comparative analysis of bacterioplankton assemblages from karenia brevis bloom and nonbloom water...

R E S E A R C H A R T I C L E

Comparativeanalysis of bacterioplanktonassemblages fromKarenia brevis bloomandnonbloomwateron thewestFlorida shelf(GulfofMexico,USA)using16S rRNAgene clone librariesKelly L. Jones, Christina M. Mikulski, Amanda Barnhorst & Gregory J. Doucette

Marine Biotoxins Program, NOAA/National Ocean Service, Charleston, SC, USA

Correspondence: Gregory J. Doucette,

Marine Biotoxins Program, NOAA/National

Ocean Service, 219 Fort Johnson Road,

Charleston, SC 29412, USA. Tel.: 11 843 762

8528; fax: 11 843 762 8700; e-mail:

[email protected]

Received 18 September 2009; revised 26

February 2010; accepted 7 May 2010.

Final version published online 2 July 2010.

DOI:10.1111/j.1574-6941.2010.00914.x

Editor: Patricia Sobecky

Keywords

Karenia brevis; harmful algal bloom; 16S rRNA

gene clone libraries; UNIFRAC; HAB-associated

bacteria; dinoflagellate.

Abstract

The brevetoxin-producing dinoflagellate, Karenia brevis, forms nearly annual

blooms off the Florida west coast, severely impacting the region’s ecology and

economy. Bacteria are often cited as either promoting or interfering with the

development of algal blooms, and thus a detailed study of the bacterioplankton

assemblages associated with K. brevis was undertaken. We developed sixteen 16S

rRNA gene clone libraries from K. brevis bloom and adjacent nonbloom water to

determine the bacterial groups present and assess the influence of K. brevis cell

number and/or depth on bacterioplankton community composition. Most nota-

bly, bacterial groups such as Rhodobacterales (Alphaproteobacteria) and Cytopha-

gales/Sphingobacteriales (Bacteroidetes), reported previously to be associated with

other harmful algal species, were often abundant in the presence of K. brevis.

Cyanobacteria frequently dominated surface samples containing no detectable

K. brevis, consistent with earlier work suggesting that these photosynthetic

organisms may be important in promoting the proliferation of these blooms by

conditioning the water. Moreover, differences in the abundance/diversity of

traditionally more rare and often undocumented phylogenetic groups (e.g.

Betaproteobacteria, Deltaproteobacteria, Chloroflexus, Firmicutes) were apparent in

bloom vs. nonbloom water. This is the first study to document the association of

these phylogenetic groups with natural K. brevis populations and suggests a

potential role for these microorganisms in K. brevis bloom dynamics.

Introduction

An increase in the frequency and global distribution of

harmful algal blooms (HABs) (Hallegraeff, 1993) over the

past several decades has fostered research aimed at identify-

ing the factors that trigger or regulate the dynamics of these

events. Bacteria are increasingly cited as influencing the

development of HABs (Doucette, 1995; Doucette et al.,

1998; Skerratt et al., 2002; Grossart et al., 2005), due in part

to their role in nutrient cycling and the production of

vitamins stimulating algal growth (Haines & Guillard,

1974; Ferrier et al., 2002), as well as more species-specific

relationships, such as promoting cyst formation (Adachi

et al., 2003), exhibiting algicidal activity (Yoshinaga et al.,

1995, 1997; Imai, 1997; Mayali & Azam, 2004), and inhibit-

ing sexual reproduction (Sawayama et al., 1991).

Karenia brevis is an unarmored dinoflagellate that pro-

liferates annually in the Gulf of Mexico, often forming dense

blooms in west Florida shelf waters. Karenia brevis also

produces a suite of potent neurotoxins known as brevetox-

ins that adversely affect a variety of organisms ranging from

fish to marine mammals, as well as humans (Steidinger

et al., 1998; Landsberg, 2002; Backer et al., 2003; Van Dolah

et al., 2003). The economic costs of these blooms have been

estimated at upwards of USD $20 million per event, due to

their impacts on several important industries such as tour-

ism and seafood (Anderson et al., 2000). A detailed investi-

gation of the bacterioplankton assemblages associated with

K. brevis blooms could identify signature and/or dominant

bacterial species or groups that may influence the popula-

tion dynamics of these blooms and possibly be diagnostic of

the phase of bloom development.

FEMS Microbiol Ecol 73 (2010) 468–485Journal compilation c� 2010 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to original US government works

MIC

ROBI

OLO

GY

EC

OLO

GY

mailto:[email protected]

The relationship between dinoflagellates and their asso-

ciated bacteria is an easily observed, yet complex association

that can be characterized as beneficial or harmful (Doucette

et al., 1998), attached or free-living (Riemann & Winding,

2001), and specific or general (Garces et al., 2007; Sapp et al.,

2007). Previous studies describing microbial communities

frequently relied on traditional ‘nonmolecular’ techniques

(Buck & Pierce, 1989; Green et al., 2004; Wichels et al.,

2004). In addition, studies that used molecular methods

(Hold et al., 2001; Schafer et al., 2002; Jasti et al., 2005) often

used unialgal cultures in nutrient-enriched medium, a

situation that ‘is quite different from the environment

experienced by a member of a complex community sub-

jected to variable natural conditions’ (Jasti et al., 2005). The

use of emerging molecular methods has allowed a more

comprehensive, high-resolution approach to describing and

understanding the ecological significance of bacterioplank-

ton assemblages associated with algal blooms. Grossart et al.

(2005) and Pinhassi et al. (2004) reported that different

bacterioplankton consortia characterized various phyto-

plankton regimes, likely as a result of the different available

particulate and dissolved organic material associated with

distinct algal species. Moreover, Fandino et al. (2001)

observed that attached and free-living bacterial flora were

represented by characteristic phylogenetic groups during a

dinoflagellate bloom, consistent with the metabolic capabil-

ities of the respective groups.

In the specific case of K. brevis, the most comprehensive

study of the associated microbial community was conducted

almost two decades ago using classical bacteriological meth-

ods (Buck & Pierce, 1989). Since then, investigators have

used molecular techniques to classify and characterize

algicidal bacteria active against this dinoflagellate (Doucette

et al., 1999; Mayali & Doucette, 2002; Roth et al., 2008a, b)

and have documented a complex relationship between

algicidal bacteria and antagonistic (against the algicidal

bacteria), potentially co-occurring microorganisms (Roth

et al., 2008a). Moreover, the isolation of algicidal bacteria

from waters with and without detectable K. brevis cells

suggests that such strains are part of the ambient microbial

community and likely increase in abundance when exposed

to their target algal cells. Such relationships may be im-

portant in the regulation of bloom dynamics, but are mostly

unclear, given the limited availability of relevant field data.

In the present study, our aim was to conduct an extensive

molecular-based survey of the bacterioplankton associated

with K. brevis through the production of 16S rRNA gene

clone libraries to identify differences in the microbial

community structure as a function of dinoflagellate cell

concentration and location in the water column, and to

characterize the association between specific phylogenetic

bacterial groups and K. brevis. To our knowledge, this work

represents the first use of molecular methods to examine the

bacterioplankton assemblages associated with natural popu-

lations of K. brevis by using environmental samples.

Clone library descriptions have, until recently, been

limited to diversity measures (Alonso-Saez et al., 2007) and

evaluations of data were largely qualitative (Cottrell &

Kirchman, 2000; Green et al., 2004); however, several statis-

tical programs that allow objective comparisons between

multiple environmental samples have emerged to deal with

the growing quantity of 16S sequence data (Hur & Chun,

2004; Schloss et al., 2004; Eckburg et al., 2005; Lozupone &

Knight, 2005; Bik et al., 2006). One such program, UNIFRAC,

was used herein to analyze our large clone library datasets.

UNIFRAC utilizes a quantitative approach, choosing a

weighted analysis to measure differences in taxon abun-

dances within a community (i.e. sample) rather than just the

presence/absence of taxa (Lozupone et al., 2007). This is

particularly important for mixed environmental samples

such as ours, with a relatively high diversity, and where

certain bacterial groups may not disappear, but change in

abundance in response to conditions at different stages of

bloom development.

Materials and methods

Sample collection

Bacteria samples were collected onboard the R/V Suncoaster in

September 2001 as part of an ECOHAB ‘process’ cruise on the

west Florida shelf and are identified herein with a ‘Process A’

or ‘Pro A’ prefix. Sampling along transects inclu-

ded stations within multiple K. brevis bloom patches (identi-

fied by satellite-based chlorophyll a fluorescence and cell

counts) or outside of bloom conditions (Fig. 1). Several

samples were obtained from the same bloom patch over a 24-

h period following the deployment of a tracking drifter (see

Van Dolah et al., 2008). Direct light microscope K. brevis cell

counts were performed at every station. One-liter samples

were collected from the surface and various bottom depths at

each station using Niskin bottles attached to a CTD/rosette

sampler. The actual depths of the bottom samples ranged

from 3 to 44 m, and were generally taken within �1 m of the

true bottom. All Process A samples were passed through an

80mm Nitex prescreen before subsampling.

Samples for total DNA analysis were collected using 0.22-

mm Sterivex filters. Briefly, 60 mL of each sample were

passed through a Sterivex (Millipore) syringe filter, followed

by 6–10 mL of DNA lysis buffer (enough to displace the

seawater sample and fill the Sterivex cartridge). The filter

was then capped at the luer-lock end and frozen vertically

with the open end up at � 20 1C. Once frozen, the open end

was sealed with parafilm and samples were transported to

the laboratory on dry ice, where they were stored at � 20 1C

until extracted.

FEMS Microbiol Ecol 73 (2010) 468–485 Journal compilation c� 2010 Federation of European Microbiological SocietiesPublished by Blackwell Publishing Ltd. No claim to original US government works

469Bacterial community associated with a K. brevis bloom

Samples for total bacterial counts were collected on 0.22-

mm black polycarbonate filters using a filtration manifold.

One milliliter of whole water was placed in a 1.5-mL

eppendorf tube prefilled with formalin (final concentration

3.7%). After 20 min, the fixed sample was passed through

the filter, and then overlaid with 1 mL of a 50% EtOH/saline

solution for 1 min. The EtOH/saline solution was removed

via filtration and the air-dried filter was stored at � 20 1C

until analyzed.

DNA extraction and PCR

A lysozyme solution (50 mg mL�1) was added to the Sterivex

filters and incubated with rotation at 37 1C for 45 min.

Proteinase K (10 mg mL�1) and sodium dodecyl sulfate

(20% w/v) were added, and filters were incubated with

rotation at 55 1C for 2 h. The lysate was removed, extracted

twice with equal amounts of phenol–chloroform–isoamyl

alcohol (25 : 24 : 1; pH 8), and extracted once with chloro-

form–isoamyl alcohol (24 : 1) using standard procedures.

The final aqueous layer was run through a Millipore

centricon filter (Billerica, MA) to concentrate the extracted

DNA and eluted with TE buffer (1 mM EDTA, 10 mM Tris-

HCl, pH 7.5). DNA was stored at � 20 1C.

Bacterial 16S rRNA gene was amplified using primers

specific for Eubacteria: 341F (50-CCTACGGGAGGCAG

CAG-30) and 907R (50-CCGTCAATTCA/CTTTGAGTTT-

30) (Muyzer et al., 1993, 1995). The PCR reactions (100mL)

were carried out under the following conditions: 1� PCR

buffer (20 mM Tris-HCl, 50 mM KCl, pH 8.4), 5.0 mM

MgCl2, 0.5mM of each primer, 200mM of each deoxyribo-

nucleotide triphosphate, 4 mL of template, and 2.5 U of Taq

polymerase (Invitrogen, Carlsbad, CA). The touchdown

PCR profile consisted of an initial denaturation for 5 min at

94 1C, followed by 11 cycles of 94 1C for 1 min, 65 1C for

1 min, and 72 1C for 1 min, where annealing temperatures

decreased by one degree every two cycles from 65 to 55 1C

for 11 cycles. Subsequently, annealing temperatures re-

mained constant at 55 1C for 10 additional cycles, followed

by a final 72 1C extension for 7 min (Muyzer et al., 1993).

PCR products were separated on a 2% agarose gel

containing SYBRs Safe DNA stain (Molecular Probes,

Eugene, OR) by electrophoresis and visualized using a

FluorChemTM 8900 Imaging System (Alpha Innotech, San

Fig. 1. Locations of field sampling stations for

ECOHAB ‘process’ cruise on the west Florida

shelf, USA, with chlorophyll a data extracted

from the CoastWatch server: http://coastwatch.

pfel.noaa.gov collected by the SeaWiFS

Orbview-2 satellite at a spatial resolution of 0.11.

The actual value/parameter is Chlorophyll a

Anomaly for September 2001. The colors

represent the D from the mean Chl-a value

(mg m�3) for the month of Sept. 2001, with the

darkest red being 135.7 mg m�3 above the mean

and the darkest blue being � 23.7 mg m�3 below

the mean. Station numbers are associated with

their corresponding position by the solid lines.


470 K.L. Jones et al.

Leandro, CA) with the accompanying ALPHAEASE FC software

(version 4.0).

Total bacterial counts

Total bacteria were enumerated on the polycarbonate filters

(one filter per sample) by directly counting 40, 6-diamidino-

2-phenylindole (DAPI, 0.1mg mL�1)-stained cells using

epifluorescence microscopy (Porter & Feig, 1980). The

correlation between the total bacterial counts and K. brevis

counts as well as the significance of bacterial counts vs.

K. brevis cells mL�1 category (high/medium vs. low/zero, see

Results) were calculated using JMP (version 5.1.2, SAS

Institute, Cary, NC).

Generation of clone libraries

A TOPO TA cloning vector kit (Invitrogen) was used to

clone PCR products and a clone library was generated for

each extracted sample. Colonies containing the PCR-

inserted plasmid were resistant to kanamycin and ampicillin,

and were positively screened on Luria–Bertani agar supple-

mented with 50mg mL�1 kanamycin.

Ninety-six colonies were randomly selected from each

clone library and cultured in Terrific broth (Sigma, St. Louis,

MO) supplemented with 100 mg mL�1 ampicillin, again

screening for positive inserts. Plasmids were extracted from

samples using a Qiagen Plasmid Mini Kit and the Qiagen

BioRobot 9600 (Qiagen, Valencia, CA). Samples were sent to

Seqwright DNA Technology Services (Houston, TX) for

sequencing on an ABI PrismTM 3730xL DNA sequencer.

The 16S rRNA gene sequences were checked for base-call

accuracy and edited, when necessary, using CHROMAS LITE

(http://www.technelysium.com.au/chromas_lite.html) and

aligned using CLUSTALX (http://bips.u-strasbg.fr/fr/Documen

tation/ClustalX/), using default parameters. Sequence align-

ments were imported into BIOEDIT (http://www.mbio.ncsu.

edu/BioEdit/bioedit.html) for further analysis. Chimeric

sequences were detected using CHIMERA_CHECK [Ribosomal

Database Project(RDP), Cole et al., 2003] and BELLEROPHON

(Huber et al., 2004). All sequences were compared against

GenBank (Altschul et al., 1990) and RDP (Cole et al., 2003)

databases for the closest phylogenetic match, identified to

the lowest possible taxonomic level, and then submitted to

GenBank with accession numbers GQ915619–GQ916503.

Clone library sequences were grouped into operational

taxonomic units (OTUs) of varying evolutionary distances

and representative sequences were chosen for each OTU

(dereplicated) using FASTGROUP II (Yu et al., 2006). OTUs

were defined by evolutionary distances of 0.05 (95% se-

quence identity, genus level), 0.03 (97% sequence identity,

species level), and 0.0 (100% sequence identity) depending

on the analysis. The Shannon diversity index and the SChao-1

estimator were used to quantify microbial diversity and

estimate the probable number of phylotypes in the source

sample, respectively, and were calculated within FASTGROUP

II. Good’s coverage [C = 1� (n1/N)] estimated the percent

coverage of each combined library by estimating the propor-

tion of phylotypes in an infinitely sized library that are

represented in the reported library. All diversity and cover-

age calculations were based on an OTU classification of 97%

sequence identity.

Neighbor-joining (NJ) trees were constructed using the

PHYLIP 3.67 package (J. Felsenstein, University of Washington,

Seattle, WA) with the Kimura 2-parameter evolutionary dis-

tance model, and were supported by 1000 bootstrap replicates

and rooted by an Archaea species, Methanobrevibacter smithii.

Trees were based on three evolutionary distances (grouped at

0.05, 0.03, and 0.0 using FASTGROUP II) and used for various

phylogenetic and statistical analyses, as described below.

Statistical analysis

The UNIFRAC web interface (Lozupone et al., 2007) integrated

clone library sequences into datasets for comparison and

analysis using common statistical tools. Per program re-

quirements, all unique sequences were represented (100%

sequence identity) to retain maximum phylogenetic infor-

mation and were used to create both a phylogenetic tree and

an environmental file relating the quantity and diversity of

OTUs within libraries. All Process A samples were combined

for initial analysis, but smaller subgroups were later analyzed

based on the initial results. A UNIFRAC value (or metric) based

on the difference between environments (the branch length

in a tree that is unique to each environment) was deter-

mined for all pairs of environments to produce a distance

matrix from which hierarchical clustering (UPGMA, based

on pairwise UNIFRAC values) and Principal Coordinate Ana-

lysis (PCoA, a dimensionality reduction technique) were

performed. A weighted, quantitative UNIFRAC algorithm was

used to include information on both abundant and rare

sequences in each environment.

Significance tests using both the UNIFRAC value described

above (weighted) and a P test (Martin, 2002), which

determines the number of parsimony changes needed to

describe the distribution of sequences between different

environments (not weighted), were also performed. Unfor-

tunately, it is difficult to compare the many different

libraries (i.e. environments) generated herein using the

UNIFRAC and the P test because the P values must be modified

(Bonferroni correction) for each pairwise comparison. This

correction markedly reduces the P value required for

significance and reduces the likelihood of demonstrating a

statistical difference between environments. Similar limita-

tions have also been reported for other programs used to

compare bacterial assemblages (e.g. Hur & Chun, 2004;

Schloss et al., 2004). UNIFRAC’s cluster environments and



http://www.technelysium.com.au/chromas_lite.html

http://bips.u-strasbg.fr/fr/Documentation/ClustalX/

http://bips.u-strasbg.fr/fr/Documentation/ClustalX/

http://www.mbio.ncsu.edu/BioEdit/bioedit.html

http://www.mbio.ncsu.edu/BioEdit/bioedit.html

PCoA are more suitable for large datasets and were used to

cluster environments in order to determine which of our

libraries represented comparable assemblages.

Community structure and phylogenetic analysis

To visually represent differences in bacterial communities

among the 16 clone libraries, a heatmap was created using

the PHYLOTEMP (http://www.phylotemp.microeco.org) pro-

gram developed by Polson (2007), whereby sample libraries

were clustered based on relative abundance data using a

Bray–Curtis similarity measurement (x-axis, hierarchical

clustering). An accompanying NJ tree, with OTUs defined

at a 0.05 evolutionary distance (95% sequence identity), was

affixed to the y-axis and annotated for large phylogenetic

groups. One representative OTU was chosen for the NJ tree

for all sequences that belonged to the same OTU (95%

similarity) using FASTGROUP II. A matrix of OTU abundance

values (normalized for library size) served as an input to the

PHYLOTEMP microbial community analysis and visualization

tool, allowing for composition-based hierarchical clustering

of libraries (Polson, 2007).

For greater resolution of relationships on a species level,

an NJ phylogenetic tree was built based on an OTU evolu-

tionary distance of 0.03 (97% sequence identity, species

level) using the method described above. Furthermore, 33

reference sequences (http://www.ncbi.nlm.nih.gov/RefSeq)

were included to annotate important phylogenetic groups

within the tree, and an Archaea species, M. smithii, was used

for rooting. Representative sequences for OTUs were arbi-

trarily chosen when FASTGROUP II sorted all sequences within

the chosen evolutionary distance (in this case 0.03). The tree

was labeled and edited using MRENT (A. Zuccon & D. Zuccon,

http://www.mrent.org/, Swedish Museum of Natural His-

tory, Stockholm, Sweden) and bootstrap values over 500 (i.e.

50%) were maintained on their respective nodes.

Results

Sample collection data

Sixteen samples obtained at nine different station locations

were analyzed for this study: seven stations (Pro A 09, 23, 25,

27, 45, 52, and 61) with corresponding surface (1.7–0.4 m)

and bottom (44.0–3.0 m) samples and two stations (Pro A 01

and 34) with surface samples (1.0 and 0.4 m) only. Tempera-

ture and salinity for surface samples ranged from 28.14 to

29.91 1C and 31.69 to 35.55 p.s.u., respectively, whereas bot-

tom samples varied from 22.42 to 28.51 1C and 33.65 to

36.53 p.s.u., respectively. Karenia brevis cell concentrations

ranged from 0 to 4910 cells mL�1 and the total bacterial counts

varied from 8.10� 105 to 1.09� 107 cells mL�1 (Table 1).

The total bacterial counts and K. brevis cell counts were

correlated significantly (nonparametric Spearman’s r for non-

normally distributed data, r= 0.74, P = 0.0010, n = 16). The

total bacterial counts for samples with � 20 K. brevis

cells mL�1 were significantly lower than for samples with

Z500 K. brevis cells mL�1 (Wilcoxon, P = 0.0023, n = 16).

Clone library descriptions and sample diversity

After removal of chimeras, chloroplasts, and other erro-

neous sequence data, there were a total of 1059 clones from

Table 1. Details of sampling stations for ECOHAB ‘process’ cruise onboard R/V Suncoaster conducted on west Florida shelf during September 2001

Station

(Process A) Date Time

Location

(Latitude/

Longitude)

Sampling

depth (m)

Temperature

at sampling

depth ( 1C)

Salinity at

sampling

depth (p.s.u.)

Karenia brevis

(cells mL�1)

Total bacterial

counts (cells mL�1)

01 Surface 9/20/2001 15 : 48 27.54/82.80 1.0 28.33 34.04 878 1.09E107

09 Surface 9/21/2001 16 : 57 27.24/82.66 1.4 29.91 34.27 4910 6.00E106

09 Bottom 9/21/2001 16 : 57 27.24/82.66 11.5 27.50 34.58 1290 7.13E106

23 Surface 9/22/2001 13 : 29 26.78/82.78 0.9 28.91 35.47 1 1.20E106

23 Bottom 9/22/2001 13 : 29 26.78/82.78 23.5 28.18 35.41 1 8.10E105

25 Surface 9/22/2001 16 : 06 26.61/83.11 0.8 28.76 35.42 0 1.02E106

25 Bottom 9/22/2001 16 : 06 26.61/83.11 28.0 23.75 36.47 0 1.18E106

27 Surface 9/22/2001 18 : 01 26.54/83.24 1.7 29.67 35.55 512 2.40E106

27 Bottom 9/22/2001 18 : 01 26.54/83.24 44.0 22.42 36.53 19 1.30E106

34 Surface 9/23/2001 5 : 12 26.80/82.37 0.4 29.58 31.69 3 3.15E106

45 Surface 9/23/2001 18 : 49 26.34/82.50 1.6 29.24 35.32 0 3.45E106

45 Bottom 9/23/2001 18 : 49 26.34/82.50 20.0 28.51 35.57 0 2.71E106

52 Surface 9/24/2001 18 : 06 27.44/82.77 1.3 28.96 34.68 4088 4.05E106

52 Bottom 9/24/2001 18 : 06 27.44/82.77 6.0 27.91 34.60 2254 3.89E106

61 Surface 9/25/2001 15 : 50 27.62/82.83 0.8 28.14 33.66 860 4.65E106

61 Bottom 9/25/2001 15 : 50 27.62/82.83 3.0 28.14 33.65 980 5.30E106

Data for temperature and salinity at sampling depth, as well as Karenia brevis cell counts (cells mL�1) and the total bacterial counts (cells mL�1) for each

sample are provided.



http://www.phylotemp.microeco.org

http://www.ncbi.nlm.nih.gov/RefSeq

http://www.mrent.org/

our 16 clone libraries (Table 2). For subsequent phylogenetic

classification and analysis, clone libraries were organized

into OTUs based on an evolutionary distance of 0.05 (genus

level) and 0.03 (species level), resulting in 313 OTUs and 374

OTUs, respectively. An evolutionary distance of 0.0 was

required for UNIFRAC analysis (Table 2), which grouped

sequences into 886 OTUs.

Using the accepted evolutionary distance of 0.03 (Rosello-

Mora & Amann, 2001; Kemp & Aller, 2004), Shannon

diversity index values ranged from 3.31 to 3.97 and the

estimated number of phylotypes in each library ranged from

approximately 101 to 555 using the Schao-1 estimator. Good’s

coverage estimated a range of 19.7–49.3% (Table 2).

UNIFRAC analysis of clone libraries

A PCoA of UNIFRAC data performed on all 16 Process A clone

libraries separated samples along the first principal compo-

nent (P1) axis predominantly by K. brevis concentration;

samples containing � 20 cells mL�1 of K. brevis grouped

separately from samples with Z500 cells mL�1, accounting

for 23.5% of the variation among environments. Exceptions

were Pro A 27 surface and Pro A 34 surface (Fig. 2a, arrows)

samples. The UNIFRAC UPGMA hierarchical clustering analy-

sis revealed the same pattern of separation that occurred

along P1 (Fig. 2b). Although the second principal compo-

nent axis explained 19.8% of the variation, the samples were

not separated by any easily defined parameter. Based on the

PCoA and UPGMA results, water samples with K. brevis cell

concentrations of 0, 1–20, 500–999, and 4 1000 cells mL�1

were operationally defined as ‘zero’, ‘low’, ‘medium’, and

‘high’ categories or environments (no samples contained

21–499 cells mL�1), each representing four of the 16 total

samples.

To elucidate the effects of depth, station location (hor-

izontal spatial scale), or other factors independent of

K. brevis cell concentration, samples were divided into

subgroups and re-evaluated using PCoA (Fig. 2c–e). Note

that the color-coded system utilized on the heatmap (see

Fig. 3) was applied to PCoA analyses and defined further

with corresponding symbols. The influence of depth was

eliminated by removing bottom samples and applying a

PCoA analysis to all surface samples (0–4910 cells mL�1). A

pattern of clustering along K. brevis concentrations (Fig. 2c)

was still apparent for both axes, with the exception of two

samples along each axis. Principal components P1 and P2

described 32.9% and 21.2% of the variability among envir-

onments, respectively. Neither UNIFRAC nor the P test re-

vealed any statistically significant differences between these

nine samples when using corrected P values. PCoA analysis

of all bottom samples (0–2254 cells mL�1; 3–44 m depth)

suggested a grouping along depth profiles (Fig. 2d). Pro A 23

bottom (1 K. brevis cell mL�1) and Pro A 25 bottom (no

K. brevis) were two of the deepest samples and grouped

together on the positive x-axis (describing 39.8% of the

variability); however, Pro A 27 bottom (no K. brevis), which

was equally deep, grouped with the more shallow environ-

ments. The bottom samples did not separate by K. brevis cell

number along P1, but the cell number may have been a

contributing factor to the separation along P2. As above, no

Table 2. Number of clones, operational taxonomic units (OTUs), and diversity indices

Station (Process A)

Karenia brevis

(cells mL�1)

Number of

clones (sequences)�OTUw (97%

identity) (groups)

Shannon

diversity index

SChao-1

index

Good’s

coverage

Good’s coverage

combined (%)

01 Surface 878 75 61 3.97 555 26.1 74.40

09 Surface 4910 62 39 3.31 187 48.4

09 Bottom 1290 79 57 3.77 362.5 37.9

52 Surface 4088 72 45 3.41 177.5 49.3

52 Bottom 2254 59 43 3.56 185.7 38.9

61 Surface 860 61 53 3.88 455.7 19.7

61 Bottom 980 71 53 3.78 238.5 35.7

27 Surface 512 48 37 3.47 300 23.8

27 Bottom 19 78 50 3.63 234.8 48.7 75.60

23 Surface 1 77 54 3.6 439.9 38.2

23 Bottom 1 47 36 3.49 101.3 34.9

25 Surface 0 71 45 3.32 333.2 43.7

25 Bottom 0 54 38 3.46 112 46.3

34 Surface 3 59 49 3.82 176 29.3

45 Surface 0 67 44 3.53 193 40.9

45 Bottom 0 79 54 3.72 192 45.6

Good’s coverage [C = 1� (n1/N)] was calculated for each station as well as for combined stations (zero/low Karenia brevis cells and medium/high K.

brevis cells; see text for explanation).�Only represents clones with clean bacterial 16S rRNA gene; chloroplasts and vectors were removed from library total.wOTUs identified at 97% sequence identity.



significant differences were demonstrated among these

seven samples based on corrected P values.

Two additional analyses were performed on paired sam-

ples (surface and bottom) with similar K. brevis cell abun-

dance. Samples with 4 1000 K. brevis cells mL�1 (high), Pro

A 09 surface and bottom and Pro A 52 surface and bottom,

grouped along their shared water column (Fig. 2e). This

parameter described 54.4% of the variation among these

environments. Samples with no K. brevis cells (zero), Pro A

25 surface and bottom and Pro A 45 surface and bottom,

also grouped along their shared water column (data not

shown), with P1 describing 45.7% of the variation among

environments. A statistically significant difference between

Pro A 09 surface and Pro A 52 surface emerged with both

UNIFRAC and the P test using a corrected P value (Po 0.05).

Once all 16 sample libraries were grouped into the

four operationally defined environments based on K. brevis

cell concentration, the zero and low environments

were separated from the medium and high environments

along P1, which explained 61.3% of the variation (Fig. 2f).

Moreover, there was a statistically significant difference

between all pairwise comparisons (except for low vs. zero)

with the P test using 1000 permutations and a corrected

P value (Po 0.05). The UNIFRAC significance test showed a

difference only between the high vs. low and the high vs.

zero pairwise comparisons (Po 0.05). Finally, by further

combining all 16 libraries into just two environments

(medium/high and zero/low), statistically significant differ-

ences (Po 0.001) were obtained using both UNIFRAC and the

P test.

Fig. 2. PCoA and UPGMA hierarchical clustering generated using UNIFRAC with OTUs defined by an evolutionary distance of 0.0 (a–f). All PCoA graphs

show the first and the second principal coordinate (P1 and P2). (a) PCoA with all 16 Pro A samples. The solid arrow is pointing to station ProA 27 surface,

and the dashed arrow is pointing to station Pro A 34 surface. (b) UPGMA clustering of all 16 Pro A samples. (c) Surface samples only (n = 9). (d) Bottom

samples only (n = 7). (e) Samples with 4 1000 K. brevis cells mL�1 only. (f) Samples combined into four environments based on K. brevis cell

concentration: high, medium, low, and zero (see text for explanation).




The 16 Process A clone libraries or environments were

clustered according to their phylogenetic composition and

grouped along the x-axis of the PHYLOTEMP heatmap (Fig. 3).

A corresponding NJ tree comprising major phylogenetic

groups was aligned to the y-axis. The tree was generated

using an evolutionary distance of 0.05 (genus level), to keep

Fig. 3. Heatmap and accompanying cluster analysis (x-axis) of all 16 Pro A samples created using the PHYLOTEMP program (http://www.phylotemp.

microeco.org) based on an evolutionary distance of 0.05 (95% sequence identity). Large phylogenetic groups are annotated on the y-axis NJ tree.

Sample names are color coded according to K. brevis cells mL�1.



OTU size to a reasonable number and yet still allow the

visualization of phylogenetic shifts within respective envir-

onments. The scale bar represents percent abundances with-

in each environment and values ranged from 1.3 to 26.9%

abundance. The hierarchical clustering of libraries revealed

two main groups, the first consisting entirely of samples

with Z500 K. brevis cells mL�1 (medium, high) and

the second consisting mainly of libraries with � 20

K. brevis cells mL�1 (zero, low), with the exception of surface

samples Pro A 61 and Pro A 27. The samples did not cluster

by depth using PHYLOTEMP.

A phylogenetic tree was constructed from 374 OTUs

based on an evolutionary distance of 0.03 (species level)

and 33 GenBank reference sequences (Fig. 4). Because PCoA

analysis (UNIFRAC) and PHYLOTEMP clustering results indicated

that phylogenetic grouping was most heavily influenced by

K. brevis concentrations, all sequences from libraries with

� 20 K. brevis cells mL�1 (zero/low, eight samples) were

combined and compared with all sequences from libraries

having Z500 cells mL�1 (medium/high, eight samples) and

paired pie charts were generated to illustrate any trends.

These graphs showed the percent abundance of each major

phylogenetic group within the combined libraries and illu-

strated compositional differences among these groups (Fig. 4).

Good’s coverage for the combined libraries was

74.4–75.6% for the zero/low and medium/high, respectively

(Table 2).

Of the 374 total OTUs, 171 OTUs were classified as

Alpha-, Beta-, and Gammaproteobacteria (Fig. 4a). The Alpha-

proteobacteria comprised 95 OTUs and represented 34% of the

zero/low libraries and 39% of the medium/high libraries,

making it the most abundant group. In contrast, only 11

OTUs were classified as Betaproteobacteria, accounting for

0.58% of the zero/low libraries and 3% of the medium/high

libraries. Finally, 65 OTUs belonged to the Gammaproteobac-

teria, comprising 10% of the zero/low libraries and 12% of the

medium/high libraries. There were notable differences be-

tween libraries in the Rhodobacterales, the Burkholderales, and

the Pseudomonadales, representing the Alpha-, Beta-, and

Gammaproteobacteria, respectively (Fig. 4a).

A total of 71 OTUs were classified as Firmicutes (31

OTUs), Actinobacteria, Chloroflexus, Fibrobacteria/Acidobac-

teria (12 OTUs), and Betaproteobacteria (17 OTUs),

although the abundances of these groups were never above

9% and most were below 3.2% (Fig. 4b). The relatively

diverse Firmicutes comprised 9% of the zero/low libraries

and 2% of the medium/high libraries, with the abundance of

all represented orders showing considerable variation. The

total sequences within the Actinobacteria, Chloroflexus, and

Fibrobacteria/Acidobacteria groups accounted for 2.2% of

the zero/low libraries and 3.2% of the medium/high libraries

and, while not overly abundant, differences were evident

within the Fibrobacteria/Acidobacteria and Chloroflexus

groups. Deltaproteobacteria sequences also made up a small

percentage of the libraries (2.8% of the zero/low libraries

and 1.4% of the medium/high libraries), but differences in

abundance within the Desulfomonadales and Bdellovibriales

were apparent.

The Cyanobacteria was the third most abundant group,

with 23 OTUs comprising 20% of the zero/low libraries and

14% of the medium/high libraries. There were clear differ-

ences in abundance within the Prochlorales and Chroococ-

cales (Fig. 4b). The Cyanobacteria were particularly

prevalent in a few of the zero/low K. brevis libraries (Fig. 3).

The most diverse group was the Cytophaga–Flavobacteria

–Bacteroidetes (CFB), represented by 107 OTUs. The CFB

group was also the second most abundant, with 21% and

22% of the sequences from the zero/low and medium/high

libraries, respectively. In particular, there were differences in

abundance within the Flavobacteriales and Cytophagales/

Sphingobacteriales among the libraries (Fig. 4b).

Discussion

The 16 clone libraries developed during this study revealed

unique bacterioplankton assemblages with sampling loca-

tion, and provided the first documentation of natural

K. brevis populations associated with high levels of taxa

such as Rhodobacterales (Alphaproteobacteria) and Cytopha-

gales/Sphingobacteriales (Bacteroidetes) that co-occur fre-

quently with other HAB species. Cyanobacteria were

frequently dominant in individual surface samples with no

detectable K. brevis, consistent with suggestions that these

photosynthetic organisms may play an important role in

providing essential vitamins and releasing bound metals

preceding bloom initiation (Glibert et al., 2009). Moreover,

fluctuations in the abundance and diversity of traditionally

more rare bacterial phylogenetic groups were apparent

among samples with low K. brevis cell counts, providing an

insight into their potential contributions toward bloom

dynamics. Our ability to delineate bacterial assemblages

according to K. brevis abundance in a statistically sound

manner was achieved by applying UNIFRAC software (Lozu-

pone et al., 2007) to the sequence dataset, while a clear

visualization of these groupings was generated using the

recently developed PHYLOTEMP program (Polson, 2007). Such

trends in the phylogenetic composition of bacterioplankton

assemblages in relation to K. brevis cell concentration are

consistent with recent evidence linking phytoplankton and

microbial community structure in other marine environ-

ments (Pinhassi et al., 2004; Wichels et al., 2004; Jasti et al.,

2005; Garces et al., 2007; Sapp et al., 2007).

Total bacterial counts and diversity indices

The bacterial abundance in the study area clearly exhibited a

positive association with K. brevis cell concentration.



Samples containing Z500 K. brevis cells mL�1 (medium/high)

showed significantly higher bacterial concentrations than

those with � 20 cells mL�1 (zero/low), averaging 5.54� 106

(� 2.6� 106) and 1.85� 106 (� 1.1� 106) bacteria mL�1, re-

spectively. Indeed, Tobe et al. (2001) and Garces et al. (2007)

reported similar correlations between the abundance of other

harmful dinoflagellates of the genus Alexandrium and certain

bacterial groups. The range of total bacterial levels in our

samples (8.10� 105–1.09� 107 cells mL�1) was well within

that expected for oligotrophic and mesotrophic aquatic sys-

tems (Glockner et al., 1999; Sekiguchi et al., 2002; Rink et al.,

2007). Because ‘bloom’ and ‘nonbloom’ waters were targeted

specifically during this study, none of our samples contained

K. brevis cell concentrations between 21 and 499 cells mL�1.

We are thus unable to provide any information on bacterio-

plankton assemblages present under these conditions.

The Shannon diversity indices calculated for the 16 clone

libraries were relatively high (mean = 3.59; 0–5 scale), did

not vary considerably among samples (Table 2), and were

similar to the values reported previously for the Sargasso

and Baltic seas, as well as the Changjiang River in China

(Sekiguchi et al., 2002; Carlson et al., 2004; Woelfel et al.,

2007). Good’s coverage revealed low ranges of coverage for

the 16 individual samples, but the values were acceptable for

Fig. 4. NJ phylogenetic tree supported by 1000 bootstrap replicates and built based on an OTU evolutionary distance of 0.03 (species level). Bootstrap

values over 500 (i.e. over 50%) were maintained on their respective nodes. Each OTU is present only once in the tree and a representative clone for

sequences within the same OTU was chosen randomly to appear on the tree. The tree was rooted by an Archaea species and reference sequences (33;

boldface type) were included to annotate important phylogenetic groups within the tree. Pie graphs were generated for each major phylogenetic group

(at the order level) by separating samples into two groups, one with Z500 Karenia brevis cells mL�1 (medium/high samples) and the other containing

� 20 K. brevis cells mL�1 (zero/low), based on UNIFRAC PCoA results. (a) Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria groups

(b) Firmicutes, Actinobacteria, Chloroflexus, Fibrobacteria/Acidobacteria, Deltaproteobacteria, CFB, and Cyanobacteria groups.



most libraries (Dang & Lovell, 2000; Kemp & Aller, 2004;

Mou et al., 2007). The Schao-1 estimate indicated that certain

locations were considerably undersampled during this study

(Table 2); however, Schao-1 biases samples that have rare

phylotypes, which is a well-documented trend within aqua-

tic systems (Kemp & Aller, 2004) and was the case herein.

UNIFRAC analysis

In order to address the challenge of comparing bacterio-

plankton assemblages from diverse environments, we

applied UNIFRAC’s PCoA analysis to objectively assess phylo-

genetic trends among our 16 west Florida shelf samples.

Sequence similarities at 95% and 97% are conventionally

used for visually interpreting large data sets (see Figs 3 and

4); however, important phylogenetic information is fre-

quently obscured by these classifications. Small differences

in bacterial sequences, especially within the rapidly evolving

16S gene, can translate into large differences in functional

groups; therefore, even a 0.03 evolutionary distance (species

level) can result in a loss of information (Fuhrman &

Campbell, 1998; Ward, 1998; Pedros-Alio, 2006). UNIFRAC

defines sequence data by an evolutionary distance of 0.0 (i.e.

100% sequence similarity), thereby maximizing the phylo-

genetic information available in a clone library. When all 16

Process A clone libraries were compiled in one PCoA

Fig. 4. Continued.



analysis, samples were separated primarily according to

K. brevis cell number, with only two exceptions. Although

pairwise comparisons revealed no statistically significant

differences (based on corrected P values) due to the large

number of environments, the grouping of samples based on

K. brevis abundance by PCoA analysis was also supported by

UNIFRAC’s UPGMA clustering and PHYLOTEMP’s heatmap clus-

tering. These patterns were further supported by statistically

significant differences for 10 out of 12 pairwise comparisons

among the zero, low, medium, and high libraries grouped

based on K. brevis cell concentration (Po 0.05) and a highly

significant difference between the combined medium/high

and zero/low libraries (Po 0.001). The Good’s overage

result for the combined libraries revealed a substantial

increase in coverage over the individual libraries (�75% vs.

o 50% coverage).

We also explored potentially influential factors other than

K. brevis cell concentration by analyzing subgroups of the

Process A samples with PCoA (UNIFRAC). For example, when

comparing only surface collections, all except two samples

still collectively grouped according to K. brevis abundance.

Interestingly, analysis of only bottom samples suggested a

grouping based on depth, with two of the deepest samples

(Pro A 23 bottom and Pro A 25 bottom) grouping together

along the P1 axis; however, a third, equally deep sample (Pro

A 27B) was not included, likely due to the absence of

Bacillales (Firmicutes), which were abundant in the other

two samples.

When paired samples (surface and bottom from the same

location) with similar K. brevis cell numbers were compared,

station location seemed to be the influential factor along P1.

Common phylogenetic groups clustering according to sta-

tion location can also be seen in the PHYLOTEMP heatmap;

however, this relationship was most pronounced for Pro A

09 and Pro A 52 surface and bottom samples (4 1000

K. brevis cells mL�1). In both of these PCoA analyses,

separation along P2 could be due largely to one unique

group being abundant in each sample: Alphaproteobacteria

SAR11 group (Pro A 25 bottom), Cytophagales/Sphingobac-

teriales (Pro A 52 bottom) (see Fig. 3).

One limitation of PCoA analyses is that the specific factor

(i.e. environmental variable) explaining the variation and

clustering of samples cannot be readily identified. Interpreta-

tion of these data is dependent on the phylogenetic analyses of

the bacterial assemblages (e.g. PHYLOTEMP) as well as the PCoA

plots. Together, the P1 and P2 axes explained between 43%

and 82% of the sequence variation, with the lowest values

generated when all 16 samples were analyzed individually,

indicating that many factors are contributing to bacterial

diversity within a sample. However, combining samples into

subgroups revealed several statistically significant trends in

bacterial community composition that were likely masked by

an abundance of one species in a given location.

The fact that cluster analysis results from both UNIFRAC

and PHYLOTEMP programs were very similar, despite being

based on different evolutionary distances, highlights the

robustness of these tests and further supports grouping

samples according to K. brevis abundance. A notable excep-

tion to this pattern was Pro A 34 surface for the UNIFRAC and

PHYLOTEMP analyses. On closer examination, it was evident

that the high number of Rhodobacterales in the Pro A 34

surface (3 K. brevis cells mL�1) sample caused it to cluster

with the medium/high libraries, either in the same large

group (UNIFRAC) or as a sister group to one of the medium/

high libraries (PHYLOTEMP).


Percent abundance values among the major phylogenetic

groups well documented as important constituents of marine

bacterioplankton assemblages were surprisingly consistent re-

gardless of K. brevis concentration (Kirchman, 2002; Scanlan &

West, 2002; Buchan et al., 2005; Wagner-Dobler & Biebl, 2006),

including the Alphaproteobacteria (34% in zero/low, 39% in

medium/high), Gammaproteobacteria (10% in zero/low, 12%

in medium/high), and CFB (21% in zero/low, 22% in

medium/high). Rather, differences in the bacterial commu-

nity composition relative to K. brevis abundance were

observed mostly within bacterial orders, perhaps indicating

that certain representatives were favored over closely related

taxa due to a given metabolic attribute(s), allowing them to

proliferate under conditions associated with the presence/

absence of K. brevis cells.

For example, Roseobacter spp. (Alphaproteobacteria) were

most abundant in samples with higher concentrations of

K. brevis, and yet present in all of our samples. The

Alphaproteobacteria are ecologically significant organisms

commonly isolated from coastal waters and identified as

mostly free living (Buchan et al., 2005); however, the

ability of Roseobacter to degrade a number of dissolved

low-molecular-weight organics such as amino acids and di-

methylsulfoniopropionate, lignins, and other aromatic com-

pounds (Buchan et al., 2000; Wagner-Dobler & Biebl, 2006)

defines its functional role within the phycosphere of several

harmful algal species, such as Prorocentrum lima, Gymnodi-

nium catenatum, and Alexandrium spp. (Lafay et al., 1995;

Green et al., 2004; Jasti et al., 2005). In addition, while

Roseobacter can subsist in nutrient poor waters (Buchan

et al., 2005), members of this genus thrive in association

with dimethylsulfoniopropionate-producing phytoplankton

(Gonzalez et al., 2000; Malmstrom et al., 2004; Garces et al.,

2007) such as K. brevis, possibly reflecting a high density of

dimethylsulfoniopropionate transporters (Malmstrom et al.,

2004) and an ability to efficiently metabolize and/or assim-

ilate this compound. Conversely, the abundance/diversity of

SAR11 and other unclassified Alphaproteobacteria was the



highest in samples with low Roseobacter (and K. brevis)

concentrations. Roseobacter thus appears to be best adapted

to dominate the Alphaproteobacteria phylum under the nutri-

tional conditions associated with K. brevis blooms considered

to be in a ‘maintenance phase’ (growth rate = 0.10–0.16 day�1,

‘slow-growing bloom,’ van Dolah et al., 2008), and may

contribute toward sustaining such populations.

Members of the Gammaproteobacteria are commonly

isolated from HABs (Green et al., 2004; Amaro et al., 2005;

Pernthaler & Amann, 2005; Sapp et al., 2007), with the traits

of this group often defined as opportunistic, resilient, and

frequently displaying potent algicidal activity (Skerratt et al.,

2002). Several investigators have speculated that this appar-

ent dominance is misleading, biased by culturing techniques

that select for this group (Eilers et al., 2000; Kisand &

Wikner, 2003). In fact, by culturing ambient bacteria

associated with a K. brevis bloom, Buck & Pierce (1989)

observed a clear dominance of Alteromonas–Pseudomonas–

Vibrio (Gammaproteobacteria) within these communities,

and yet our libraries contained only three Vibrio clones. Our

data revealed that Gammaproteobacteria shared similar, but

low, relative abundances within our zero/low and medium/

high libraries (10% and 12%, respectively). Moreover, un-

like the other major phylogenetic groups, orders within the

Gammaproteobacteria phylum showed few distinct trends,

with one notable exception being the greater presence of

Pseudomonas within zero/low libraries. Our results are

similar to those of Garces et al. (2007), who reported that

Gammaproteobacteria were present in consistently low pro-

portions throughout the different phases of Alexandrium

spp. blooms in the NW Mediterranean Sea. However,

Hasegawa et al. (2007) encountered a dominant and diverse

group of Gammaproteobacteria attached to Alexandrium

fundyense cells in the Gulf of Maine, USA, that was not

present in the surrounding water (free-living or attached to

particles), suggesting a taxon-specific selective capability of

the phycosphere.

The similarities in Gammaproteobacteria representation

between the zero/low and the medium/high libraries docu-

mented herein may reflect the fact that material for the latter

was collected during the ‘maintenance phase’ of a K. brevis

bloom, as noted above (van Dolah et al., 2008). Gammapro-

teobacteria, particularly those within the Alteromonas–

Pseudomonas–Vibrio group, are often described as r-strate-

gists in marine systems (Weinbauer et al., 2006), colonizing

quickly and showing large fluctuations in abundance as

more specialized taxa begin to grow in response to the

establishment of unique, but relatively constant conditions.

It appears that the environmental conditions associated with

either the zero/low or the medium/high (i.e. maintenance

bloom phase) libraries were sufficiently transient to favor

the rapid colonization and growth of these Gammaproteo-

bacteria taxa.

Members of the CFB are a highly diverse group of bacteria

able to degrade high-molecular-weight organics such as

cellulose, chitin, and pectin (Kirchman, 2002), and are often

considered an important component of the microbial loop

associated with coastal phytoplankton blooms (Pinhassi

et al., 2004; Jasti et al., 2005; Rooney-Varga et al., 2005), sea

ice communities, freshwater biofilms, and marine sediments

(Kirchman, 2002). This group, represented mostly within

marine systems by the Cytophaga–Flavobacteria class (Cot-

trell & Kirchman, 2000; Fandino et al., 2001; Kirchman,

2002; Malmstrom et al., 2004; Celussi & Cataletto, 2007),

has also been proposed as a potentially important regulator

of HAB dynamics based on its algicidal activity and antag-

onism toward other bacteria (Doucette et al., 1999; Mayali &

Doucette, 2002; Roth et al., 2008a, b). Our libraries reflected

a similar pattern of diversity for both the zero/low and the

medium/high libraries, with the CFB comprising �20% of

the total abundance, but representing 107 OTUs – the most

of any phylogenetic group. Within the CFB, the Cytopha-

gales/Sphingobacteriales were most abundant in the med-

ium/high libraries, whereas the Flavobacteriales were more

common in the zero/low libraries. These two bacterial

orders are considered to have no major differences in

metabolic characteristics (Fandino et al., 2005), and yet our

data suggest that the former are more successful when

provided a source of K. brevis-derived organics. Several

bacteria algicidal toward K. brevis have been isolated:

Formosa spp. S03 (Roth et al., 2008b), Flavobacterium 5N-3

(Adachi et al., 2002), and Flavobacterium 41-DBG2 (Mayali

& Doucette, 2002). All of these taxa fall within the order

Flavobacteriales and, indeed, Formosa spp. S03 and

Flavobacterium 41-DBG2 were identified within our clone

libraries in the presence and absence of K. brevis cells,

although in very small abundances. While such bacteria are

apparently part of the ambient bacterial assemblage of the

west Florida shelf as suggested by Doucette et al. (1999),

their role in regulating K. brevis bloom dynamics remains

uncertain.

Cyanobacteria are ubiquitous throughout coastal waters,

and yet little is known about their relationship with HAB

species and how they interact with these organisms across

their similar niches in the upper water column. It has been

postulated that Trichodesmium (Order: Oscillatoriales) con-

ditions Gulf of Mexico waters for K. brevis blooms by fixing

elemental nitrogen following aeolian iron input, coupled

with the release of this nitrogen in dissolved organic form

available for uptake by the dinoflagellate (Lenes et al., 2001).

In addition, some dinoflagellates, including K. brevis, have

been shown to ingest Synechococcus as a potential prey

source in nutrient-poor environments (Jeong et al., 2005;

Glibert et al., 2009). Cyanobacteria composed 20% of our

zero/low libraries, and 14% of our medium/high libraries,

with Synechococcus (Order: Chroococcales) dominating in



both cases; however, Prochlorococcus (Order: Prochlorales)

was also common in the zero/low libraries. While abundant,

this clade was very low in diversity: only 23 OTUs (evolu-

tionary distance of 0.03) formed this entire group. The

highest representation of Cyanobacteria (38%) occurred in

surface samples Pro A 23 and 25, which contained little to

no K. brevis cells and were collected during the day. In

contrast, surface samples obtained at similar times (Pro A

52, 01, 61) with higher K. brevis concentrations showed far

fewer Cyanobacteria (average 14%), suggesting that the

presence of K. brevis and/or its associated bacterial assem-

blage at the very least do not promote the growth of ambient

cyanobacterial populations.

Our extensive clone libraries provided a unique opportu-

nity to examine not only phylogenetic groups commonly

associated with algal blooms (Alphaproteobacteria, Gamma-

proteobacteria, and CFB group) but also yielded an insight

into the dynamics of smaller phylogenetic groups whose

roles within marine systems are not as well understood.

Betaproteobacteria and Deltaproteobacteria are almost never

encountered in culture-based studies of coastal environ-

ments and are rarely detected even with molecular methods

[e.g. denaturing gradient gel electrophoresis (DGGE),

FISH]; nonetheless, both groups fluctuated in abundance

and diversity among our zero/low and medium/high li-

braries. Betaproteobacteria ranged fivefold from 0.59% to

3.03% in the zero/low vs. medium/high libraries, respec-

tively, and yet were represented by only one unclassified

taxon in the latter. Deltaproteobacteria was reduced in both

abundance and diversity in medium/high compared with

zero/low libraries. In addition, Chloroflexus, well known for

its association with cyanobacteria in microbial mats and

utilization of organics released by the latter (van der Meer

et al., 2003), was present only in medium/high libraries,

where Synechococcus was the dominant cyanobacterium.

The abundance of Fibrobacteria/Acidobacteria was the lowest

in these libraries, but has been reported previously from

other marine environments (e.g. Zaballos et al., 2006 and

references therein). The abundance of Firmicutes was also

lower in the medium/high vs. the zero/low libraries,

although the diversity was similar. Such patterns demon-

strate that microbial complexity potentially important in

regulating HAB dynamics is reflected in both abundant and

more rare phylogenetic groups, the latter possibly under-

estimated by other methods used to describe microbial

communities (i.e. DGGE, FISH).

Clone library data -- implications and next steps

While PCR-based clone libraries provide detailed informa-

tion on microbial community structure, they can be criti-

cized for their potential to bias interpretations (Farrelly

et al., 1995; Wintzingerode et al., 1997; Cottrell & Kirchman,

2000; Moeseneder et al., 2005). For example, Cottrell &

Kirchman (2000) found that PCR-based clone libraries

favored amplification of Alphaproteobacteria and selected

against the CFB group, leading them to conclude that

group-specific fluorescent probes (FISH) reflected the true

bacterial assemblage more accurately. However, whole-gen-

ome random shotgun sequencing in the Sargasso Sea

revealed Alphaproteobacteria (mainly SAR11) to be the

dominant bacterial group independent of PCR bias and in

the presence of CFB representatives (Venter et al., 2004).

Moreover, there are caveats regarding the hybridization

efficiency of the universal Eubacterial probe EUB338

(Bouvier & del Giorgio, 2003; Pernthaler & Amann, 2005),

limitations among available group-specific probes (Siyam-

balapitiya & Blackall, 2005; Alonso-Saez et al., 2007), and

the potential to obscure finer taxonomic details by focusing

on these broad phylogenetic groupings. Sequencing of

DGGE bands, considered to represent single phylotypes, is

another approach used commonly to assess bacterial com-

munity composition (Fandino et al., 2001; Rooney-Varga

et al., 2005; Sala et al., 2005; Celussi & Cataletto, 2007; Rink

et al., 2007). Nonetheless, cloning and sequencing of single

DGGE bands from our material yielded from 8 to 13 distinct

phylotypes (data not shown), consistent with the findings

reported by other investigators (Gasser, 1998; Jackson et al.,

2000; Sekiguchi et al., 2002; Kisand & Wikner, 2003) and

leading us to adopt the clone library strategy for assessing

phylogenetic diversity. Our clone libraries were generated

using primer sets shown to be less discriminative than others

(Muyzer et al., 1995), and yet still contained fragments

large enough to be phylogenetically informative for the 16

samples examined.

Clone libraries have been used extensively to describe

microbial communities as well as to infer ecological function

of the species represented; however, the targeted 16S rRNA

gene does not necessarily correspond with metabolic activ-

ity. Unfortunately, the sample collection/extraction proto-

cols we used did not allow for RNA extraction and thus a

more direct assessment of metabolic activity. Nonetheless,

recent reports suggest that taxa represented in 16S rRNA

gene libraries are similar to those described in correspond-

ing 16S rRNA libraries (Kamke et al., 2010; Logue &

Lindstrom, 2010), and that prevailing environmental condi-

tions select against bacteria that do not fill an ecological

niche in the existing community. Further, Arotsker et al.

(2009) noted that their 16S rRNA gene libraries reflected the

same seasonal shifts in Vibrio spp. associated with coral

black band disease as revealed by culturing Vibrio spp. from

the same corals used to generate their libraries. Thus, while

we acknowledge the benefits of rRNA-based analyses, we

would argue that our 16S rRNA gene clone libraries are a

reasonable representation of the metabolically active frac-

tion of the interrogated bacterioplankton assemblage.



Here, we report the most comprehensive molecular-based

survey to date of bacterioplankton assemblages associated

with the HAB species K. brevis. The predominant influence

of dinoflagellate cell abundance on microbial community

structure was demonstrated clearly, with few exceptions.

Positive associations between K. brevis and certain bacterial

groups were documented and suggest the possibility of a

functional significance based on known bacterial metabolic

traits. To what degree these associations are specific to

K. brevis vs. a more general response to the availability of

algal-derived organics, and whether the bacterial taxa in-

volved are consistent/predictable from year to year (e.g.

Fuhrman et al., 2006), are important questions to address

and should include a focus on the attached flora inhabiting

the phycosphere (e.g. Hasegawa et al., 2007). The potential

for such interactions to affect either positively or negatively

the in situ growth and/or the toxicity of this dinoflagellate,

especially if certain bacteria can be identified as critical

players in field populations, also needs to be examined

across a wider range of cell concentrations and growth rates

(i.e. bloom stages) than was possible during this study.

Obtaining this information represents a difficult challenge,

and yet is essential for defining the true role of bacteria in

regulating various phases of bloom development/decline

and assessing the feasibility of developing HAB management

strategies involving these microorganisms.

Acknowledgements

We wish to thank Dr S. Polson for the analysis of our data

using his recently developed PHYLOTEMP statistical tool to

produce our heatmap. We also thank Kimberly Nowocin for

her expert assistance in the design/editing of the figures and

Dr L. Kracker for providing the sample site map. In

addition, we acknowledge Lara Adams, Stephanie Brunelle,

and Laura Webster for providing valuable edits to this

manuscript, as well as two anonymous reviewers for their

insightful comments. Samples for this work were collected

during the ECOHAB Florida Regional Field Program

cruises. Support for this research was provided by the US

ECOHAB Program sponsored by National Oceanic and

Atmospheric Administration (NOAA), the US EPA, NSF,

NASA, and ONR, and by NOAA/NOS operational funds.

This is contribution number 335 from the US ECOHAB

Program.

Disclaimer

This publication does not constitute an endorsement of any

commercial product or intend to be an opinion beyond

scientific or other results obtained by the NOAA. No

reference shall be made to NOAA, or this publication

furnished by NOAA, to any advertising or sales promotion,

which would indicate or imply that NOAA recommends

or endorses any proprietary product mentioned herein, or

which has as its purpose an interest to cause the adver-

tised product to be used or purchased because of this

publication.

Authors’contribution

K.L.J. and C.M.M. contributed equally to this work.

References

Adachi M, Fukami K, Kondo R & Nishijima T (2002)

Identification of marine algicidal Flavobacterium sp. 5N-3

using multiple probes and whole-cell hybridization. Fish Sci

68: 713–720.

Adachi M, Kanno T, Okamoto R, Itakura S, Yamaguchi M &

Nishijima T (2003) Population structure of Alexandrium

(Dinophyceae) cyst formation-promoting bacteria in

Hiroshima Bay, Japan. Appl Environ Microb 69: 6560–6568.

Alonso-Saez L, Balague V, Sa EL, Sanchez O, Gonzalez JM,

Pinhassi J, Massana R, Pernthaler J, Pedros-Alio C & Gasol JM

(2007) Seasonality in bacterial diversity in north-west

Mediterranean coastal waters: assessment through clone

libraries, fingerprinting and FISH. FEMS Microbiol Ecol 60:

98–112.

Altschul SF, Gish W, Miller W, Myers EW & Lipman DJ (1990)

Basic local alignment search tool. J Mol Biol 215: 403–410.

Amaro AM, Fuentes MS, Ogalde SR, Venegas JA & Suarez-Isla BA

(2005) Identification and characterization of potentially algal-

lytic marine bacteria strongly associated with the toxic

dinoflagellate Alexandrium catenella. J Eukaryot Microbiol 52:

191–200.

Anderson DM, Kaoru Y & White AW (2000) Estimated annual

economic impacts from harmful algal blooms (HABs) in the

United States. Woods Hole Oceanogr. Inst. Tech. Rept., Woods

Hole, MA, WHOI-2000-11.

Arotsker L, Siboni N, Ben-Dov E, Kramarsky-Winter E, Loya Y &

Kushmaro A (2009) Vibrio sp. as a potentially important

member of the Black Band Disease (BBD) consortium in Favia

sp. corals. FEMS Microbiol Ecol 70: 515–524.

Backer LC, Fleming LE, Rowan A et al. (2003) Recreational

exposure to aerosolized brevetoxins during Florida red tide

events. Harmful Algae 2: 19–28.

Bik EM, Eckburg PB, Gill SR, Nelson KE, Purdom EA, Francois F,

Perez-Perez G, Blaser MJ & Relman DA (2006) Molecular

analysis of the bacterial microbiota in the human stomach.

P Natl Acad Sci USA 103: 732–737.

Bouvier T & del Giorgio PA (2003) Factors influencing the

detection of bacterial cells using fluorescence in situ

hybridization (FISH): a quantitative review of published

reports. FEMS Microbiol Ecol 44: 3–15.

Buchan A, Collier LS, Neidle EL & Moran MA (2000) Key

aromatic-ring-cleaving enzyme, protocatechuate 3,



4-dioxygenase, in the ecologically important marine

Roseobacter lineage. Appl Environ Microb 66: 4662–4672.

Buchan A, Gonzalez JM & Moran MA (2005) Overview of the

marine Roseobacter lineage. Appl Environ Microb 71:

5665–5677.

Buck JD & Pierce RH (1989) Bacteriological aspects of Florida red

tides: a revisit and newer observations. Estuar Coast Shelf S 29:

317–326.

Carlson CA, Giovannoni SJ, Hansell DA, Goldberg SJ, Parsons R

& Vergin K (2004) Interactions among dissolved organic

carbon, microbial processes, and community structure in the

mesopelagic zone of the northwestern Sargasso Sea. Limnol

Oceanogr 49: 1073–1083.

Celussi M & Cataletto B (2007) Annual dynamics of

bacterioplankton assemblages in the Gulf of Trieste (Northern

Adriatic Sea). Gene 406: 113–123.

Cole JR, Chai B, Marsh TL et al. (2003) The Ribosomal Database

Project (RDP-II): previewing a new autoaligner that allows

regular updates and the new prokaryotic taxonomy. Nucleic

Acids Res 31: 442–443.

Cottrell MT & Kirchman DL (2000) Community composition of

marine bacterioplankton determined by 16S rRNA gene clone

libraries and fluorescence in situ hybridization. Appl Environ

Microb 66: 5116–5122.

Dang H & Lovell CR (2000) Bacterial primary colonization and

early succession on surfaces in marine waters as determined by

amplified rRNA gene restriction analysis and sequence analysis

of 16S rRNA genes. Appl Environ Microb 66: 467–475.

Doucette GJ (1995) Interactions between bacterial and harmful

algae: a review. Nat Toxins 3: 65–74.

Doucette GJ, Kodama M, Franca S & Gallacher S (1998) Bacterial

interactions with harmful algal bloom species: bloom ecology,

toxigenesis, and cytology. Physiological Ecology of Harmful

Algal Blooms, Vol G 41 (Anderson D, Cembella AD &

Hallegraeff GM, eds), pp. 619–647. Springer-Verlag, Berlin.

Doucette GJ, McGovern ER & Babinchak JA (1999) Algicidal

bacteria active against Gymnodinium breve (Dinophyceae).

I. Bacterial isolation and characterization of killing activity.

J Phycol 35: 1447–1454.

Eckburg PB, Bik EM, Bernstein CN, Purdom E, Dethlefsen L,

Sargent M, Gill SR, Nelson KE & Relman DA (2005) Diversity

of the human intestinal microbial flora. Science 308:

1635–1638.

Eilers H, Pernthaler J, Glockner FO & Amann R (2000)

Culturability and in situ abundance of pelagic bacteria from

the North Sea. Appl Environ Microb 66: 3044–3051.

Fandino LB, Riemann L, Steward GF, Long RA & Azam F (2001)

Variations in bacterial community structure during a

dinoflagellate bloom analyzed by DGGE and 16S rDNA

sequencing. Aquat Microb Ecol 23: 119–130.

Fandino LB, Riemann L, Steward GF & Azam F (2005)

Population dynamics of Cytophaga–Flavobacteria during

marine phytoplankton blooms analyzed by real-time

quantitative PCR. Aquat Microb Ecol 40: 251–257.

Farrelly V, Rainey FA & Stackebrandt E (1995) Effect of genome

size and rrn gene copy number on PCR amplification of 16S

rRNA genes from a mixture of bacterial species. Appl Environ

Microb 61: 2798–2801.

Ferrier M, Martin JL & Rooney-Varga JN (2002) Stimulation of

Alexandrium fundyense growth by bacterial assemblages from

the Bay of Fundy. J Appl Microbiol 92: 706–716.

Fuhrman JA & Campbell L (1998) Microbial microdiversity.

Nature 393: 410–411.

Fuhrman JA, Hewson I, Schwalbach MS, Steele JA, Brown MV &

Naeem S (2006) Annually reoccurring bacterial communities

are predictable from ocean conditions. P Natl Acad Sci USA

103: 13104–13109.

Garces E, Vila M, Rene A, Alonso-Saez L, Angles S, Luglie A,

Maso M & Gasol JM (2007) Natural bacterioplankton

assemblage composition during blooms of Alexandrium spp.

(Dinophyceae) in NW Mediterranean coastal waters. Aquat

Microb Ecol 46: 55–70.

Gasser R (1998) What’s in that band? Int J Parasitol 28: 989–996.

Glibert PM, Burkholder JM, Kana TM, Alexander J, Skelton H &

Shilling C (2009) Grazing by Karenia brevis on Synechococcus

enhances its growth rate and may help to sustain blooms.

Aquat Microb Ecol 55: 17–30.

Glockner FO, Fuchs BM & Amann R (1999) Bacterioplankton

compositions of lakes and oceans: a first comparison based on

fluorescence in situ hybridization. Appl Environ Microb 65:

3721–3726.

Gonzalez JM, Simo R, Massana R, Covert JS, Casamayor EO,

Pedros-Alio C & Moran MA (2000) Bacterial community

structure associated with a dimethylsulfoniopropionate-

producing North Atlantic algal bloom. Appl Environ Microb

66: 4237–4246.

Green DH, Llewellyn LE, Negri AP, Blackburn SI & Bolch CJS

(2004) Phylogenetic and functional diversity of the cultivable

bacterial community associated with the paralytic shellfish

poisoning dinoflagellate Gymnodinium catenatum. FEMS

Microbiol Ecol 47: 345–357.

Grossart H-P, Levold F, Allgaier M, Simon M & Brinkhoff T

(2005) Marine diatom species harbour distinct bacterial

communities. Environ Microbiol 7: 860–873.

Haines KC & Guillard RRL (1974) Growth of vitamin B12-

requiring marine diatoms in mixed laboratory cultures with

vitamin B12-producing marine bacteria. J Phycol 10: 245–252.

Hallegraeff GM (1993) A review of harmful algal blooms and

their apparent global increase. Phycologia 32: 79–99.

Hasegawa Y, Martin JL, Giewat MW & Rooney-Varga JN (2007)

Microbial community diversity in the phycosphere of natural

populations of the toxic alga, Alexandrium fundyense. Environ

Microbiol 9: 3108–3121.

Hold GL, Smith EA, Rappe MS, Maas EW, Moore ERB, Stroempl

C, Stephen JR, Prosser JI, Birkbeck TH & Gallacher S (2001)

Characterisation of bacterial communities associated with

toxic and non-toxic dinoflagellates: Alexandrium spp. and

Scrippsiella trochoidea. FEMS Microbiol Ecol 37: 161–173.



Huber T, Faulkner G & Hugenholtz P (2004) Bellerophon; a

program to detect chimeric sequences in multiple sequence

alignments. Bioinformatics 20: 2317–2319.

Hur I & Chun J (2004) A method for comparing multiple

bacterial community structures from 16S rDNA clone library

sequences. J Microbiol 42: 9–13.

Imai I (1997) Algicidal ranges in killer bacteria of direct attack

type for marine phytoplankton. Bull Plankt Soc Jpn 44: 3–9.

Jackson CR, Roden EE & Churchill PF (2000) Denaturing

gradient gel electrophoresis can fail to separate 16S rDNA

fragments with multiple base differences. Mol Biol Today 1:

49–51.

Jasti S, Sieracki ME, Poulton NJ, Giewat MW & Rooney-Varga JN

(2005) Phylogenetic diversity and specificity of bacteria closely

associated with Alexandrium spp. and other phytoplankton.

Appl Environ Microb 71: 3483–3494.

Jeong HJ, Park JY, Nho JH, Park MO, Ha JH, Seong KA, Jeng C,

Seong CN, Lee KY & Yih WH (2005) Feeding by red-tide

dinoflagellates on the cyanobacterium Synechococcus. Aquat


Kamke J, Taylor MW & Schmitt S (2010) Activity profiles for

marine sponge-associated bacteria obtainted by 16S rRNA vs

16S rRNA gene comparisons. ISME J 4: 498–508.

Kemp PF & Aller JY (2004) Bacterial diversity in aquatic and

other environments: what 16S rDNA libraries can tell us.

FEMS Microbiol Ecol 47: 161–177.

Kirchman D (2002) The ecology of Cytophaga–Flavobacteria in

aquatic environments. FEMS Microbiol Ecol 39: 91–100.

Kisand V & Wikner J (2003) Combining culture-dependent and -

independent methodologies for estimation of richness of

estuarine bacterioplankton consuming riverine dissolved

organic matter. Appl Environ Microb 69: 3607–3616.

Lafay B, Ruimy R, Rausch de Traubenberg C, Breittmayer V,

Gauthier MJ & Christen R (1995) Roseobacter algicola sp. nov.,

a new marine bacterium isolated from the phycosphere of the

toxin-producing dinoflagellate Prorocentrum lima. Int J Syst

Bacteriol 45: 290–296.

Landsberg JH (2002) The effects of harmful algal blooms on

aquatic organisms. Rev Fish Sci 10: 113–390.

Lenes JM, Darrow BP, Cattrall C et al. (2001) Iron fertilization

and the Trichodesmium response on the West Florida Shelf.

Limnol Oceanogr 46: 1261–1270.

Logue JB & Lindstrom ES (2010) Species sorting affects

bacterioplankton community composition as determined by

16S rDNA and 16S rRNA fingerprints. ISME J 4: 729–738.

Lozupone C & Knight R (2005) UniFrac: a new phylogenetic

method for comparing microbial communities. Appl Environ

Microb 71: 8228–8235.

Lozupone CA, Hamady M, Kelley ST & Knight R (2007)

Quantitative and qualitative b diversity measures lead to

different insights into factors that structure microbial

communities. Appl Environ Microb 73: 1576–1585.

Malmstrom RR, Kiene RP & Kirchman DL (2004) Identification

and enumeration of bacteria assimilating

dimethylsulfoniopropionate (DMSP) in the North Atlantic

and Gulf of Mexico. Limnol Oceanogr 49: 597–606.

Martin AP (2002) Phylogenetic approaches for describing and

comparing the diversity of microbial communities. Appl

Environ Microbiol 68: 3673–3682.

Mayali X & Azam F (2004) Algicidal bacteria in the sea and their

impact on algal blooms. J Eukaryot Microbiol 51: 139–144.

Mayali X & Doucette GJ (2002) Microbial community

interactions and population dynamics of an algicidal

bacterium active against Karenia brevis (Dinophyceae).

Harmful Algae 1: 277–293.

Moeseneder MM, Arrieta JM & Herndl GJ (2005) A comparison

of DNA- and RNA-based clone libraries from the same marine

bacterioplankton community. FEMS Microbiol Ecol 51:

341–352.

Mou X, Hodson RE & Moran MA (2007) Bacterioplankton

assemblages transforming dissolved organic compounds in

coastal seawater. Environ Microbiol 9: 2025–2037.

Muyzer G, Waal EC & Uitterlinden AG (1993) Profiling of

complex microbial populations by denaturing gradient gel

electrophoresis analysis of polymerase chain reaction-

amplified genes coding for 16S rRNA. Appl Environ Microb 59:

695–700.

Muyzer G, Teske A, Wirsen CO & Jannasch HW (1995)

Phylogenetic relationships of species and their identification in

deep-sea hydrothermal vent samples by denaturing gradient

gel electrophoresis of 16S rDNA fragments. Arch Microbiol

164: 165–172.

Pedros-Alio C (2006) Marine microbial diversity: can it be

determined? Trends Microbiol 14: 257–263.

Pernthaler J & Amann R (2005) Fate of heterotrophic microbes in

pelagic habitats: focus on populations. Microbiol Mol Biol R 69:

440–461.

Pinhassi J, Sala MM, Havskum H, Peters F, Guadayol O, Malits A

& Marrase C (2004) Changes in bacterioplankton composition

under different phytoplankton regimens. Appl Environ Microb

70: 6753–6766.

Polson SW (2007) Comparative analysis of microbial community

structure associated with acroporid corals during a disease

outbreak in the Florida Reef Tract. PhD Thesis, Molecular and

Cellular Biology and Pathobiology Program, Marine

Biomedicine and Environmental Sciences, Medical University

of South Carolina, Charleston, SC.

Porter KG & Feig YS (1980) The use of DAPI for identifying and

counting aquatic microflora. Limnol Oceanogr 25: 943–948.

Riemann L & Winding A (2001) Community dynamics of free-

living and particle-associated bacterial assemblages during a

freshwater phytoplankton bloom. Microb Ecol 42: 274–285.

Rink B, Seeberger S, Martens T, Duerselen C-D, Simon M &

Brinkhoff T (2007) Effects of phytoplankton bloom in a

coastal ecosystem on the composition of bacterial

communities. Aquat Microb Ecol 48: 47–60.

Rooney-Varga JN, Giewat MW, Savin MC, Sood S, LeGresley M &

Martin JL (2005) Links between phytoplankton and bacterial



community dynamics in a coastal marine environment.


Rosello-Mora R & Amann R (2001) The species concept for

prokaryotes. FEMS Microbiol Rev 25: 39–67.

Roth PA, Mikulski CM & Doucette GJ (2008a) The influence of

microbial interactions on the susceptibility of Karenia species

to algicidal bacteria. Aquat Microb Ecol 50: 251–259.

Roth PB, Twiner MJ, Mikulski CM, Barnhorst AB & Doucette GJ

(2008b) Comparative analysis of two algicidal bacteria active

against the red tide dinoflagellate Karenia brevis. Harmful

Algae 7: 682–691.

Sala MM, Balague V, Pedros-Alio C, Massana R, Felipe J, Arin L,

Illoul H & Estrada M (2005) Phylogenetic and functional

diversity of bacterioplankton during Alexandrium spp.

blooms. FEMS Microbiol Ecol 54: 257–267.

Sapp M, Schwaderer AS, Wiltshire KH, Hoppe H-G, Gerdts G &

Wichels A (2007) Species-specific bacterial communities in the

phycosphere of microalgae? Microb Ecol 53: 683–699.

Sawayama S, Sako Y, Ishida Y, Niimura K, Abe A & Hiroishi S

(1991) Purification and structure determination of the

bacterial mating inhibitor for Chlamydomonas reinhardtii and

Alexandrium catenella. Nippon Suisan Gakk 57: 307–314.

Scanlan DJ & West NJ (2002) Molecular ecology of the marine

cyanobacterial genera Prochlorococcus and Synechococcus.


Schafer H, Abbas B, Witte H & Muyzer G (2002) Genetic diversity

of ‘satellite’ bacteria present in cultures of marine diatoms.


Schloss PD, Larget BR & Handelsman J (2004) Integration of

microbial ecology and statistics: a test to compare gene

libraries. Appl Environ Microb 70: 5485–5492.

Sekiguchi H, Watanabe M, Nakahara T, Xu B & Uchiyama H

(2002) Succession of bacterial community structure along the

Changjiang River determined by denaturing gradient gel

electrophoresis and clone library analysis. Appl Environ Microb

68: 5142–5150.

Siyambalapitiya N & Blackall LL (2005) Discrepancies in the

widely applied GAM42a fluorescence in situ hybridisation

probe for Gammaproteobacteria. FEMS Microbiol Lett 242:

367–373.

Skerratt JH, Bowman JP, Hallegraeff G, James S & Nichols PD

(2002) Algicidal bacteria associated with blooms of a toxic

dinoflagellate in a temperate Australian estuary. Mar Ecol Prog

Ser 244: 1–15.

Steidinger KA, Vargo GA, Tester PA & Tomas CR (1998) Bloom

dynamics and physiology of Gymnodinium breve with

emphasis on the Gulf of Mexico. Physiological Ecology of

Harmful Algal Blooms, Vol G 41 (Anderson D, Cembella AD &

Hallegraeff GM, eds), pp. 133–153. Springer-Verlag, Berlin.

Tobe K, Ferguson C, Kelly M, Gallacher S & Medlin LK (2001)

Seasonal occurrence at a Scottish PSP monitoring site of

purportedly toxic bacteria originally isolated from the toxic

dinoflagellate genus Alexandrium. Eur J Phycol 36: 243–256.

Van der Meer MTJ, Schouten S, Sinninghe Damste JSS, de Leeuw

JW & Ward DM (2003) Compound-specific isotopic

fractionation patterns suggest different carbon metabolisms

among Chloroflexus-like bacteria in hot-spring microbial mats.

Appl Environ Microb 69: 6000–6006.

Van Dolah FM, Doucette GJ, Gulland D, Rowles T & Bossart GD

(2003) Impacts of algal toxins on marine mammals. Toxicology

of Marine Mammals (Vos JG, Bossart GD, Fournier M &

O’Shea TJ, eds), pp. 247–269. Taylor & Francis, New York.

Van Dolah FM, Leighfield TA, Kamykowski D & Kirkpatrick GJ

(2008) Cell cycle behavior of laboratory and field populations

of the Florida red tide dinoflagellate, Karenia brevis. Cont Shelf

Res 28: 11–23.

Venter JC, Remington K, Heidelberg JF et al. (2004)

Environmental genome sequencing of the Sargasso Sea. Science

304: 66–74.

Wagner-Dobler I & Biebl H (2006) Environmental biology

of the marine Roseobacter lineage. Annu Rev Microbiol 60:

255–280.

Ward D (1998) A natural species concept for prokaryotes. Curr

Opin Microbiol 1: 271–277.

Weinbauer MG, Christen R & Hofle MG (2006) The response of

Vibrio- and Rhodobacter-related populations of the NW

Mediterranean Sea to additions of dissolved organic matter,

phages, or dilution. Microb Ecol 51: 336–344.

Wichels A, Hummert C, Elbrachter M, Luckas B, Schutt C &

Gerdts G (2004) Bacterial diversity in toxic Alexandrium

tamarense blooms off the Orkney Isles and the Firth of Forth.

Helgoland Mar Res 58: 93–103.

Wintzingerode FV, Gobel UB & Stackebrandt E (1997)

Determination of microbial diversity in environmental

samples: pitfalls of PCR-based rRNA analysis. FEMS Microbiol

Ecol 21: 213–229.

Woelfel J, Schumann R, Adler S, Hubener T & Karsten U (2007)

Diatoms inhabiting a wind flat of the Baltic Sea: species diversity

and seasonal succession. Estuar Coast Shelf S 75: 296–307.

Yoshinaga I, Kawai T, Takeuchi T & Ishida Y (1995) Distribution

and fluctuation of bacteria inhibiting the growth of a marine

red tide phytoplankton Gymnodinium mikimotoi in Tanabe

Bay. Fish Sci 61: 780–786.

Yoshinaga I, Kawai T & Ishida Y (1997) Analysis of algicidal

ranges of the bacteria killing the marine dinoflagellate

Gymnodinium mikimotoi isolated from Tanabe Bay, Wakayama

Pref., Japan. Fish Sci 63: 94–98.

Yu Y, Breitbart M, McNairnie P & Rohwer F (2006) FastGroupII:

a web-based bioinformatics platform for analyses of large 16S

rDNA libraries. BMC Bioinform 7: 57.

Zaballos M, Lopez-Lopez A, Ovreas L, Galan Bartual S, D’Auria

G, Alba JC, Legault B, Pushker R, Daae FL & Rodrıguez-

Valera F (2006) Comparison of prokaryotic diversity at

offshore oceanic locations reveals a different microbiota

in the Mediterranean Sea. FEMS Microbiol Ecol 56:

389–405.



comparative analysis of bacterioplankton assemblages from karenia brevis bloom and nonbloom water...

Documents