Introduction

The stock species of algae used in aquaculture are maintained carefully to avoid contamination by bacteria, ciliates, or other algae. In spite of these precautions, the algal cultures are generally not axenic and unknown bacteria are often associated with them. In large-volume cultures other bacteria can be easily introduced through the air or by contact. However, bacteria are considered as innocuous except when Vibrio sp. are present. Such bacteria could have a harmful effect [10, 17, 20, 26] or, conversely, a beneficial influence [10, 34]. Few authors have studied the bacteria associated with algae used for larval rearing in aquaculture [31, 25, 10]. According to Prieur [31] the concentration of culturable bacteria varied from 106 to 2 × 108 mL−1 of algal culture and the species identified by phenotypic characters were diverse, apparently without any dominant bacterial type. However the method used, i.e., numerical taxonomy, did not allow affiliation of the bacteria to genera, groups or families. That bacteria could influence algal cultures was first demonstrated by Fukami et al. [10], who observed a positive effect with some bacteria, especially a strain of an Alcaligenes sp. on growth of the diatom Chaetoceros ceratosporum. Suminto and Hirayama [34] also noted a significant and beneficial effect of a Flavobacterium sp. on the culture of Chaetoeceros gracilis. However, this isolate did not influence Isochrysis galbana or Pavlova lutheri during the exponential phase of growth but it prolonged the viability of cells in stationary phase. On the other hand, a strain of a Flavobacterium sp., isolated during a red tide, exhibited an algicidal effect on Gymnodinium mikimoto [10], whereas a Pseudomonas sp. was reported to cause mortality of Prorocentrum micans [20]. Last, an isolate related to Micrococcus sp. caused mortality of the harmful dinoflagellate Cochlodinium polykrikoides [26]. Such adverse effect of bacteria on algae used in aquaculture could cause problems including poor growth and flocculation, although no definite reports are known of this.

When bivalve larvae show slow growth or do not feed, various causes are suspected such as the quality of seawater, eggs or microalgae. Microalgal culture are considered to have a high nutritional value when cells appear healthy. However, current observations indicate that this does not guarantee good nutritional value. For example, a culture of Isochrysis galbana regularly disrupted growth of scallop (Pecten maximus) larvae even through the algal culture did not exhibit any unusual features (René Robert, personal communication). This was considered to be due to a Cytophaga sp. which was present in the culture. The impact of bacteria on larvae may depend on their concentration, and bacteria associated with algae may be more numerous than those present in sea water. The inlet seawater in an aquaculture site surveyed by Prieur [31] generally contained between 102 and 103 culturable bacteria per mL, whereas bacteria associated with algal cultures ranged from 103 to 105 mL−1 in rearing tanks [31]. The impact of bacterial growth in larval rearing tanks can be substantial. But before determining the relationship between larval rearing problems and bacteria associated with the microalgae, an analysis of the microalgal bacterial flora is an essential step. In this study, bacteria associated with algal cultures from six hatcheries were enumerated and identified by molecular methods as part of a collaborative European programme.

Materials and Methods

Algal Cultures

Algal cultures were provided by commercial hatcheries located in Spain (Tinamenor S.A.), the United Kingdom (Seasalter Shellfish, Whitstable, from two hatcheries at Reculver and Walney), Guernsey (Guernsey Sea Farms), Ireland (Lissadel Shellfish Co.), and France (Satmar). The methods of algal cultivation differed at the different hatcheries. Algae for larval nutrition were produced in continuous cultures, without temperature regulation, and depending partially on solar light in the UK, Spain and Ireland, whereas in France algae were grown in batch cultures with only artificial light. Enriched seawater based on the Conway medium [35] was used in the Satmar and Guernsey hatcheries, and f/2 medium [16] in the other hatcheries. The concentration and duration of algal cultures varied according to hatcheries and seasons. Seawater used for culture medium was sterilized in large volumes with chlorine and neutralization with equimolar or excess sodium thiosulfate in the Satmar hatchery and by pasteurization in the other hatcheries. In small volumes (<20 L) seawater was autoclaved.

The diet distributed to bivalve (mainly oyster) larvae was composed of a mixture of two to three microalgae in equal quantities of cells mL−1 : one or two flagellates such as Pavlova lutheri, Isochrysis galbana, and one diatom, such as Skeletonema costatum, Chaetoceros calcitrans, Chaetoceros muelleri or Chaetoceros gracilis. Tetraselmis suecica and Rhodomonas sp. were given to larvae near metamorphosis in some hatcheries. The quantity of algae distributed daily to larvae depended on the rearing method and the age of larvae and ranged from 5 × 103 to 2 × 104 cells per larva.

In each hatchery, 1 L of algal culture used as larval food was sampled routinely and sent by mail at 4°C, taking a maximum of 48 h to arrive at the laboratory. On receipt, a portion of each algal sample was fixed with formaldehyde (1% v/v final concentration) for counting algal cells using a hemacytometer (Malassez cell) and bacterial cells by epifluorescence microscopy.

Testing Microalgae in Larval Bioassays

Tests were performed to ensure that transport did not cause deterioration of algal cultures. Tests were set up in triplicate in 2-L beakers, each containing 20,000 Crassostrea gigas larvae, previously obtained by stripping the gonads of mature animals. Seawater was filtered to 1 μm and maintained at 25°C.

Each sample of algae was sufficient to feed the larvae three times. During that period algae were kept in their original containers, illuminated but not aerated. A control culture was fed on algae grown at the Brest facilities.

One week later, at the end of the experiment, each batch of larvae was sampled and size (major length) measured automatically using image analysis processing software.

Counting of Bacteria

Bacteria were counted on a epifluorescence microscope following staining with 4′,6-diamidino-2-phenylindole (DAPI) according to [30]. In addition algal samples were serially diluted 10-fold and plated on Marine agar 2216 E (Difco) and TCBS (Thiosulfate Citrate Bile Sucrose) agar (Difco). Colonies were counted after incubation for 2 days for TCBS agar and 2 weeks for Marine agar at 20°C.

Isolation, Selection and Identification of Bacterial Strains

On plates containing 30 to 200 colonies, three to six single colonies per distinct morphological type were picked and purified for identification of bacterial strains. When it was impossible to distinguish different types of colony or when they were too diversified, 20 to 25 colonies were collected at random.

DNA was extracted and purified by a simple previously described method [12]. Briefly, bacterial cells from a overnight culture were collected and resuspended in 200 μL lysis buffer (10 mM tris/HCl, pH 8.0, 1 mM EDTA, 1% Triton X-100) and boiled for 5 min. DNA was purified by a single chloroform extraction. DNA was amplified using a Ready-to-go RAPD (Randomly Amplified Polymorphic DNA) kit (Amersham) according to the manufacturer’s instructions. To ensure that colonies with identical morphology corresponded to the same isolate, DNA was amplified twice via RAPD using two different primers (primer 1283: 5′-GCGATCCCCA-3′ and primer 3: 5′-GTAGACCCGT-3′). The RAPD products were separated by electrophoresis in 2% agarose gels in TAE buffer containing 0.5 μg mL−1 ethidium bromide. In most cases, colonies with the same morphology corresponded to only one fingerprint, but sometimes to two or even three different fingerprints. In these cases the percentage of these colony types was divided equally by the number of related fingerprints.

For each algal sample one to three bacteria per RAPD pattern were identified by partial sequencing of 16S rDNA (about 700 bases). Sequencing PCR reaction was performed in 50 μL final volume, with ~50 ng DNA, 250 μM of each dNTP, 1.5 M MgCl2, 1.5 U Taq polymerase (Appligene), 100 ng of each primer, 342f (5′-TACGGGAGGCAGCA-3′) and 1114r (5′-AGGGTTGCGCTCGTTG-3′) or in some cases 8 f (5′-AGAGTTTGATCCATGGC-3′) and 925 r (5′-AAACTCAAATGAATTGACGG-3′), and double-distilled water (Milli-Q). The PCR temperature profile used was 95°C for 3 min, followed by 30 cycles consisting of 94°C for 30 s, 52°C for 1 min, 72°C for 90 s with a final extension step at 72°C for 5 min. PCR products were purified according to Gauthier et al. [12]. Briefly, Taq polymerase was removed by extraction with chloroform (2 volumes). The amplicons were precipitated by adding 0.6 M polyethylene glycol (PEG) and 2 M NaCl, and rinsed with 70% ethanol. Sequencing was performed by the Genome Express Company with an automatic sequencer (Grenoble, France).

Phylogenetic Analysis

The 16S rRNA gene sequences were aligned with the nearest reference sequences, obtained from Blast analysis, with the ClustalW and Megalign programmes (Dnastar, Madison, WI). Phylogenetic trees were built with the Phylowin program [11] using the neighbor-joining method. A strain was affiliated into a group or genus if it was located by phylogenetic analysis inside this group or genus. An isolate was affiliated to a reference bacterium when their 16S rDNA sequences >97% identical.

Nucleotide Sequence Accession Numbers

The nucleotide sequences of the isolates in this study have been deposited in the EMBL and GenBank data banks. The accession numbers of the isolates and relevant bacterial sequences those found in the GenBank database are given in Figs. 2 and 3.

Results

Larval Bioassays

Tests showed that the 24 to 48 h transport did not cause deterioration in the algal quality. The algal culture did not collapse during storage and larval survival was >90% at the end of the experiments. Comparison of the oyster larval size on day 9 showed a slight, but not significant, reduction of growth with algae from Satmar, compared with the control, and a significant improvement in one case (Tinamenor).

Bacterial Counts

Plate counts and direct counts were closely related, with the ratio between direct counts and plate counts being <2.5 in 14 cases out of 19 (Table 1). In eight out of these 14 cases, plate counts were greater than direct counts, especially when bacteria were less numerous than algal cells. For the others (five out of 19) the ratios ranged from 3.7 to 12.1. Considering the highest counts between direct counts and plate counts, the bacterial concentrations were <10 times those of the algal cell concentrations in all except four cases. Bacterial concentrations and/or the algal:bacterial cell ratios were on average significantly less in P. lutheri and I. galbana than in the diatoms (Fig. 1), T. suecica and Rhodomonas sp. TCBS plate counts were often below the level of detection (10 CFU mL−1), except for all algal cultures received from Reculver, where TCBS counts ranged from 9.0 × 10 to 5.2 × 104 CFU mL−1 (data not shown).

Figure 1
figure 1

Concentrations of bacteria and algae in different algal cultures. The first letter indicates the hatchery, L: Lissadel; S: Satmar; R: Reculver; G: Guernsey; W: Walney; T: Tinamenor. The two following letters indicate the algal culture, Ig: Isochrysis galbana; Pl: Pavlova lutheri; Cc: Chaetoceros calcitrans; Cp: C. pumilum; Cm: C. muelleri; Ts: Tetraselmis suecica; Rs: Rhodomonas sp.

Full size image
Table 1 Ratio between direct count (DC) and plate count (PC):DC/PC
Full size table

Characterization of the Bacterial Population of Algal Cultures

Estimates of the proportion of identified bacteria per algal strain (Table 2) depended on the ratio of culturable bacteria, the success of collection, amplification, and sequencing. The poor recovery of bacteria associated with I. galbana was mainly due to the inability to regrow many of the dominant isolates. For C. gracilis, C. calcitrans, and S. costatum only a small fraction of bacterial population was identified because culturable bacteria were a small proportion of the total bacteria.

Table 2 Percentage of identified bacteria
Full size table

In nine out of 19 cases one strain represented at least 50% of the bacterial flora, but in all algal cultures except one (S. costatum, Guernsey), there were at least three bacterial strains per algal culture, including strains which did not regrow. No relationship was found between diversity and bacterial concentration. The phylogenetic trees show the affiliation of representative bacteria isolated from the algal cultures (Figs. 2 and 3). The isolates belonged to a wide range of genera or groups.

Figure 2
figure 2

Neighbor-joining tree (pair gap removal, mean length 786) representing the phylogenetic relationship among partial sequences of bacteria (boldface) isolated on solid medium from the algal cultures and α- and β-Proteobacteria from the EMBL sequences data bank. Bootstrap values (500 replicates) >50% are indicated. Accession numbers of EMBL sequences are given following the name of the isolate. Isolates are identified by three letters and one figure. The first letter indicates the hatchery and the two following letters indicate the algal culture (see legend to Fig. 1). The number indicates the serial number of an isolate on agar plate.

Full size image
Figure 3
figure 3

Neighbor-joining tree (pair gap removal, mean length 867) representing the phylogenetic relationship among partial sequences of bacteria (boldface) isolated on solid medium from the algal cultures and γ-Proteobacteria, CFB group, and Actinobacteria, from the EMBL databank sequences. For other legends see Fig. 2.

Full size image

The Rhodobacter group is well represented and most of isolates in this group are clustered in four genera: Ruegeria, Roseobacter, Sulfitobacter and Prionitis (Fig. 2). The other isolates placed inside α-Proteobacteria are scattered in four groups: Hyphomonas, Zymomonas, Rhizobiaceae, and Azospiririllum (Fig. 2). Two isolates belonging to the same species close to Zooglea ramigera were members of β-Proteobacteria (Fig. 2).

γ-Proteobacteria were weakly represented and very dispersed with only one isolate per genus (Fig. 3). Conversely in the CFB group, most isolates were grouped with, and closely related to, Muricauda ruestringensis. Some Gram-positive bacteria were also present. Three isolates had at least 99% similarity with reference bacteria: WRs-7 (Bacillus horikoshii), Tig-5 (Agrobacterium sanguineum), LCm-3 (Alteromonas macleodii); 13 were almost identical (99% homology) to environmental isolates. Several (Table 3) were clustered with bacteria associated with harmful dinoflagellates, Alexandrium spp. culture [6], Scrippsiella trochoidea [17], and Aureococcus anophagefferens [4]. Thus, LSc-10 and LSc-11-3 are closely related to Scripps-101; Spl-1, GIg-3, and GPl-3 to Roseobacter 253-16; WTs-4 to ATAM173a-17 (Table 4); LTs7 to Roseobacter 253-13; WRs1 to C116-18; SCc-7 to Alteromonas 253-19 and RSc-1; and LSc-7 to Flavobacterium 7BT (Fig. 3).

Table 3 Isolates from the EMBL data bank most closely to bacterial strains associated with microalgal cultures
Full size table
Table 4 Abundance of culturable bacteria associated with different algal species
Full size table

Some others were also closely related to recently identified marine bacteria. RSc-10 and RTs-10 were affiliated with Arctic R7697 and WRs-1 with Caulobacter C116-18. Some isolates may constitute new species and some others may be new genera when their branches are deep, for example GPl-9, LCm-7, SCc-11, SIg-11, and GPl-11. The Gram-positive bacteria, LCm-11, LIg-3, WRs-7 and RSC-6, were only related to terrestrial bacteria.

Isolates belonging to the Rhodobacter group represented 36% of bacterial flora and 50.6% of them were members of Roseobacter clade (Table 4). Ten bacterial species (numbered 1 to 10, Table 3) comprised half of the microbiota in terms of abundance. Four strains (LTs-2, WRs-12, RIg-3, TCg-9), species 1 (Table 3), grouped close to the isolate Hstpl 95 (Fig. 2, Rhodobacter group), were present in four hatcheries and associated with four different algal species. Another three isolates (GPl-3, GIg-3, SPl-1), species 6 (Table 3), similar to Roseobacter 253-16, colonized two algal species in two different hatcheries. Three strains (WTs-7, WPl-5, GIg-12), species 7 (Table 3), clustered with the isolate JP63-1, were present in three algal species and two hatcheries. Five strains (LSc-7, RSc-1, Tig-1, Wig-7, SCc-2), species 10 (Table 3), closely related to Muricauda ruestringensis, were associated with three diatom cultures and two I. galbana cultures. They were present in five out of six hatcheries. The other bacterial species were less ubiquitous but they were harbored by two algal species and/or by two hatcheries at least.

Analysis of Bacterial Population by Hatchery

(See Table 5).

Table 5 Bacterial populations by hatchery and algal culture
Full size table

Lissadel

Bacteria were moderately numerous and were at the same concentration or up to four times (T. suecica) greater than algal cells. Most of the dominant bacteria could be identified and showed little diversity. In all cultures two strains constituted >78% of the bacterial population identified. The bacterial populations were very different from one algal culture to another. The C. muelleri culture exhibited an atypical bacterial community, constituted by a strain of Alteromonas macleodii (LCm-3) and an Actinobacterium (LCm-11).

Satmar

In C. calcitrans and I. galbana cultures bacterial cells were at the same level as algal cells, while P. lutheri harboured very few bacteria. Bacterial communities of the first two were relatively diverse with at least three bacteria in the dominant flora. In P. lutheri, a Ruegeria sp. (SPl-1) constituted the vast majority of the bacteria.

Guernsey

The ratio of bacterial cells to algal cells was greater than in other hatcheries, except for Reculver, and the bacterial populations showed little diversity. S. costatum was colonised by only one bacterium, a Gelidibacter sp. (GSc-1), and at high concentration. The culturable bacteria associated with P. lutheri at the Satmar and Guernsey hatcheries were mainly composed of strains (SPl-1, GPl-3) belonging to the same cluster as Ruegeria 253-16 (99% similarity). However, strain GPl-3 was 39 times more numerous and the ratio of bacterial cells to algal cells (direct counts) was 112 times higher. A similar strain (Gig-3) and two others were also established in the I. galbana culture.

Tinamenor

In I. galbana, although algal cells were at high density, bacteria numbers remained moderate, in contrast to C. gracilis where they reached high concentrations. The predominant bacteria were Agrobacterium sanguineum (Tig-5) in I. galbana and a Roseobacter sp. (TCg 9) as also found in other hatcheries in C. gracilis.

Walney

The direct counts were very close to the plate counts. Concentrations of bacteria were average, except in Rhodomonas sp. and T. suecica cultures where they exceeded 10 bacterial cells per algal cell. However, in comparison with these algae which have large cells, bacteria represented little in terms of biomass. In T. suecica, a Roseobacter sp. constituted 60% of the bacterial population, and an unidentified bacterium colonized the I. galbana culture. No bacterium was common between the different algal cultures, except for a Sphingomonas in T. suecica (WTs-7) and P. lutheri (WPl-5). However, the algal cultures shared the highest number of identical bacteria with other hatcheries (Figs. 2 and 3).

Reculver

The bacterial population was high compared to algal concentration, and of the same magnitude as in the Guernsey hatchery. The bacterial population was diverse except in the T. suecica culture, which was dominated by a Rhizobium strain (RTs2) also isolated from the I. galbana culture. The bacterial population identified from S. costatum was different from the others with isolates RSc-1 (Muricauda sp.) and RSc-6 (Kocuria sp.) forming the dominant population. Two isolates, RSc-10 in C. gracilis and RTs-10 in T. suecica, were identical but low in abundance.

Discussion

In open and oligotrophic ecosystems, bacterioplankton communities are generally highly diverse, and this diversity often prevents any correspondence between unculturable and culturable bacteria at species level [9]. When a sample of seawater was incubated in a tank with added organic matter, the diversity of the bacterial populations dropped, the percentage of culturable bacteria increased markedly, and the similarity between unculturable and culturable bacteria increased [21]. Similar trends occurred naturally in biotopes enriched in organic matter. During an algal bloom, bacterial populations increased as well as the proportion of culturable bacteria [28]. In aquaculture systems, which are generally richer in organic matter than natural seawater, the proportion of culturable bacteria may be much more important [18] than in the open sea. In a closed volume such as a container of algal culture, a significant fraction of the unculturable bacterial cells might be identical to culturable bacteria. In this study, bacteria cultured on Marine Agar were estimated to represent from 10% to >60% of the total bacterial communities. In some cases, the poor recovery was not always due to lack of culturability. Collection of some dominant colonies, particularly those from the I. galbana culture, failed because of failure to regrow the tiny colonies.

Despite the imprecise direct counts and selectivity of the Zobell medium, recovery of bacteria appeared high. Better results might be obtained by using a medium poorer in organic matter as tested for the culture of bacterioplankton [15]. However, several authors used Marine Agar as a reference medium to study bacterioplankton [21]. Culturability of marine bacteria has been discussed for many years without reaching clear conclusions [9, 28]. Although methods such as dilution (MPN, most-probable-number) optimize the detection of viable cells, the ratio often remains below 1% [13]. Bernard et al. [5] reported a greater proportion of marine bacteria growing on Marine Agar, from 0.43 to 3.52% of total bacteria, whereas the viable cell concentrations reached 10.8 to 22.2%. Nevertheless, cultivation reduces the diversity of bacterial populations, and selection of bacterial genera by Marine Agar was clearly seen in DGGE analysis performed by these authors.

Direct counts of bacteria were probably underestimated, especially when the bacterial concentration was low, as some bacteria could be attached to algal cells. Up to 66% of the total bacterial biomass has been found attached to particles of >10 μm, including phytoplankton [36]. Algal cells can also mask bacteria because they are larger and highly fluorescent. The concentration of algal cells could not be reduced by dilution since bacteria would not then be sufficiently concentrated for counting. Bacterial clumps, which were sometimes observed, could distort counting especially if they were dissociated during dilution. These different biasing factors could explain why in eight cases the plate counts were greater than direct counts. Consequently, it was impossible to define accurately the true concentration of bacteria associated with microalgae and the proportion of culturable bacteria. When bacterial cell concentrations were greater than those of algal cells and when direct counts were higher than the plate counts, this inaccuracy may be reduced. Although the direct counts were probably underestimated, bacterial concentrations appeared to fluctuate around 107 cells mL−1, except in some cases where they reached concentrations as high as 5 × 108 cells mL−1.

The bacterial flora associated with algal cultures, as identified by partial sequences of 16S rDNA, was of low diversity. Greater than 80% were composed of two or three strains, and in 9 out of 19 algal cultures one strain comprised >50% of the bacteria.

High concentrations of bacteria were not correlated with low species diversity. For example, at the Satmar hatchery the bacterial flora growing in the P. lutheri culture was of relatively low concentration (2.8 × 105 mL−1) but it was dominated by one strain of Roseobacter (SPl-1). Conversely, bacterial populations of S. costatum cultures from Reculver were of high concentration (5.3 × 108 cell mL−1), but they remained relatively diverse.

The strains which colonized the microalgae cultures belonged principally to the Rhodobacter and CFB groups, which are frequently encountered in seawater. Roseobacter, Ruegeria, Sulfitobacter, and Antarctobacter constitute an important part of the bacterioplankton found in marine ecosystems [14, 39, 40]. Members of the CFB group are also ubiquitous in open oceans and coastal seawater [3, 8, 9]. However, enrichment of these last clusters seemed to occur in algal cultures [32], in comparison with γ-Proteobacteria and β-Proteobacteria, which were poorly represented. Culturability alone can not explain this trend. Indeed, according to Eilers et al. [9], bacterial culture favors members of the γProteobacteria and Roseobacter, compared to members of the CFB group.

Gram-positive bacteria, which occurred at low level in some algal cultures, were probably not native to the marine environment but are suspected of being conveyed by tap water used to rinse the algal culture transport vessel.

The spectrum of genera and species in flagellate cultures seemed to increase when the number of analyses increased (Table 4). All groups which were identified in this study were present in I. galbana cultures (six samples). In P. lutheri cultures (three samples) only two were missing. However no members of the γ-Proteobacteria and CFB group occurred in T. suecica (three samples). S. costatum (three samples) appeared still more selective since members the CFB group colonized the cultures almost exclusively. For other algae conclusions were difficult because only one analysis was done.

Among the bacterial groups that colonized the algal cultures, some species appeared particularly well adapted to the conditions of algal culture. For example the high similarity with sequences of bacteria isolated from potentially toxic flagellate/picoplankton cultures was remarkable [4, 6, 17]. The isolates clustered around Muricauda ruestrengensis colonized several algal cultures in the same hatchery or in different facilities. Several cultures of flagellate/picoplankton cultures harbored the same bacterial species. Diatoms had differing bacterial populations. Few isolates were shared with flagellate cultures. The bacterial species were less diverse and did not include Rhizobiaceae, Rhodospirillaceae, Zymomonas, and Hyphomonas. The S. costatum culture harbored members of CFB group γ- and a few α-proteobacteria. The absence of other groups might be caused by production of antibacterial substances [24] and/or by promoting, with exudates rich in polysaccharides [37], particular bacterial strains, such as members of Cytophagacaea. Indeed the latter can become dominant in ecosystems rich in polysaccharides, and they degrade them very efficiency [19]. The ability to produce extracellular polymeric substances appeared very widespread in all diatom species [37]. On the other hand, different factors such as light, CO2, mineral salt concentration, and organic matter released by microalgae may favor the establishment of certain species and then their redundancy. Zubkov et al. [40] and Ansede et al. [1] showed that the dimethylsulfoniopropionate released by phytoplankton during blooms was mainly metabolized by Roseobacter-related species. That would explain the predominance of this group during algal blooms and in algal cultures.

The Rhodobacter group encompassed chemoautotroph and photosynthetic species and others which have widely differing metabolisms [14, 39]. Some isolates such as RIg-3, WRs-12, LTs-2, and TCg-9 clustered with HstpL95 (97% similarity) could be photosynthetic and others might fix nitrogen as suggested by Pereg et al. [27]. Bacteria can modify algal metabolism as demonstrated with harmful algae by Brinkmeyer et al. [6]. In this case the presence of bacteria increased the yield of toxin. The nutritional relationship is complex, but a synergism between algal and bacterial cells is generally observed [15].

The presence of certain strains could have negative consequences. Isolates LTs-7 and SCc-7 clustered, respectively, with strains 253-13 and 253-19, which were associated with toxic algae and were suspected to produce toxin [6]. Although no experiment has yet been performed with these bacteria, they might have similar detrimental effect to those of Gyrodinium aureolum [33, 22] and Alexandrium tamarense [22] on bivalve juveniles and larvae.

Some bacteria, including members of the CFB and Pseudomonas spp., can significantly affect or stimulate algae, especially diatoms [10, 34]. Although in this study neither members of the CFB group nor γ-Proteobacteria were affiliated with these strains, it would be interesting to test these isolates one by one or in combination in axenic diatoms and flagellates to observe their effect on algal and larval cultures.

Bacteria could also promote larval rearing success. The role of bacteria in larval nutrition must be minor compared to algae since the bacterial biomass is low and the retention efficiency of larvae is low for particles below 1–2 μm. However, bacteria can complement the diet with growth factors, as reported by Moal et al. [23], and improve larval growth significantly, even at low concentrations [7].

There are few sources of algal strains for hatcheries, and the method of algal maintenance was not taken into account in analysis of the bacterial populations. This could represent a potential risk if associated bacteria affected larval or algal cultures. Although the suspected bacteria have yet to be tested, it is recommended that hatcheries check the bacterial diversity regularly on solid medium or change the source of algae frequently. In the future, algae could be deliberately associated with bacteria selected for their innocuous or beneficial effect. For example, Avendano et al. [2] used axenic algae combined with probiotic bacteria to improve the survival of scallop (Argopecten purpuratus) larvae. If this method is applied, the problem will be to maintain these beneficial bacteria in algal culture since different bacteria can easily colonize batches of algae in the actual conditions of culture.

Several features of bacterial populations can be drawn from this study. First a large proportion of bacterial populations associated with algal cultures were easily culturable. Use of specific media would probably allow culture of other groups including cyanobacteria and other autotrophs.

They are not so diverse, dominated by Roseobacter and CFB groups. Although a wide range of bacterial species could occur in algal culture the most frequently encountered belonged to 10 species, some of which were also found by other authors.

Bacteria showed little overlap between diatom cultures and flagellate cultures, i.e., they appeared to harbor distinct bacterial populations. That proved the importance of interaction between bacteria and microalgae. However, some bacterial species were present in both types of phytoplankton cultures, thus showing that their difference was not total.