Phylogenetic and functional diversity of bacterioplankton during Alexandrium spp. blooms

Abstract

The phylogenetic and functional diversity of the bacterioplankton assemblage associated with blooms of toxic Alexandrium spp. was studied in three harbours of the NW Mediterranean. Denaturing gradient gel electrophoresis and DNA sequence analysis revealed the presence of a bacterium within the Roseobacter clade related to the presence of Alexandrium cells. Phylogenetic diversity was affected by the presence of Alexandrium spp., geographic situation and seasonality. In contrast, functional diversity, assessed with Biolog plates, was clearly affected by seasonality, but not by the presence of Alexandrium, indicating that the presence of the bacterium associated with the blooms was not enough to modify the metabolic pattern of the bacterioplankton assemblage.

Alexandrium, Bacterioplankton, Phylogenetic diversity, Functional diversity, BIOLOG, DGGE

1 Introduction

Marine dinoflagellates of the genus Alexandrium include a number of species that produce paralytic shellfish poisoning toxins (PSTs). Blooms of toxic species of Alexandrium are common in coastal areas and occasionally cause losses to tourism, fisheries and aquaculture. Interactions between algae and bacteria are among the natural factors that might play a role in harmful algal bloom (HAB) dynamics [1]. Reported interactions of bacteria with Alexandrium species include promotion of ecdysis [2] and cyst formation [3], and algicidal effects [4]. The potential contribution of bacteria to Alexandrium spp. toxicity has been actively studied in the recent years, but is still a matter of debate [5–9]. It has been shown that Proteobacteria, particularly those of the Roseobacte r clade, dominated the microbiota of A. catenella cultures [7] and, together with Alteromonas, of several other Alexandrium spp. cultures [10,11]. These interactions, observed in algal cultures, suggest a distinct composition of the bacterioplankton during HAB episodes. However, information on the composition of the bacterial assemblage during HABs is still very limited [12].

Changes in bacterioplankton structure may be important to carbon and nutrient flows, since the former is likely to affect functional diversity of the microbial community. Dinoflagellate blooms provide massive inputs of dissolved organic carbon (DOC) into coastal waters, triggering a response of the microbial community in cell number and activity [13]. However, in a study with algal cultures, Chen and Wangersky [14] observed that the DOC released from A. tamarense was not used by bacteria. Up to now, no information has been reported on bacterial utilization of carbon sources during Alexandrium spp. blooms.

Blooms of Alexandrium species are common in harbours along the NW Mediterranean [15]. We have studied phylogenetic and functional diversity of the bacterioplankton assemblage in three harbours of the Catalan coast, including two HAB episodes of Alexandrium catenella and A. minutum. In the present study, we address the following questions: (1) Is the phylogenetic composition of the bacterioplankton assemblage during HABs of the dinoflagellate A. catenella and A. minutum different compared to that in absence of HABs? (2) Is any species in the bacterioplankton assemblage reacting peculiar to harmful algal bloom situations? (3) Are the shifts in bacterial phylogenetic diversity associated with changes in functional diversity? Phylogenetic diversity was analyzed by denaturing gradient gel electrophoresis (DGGE) of 16S rRNA gene fragments. The most relevant bands in the gels were sequenced. Bacterial utilization of carbon sources was assessed by analyzing the community-level physiological profiles (CLPP) of the utilization of the 31 carbon sources in Biolog-ECO plates.

2 Materials and methods

2.1 Sampling and cultivation

Surface seawater was sampled from three harbours along the Catalan coast (NW Mediterranean). The Tarragona Harbour is located 100 km south of Barcelona and was sampled four times during a bloom of A. catenella in June 2001 and once in February 2003, when no toxic dinoflagellates were detected. The Arenys de Mar Harbour is located 40 km north of Barcelona and was sampled six times during a bloom of A. minutum in January–February 2002 and once in February 2003, when no toxic dinoflagellates were detected. Additionally, during 2001 and 2002, we took monthly samples from the Barcelona Harbour along two seasonal cycles in which toxic dinoflagellate blooms were not found. The Barcelona Harbour was used as a control of harbour without toxic dinoflagellates, and special attention was given to the periods coinciding with the bloom episodes in Arenys de Mar and Tarragona (June–July 2001 and January–February 2002). Phylogenetic analyses were performed on the bacterial assemblages of the harbour samples and on those found in the cultures of two dinoflagellate strains (A. minutum and A. catenella, kept in the culture collection of the Institut de Ciències del Mar). A. minutum was isolated from the Arenys de Mar Harbour in May 1996 and A. catenella from the Tarragona Harbour in June 1998. Cultures were grown in Guillards f/2 medium without silicate-enriched seawater, and maintained at 360–420 μE m⁻² s⁻¹ (12 h light:dark cycle) light intensity and a temperature of 25°C. Carbon substrate utilization was determined by means of BIOLOG plates in the subset of samples shown in Table 1, which will constitute the basic data used in the present paper.

1

Properties of samples used to investigate differences in phylogenetic and functional diversity of bacterioplankton

Harbour	Date	TEMP	CHLa	BACT	TIN	SRP	N:P	ALEX
T	08 Jun 2001	22.8	367	1.5 × 1010	21.6	1.07	20.2	3.20 × 107
T	12 Jun 2001	22.8	155	1.1 × 1010	17.0	0.13	136.2	1.50 × 107
T	15 Jun 2001	22.3	17	8.1 × 109	11.8	0.21	56.0	2.70 × 106
T	19 Jun 2001	21.7	20	5.4 × 109	7.2	0.18	40.0	ND
A	18 Jan 2002	12.5	10.3	8.6 × 108	−	−	−	7.28 × 105
A	21 Jan 2002	12.5	−	1.1 × 109	−	−	−	7.25 × 105
A	29 Jan 2002	12.5	13.4	3.7 × 109	1.6	0.25	6.2	2.55 × 106
A	01 Feb 2002	12.3	13.8	1.5 × 109	2.2	0.34	6.7	1.45 × 106
A	05 Feb 2002	12.5	8.5	1.7 × 109	3.0	0.18	16.8	1.85 × 106
A	08 Feb 2002	12.6	4.7	1.5 × 109	2.3	0.15	15.3	2.15 × 106
B	05 Jun 2001	20.4	7.5	2.3 × 109	5.9	0.17	35.9	ND
B	15 Jun 2001	21.6	37.2	9.0 × 109	2.5	0.02	123.4	ND
B	20 Jun 2001	22.9	1.4	2.3 × 109	18.2	1.14	16.0	ND
B	12 Jul 2001	24.9	1.6	3.6 × 109	1.7	0.02	84.7	ND
B	16 Jan 2002	11.0	1.9	2.0 × 109	50.5	1.76	28.7	ND
B	30 Jan 2002	12.8	11.7	3.9 × 109	20.6	0.53	39.2	ND
B	14 Feb 2002	12.5	3.9	1.8 × 109	23.2	1.09	21.4	ND
B	28 Feb 2002	14.2	8.3	1.9 × 109	22.6	1.03	22.0	ND

TEMP: temperature,°C; CHLa: chlorophyll a, μg l⁻¹; BACT: bacterial concentration, cells l⁻¹; TIN: total inorganic nitrogen, μM; SRP: soluble reactive phosphorus, μM; N:P: TIN/SRP ratio; ALEX: Alexandrium abundance, cells l⁻¹. ND: not detected. (–): not available.

2.2 Environmental parameters

Chlorophyll a concentration was determined fluorimetrically according to Yentsch and Menzel [16]. A volume of 100 ml was filtered through Whatman GF/F filters. The filters were kept frozen at −20°C for at least 2 h and pigments were extracted in 6 ml 90% acetone during 24 h at 4°C in the dark. Fluorescence of the extracts was measured in a Turner Designs fluorometer.

Concentrations of soluble reactive phosphorus (SRP), nitrate, nitrite and ammonia were measured according to Grasshoff et al. [17]. Total inorganic nitrogen (TIN) was calculated by adding the molarity of NO₃, NO₂ and NH₄, and the N:P ratio was calculated by dividing TIN values by SRP values.

Bacterial numbers were obtained by epifluorescence microscopy (Zeiss Axioplan, 100× objective) after fixation with 2% formaldehyde (final concentration) and DAPI (final concentration 0.5 mg ml⁻¹) staining of 5 ml seawater filtered onto 0.2 μm black polycarbonate filters [18].

For Alexandrium spp. counts, samples of 120 ml were fixed with formalin–hexamine solution (0.4%, final concentration). Phytoplankton cells were counted under the inverted microscope [19], using 50 cm³ settling chambers.

2.3 Phylogenetic diversity

Microbial biomass was collected on a 0.2 μm Sterivex filter unit (Durapore, Millipore) by filtering around 2–10 l of seawater, or 100 ml of the phytoplankton cultures. With a peristaltic pump, the samples were filtered in succession through a 3 μm polycarbonate filter (Millipore) and the Sterivex unit. The Sterivex units were filled up with 1.8 ml of lysis buffer and kept at −70°C. Microbial biomass was digested with lysozyme, proteinase K and sodium dodecyl sulphate, and the nucleic acids were extracted by phenolization and concentrated in a Centricon-100 (Millipore), as described in Schauer et al. [20]. The quality of the recovered DNA was checked by agarose gel electrophoresis. Nucleic acid extracts were stored at −70°C.

2.3.1 PCR and DGGE

One microliter of the DNA extract was used as template for polymerase chain reaction (PCR) amplification of bacterial 16S rRNA. We used the bacterial specific primer 358f (5′-CCT ACG GGA GGC AGC AG-3′) with a 40 bp GC-clamp, and the universal primer 907rM (5′-CCG TCA ATT CMT TTG AGT TT-3′), which amplify a 550 bp DNA fragment of bacterial 16S rRNA [21]. Reaction conditions and the PCR program used were identical to the ones described in Schauer et al. [20], with the only difference that the reverse primer used here had an extra ambiguity. PCR products were verified and quantified by agarose gel electrophoresis with a standard in the gel (Low DNA Mass Ladder, Invitrogen). DGGE [21] was carried out with a DGGE-2000 system (CBS Scientific Company) as described in Schauer et al. [20]. A 6% polyacrylamide gel with a gradient of DNA-denaturant agent was casted by mixing solutions of 40% and 80% denaturing agent (100% denaturing agent is 7 M urea and 40% deionized formamide). Eight hundred nanograms of PCR product were loaded for each sample and the gel was run at 100 V for 16 h at 60°C in 1× TAE buffer (40 mM Tris [pH 7.4], 20 mM sodium acetate, 1 mM EDTA). The gel was stained with the nucleic acid stain SybrGold (Molecular Probes) and visualized with UV in a Fluor-S MultiImager (Bio-Rad) with the Multi-Analyst software (Bio-Rad). High-resolution images (1312 × 1034 pixels, 12-bits dynamic range) were saved as computer files.

2.3.2 Sequencing and phylogenetic analysis

DGGE bands were excised from the gel, resuspended in 20 μl of MQ sterile water and left at 4°C overnight. An aliquot of 5 μl was used for PCR reamplification with the original primer set. Ten microlitres of the PCR product was loaded again in a DGGE gel to confirm the position of the bands. The remaining PCR product was purified with the PCR Purification kit (Qiagen) and sequenced with a Big Dye Terminator Cycle Sequencing kit (v.3) (PE Biosystems) and an ABI 3100 (Applied Biosystems) automated sequencer. The sequences obtained were compared with public DNA database sequences by using BLAST [22]. The partial sequences of 16S rRNA genes obtained in this study were deposited in GenBank with the Accession Nos. DQ008452–DQ008468

2.3.3 Quantitative analysis of DGGE

Digitized DGGE images were analyzed with the Diversity Database software (Bio-Rad). The software identifies the different bands, the contribution of each band to the total intensity in one lane and also bands which are in the same position. The sequenced bands also helped to identify these positions. This information was used to construct a matrix comparing lanes and positions, taking into account the relative contribution of each band (in percentage) to the total intensity of the lane. Using the software Statistica 6.0, a principal component analysis (PCA) was performed with the sub-matrix, comprising the intensity of the 12 bands that were present at least in 4 of the gels. PCA is a statistical tool that reduces the number of variables in a data matrix. The new variables are projected onto new axes or principal components which account for a certain amount of variance in the data. The first two principal components account for the highest variance and, therefore, it is their scores that are generally plotted. Similar samples appear close to each other in the PCA diagram.

2.4 Functional diversity

Biolog-EcoPlate™ (Biolog Inc.) microplates were used to determine differences in the functional diversity of the bacterioplankton assemblages. Biolog-EcoPlates contain 31 individual carbon sources with tetrazolium violet. Substrate catabolism is measured as the reduction of tetrazolium violet to a colored formazan that can be quantified photometrically. The resultant patterns are a function of the original assemblage inoculated into the wells. After inoculation of 150 μl in each well, samples were incubated at room temperature during six days and then kept at −20°C until further use. Previous tests showed that freezing the plate after incubation did not have an effect on the measured absorbance. Changes in color development were measured using a spectrophotometric microplate reader (ELX800 BIOTEK Instruments, Inc. Winooski, Vermont, USA) at a wavelength of 590 nm. In order to determine the differences among the community level physiological profiles (CCLP), principal component analysis (PCA) was carried out with the software Statistica v. 6.0, as described for the quantitative analysis of DGGE fingerprints. The mean color development of the three triplicate wells for each substrate was calculated and mean absorbance of the blanks (with only water) was subtracted. To reduce the influence of the color development rate among the plates, the absorbance value for each substrate was transformed as in Garland and Mills [23], by dividing the mean absorbance for each substrate by the total color development of the plate. Negative responses were set to zero before the statistical analysis.

3 Results

During the periods June 2001 and January-February 2002 a total of 18 samples was obtained from the surface of three harbours in the Catalan coast. The bloom in the Tarragona Harbour (T) started on June 8 (2001) and was characterized by very high chlorophyll a (367 μg l⁻¹) and A. catenella (3.2 × 10⁷ cells l⁻¹) concentrations. Both decreased with time and Alexandrium was not detected on the last day of sampling (June 19). Phytoplankton composition during the bloom was dominated exclusively by A. catenella, but on June 15, also some small dinoflagellates appeared, and on June 19, it consisted of diatoms mainly, especially Thalassionema nitzchioides and the dinoflagellate Prorocentrum triestinum. The bloom in the Arenys de Mar Harbour (A) took place in January–February 2002 and was less intense, with maximal chlorophyll a concentrations of 13.8 μg l⁻¹ and maximal A. minutum concentrations of 2.55 × 10⁶ cells l⁻¹. Although phytoplankton in the Arenys de Mar Harbour was dominated by A. minutum, also Chaetoceros spp. and Prorocentrum micans were found in high numbers. In the Barcelona Harbour (B), chlorophyll a concentrations during June–July (2001) and January–February (2002) varied between 1.4 and 37.2 μg l⁻¹, but Alexandrium cells were not detected (less than 20 cells l⁻¹). The phytoplankton community was dominated by the dinoflagellate Scripsiella spp. and the diatoms Chaetoceros spp. and Cerataulina pelagica. The three sites exhibited similar surface water temperatures during the sampling periods, 20.4–24.9°C in June and 11–14.2°C in January.

Bacterial concentration was higher in harbour T (5.4 × 10⁹–1.5 × 10¹⁰ cells l⁻¹) than in B or A (8.6 × 10⁸–3.9 × 10⁹ cells l⁻¹), with the exception of June 15 in B (9.0 × 10⁹ cells l⁻¹), when a very high chlorophyll a concentration was also recorded (37.2 μg l⁻¹).

The highest inorganic nutrient concentrations, soluble reactive phosphorus (SRP) and total inorganic nitrogen (TIN) were found in the Barcelona Harbour in January (SRP: 0.53–1.76 μM; TIN: 20.6–50.5 μM).

3.1 Phylogenetic diversity

For the establishment of phylogenetic diversity, two DGGE gels were run. The first one (Gel 1) included all the samples from the blooms in Arenys de Mar and Tarragona, and 6 non-bloom samples from the Barcelona Harbour taken in June 2001 and January 2002. A second DGGE gel (Gel 2) was run in order to compare the band patterns in periods without blooms, with the samples taken on February 3, 2003 from Tarragona (T) and Arenys de Mar (A; in two locations: a and b), and with samples from the cultures of A. catenella and A. minutum. For comparison with Gel 1, we added a sample taken on February 8, 2002 during the A minutum bloom in Arenys de Mar.

Analysis of Gel 1 (Fig. 1) resulted in a total of 145 detectable bands in 22 different positions. The number of bands per sample varied between 6 and 13 (mean 9). A total of 12 bands was excised and successfully sequenced (see number and position in Fig. 1). These bands accounted for a mean of 70% of the total band intensity in each lane. In order to determine their closest sequence in public databases, a BLAST search was used [22]. The sequence of the bands showed high similarity (99–100%) to those of clones obtained from marine environments (Table 2), and lower similarity with cultured strains (96.9–97.8%). Several bands migrated together in the gel: BH1, BH29 and BH30 (sequence similarities 99.6–99.8%); BH2, BH7, BH31 (sequence similarities above 99.8%); BH3 and BH27 (sequence similarity 98.9%) and BH4 and BH5 (sequence similarity 100%). These high similarities among sequenced bands (above 99.6%) resulted in 6 positions in the gel, three corresponding to Bacteroidetes, two corresponding to Alphaproteobacteria, both belonging to the Roseobacter clade, and one belonging to the Actinobacteria group.

1

DGGE gel showing PCR-amplified bacterial 16S rRNA gene fragments for the 19 samples taken during the periods June 2001 in the Tarragona and Barcelona Harbours, and January–February 2002 in the Arenys de Mar and Barcelona harbours. Horizontal filled bars indicate the presence, and empty bars the absence of Alexandrium spp. cells in the samples.

2

Relative intensity and phylogenetic affiliation of the sequences obtained in the DGGE bands shown in Figs. 1 and 2

Band no.	Closest match	GenBank Accession Number	Sequence similarity (%) (no. of bases)	Taxonomic group	Closest cultured match GenBank Accession Number Sequence similarity (%)
Harbours
BH1	Clone Flo-17	AY684339	99.4 (511)	Alphaproteobacteria	Roseobacter sp. AY536579 (97.2)
BH29	Clone Flo-17	AY684339	98.6 (495)	Alphaproteobacteria	Roseobacter sp. AY536559 (98.2)
BH30	Clone Flo-17	AY684339	98.8 (513)	Alphaproteobacteria	Roseobacter sp. AY536559 (97.3)
BH2	Clone NAC11–3	AF245632	100 (510)	Alphaproteobacteria	Roseobacter sp. AY136132 (96.9)
BH7	Clone NAC11-3	AF245632	100 (512)	Alphaproteobacteria	Roseobacter sp. AY136132 (96.9)
BH31	Clone NAC11–3	AF245632	99.8 (490)	Alphaproteobacteria	Roseobacter sp. AY136132 (96.9)
BH18	Clone NAC11–3	AF245632	100 (452)	Alphaproteobacteria	Roseobacter sp. AY136132 (96.7)
BH21	Clone NAC11–3	AF245632	100 (452)	Alphaproteobacteria	Roseobacter sp. AF124521 (96.9)
BH3	Clone BY-66	AJ298377	99.1 (457)	Bacteroidetes	Flavobacterium sp. AF493679 (92.8)
BH27	Clone BY-66	AJ298377	99.6 (479)	Bacteroidetes	Flavobacterium sp. AF493679 (91.6)
BH4	Clone BY-65	AJ298376	99.8 (508)	Bacteroidetes	Flexibacter sp. AB058905 (97.8)
BH5	Clone BY-65	AJ298376	99.8 (519)	Bacteroidetes	Flexibacter sp. AB058905 (96.9)
BH8	Clone PI-RT97	AY580355	99.8 (490)	Actinobacteria	Actinobacterium sp. AJ565416 (98.6)
BH16	Clone SL-6a	AY337037	92.5 (512)	Bacteroidetes	Cryomorphaceae bacterium AB176674 (91.5)
BH17	Clone EBAC36H07	AF268235	99.8 (506)	Prasinophycea chloroplast	Mantoniella squamata X90641 (99)
BH19	Clone SIC B8018	AF277557	99.4 (491)	Bacteroidetes	Flexibacter tractuosus AB078070 (92.5)
BH20	Clone MB11E04	AY033305	98.2 (453)	Bacteroidetes	Flavobacterium sp. AJ623288 (92.9)
Cultures
BH12	Cellulophaga sp. RE2–13	AF539758	98.7 (513)	Bacteroidetes	Cellulophaga sp. AF539758 (98.7)
BH25	Cellulophaga sp. RE2-13	AF539758	98.8 (415)	Bacteroidetes	Cellulophaga sp. AF539758 (98.8)
BH13	Clone D019	AF177558	97.9 (514)	Alphaproteobacteria	Roseobacter sp. AY332662 (96.9)
BH14	Paracoccus versutus	AY014174	97.0 (492)	Alphaproteobacteria	Paracoccus versutus AY014174 (97.0)
BH15	Clone BTM7	AY193269	97.8 (495)	Alphaproteobacteria	Ruegeria atlantica AF124521 (96.0)
BH22	Bacteroidetes bacterium KMM 3902	AY187689	93.6 (482)	Bacteroidetes	Algibacter lectus AY187689 (93.6)
BH23	Clones 68	AY171354	95.1 (490)	Bacteroidetes	Cellulophaga pacifica AB100842 (92.3)
BH24	Clone AP-27	AY145564	100 (488)	Alphaproteobacteria	Roseobacter sp. AY536559 (99.6)
BH26	Ruegeria sp. AS-36	AJ391197	100 (516)	Alphaproteobacteria	Ruegeria sp. AJ391197 (100)

In Gel 2, 15 bands were excised and successfully sequenced (Fig. 2, Table 2). Bands BH16 to BH21 were found in the samples obtained in Arenys de Mar and in Tarragona in February 2003. These bands corresponded to three Bacteroidetes and two Alphaproteobacteria. Band BH17, identified as a Prasynophycean chloroplast, was found only in both locations of Arenys de Mar, but not in Tarragona. Both Alphaproteobacteria were identical, and also to BH2 and BH7 in Gel 1 (Fig. 1 and Table 2). As in Fig. 1, this sequence was found in all the three field samples.

2

DGGE gel showing PCR-amplified bacterial 16S rRNA gene fragments for the samples taken in February 2003 in the Arenys de Mar Harbour (two locations, a and b) and the Tarragona Harbour, for two algal cultures: A. minutum and A. catenella. The sample from Arenys de Mar taken on the 8th of February 2002 was also included for comparison with the gel in Fig. 1. Horizontal filled bars indicate the presence, and empty bars the absence of Alexandrium spp. cells in the environmental samples. Dashed bars indicate samples obtained from algal cultures.

Different profiles were detected for each dinoflagellate culture. The A. minutum culture presented four bands. Two bands showed the closest similarity with cultured species: Cellulophaga sp. (98.7%) and Paracoccus versutus (97.0%) (belonging to Bacteroidetes and Alphaproteobacteria, respectively) and the other two showed the closest match with uncultured clones of Alphaproteobacteria and the closest cultured matches with Roseobacter sp. (96.9%) and Ruegeria atlantica (96.0%) (see accession numbers of these matches in Table 2). Five bands were sequenced from the A.catenella lane and three of them showed the closest match with cultured bacteria: Algibacter lectus (93.6%), Cellulophaga sp. (98.8%) and Ruegeria sp. (100%). The other two showed the highest match with uncultured species and corresponded to a Bacteroidetes (95.1%) and Alphaproteobacteria (100%), with the closest cultured match with Cellulophaga pacifica (92.3%) and Roseobacter sp. (99.6%), respectively.

Two main positions dominated the gels and together contributed 11–62% (mean 32%) of the total intensity of the lanes in Fig. 1. They were sequenced three times from different lanes and since similarity among the bands was above 99.6%, we consider them as identical bands. Hereafter, they will be called BH1 (including BH1, BH29 and BH30) and BH2 (including BH2, BH7, BH31) (Fig. 1). Both belonged to the Alphaproteobacteria and showed the closest match with Roseobacter sp. (see accession numbers in Table 2), but similarity between them was low, 96.1%. Band BH1 presented the highest similarity with clone Flo-17, while BH2 showed the best match (99.8–100%) with clone NAC11–3, originally isolated from the Atlantic ocean during an algal bloom [24] and with a widespread presence in marine environments. Both bands showed a different pattern of intensities in the gel: while BH2 could be found in all harbour samples, BH1 was more evident in samples with presence of Alexandrium cells (Figs. 1 and 2). The ratio between the relative intensity of BH1 vs. BH2 (Fig. 3) was clearly higher in Tarragona in the presence (5.0–2.0) than in the absence (0.4–0) of A. catenella, and in Arenys de Mar it was higher in the presence (2.0–0.4) than in the absence of A. minutum. In Barcelona, where HABs were not recorded, the ratio between both bands was lower than in the harbours with toxic dinoflagellates (mean = 0.11), with a maximum value of 0.4 on the 4th of May. The ratio for most of the samples was <0.2.

3

Ratio between relative intensity of bands BH1 and BH2 in the Tarragona, Arenys de Mar and Barcelona Harbours. Arrows indicate the absence of band BH1 in the gel. Horizontal filled bars indicate the presence, and empty bars the absence of Alexandrium spp. cells in the samples. The dashed line shows a value of 0.4 for the ratio.

The ratio BH1 vs. BH2 increased with increasing chlorophyll a concentration (Fig. 4) in the three harbours sampled. Generally, chlorophyll a was higher during the Alexandrium blooms than in absence of blooms. However, for the range of chlorophyll a concentrations between 1.65 μg l⁻¹ (log = 0.5) and 4.95 μg l⁻¹ (log = 1.6), that comprised samples with and without Alexandrium cells from the three harbours, the ratio BH1/BH2 was higher in the presence of Alexandrium (n= 6, mean =−0.128) than in its absence (n= 17, mean =−1.06). A t-test revealed that this difference was significant (t-value =−3.57, p= 0.002).

4

Log–log plot of the ratio of the relative intensity of bands BH1/BH2 vs. chlorophyll a concentration in the three harbours: ◻: Tarragona; ∘: Arenys de Mar; △: Barcelona. Filled symbols: Alexandrium spp. presence; Empty symbols: Alexandrium spp. absence.

The relationships among the genetic composition of the bacterial assemblages during the bloom periods of June 2001 and January 2002 (DGGE in Fig. 1) were assessed with a principal component analysis of the data matrix formed by the relative intensity of the 12 bands that were present at least in four of the gels. PC1 explained 33%, and PC2 19.4% of the variance (Fig. 5(a)). Samples taken in the same harbour and period of sampling were included in closed lines drawn by eye. As can be seen in Fig. 5(a), the samples taken in Tarragona showed a tendency to more negative scores on PC1 than the samples from Arenys or from Barcelona. This can be explained by the high intensity in the Tarragona samples (specially in those from June 13 and 15) of band BH1 and another unsequenced band, which present negative loadings on PC1 (Table 3). In contrast, Band BH5, with positive loading on PC1, was stronger in the Arenys and Barcelona (winter and June) samples. The stronger intensity of band BH1 in bloom samples (negative PC1 loading) and the positive loadings on PC2 of bands BH8 and BH4 in winter samples (Table 3) explain the grouping of the Arenys bloom samples, which are situated in the upper left part of the graph, in a zone of relatively low scores on PC1 (<1.2) and positive scores on PC2. Generally, samples taken in June were situated below samples taken in January, what could be explained by three bands that appeared associated with the June samples: two unsequenced bands and BH3.

5

Plot of the scores for PC1 against PC2 of the data obtained during the periods June–July 2001 and January–February 2002 in the three harbours for: (a) matrix of relative intensity of bands obtained from the DGGE gel in Fig. 1. (b) Matrix of relative absorbance of carbon sources in the Biolog-Eco plate. Lines were drawn by eye in order to show the grouping of the data according to the period and harbour sampled.

3

Correlation coefficients (loadings) of selected DGGE bands and BIOLOG substrates with the two principal components (PC1 and PC2) of the respective principal component analyses (sampling dates listed in Table 2; scores shown in Fig. 5)

Variable	PC1	PC2
DGGE band
BH5	0.60	−0.20
BH1	−0.76	0.20
BH4	0.36	0.58
BH8	0.16	0.56
BH3	−0.46	−0.55
Biolog substrate
l-Asparagine	0.83	−0.17
l-Phenylalanine	0.67	0.13
γ-Hydroxybutyric acid	0.66	0.48
α-Ketobutyric acid	0.56	−0.15
Phenylethylamine	0.55	0.21
Glycyl-l-Glutamic acid	0.55	−0.46
l-Arginine	0.54	−0.03
l-Serine	0.53	−0.53
d-Galacturonic acid	0.51	−0.47
d-Galactonic acid γ-lactone	−0.68	−0.00
N-acetyl-d-glucosamine	−0.59	0.21
α-Cyclodextrine	−0.56	−0.57
Pyruvic acid methyl ester	−0.28	0.85
2-Hydroxy benzoic acid	0.03	0.81
Itaconic acid	0.42	0.55

Only variables (DGGE bands or BIOLOG substrates) with loadings of absolute value >0.5 (highlighted in bold) on one of the principal components are shown.

3.2 Functional diversity

PCA was carried out on the 18 transformed Biolog data sets from the three harbours in the two periods of study: June 2001 and January 2002 (Fig. 5(b)). PC1 accounted for 19% and PC2 for 15% of the variance in the data. The samples showed a distinct separation for sampling period, but not for harbours. In order to relate the utilization of individual carbon sources to the differences in the sole carbon source utilization patterns, correlations between the substrate variables and the PCs were examined (Table 3). The higher the correlation, the more important that substrate was for differentiating among samples, without implying that the carbon source was highly utilized. The most important substrates for differentiating between sole carbon source utilization were those with factor loading >0.5.

An examination of Fig. 5(b) and Table 3 shows which carbon substrates were most important for the differentiation between bacterial assemblages. PC1 clearly differentiated the data according to the period of sampling. Samples taken in January always had higher PC1 values than samples taken in June. The substrates that positively correlated with PC1, i.e. that were more used in January than in June, were five amino acids (l-asparagine, l-phenylalanine, glycyl-l-glutamic acid, l-arginine and l-serine), three carboxylic acids (γ-hydroxybutyric acid, α-ketobutyric acid and d-galacturonic acid) and one amine (phenylethylamine). Only three carbon sources showed negative correlation with PC1, i.e. higher utilization in June than in January: two carbohydrates (d-galactonic acid, γ-lactone and N-acetyl-d-glucosamine) and the polymer α-cyclodextrine. This indicates that the bacterial assemblage in the samples taken in January used relatively more amino acids than the assemblage sampled in June.

PC2 did not allow a clear separation of the data, although generally most of the samples from Tarragona and Arenys were located below those from Barcelona. The carbon sources that showed positive correlation with PC2 were a pyruvic acid methyl ester, 2-hydroxy benzoic acid and itaconic acid, and negative correlation was observed with α-cyclodextrine and l-serine. No correlation was found between PC2 and any of the environmental parameters measured.

4 Discussion

4.1 Phylogenetic diversity

From the three harbours sampled in this study, the phylogenetic composition of bacterioplankton was different in the presence and absence of blooms of the toxic dinoflagellates A. catenella and A. minutum. Overall, the taxonomic composition of bacterioplankton in the harbours was dominated by the Bacteroidetes, the Alpha subclass of Proteobacteria and the Actinobacteria. DGGE gels revealed that two bands, both related to Roseobacter, accounted for most of the intensity in each lane (11–62%, mean 32%). The significance of the Roseobacter clade in marine biogeochemical processes has been recognized [24,25] and it is considered to be very diverse, both physiologically and geographically. The genus Roseobacter has been found in bacterial isolates fromAlexandrium spp. [10,11] and from A. catenella[7] in cultures. In the environment, DGGE analysis has reported the presence of Roseobacter spp. during blooms of A. tamarense off the Orkney Isles [12].

The two main bands (BH1 and BH2) in the harbour samples, both related to Roseobacter, showed low mutual similarity (96.1%) and a different pattern of intensities in the DGGE gels. Band BH2 appeared in all the field samples, while the relative intensity of BH1 was higher in samples where Alexandrium blooms were detected. The ratio between the relative intensity of these two bands (BH1 vs. BH2) was significantly higher in samples with (5.0–0.4) than without (0.4–0) Alexandrium cells.

Band BH1, which appeared associated to the presence of Alexandrium cells, had clone Flo-17 as closest match. This clone was isolated during the decay of a Skeletonema (non-toxic diatom) bloom in a mesocosm experiment [26]. This could indicate a possible relation between the chlorophyll a concentration and the presence of the band. However, in the range of chlorophyll a concentrations that comprised both samples with and without Alexandrium (1.65–4.95 μg l⁻¹), the ratio BH1/BH2 was significantly higher in the presence of Alexandrium cells. This would support the conclusion that, although band BH1 was not exclusively associated with Alexandrium spp. blooms, it was far more important during these blooms. This relationship has also been suggested for the bacteria isolated during a toxic bloom of A. tamarense[12]. Moreover, 16S rRNA-targeted probes for bacteria, isolated during an A. tamarense bloom, detected Roseobacter spp. in high numbers in the water column when Alexandrium spp. cells were both present and absent in the same area [27]. The potential contribution of the bacteria associated with Alexandrium spp. to algal toxicity is an open question. The only study performed in situ during A. tamarense blooms [12] revealed no contribution to toxicity by any of the isolated bacteria, among which several members of the Roseobacter clade were found.

Band BH1 was not detected in the bacterial assemblage of our Alexandrium cultures. Sequence analysis of excised DGGE bands showed a completely different bacterial composition in the cultures than in the harbours, since none of the bands in the cultures were found in the harbours. Most of the bands in the cultures had closest match with microorganisms with known surface-associated life histories (e.g. the Bacteroidetes phylum, and some members of the Roseobacter clade including Ruegeria algicola) as Hold et al. [10] found in Alexandrium cultures. This deserves special attention, since toxigenic bacteria apparently need adhesion to algal cells to exhibit toxicity [28]. However, the conditions utilized for the algal cultures, such as high N and P concentrations, and especially the abiotic forces exerted by the containers may be, at least partially, responsible for the different composition of the bacterial assemblage in the environment and in the algal cultures.

The PCA analysis revealed a clear influence of the presence of toxic dinoflagellates on the composition of the bacterial assemblage, based mainly on the presence of band BH1 during the blooms. However, geographical origin and seasonality also appeared to be important factors in determining differences in the phylogenetic composition of bacterioplankton. Although the seasonal patterns of the distribution of samples in the PCA could be partly attributed to bands that were not sequenced, several sequenced bands had an important role, showing different intensities according to the period of study. Band BH3, related to Bacteroidetes, with Flavobacterium sp. as the closest cultured match, was more important in the samples taken in June. From the two bands related to Flexibacter sp., BH4 predominated the samples taken in January, while BH5 was associated with the Arenys and Barcelona samples. The closest match to these bands was originally found in a mesocosm experiment in our area of study, the NW Mediterranean coast [29]. The differences in bacterioplankton composition during different periods of the year found in this study agree with previous observations of the seasonal succession of bacterioplankton in the Catalan coast [20,30].

4.2 Functional diversity

We hypothesized that changes in the phylogenetic composition would be partly responsible for changes in bacterial metabolism. Contrary to phylogenetic diversity, however, functional diversity did not seem to be affected by the presence of toxic dinoflagellates or phytoplankton in general. PCA analysis did not reveal clear differences in the substrates utilized in the Biolog plates, in the presence or absence of Alexandrium cells.

Biolog plates have been utilized to analyze functional diversity in many environments [31], including marine bacterioplankton [32,33]. Although some criticisms of the method have arisen [34], for example that it is based on a culture response and might not be related to the actual utilization of carbon sources, Biolog plates are a useful tool for differentiating between microbial communities according to their metabolic capabilities. This approach has revealed a similar functional diversity of attached vs. free-living bacteria during freshwater phytoplankton blooms [35] and has provided information to identify bacterial assemblages involved in the clearance of paralytic shellfish toxins (PSTs) in mollusks [36]. We still have a limited knowledge on bacterioplankton functional diversity during HABs, and only Green et al. [37] reported on functional diversity of culturable bacteria in Gymnodinium catenatum cultures: an Alphaproteobacteria capable of aerobic anoxygenic photosynthesis and a Gammaproteobacteria capable of hydrocarbon utilization.

PCA of the data of sole carbon sources utilization in the Biolog plates did not show a separation according to the presence or absence of toxic dinoflagellates, as for the phylogenetic diversity. Probably the presence of band BH1, clearly associated with Alexandrium, was not enough to cause detectable changes in the functional diversity of the whole bacterial assemblage. Chen and Wangersky [14] compared bacterial decay of DOC from senescent cultures of phytoplankton and observed that the DOC released by A. tamarense was bacteria resistant, in contrast to DOC from other algal species tested. They suggested that bacteria associated with these red tides subsist on material coming from sources other than the dinoflagellates. These results agree with our findings that the bacterial assemblages did not vary their ability to utilize carbon sources in the presence or absence of toxic Alexandrium spp. Some phytoplankton cells are known to release bacteriostatic compounds [38] and their effects have been detected during freshwater blooms [39] and in the marine environment [40]. If Alexandrium produced these compounds, this could explain the weak correlation between bacteria and chlorophyll a concentration in our samples (n= 18, log:log y=−13.77 + 1.55x, r²= 0.60). However, there is still a lack of knowledge on DOC composition in marine environments. Information of DOC dynamics and composition during HABs is needed in order to elucidate the carbon sources available for bacterioplankton during bloom events.

Seasonality appeared to be the main factor driving functional diversity of bacterial assemblages in the harbours studied. This difference was mainly based on a higher utilization of several amino acids in January. Amino acids are both carbon and nitrogen sources for bacteria. The N:P ratio of the waters was higher in June (63) than in January (19), (t-test: n= 16, p= 0.018). Probably, bacterial metabolism in January was more focused on utilizing carbon sources, which could supply them not only with carbon but also with nitrogen, such as amino acids. In an attempt to link phylogenetic and functional diversity, it would be desirable to relate the higher amino acid utilization in January with the predominance of band BH4, close to Flexibacter sp. However, isolation of the bacterium and further metabolic tests would be needed.

5 Conclusion

Phytoplankton blooms provide large inputs of organic matter that can be used by bacteria, triggering shifts in bacterial activity and metabolic and phylogenetic diversity [13]. However, phytoplankton composition of the blooms has an important role in determining the diversity of the associated bacteria [41]. In this study, we have observed changes in phylogenetic diversity of the bacterioplankton assemblage during blooms of the toxic dinoflagellate Alexandrium spp. Particularly, a Roseobacter spp. bacterium appeared to be associated with the presence of both A. minutum and A. catenella. Changes in phylogenetic diversity were not paralleled by changes in functional diversity, which showed only differences between seasons.

The influence of toxic algal blooms seems to be elusive: changes in the phylogenetic composition are not necessarily translated into different metabolic capabilities. However, this study has identified a bacterium that may play a role in the presence of toxic algal blooms. Understanding of the algal–bacterial relationships will likely demand detailed analysis at the single species level.

Acknowledgements

L. Cros is acknowledged for providing the phytoplankton cultures. This work was supported by the projects BIOHAB (EVK3-CT99-00015), BASICS (EVK3-CT2002-00078) and PROCAVIR (CTM 2004-04404-C02–01/MAR) and by a post-doctoral contract from the Spanish Ministry of Science and Education to M.M.S.