The growth responses of the 20 purified strains were

evaluated by microscopic enumeration and plating

efficiency. After 20 days of incubation, the mean number

of cyanobacterial CFU ranged from 1� 106 to

15.3� 106CFU/mL (Table 2). Nostoc CENA54 pre-

sented 15 times lower CFU/mL than N. muscorum

CENA61. Macroscopical colonies were visible from 5

days (N. muscorum CENA61 and CENA18, Fischerella

sp. CENA19) to 14 days. Short filaments (two or four

cells) of N. muscorum CENA61 gave rise to discrete

colonies with an efficiency of 100%, while five strains

(N. muscorum CENA18, Fischerella sp. CENA19,

Cylindrospermum sp. CENA33, N. piscinale CENA21,

N. punctiforme CENA48) were plated with efficiencies

ranging from 50% to 70%. Two Nostoc strains

(CENA54 and CENA8) presented plating efficiency of

around 40%. Plating efficiencies of 10–29% were

obtained for seven cyanobacterial strains (N. muscorum

CENA14, Fischerella sp. CENA16, Calothrix sp.

CENA40, N. muscorum CENA2, Calothrix sp. CENA7,

N. commune CENA5 and N. calcicola CENA23). Five

cyanobacterial strains (Nostoc sp. CENA46, A. ambigua

CENA22, Anabaena sp. CENA 28 and CENA51, and

Cylindrospermum sp. CENA55) were plated with effi-

ciencies below 10% under the same conditions.

3.3. Cell growth

Growth was evaluated for five strains of cyanobacter-

ia (N. muscorum CENA18 and CENA61, N. piscinale

CENA21, Cylindrospermum sp. CENA33 and Fischer-

ella sp. CENA19) (Table 2). Chl a has been used by

several researchers (Mallik and Rai, 1994; Meeks et al.,

ARTICLE IN PRESS

Table 2

Colony-forming units (CFU), chlorophyll a content and plating efficiency of heterocyst-forming cyanobacteria

Isolates Origin 106CFUml�1 mgChl aml�1 Plating efficiency (%)

Nostoc muscorum CENA61 Negro River 15.370.65 14.570.52 100

Nostoc muscorum CENA18 Japura River 14.470.53 10.3470.75 70

Fischerella sp. CENA19 Mari 7.570.43 5.6470.57 62

Cylindrospermum sp. CENA33 Manacapuru 12.470.46 6.8670.28 50

Nostoc piscinale CENA21 Mari 10.470.66 2.9370.30 50

Nostoc punctiforme CENA48 Santa Luzia 9.570.44 6.7570.71 50

Nostoc sp. CENA54 Obidos (righ margin) 1.070.08 7.8270.75 41

Nostoc sp. CENA8 Santo Antonio do Ic-a 11.571.11 2.3271.02 40

Nostoc muscorum CENA14 Xibeco 9.571.32 3.5171.19 29

Fischerella sp. CENA16 Tupe 3.270.80 6.5570.65 29

Calothrix sp. CENA40 Madeira River 6.470.44 4.4070.55 24

Nostoc muscorum CENA2 Vargem Grande 6.270.36 2.8170.41 23

Calothrix sp. CENA7 Santo Antonio do Ic-a 3.270.10 2.7770.84 13

Nostoc commune CENA5 Santo Antonio do Ic-a 5.370.53 7.2370.51 12

Nostoc calcicola CENA23 Mari 2.370.61 1.8470.19 10

Nostoc sp. CENA46 Paura 4.570.44 5.2770.87 7

Anabaena sp. CENA51 Caldeirao 4.070.50 2.0970.37 6

Anabaena ambigua CENA22 Mari 3.671.15 3.6471.07 5

Cylindrospermum sp. CENA55 Obidos (left margin) 3.570.46 5.1570.37 3

Anabaena sp. CENA28 Itapeua 1.770.79 2.2870.43 2

Each value is the mean 7S.D. of three independent determinations.

0

5

10

15

20

25

0 48 96 144 192 240 288

N. muscorum CENA18

Fischerella sp. CENA19

N. piscinale CENA21

Cylindrospermumsp. CENA33N. muscorum CENA61

Time in hours

µg C

hla

/ml

Fig. 2. Growth curve based on chlorophyll a contents of five

cyanobacterial strains grown in AA/4 medium at 28 1C. Data

are mean 7S.D. of three independent determinations.

M.F. Fiore et al. / Water Research 39 (2005) 5017–5026 5021

1983; Thiery et al., 1991) for estimating cyanobacterial

biomass because cell quantification cannot be uniformly

applied to morphologically diverse cyanobacteria and

many species produce compact, aggregated colonies.

Under the experimental conditions, 12-day-old N.

muscorum CENA61 cells showed the highest growth

yield (21.7 mgChl amL�1) and N. piscinale the lowest

growth yield (12.3 mgChl amL�1). All strains showed an

initial lag phase (ranging from 48 to 96 h) with N.

muscorum CENA18 and Cylindrospermum sp. CENA33

exhibiting the shortest lag of 48 h (Fig. 2). This phase is

dependent on the growth conditions of the inoculum

(age, temperature) and the nature of the medium.

Fischerella sp. CENA19 displayed an intense green

appearance in the liquid AA/4 medium and grew as a

dense, highly branched mat, consisting of thick main

filaments out of which more slender and tapering branch

filaments grew. N. muscorum CENA18 growing in liquid

AA/4 medium had a brownish appearance probably due

to the production of brown sheaths (Geitler, 1932) that

enfold trichomes forming a common matrix with a firm

surface pellicle (biofilm). Single filaments from these two

strains were not easily obtained and the treatment varied

(sonication, vortexing, syringe disruption) depending on

the degree of aggregation in order to produce dispersed

cell suspensions. Cylindrospermum sp. CENA33 had a

dark green appearance and a confluent mucilage holding

many trichomes together to form a loose aggregate

colony in liquid AA/4 medium and cell suspensions were

easier to obtain. N. muscorum CENA61 had a green

appearance and produced gelatinous colonies. The

gelatinous material was water soluble and became

dispersed through the surrounding medium making the

liquid culture highly viscous and even gelatinous. N.

piscinale CENA21 had a green appearance and pro-

duced gelatinous colonies that were not soluble in the

medium and also formed a surface pellicle (biofilm).

ARTICLE IN PRESSM.F. Fiore et al. / Water Research 39 (2005) 5017–50265022

3.4. Nitrogenase activity

The N2 fixation capacity of the heterocysts cyanobac-

teria showed distinct differences among species and/or

strains (Table 3). The order of nitrogen fixing activity

(nmolesC2H4 � mgChl a�1 h�1) for the isolates grown

under the same environmental conditions was N. piscinale

CENA214N. muscorum CENA184Cylindrospermum

sp. CENA334Fischerella sp. CENA14N. muscorum

CENA61. Nostoc strains showed both the highest and

lowest nitrogenase activity. Nitrogenase activity of N.

piscinale CENA21 was 50% and 30% higher than N.

muscorum CENA61 and CENA18, respectively.

3.5. Phylogenetic analysis

Nearly complete nucleotide sequences (1338–1432 bp)

of the 16S rRNA covering base positions 27-1494

(corresponding to E. coli numbering) were determined

for eight cyanobacterial strains. The 16S rRNA

sequences of the five isolated Amazonian strains showed

different levels of identity when compared to cyanobac-

terial sequences on public databases. The BLAST

analysis of the 16S rRNA sequence from the true-

branching Fischerella sp. CENA19, identified by the cell

morphology system, compared favorably (95% identity)

with the genus Fischerella sp. 1711 (AJ544076). Further-

more, the sequences from N. piscinale CENA21, N.

muscorum CENA18 and N. muscorum CENA61 shared

94%, 95% and 96% identity to Nostoc sp. PCC7120

(NC003272). The alignment of the 16S rRNA gene

sequences among the three Nostoc strains isolated in this

study revealed lower than 95% sequence identities. The

BLAST analysis of Cylindrospermum CENA33 showed

95% sequence identity to Cylindrospermum PCC7417

(AJ133163). For the three other strains obtained from a

culture collection and sequenced in this study, only N.

commune UTEX584 identified by the cell morphology

system compared favorably (96% identity) with the genus

Nostoc sp. strain PCC73102 (AF027655). A. flos aquae

UTCC64 shared 98% identity with genus Nostoc

PCC7120 and P. autumnale UTEX1580 showed low

identity (86%) to P. muscicola M-221 (AB003165).

Table 3

Nitrogenase activity of five cyanobacterial strains measured

using acetylene reduction assay

Strains nmoles C2H4 mgChl a�1 h�1

Nostoc muscorum CENA18 16.18873.59

Nostoc piscinale CENA21 22.2367 2.98

Nostoc muscorum CENA61 10.99771.33

Cylindrospermum CENA33 13.84972.16

Fischerella CENA19 11.50772.23

Phylogenetic relationships were investigated with

other cyanobacteria from Genbank using the parsimony

analysis (Fig. 3). All the heterocyst-forming cyanobac-

terial strains formed a monophyletic cluster in the tree.

There are two groups within the cluster: 1. Strains

characterized by solitary trichomes belonging to the

Nostocales order, and 2. Strains characterized by

branched trichomes forming the Stignomatales order.

The P. autumnale UTEX1580 genus assigned to the

Oscillatoriales order, which consists of filamentous non-

heterocyst-forming cyanobacteria, grouped with Chlor-

ococcidiopsis thermalis PCC7203 (AB039005) belonging

to the Pleurocapsales order, despite distinct morpholo-

gical differences, such as unicellular cells. However, the

bootstrap percentage supporting this clade (cluster) is

lower than 50% and the percentage of sequence identity

between these two strains is also low (88%).

4. Discussion

Heterocyst-forming cyanobacteria were isolated from

17 geographically diverse sites along the Brazilian

Amazon floodplain, 16 from white-water rivers

(‘varzea’) and one from a black-water river (‘igapo’).

Five genera (15 species) of N2-fixing cyanobacteria

were identified. Four of these (Nostoc, Calothrix,

Cylindrospermum and Fischerella), to our knowledge,

have not previously been isolated from the Brazilian

Amazon region. Nostoc strains were the most common

heterocyst-forming cyanobacteria (isolated from

17 locations) followed by Fischerella (13 locations),

Calothrix (five locations) and Cylindrospermum and

Anabaena (three locations). The dominance of the

Nostoc genus has also been reported in other terrestrial

habitats (Olson et al., 1998). The predominance of

Nostoc in certain habitats has been attributed to its

ability to remain desiccated for months to years, then

recover metabolic activity within hours to days after

rehydration with water; its ability to screen damaging

UV light in terrestrial and shallow benthic habitats; its

resistance to grazing probably due to production of a

large amount of sheath material and the formation of

colonies that are too large for many algivores to

consume (Dodds et al., 1995).

The 16S rRNA gene sequences obtained from the five

Amazonian isolates do not perfectly, but only closely,

match previously known cyanobacterial gene sequences.

The BLAST analysis of the Fischerella CENA19, Nostoc

CENA18, CENA61, CENA21 and Cylindrospermum

CENA33 16S rRNA gene resulted in agreement at the

genus level with phenotypic characteristics. The three

Amazon Nostoc isolates compared favorably with

Nostoc PCC7120. This strain had its genome sequenced

(Kaneko et al., 2001) as Anabaena sp. PCC7120.

Taxonomic confusion was established for this strain

ARTICLE IN PRESS

51

96

98

98

63

62

55

55

89

76

53

0.1 98

Nostoc sp. PCC7120

Anabaena flos-aquae UTCC64

Cylindrospermum sp. PCC7417

Nostoc muscorum CENA61

Nostoc muscorum CENA18

Nostoc commune UTEX584

Nostoc sp.PCC73102

Nostoc sp. SAG2028

Nostoc piscinale CENA21

Cylindrospermum sp. CENA33

Fischerella muscicola SAG2027

Nostochopsis lobatus 92.1

Fischerella sp. 1711

Westiellopsis prolifica SAG16.93

Fischerella sp. CENA19

Fischerella muscicola PCC7414

Chlorogloeopsis sp.PCC7518

Leptolyngbya sp. PCC7375

Synechococcus sp. PCC6301

Microcystis aeruginosa PCC7941

Phormidium autumnale UTEX1580

Chlorococcidiopsis thermalis PCC7203

Phormidium mucicola M-221

Gloeobacter violaceus PCC7421

Escherichia coli K12

Fig. 3. A phylogenetic tree of the cyanobacterial species identified by 16S rRNA gene sequencing. The 16S rRNA gene sequence from

Escherichia coli strain K12 was treated as the outgroup. Numbers at the nodes represent percentage bootstrap values of 1000

resamplings. Sequences shown in bold were generated during this study. The scale is the expected number of substitutions per position.

Nucleotide sequences obtained in this study have been deposited in GenBank under accession numbers AY039703,

AY218827–AY218833.

M.F. Fiore et al. / Water Research 39 (2005) 5017–5026 5023

because it was originally described as N. muscorum

(Adolph and Haselkorn, 1971) and renamed by Rippka

et al. (1979) as Anabaena sp. PCC7120, but based on

DNA–DNA hybridization (Lachance, 1981) it was

renamed again as Nostoc sp. PCC7120 (Rippka and

Herdman, 1992). Although 16S rRNA sequences of

Amazon strains were related to sequences deposited in

public data banks, sequence identities were lower than

97.5%, which indicates that these strains belong to

different species (Stackebrandt and Goebel, 1994). Thus

the Amazon isolates may represent novel species within

the Cyanobacteria division.

The phylogenetic tree constructed on the basis of the

16S rRNA gene sequences showed that Fischerella sp.

CENA19, belonging to the Stigonematales order,

clustered within the Nostocales order, despite distinct

morphological differences, such as branched trichomes.

This result supports the monophyletic cluster of hetero-

cyst-forming cyanobacteria as previously reported

(Gugger and Hoffmann, 2004; Henson et al., 2004;

Lyra et al., 2001; Turner, 1997; Wilmotte, 1994;

Wilmotte and Herdman, 2001). An internally incoherent

clade, as observed for P. autumnale UTEX1580 grouped

with C. thermalis PCC7203, can be expected since

ARTICLE IN PRESSM.F. Fiore et al. / Water Research 39 (2005) 5017–50265024

analyses of 16S rRNA sequences from cyanobacteria

have shown that many previously described genera are

not phylogenetically coherent and are in need of

taxonomic revision (Honda et al., 1999; Turner et al.,

1999).

The isolates obtained demonstrated that the Amazon

floodplain supports a diverse assemblage of culturable

N2-fixing cyanobacteria. The pH values from the natural

habitats of cyanobacterial isolates may have restricted

the cyanobacterial population, since the MPN and the

number of isolates was lower in the tributary sediments

which were more acid than the main channel sediments.

However, the water composition may have also influ-

enced the cyanobacterial population, since the only

sample collected in the ‘igapo’ (black-water) showed

high MPN and number of isolates, despite the low pH.

In general, cyanobacteria are more abundant in neutral

or slightly alkaline environments (pH 7.0–10) and even

some species encountered in mildly acid habitats (pH

5.0–6.0) are generally acid tolerant rather than acid-

ophilic (Rippka et al., 1981). The few representative

species isolated (15 species) may indicate that the

technique used was limiting for some taxa since the

diversity of cyanobacteria in different environments is

normally greater than that revealed here. It is possible

that only a small percentage of cyanobacterial species

were cultured, since cyanobacteria from some habitats

and those of certain taxonomic groups prove to be

difficult to culture with existing methods (Castenholz,

1988). Molecular investigations also showed that cya-

nobacterial populations in natural environments have

been underestimated using current isolate purification

and culturing methods by introducing the problem of

strain selectivity (Ward et al., 1998). However, despite

the limitation of culturing methods, the diversity of

cyanobacterial isolates obtained from the Amazon

floodplain provides a potential source for future

research, including bioprospecting, searching for biolo-

gically active secondary metabolites, phylogenetic stu-

dies, cyanobacterial genomics and the role of these

microorganisms in the ecology of the Amazon region.

The few studies found in the literature regarding the

distribution of cyanobacterial communities in the

Amazon identified a small number of heterocyst-

forming species occurring in water and in periphyton

with the majority being planktonic non-heterocyst-

forming species. A study of the algal communities at

Castanho Lake in the Amazon region showed that

maximum taxon numbers occurred in the falling- and

low-water periods (Uherkovich and Schmidt, 1974). Two

hundred and nine algal taxa were described with 19 of

these being cyanobacterial species with occasional blooms

of Anabaena hasalii, Aphanizomenon flos-aquae and

Oscillatoria limosa. The investigation of the communities

of periphyton in nine habitats of seven rainforest streams

near Manaus showed that of 329 algal taxa, 37 were

cyanobacterial species (Uherkovich and Franken, 1980).

Samples from the algal community of the Tapajos River

showed the presence of M. aeruginosa and O. limosa

(Uherkovich, 1976). In a study conducted at the

Camaleao Lake (Marchantaria Island, Solimoes River),

262 species of algae were described, nine of which belong

to the Cyanobacteria (Rodrigues, 1994).

The cyanobacterial strains isolated in this study are

heterocyst-forming terrestrial types and were found in

floodplain sediments without vegetation during the early

dry season. Some species of the most abundant genus

isolated, Nostoc, also showed the highest N2 fixation

rates. The differences in N2 fixation among the strains

investigated could be caused by variability in their N2-

fixing capacity. The N2 fixation rate is influenced by the

number of heterocysts present in the trichome, which in

turn varies depending on the cyanobacterial species and

prevailing environmental conditions (Wolk et al., 1994).

The nitrogenase enzyme complex requires about 20 gene

products for its synthesis and assembly (Dean and

Jacobson, 1992). The different arrangements of the

nitrogenase gene cluster and different copy numbers of

some of these genes among diazotrophic cyanobacteria

(Fay, 1992; Golden et al., 1985; Meeks et al., 2001;

Saville et al., 1987; Singh and Stevens, 1992), although

not yet evaluated, may also affect the N2-fixing

efficiency of the whole culture.

In this study we isolated and identified free-living N2-

fixing cyanobacteria present in the Amazon floodplain

ecosystems. The finding that these microorganisms

colonize the 17 different sites suggests they may be

partially responsible for the high rates of N2 fixation

observed in the floodplain system. A more comprehen-

sive database on the cyanobacterial population and

distribution in the Amazon floodplain, along with

information regarding in situ contributions to N cycling,

will allow for more precise evaluation of its contribu-

tions to new N inputs and its importance for the

Amazon biogeochemistry.

Acknowledgements

We thank the Isotope Ecology Laboratory at CENA/

USP for collecting sediment samples. Sincere apprecia-

tion is expressed to Dr. C. L. Sant’Anna and Dr. M.T.P.

Azevedo (Phycology Laboratory, Botany Institute,

Sao Paulo, Brazil) for assistance with cyanobacteria

taxonomy.

M.F.F. was supported by a graduate scholarship from

the Fundac- ao de Amparo a Pesquisa do Estado de Sao

Paulo (FAPESP), Brazil. Research by J.T.T. and H.L.

was supported by individual Discovery grants from the

National Sciences and Engineering Research Council

(NSERC) of Canada. B.A.N and J.C.N were funded by

the Australian Research Council.

ARTICLE IN PRESSM.F. Fiore et al. / Water Research 39 (2005) 5017–5026 5025

References

Adolph, K.W., Haselkorn, R., 1971. Isolation and character-

ization of a virus infecting the blue-green alga Nostoc

muscorum. Virology 46, 200–208.

Allen, M.B., 1968. Simple conditions for growth of unicellular

blue-green algae on plates. J. Phycol. 4, 1–4.

Allen, M.B., Arnon, D.I., 1955. Studies on nitrogen-fixing blue-

green algae. I. Growth and nitrogen fixation by Anabaena

cylindrica Lemm. Plant Physiol. 30, 366–372.

Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang,

Z., Miller, W., Lipman, D.J., 1997. Gapped BLAST and

PSI-BLAST: a new generation of protein database search

programs. Nucleic Acids Res. 25, 3389–3402.

Anagnostidis, K., Komarek, J., 1990. Modern approach to the

classification system of Cyanophytes, 5—Stigonematales.

Arch. Hydrobiol. Suppl. 86, Algolog. Stud. 59, 1–73.

Bergman, B., Gallon, J.R., Rai, A.N., Stal, L.J., 1997.

N2-fixation by non-heterocystous cyanobacteria. FEMS

Microbiol. Rev. 19, 139–185.

Capone, D.G., Zehr, J.P., Paerl, H.W., Bergman, B., Carpen-

ter, E.J., 1997. Trichodesmium, a globally significant marine

cyanobacterium. Science 276, 1221–1229.

Castenholz, R.W., 1988. Culturing methods for cyanobacteria.

Meth. Enzymol. 167, 68–93.

Dean, D.R., Jacobson, M.R., 1992. Biochemical genetics of

nitrogenase. In: Stacey, G., Burris, R.H., Evans, H.J. (Eds.),

Biological Nitrogen Fixation. Chapman & Hall, New York,

pp. 763–834.

Devol, A.H., Forsberg, B.R., Richey, J.E., Pimentel, T.P., 1995.

Seasonal variation in chemical distributions in the Amazon

(Solimoes) River: a multiyear time series. Global Biochem.

Cy. 9, 307–328.

Dodds, W.K., Gudder, D.A., Mollenhauer, D., 1995. The

ecology of Nostoc. J. Phycol. 31, 2–18.

Fay, P., 1992. Oxygen relations of nitrogen fixation in

cyanobacteria. Microbiol. Rev. 56, 340–373.

Fiore, M.F., Moon, D.H., Tsai, S.M., Lee, H., Trevors, J.T.,

2000. Miniprep DNA isolation from unicellular and

filamentous cyanobacteria. J. Microbiol. Meth. 39, 159–169.

Forsberg, B.R., Devol, A.H., Richey, J.E., Martinelli, L.A.,

Santos, H., 1988. Factors controlling nutrient concentrations

in Amazon floodplain lakes. Limnol. Oceanogr. 33, 41–56.

Furch, K., 1984. Water chemistry of the Amazon basin: the

distribution of chemical elements among freshwaters. In:

Sioli, H. (Ed.), The Amazon. Dr. W. Junk Publishers,

Dordrecht, pp. 167–200.

Geitler, L., 1932. Cyanophyceae. In: Kolkwitz, R. (Ed.),

Rabenhorst’s Kryptogamenflora von Deutschland. Oster-

reich und der Schweiz, vol. 14. Leipzig Akademische

Verlagsgesellschaft.

Golden, J.W., Robinson, S.J., Kaselkorn, R., 1985. Rearrange-

ment of nitrogen fixation genes during heterocyst differ-

entiation in the cyanobacterium Anabaena. Nature 314,

410–423.

Gugger, M.F., Hoffmann, L., 2004. Polyphyly of the true

branching cyanobacteria (Stigonematales). Int. J. Syst.

Evol. Microbiol. 54, 349–357.

Hardy, R.W.F., Burns, B.C., Holsten, R.D., 1973. Application

of the acetylene-ethylene assay for measurement of nitrogen

fixation. Soil Biol. Biochem. 5, 47–81.

Haselkorn, R., Buikema, W.J., 1992. Nitrogen fixation by

cyanobacteria. In: Stacey, G., Burris, R.H., Evans, H.J.

(Eds.), Biological Nitrogen Fixation. Chapman & Hall,

New York, pp. 166–190.

Henson, B.J., Watson, L.E., Barnum, S.R., 2004. Molecular

differentiation of the heterocystous cyanobacteria Nostoc

and Anabaena, based on complete NifD sequences. Curr.

Microbiol. 45, 161–164.

Honda, D., Yokota, A., Sugiyama, J., 1999. Detection of seven

major evolutionary lineages in cyanobacteria based on the

16S rRNA gene sequence analysis with new sequences of

five marine Synechococcus strains. J. Mol. Evol. 48,

723–739.

Iron, G., Junk, W.J., Mello, J.A.S.N., 1997. The large central

amazonian river floodplains near Manaus: geological,

climatological, hydrological and geomorphological aspects.

In: Junk, W.J. (Ed.), The Central Amazon Floodplain,

Ecological Studies, vol. 126. Springer, Berlin, Heidelberg,

pp. 24–46.

Junk, W.J., 1970. Investigation on the ecology and produc-

tion—biology of the ‘‘floating meadows’’ (Paspalo-echino-

choletion) on the middle Amazon, Part I: the floating

vegetation and its ecology. Amazoniana 2, 449–496.

Junk, W.J., 1984. Ecology of the varzea floodplains of Amazon

white-water rivers. In: Sioli, H. (Ed.), The Amazon. Dr.

W. Junk Publishers, Dordrecht, pp. 271–293.

Kaneko, T., Nakamura, Y., Wolk, C.P., Kuritz, T., Sasamoto,

S., Watanabe, A., Iriguchi, M., Ishikawa, A., Kawashima,

K., Kimura, T., Kishida, Y., Kohara, M., Matsumoto, M.,

Matsuno, A., Muraki, A., Nakazaki, N., Shimpo, S.,

Sugimoto, M., Takazawa, M., Yamada, M., Yasuda, M.,

Tabata, S., 2001. Complete genomic sequence of the

filamentous nitrogen-fixing cyanobacterium Anabaena sp.

strain PCC 7120. DNA Res. 8, 205–213.

Komarek, J., Anagnostidis, K., 1989. Modern approach to the

classification system of Cyanophytes, 4—Nostocales. Arch.

Hydrobiol. Suppl. 823, Algolog. Stud. 56, 247–345.

Lachance, M.-A., 1981. Genetic relatedness of heterocystous

cyanobacteria by deoxyribonucleic acid deoxyribonucleic

acid reassociation. Int. J. Syst. Bacteriol. 31, 139–147.

Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H.,

Yadhukumar, Buchner, A., Lai, T., Steppi, S., Jobb, G.,

Forster, W., Brettske, I., Gerber, S., Ginhart, A.W., Gross,

O., Grumann, S., Hermann, S., Jost, R., Konig, A., Liss, T.,

Lussmann, R., May, M., Nonhoff, B., Reichel, B., Strehlow,

R., Stamatakis, A., Stuckmann, N., Vilbig, A., Lenke, M.,

Ludwig, T., Bode, A., Schleifer, K.H., 2004. ARB: a

software environment for sequence data. Nucleic Acids

Res. 32, 1363–1371.

Lyra, C., Suomalainen, S., Gugger, M., Vezie, C., Sundman, P.,

Paulin, L., Sivonen, K., 2001. Molecular characterization of

planktic cyanobacteria of Anabaena, Aphanizomenon, Mi-

crocystis and Planktothrix genera. Int. J. Syst. Evol.

Microbiol. 51, 513–526.

Mallik, N., Rai, L.C., 1994. Removal of inorganic ions from

wastewaters by immobilized microalgae. World J. Micro-

biol. Biotech. 10, 439–443.

Martinelli, L.A., Victoria, R.L., Trivelin, P.C.O., Devol, A.H.,

Richey, J.E., 1992. 15N natural abundance in plants of the

Amazon River floodplain and potential atmospheric

N2 fixation. Oecologia 90, 591–596.

ARTICLE IN PRESSM.F. Fiore et al. / Water Research 39 (2005) 5017–50265026

Meeks, J.C., Castenholz, R.W., 1971. Growth and photosynth-

esis in an extreme thermophile, Synechococcus lividus

(Cyanophyta). Arch. Microbiol. 78, 25–41.

Meeks, J.C., Wycoff, K.L., Chapman, J.S., Enderlin, C.S.,

1983. Regulation of expression of nitrate and dinitrogen

assimilation by Anabaena species. Appl. Environ. Micro-

biol. 45, 1351–1359.

Meeks, J.C., Elhai, J., Thiel, T., Poots, M., Larimer, F.,

Lamerdin, J., Predki, P., Atlas, R., 2001. An overview of the

genome of Nostoc punctiforme, a multicellular, symbiotic

cyanobacterium. Photosynth. Res. 70, 85–106.

Neilan, B.A., Jacobs, D., Del Dot, T., Blackall, L.L., Hawkins,

P.R., Cox, P.T., Goodman, A.E., 1997. rRNA sequences

and evolutionary relationships among toxic and nontoxic

cyanobacteria of the genus Microcystis. Int. J. Syst.

Bacteriol. 47, 693–697.

Olson, J.B., Steppe, T.F., Litaker, R.W., Paerl, H.W., 1998.

N2-fixing microbial consortia associated with the ice cover

of lake Bonney, Antarctica. Microb. Ecol. 36, 231–238.

Postgate, J.R., 1969. Viable counts and viability. In: Norris,

J.R., Ribbons, D.W. (Eds.), Methods in Microbiology.

Academic Press, New York, pp. 611–662.

Rippka, R., Herdman, H., 1992. Pasteur Culture Collection of

Cyanobacteria: Catalogue and Taxonomic Handbook. I.

Catalogue of Strains. Institut Pasteur, Paris.

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., Waterbury, J.B., Stanier, R.Y., 1981. Isolation and

purification of cyanobacteria: some general principles. In:

Starr, M.P., Stolp, H., Truper, H.G., Balows, A., Schlegel,

H.G. (Eds.), The Prokaryotes. Springer, Berlin,

pp. 212–220.

Rodrigues, M.S., 1994. Biomass and phytoplankton production

in Camaleao lake (Marchantaria Island, Amazon). Ph.D.

thesis. Universidade Federal do Amazonas/Instituto Nacio-

nal de Pesquisas da Amazonia, Manaus.

Saville, B., Straus, N., Coleman, J.R., 1987. Contiguous

organization of nitrogenase genes in a heterocystous

cyanobacterium. Plant Physiol. 85, 26–29.

Singh, R.K., Stevens Jr., E., 1992. Cloning of the nifHDK genes

and their organisation in the heterocystous cyanobacterium

Mastigocladus laminosus. FEMS Microbiol. Lett. 94,

227–234.

Sioli, H., 1975. Tropical river: The Amazon. In: Whitton, B.A.

(Ed.), River Ecology. U.C. Press, Berkeley, pp. 461–487.

Solheim, B., Johanson, U., Callaghan, T.V., Lee, J.A., Gwynn-

Jones, D., Bjorn, L.O., 2002. The nitrogen fixation potential

of Arctic cryptogram species is influenced by enhanced UV-

B radiation. Oecologia 133, 90–93.

Stackebrandt, E., 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.

Swofford, D.L., 2000. PAUP*: phylogenetic analysis using

parsimony (and other methods), version 4.0. Sinauer

Associates, Sunderland, MA.

Thiery, I., Nicolas, L., Rippka, R., Marsac, N.T., 1991.

Selection of cyanobacteria isolated from mosquito breeding

sites as a potential food source for mosquito larvae. Appl.

Environ. Microbiol. 57, 1354–1359.

Turner, S., 1997. Molecular systematics of oxygenic photosyn-

thetic bacteria. Pl. Syst. Evol. [Suppl.] 11, 13–52.

Turner, S., Pryer, K.M., Miao, V.P.W., Palmer, J.D., 1999.

Investigating deep phylogenetic relationships among cyano-

bacteria and plastids by small subunit rRNA sequence

analysis. J. Eukaryot. Microbiol. 46, 327–338.

Uherkovich, G., 1976. Algen aus den Flussen Rio Negro und

Rio Tapajos. Amazoniana 5, 465–515.

Uherkovich, G., Franken, M., 1980. Aufwuchsalgen aus

zentralamazonischen Regenwaldbachen. Amazoniana 7,

49–79.

Uherkovich, G., Schmidt, G.W., 1974. Phytoplanktontaxa in

dem zentralamazonischen Schwemmlandsee Lago do Cas-

tanho. Amazoniana 5, 243–283.

Victoria, R.L., Martinelli, L.A., Richey, J.E., Devol, A.H.,

Forsberg, B.R., Ribeiro, M.N.G., 1989. Spatial and

temporal variations in soil chemistry on the Amazon

floodplain. GeoJournal 19, 45–52.

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.

Wilmotte, A., 1994. Molecular evolution and taxonomy of the

cyanobacteria. In: Bryant, D.A. (Ed.), The Molecular

Biology of Cyanobacteria. Kluwer Academic, Dordrecht,

pp. 1–25.

Wilmotte, A., 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. 2nd ed., vol. 1. Springer, New

York, pp. 487–493.

Wissmar, R.C., Richey, J.E., Stallard, R.F., Edmond, J.M.,

1981. Plankton metabolism and carbon processes in the

Amazon River, its tributaries, and floodplain waters, Peru-

Brazil, May–June 1977. Ecology 62, 1622–1633.

Wolk, C.P., Wojciuch, E., 1973. Simple methods for plating

single vegetative cells of, and for replica-plating, filamentous

blue-green algae. Arch. Mikrobiol. 91, 91–95.

Wolk, P., Ernst, A., Elhai, J., 1994. Heterocyst metabolism and

development. In: Bryant, D.A. (Ed.), The Molecular

Biology of Cyanobacteria. Kluwer, Dordrecht, pp. 769–823.