Skip to main content
European Commission logo
English English
CORDIS - EU research results
Content archived on 2024-06-18

Microbes as positive actors for more sustainable aquaculture

Final Report Summary - PROMICROBE (Microbes as positive actors for more sustainable aquaculture)

Executive Summary:

A major bottle neck in aquaculture is the juvenile production, that is characterized by unpredictable larval survival and larval/juvenile quality caused by negative microbial interference.

In order to understand the interactions between the host and its microbial community (MC) and to use these interactions to the benefit of the fish, the project wants to disentangle the complex interplay between the different biological components of the aquaculture ecosystem.

The major objectives are:

• How does the microbial community evolve as the host progresses through its life cycle?
• How stable is the microbial community in relation to perturbation caused by changes in environmental conditions and how resilient is the microbial community?
• What is the effect of micro-organisms on the host metabolism, its disease susceptibility and viability?
• Considering that some environmental factors (e.g. salinity, feed composition) have a major influence on the MC composition, to what degree is it possible to influence or steer the MC composition and activity?
• To what extend can microbes present in aquaculture rearing systems be (re-) used to retain organic wastes and nutrients, and thus reduce the impact on the environment?

The study of the establishment, stability and resilience of gut microbial communities, using cod, sea bass or tilapia revealed that for cod larvae no significant differences in microbiota were detected for samples from 17, 32, and 61 dph that could be attributed to the different feeding regimes. For the sea bass there were differences between the bacterial communities sampled before and after weaning, and with the two feed sequences. The diversity increased significantly between 30 and 60dph, but it did not differ between regimes. For tilapia, the system set-up had a clear effect on the microbial communities in water and gut. It was evident in all 3 species that the waterMC had a more pronounced effect on the gutMC than feed or feeding regimes.

In order to study the functionality of microbial communities at host level, a disinfection protocol for each species was established. The protective effect of putative probiotics, added at increasing species richness (between 1 strain and 10 strains) towards challenged cod or sea bass larvae revealed that species richness within the range of 1 to 10 did not have an effect on the protection of challenged cod or sea bass larvae.

Through the ‘in system’-production of microorganisms coagulated into bioflocs, the impact on the environment can be reduced, while the bioflocs themselves can be used as an additional feed source for fish. This was only tested on tilapia in this project.

The overall outcome is that the gut MC of fish larvae can be steered through the MC present in the water, while the latter can be influenced through the production system (being it RAS or the use of matured water). Both above mentioned systems should be preferred over flow-through set ups from a microbial community management as well as from an environmental point of view.

Project Context and Objectives:

European aquaculture industry is facing a series of bottlenecks limiting the production. To further develop aquaculture, the major bottlenecks need to be systematically removed. With respect to microbial interference, we need to make use of the natural mutualistic symbiotic relationships that have evolved over millions of years between the host and the microbial community. Hence, we need to understand the mutual and reciprocal interactions between them and use these interactions to the benefit of the viability and robustness of the fish under aquaculture conditions. This “join them” approach is contradictory to the traditional “beat them” strategy generally applied in microbial management used in human medicine, agriculture and aquaculture.

The project has started from the concept that the host microbial community (MC) is influenced by the host itself and by the microbial community of the system in which the host is living. It is anticipated that there is a reciprocal interaction between the different compartments of the system in every stage of the life cycle as laid out in Fig. 1 (horizontal arrows). The strength and the nature of those interactions can depend on the life cycle stage. Apart from these horizontal interactions, it is anticipated that microbial communities evolve as the host grows or the system changes. A certain degree of top-down conditioning belongs to the possibilities. This would mean that the early colonisation of the host by certain micro-organisms, might condition the host response and determine the microbial community composition in later life stages. In addition the environment (e.g. temperature, feed, water quality, sanitation measurements) will determine to a great extent the composition and activity of the microbial communities in the different compartments. The environment can be modulated through the composition of the feed or through compounds that directly influence the activity of micro-organisms. On the other hand nutritional compounds or immunomodulators might have a direct influence on the host, modulating the microbial community of both host and system.

Fig 1. Conceptual framework of PROMICROBE. Every life stage carries its associated microbial community (MCe: microbial community of the intestine or attached to outer surface as for instance for eggs). The MCe is to an undefined degree in equilibrium with the MC of the system (MCs). The host can influence the MCe through its antimicrobial activities as determined by its immunological capacities in every life stage (horizontal arrows). Yet the MCe is probably also in equilibrium or influenced by the microbial community of the system (MCs). The composition and activities of microbial communities (MCe or MCs) can be modulated to a certain degree, influencing the host. The host can be influenced either by direct interference with its immune system or through nutrition, modulating indirectly the MC. Thick arrows in this scheme are the subject of research activities in this project. (imm: immunomodulators; nutri: nutrition)


The aquatic environment has a very strong influence on the microbial community composition, its activity and hence on the aquaculture organisms.

In order to document and understand several of the arrows of figure 1, the following main objectives were put forward:

• How does the microbial community evolve as the host progresses through its life cycle?
• How stable is the microbial community in relation to perturbation caused by changes in environmental conditions and how resilient is the microbial community?
• What is the effect of micro-organisms on the host metabolism, its disease susceptibility and viability?
• Considering that some environmental factors (e.g. salinity, feed composition) have a major influence on the MC composition, to what degree is it possible to influence or steer the MC composition and activity?
• To what extend can microbes present in aquaculture rearing systems be (re-) used to retain organic wastes and nutrients, and thus reduce the impact on the environment?

Project Results:


Aquaculture is still facing a number of challenges that need to be systematically dealt with. One bottleneck is the juvenile production, that is characterized by unpredictable larval survival, larval/juvenile quality and robustness caused by negative microbial interference.

In order to understand the mutual and reciprocal interactions between the host and its microbial community and to use these interactions to the benefit of the fish, this project wants to disentangle the complex interplay between the different biological components of the aquaculture ecosystem.

These are the major objectives that will be addressed experimentally:

• How does the microbial community evolve as the host progresses through its life cycle?
• How stable is the microbial community in relation to perturbation caused by changes in environmental conditions and how resilient is the microbial community?
• What is the effect of micro-organisms on the host metabolism, its disease susceptibility and viability?
• Considering that some environmental factors (e.g. salinity, feed composition) have a major influence on the MC composition, to what degree is it possible to influence or steer the MC composition and activity?
• To what extend can microbes present in aquaculture rearing systems be (re-) used to retain organic wastes and nutrients, and thus reduce the impact on the environment?

In order to study of the establishment, stability and resilience of gut microbial communities first, protocols for harmonization of DNA isolation, PCR and DGGE and PCR-DGGE inter-calibration has been attempted. However, the test revealed that comparisons between labs was not good enough for putting various gels from different labs together in one joint data file. Moreover, even joining different gels from the same lab into one data file is not trivial. As a consequence several of the deliverables for WP1 had to be reformulated.

The experiments to obtain materials for characterization of the microflora of cod, tilapia and sea-bass as a function of developmental stage and holding regime have been completed (Tasks 1.2-1.4). Due to the problems with developing a harmonized PCR-DGGE protocol and joining all data in one data file the analysis was put on hold, and was therefore delayed. For the cod experiment three different feeding regimes were used until 17 days post hatching (dph), where after they were identical and fed Artemia and later formulated feed. For tilapia two different recirculation systems were used, with replication of one to include within systems variation. For sea bass larvae that were fed from 7dph onwards either with formulated feed or with live Artemia nauplii until 30-35 dph when they were weaned. From all three experiments DNA has been extracted from individual larvae, water and feed for PCR-DGGE analysis of the composition of the bacterial community. For cod larvae no significant differences in microbiota were detected for samples from 17, 32, and 61 dph that could be attributed to the different feeding regimes. However, at all these sampling days, the microbiota of larvae from some of the replicate tanks was significantly different from the other replicates. Only at 8 dph the larvae microbiota was found to differ significantly between two of the feeding regimes. Within tanks the microbiota of individual larvae appeared more similar at 17 and 32 dph. DNA sequencing showed that proteobacteria (alpha-, epsilon-, and gamma-) dominated at Day 8, 17, 32, and 61. Firmicutes, Bacteriodetes, and Cyanobacteria were also found. For the sea bass there were differences between the bacterial communities sampled before and after weaning, and with the two feed sequences. The diversity increased significantly between 30 and 60dph, but it did not differ between regimes. Some bands were sequenced, and their SIMPER analysis indicated that Arcobacter sp. was the main taxonomic unit responsible for the difference of similarity of the two dietary groups at 30 dph, but it was also strongly dominant. At 60 dph, there was a shift to various Firmicutes, and the dissimilarity between communities was mainly due to two species within Staphylococcus. This dissimilarity was, however, mainly between tanks and not between feeding regimes. For tilapia variation in gut microbiota between individuals grown within the same tank was similar to the variation between individuals grown in different tanks but receiving the same treatment. The gut and water microbiota changed over time. Moving window analysis based on Pearson correlation showed a gradual decrease in percentage similarity over time. Non-metric multidimensional scaling (NMDS) indicated that the microbiota changed over time, and this was confirmed by analysis of similarity (ANOSIM). Gut microbial communities (gut-MC) from fish reared in RAS also showed less tank-to-tank, and between replicate systems, variation. There was only a weak correlation between feed-MC and water-MC, feed-MC and gut-MC, and water-MC with gut-MC. The low overlap was confirmed by NMDS, ANOSIM and principal components analysis (PCA). The system set-up had a clear effect on the microbial communities in water and gut.

Due to delays and problems for the above tasks, the synthesis on the characterisation of the gastro intestinal microbiota of reared fish (Task 1.5) is delayed and the goals had to be modified. Based on a thorough process during the last six months consensus is established for 1) new questions to address, 2) data processing, 3) data presentation, and 4) statistical analysis of data. An outline of the report with responsible persons and partners for each part has also been developed.

For sea bass, the individual microbiota in faeces has been characterised over time (Task 1.6). Even with apparently the same rearing conditions, the fish maintained in some tanks differed in terms of similarity and diversity. The effect of the origin of the fish, or that of their previous rearing conditions and promiscuity, were observed still after mixing them in new tanks. The high individual variability in the faecal microbiota seemed to be regulated by endogenous factors and by tank promiscuity. The resilience of the microbial community (Task 1.7) was evaluated after a perturbation consisting of an oxytetracycline (antibiotic) treatment for 10 days, and was compared with fish kept in a new tank without antibiotic treatment. Faeces were collected from anesthetized individuals by stripping, and microbiota in faeces was characterized by PCR-DGGE. The community composition was followed up four dates after treatment, during two months. The effect of the antibiotic treatment appeared moderate in terms of intra-group similarity, and resulted in decrease diversity. Further analyses are required to complete the task. To compare the autochthonous (mucus associated) microbiota and the faeces microbiota (Task 1.8) individuals were euthanized for collection of microbes from the intestinal wall and from the faeces, and PCR-DGGE was used to characterise the microbiota. Preliminary data indicated that the mucosal community was more similar among individuals, and less diverse than the faecal community.

For the study on the functionality of microbial communities at host level (WP2), a disinfection protocol for tilapia needed to be established. The protocol to obtain axenic tilapia is more elaborate than the one of the sea bass as the tilapia eggs are most of the times contaminated with fungi next to bacteria. The protocol comprises two consecutive disinfections; the first with hydrogen peroxide followed 24 hrs later by the second with sodium hypochlorite. The eggs are incubated in artificial fresh water containing a mixture of ampicillin, rifampicin, kanamycin, trimethoprim, amphotericin-B, fluorescent brightener 28 and zymolyase. The eggs are put in autoclaved 0.5L bottles with filtered (0.2 µm) air in- and outlet at 28°C. Unfertilised eggs are removed daily from the incubation bottles. Antibiotics and antifungals addition into the larvae incubation medium needs to be repeated every 48 hours in order to maintain axenic conditions.

The existing gnotobiotic food chain sea bass-Artemia, was adapted to use compound diet from start-feeding. This will allow studying the effects of the composition of compound diets on the digestive and immune system. The compound diet was γ-irradiated with a Co-60 source at different doses: 3, 5, 7 and 25 KGray. Only the irradiation at 25 KGray resulted in feed without bacteria (culturable and non-culturable). The irradiation had no effect on the fatty acid content and composition. However, it did have negative effects on the vitamin content. As a lower strength than 25KGray did not result in bacteria free diets, another way to disinfect without destroying the feed ingredients was sought after. A mixture of 4 antibiotics was able to kill all bacteria in the compound feed particles.

Small sterile vials, each containing 12 larvae, were mounted on a rotor to keep the compound diet in suspension. Two rotation speeds were tested: 1 and 4 rpm. The vials were placed perpendicularly to the axis of the rotor or tangential.

The compound diet of 75 µm, γ-irradiated at 25 KGray, fed every 2 days from day 5 after hatching onwards at 0.1 mg dry weight of diet for each vial, with vials mounted on a rotor at 4 rpm and a tangential rotation axis led to a survival of 60-70% by day after hatching 11 (DAH11) at 16°C.

In order to study the effect of selected microorganisms or feed compounds on the immune system, a challenge test with Listonella anguillarum was set up. The window of infection was determined by challenging the larvae either at DAH 7, or 8 or 9 by addition of 105 CFU of L. anguillarum/mL to the water. Non-challenged fish were used as a positive control, while fish challenged on DAH 3 (normal procedure when Artemia was used) were used as negative control. Fish challenged on DAH 7 or 8 had a significantly lower survival than the positive control (non-challenged) fish four and six days after infection. As we want to study the effects of the feed composition, a challenge on DAH 8 was preferred over a challenge on DAH 7.

The protocol with feeding compound diet every two days from DAH 5 onwards and challenging the fish on DAH 8 with 105 CFU/mL of L. anguillarum will be applied in experiments during this project.

However, as the vitamins were destroyed due to the high irradiation, experiments were stopped.

Instead, the effect of heat shock protein (HSP) induction in sea bass larvae by a variety of protocols and its effect on the robustness of the larvae verified by exposure to stressors (biotic and/or abiotic) was tested.

Heat shock proteins (HSPs) are a universal response to stress. The production of these proteins leads to a faster and more pronounced answer towards all kinds of stressors. This includes the attack of pathogenic bacteria. Through the controlled induction of HSP production, it has been found that Artemia are better protected to certain pathogenic bacteria (outside this project). The same chemicals have been tested at different concentrations with sea bass larvae. Fish larvae were sampled after 2 different exposure times.

The exposure to Tex-OE (an Opuntia derived chemical) did not lead to HSP induction even when used up to 40 mg/L. Myricetine had a tendency to be toxic at the highest tested level of 60 mg/L. Hence, it was tested at 15 and 30 mg/L. However, there was no protection from the Vibrio anguillarum HI-610 strain. The tested myricetine concentrations did not lead to HSP induction either.

During the second reporting a second challenge test using active metabolising cells (AMC) was set up. The Vibrio were kept active by addition of trypton (10 mg/L) or the equivalent of 300 homogenised sea bass larvae/L. The use of the AMC may be more close to the actual situation in hatcheries.

The effect of single probiotic candidates and mixtures thereof, with a species diversity ranging from two to four, have been added to sea bass larvae that afterwards were challenged (V. anguillarum HI-610) or not. The single probiotic candidates had no significant effect on the non-challenged sea bass larvae. Only Acinetobacter sp. KF-3 (code: B7) had a slightly positive effect on the survival of the sea bass larvae. The survival of the sea bass larvae after addition of a mixture of 4 bacteria (Cytophaga sp., Phenylobacterium sp., Paracoccus denitrificans and Uncultured Acinetobacter sp. KF-3) was significantly lower compared to the axenic control on DAH 9 and 11. The sea bass experiments on the effect of the probiotic diversity on the survival of challenged larvae needs to be continued.

In the cod experiments, no effect of species richness, ranging from 1 to 5, could be detected on survival after challenge with the pathogen V. anguillarum HI-610, but there was a significant effect of bacteria absence or presence. Two of the probiotic candidates tested (Rugeria sp. and Pseudoalteromonas sp.) seemed to have a protective ability against the pathogen.

The effect of single probiotic candidates, able to degrade quorum sensing molecules, and mixtures thereof, with a species diversity ranging from two to five, have been added to sea bass larvae that afterwards were challenged (V. anguillarum HI-610) or not.

Pseudoalteromonas sp. and Lysinibacillus sp. had a positive effect on the survival of the non-challenged sea bass larvae until DAH 11. However, the survival decreased sharply thereafter. Finally, only the larvae that received F5 (unidentified) had a significantly higher survival than the axenic larvae by DAH 15.

The experiment with addition of single probiotic candidates followed by a challenge with V. anguillarum HI-610 was run twice. The results could not be repeated; Bacillus thuringiensis (LCDR16) resulted once in the best survival of the challenged sea bass larvae, while in the other experiment, this probiotic candidate resulted in the worst survival. A similar observation was made for F5 (unidentified). When 10 mg/L trypton was added to have AMC, none of the single probiotic candidates had any significant positive effect on the challenged sea bass larvae. Bacillus thuringiensis (LCDR16) and Pseudoalteromonas sp. (F1) even had a negative effect on the challenged sea bass larvae.

Random mixtures of the quorum sensing degrading probiotic candidates were made with a species richness ranging between two and five. No one mixture, no matter its species diversity was able to result in a better survival of the challenged sea bass larvae than the challenged sea bass larvae without addition of probiotic candidates.

Cod experiments were conducted with the same five quorum sensing degrading probiotic candidates. There was a clear effect of adding quorum sensing degraders to the cod larvae water before the challenge. The time until 50% mortality (LT50) was doubled for the flasks with QS degraders compared to bacteria free cod larvae. There was no coherent effect within the richness gradient. However, the larvae without QS degraders present before challenge were more vulnerable to the pathogen, and more than one QS degrader present resulted in some more robustness than only one QS degrader present. An extra experiment was performed to test a species richness of 6 and 10, but no effect was detected between the survival of the challenged cod larvae in both treatments.

The immunostimulating effect of 3 yeast types with a different cell wall composition (WT, mnn-9 and CHS 3) and 3 commercially available immunostimulating compounds (β-glucan, propylene glycol alginate and MacroGard®) has been tested on challenged sea bass larvae in 2 experiments. The yeast cells were added at 3 different densities: 105, 106 and 107 cells/mL, while the compounds were added at 0.1 g/L and at 0.25 g/L. The higher and highest yeast density had a negative effect on the survival of the challenged larvae irrespective of the yeast type used. The β-glucan and propylene glycol alginate were able to protect the challenged sea bass larvae at both tested concentrations till DAH 11 (= 4 days post infection) in 1 experiment. However, this was not the case in the next.

The protocol to extract RNA from sea bass larvae has been optimised by partner 7. It was found that shredding the sea bass larvae with metal beads was the better way to pulverise the larvae. RNA was extracted from samples containing between 1 and 10 individuals. Good quality RNA could be extracted from even one individual leading to enough material for cDNA-AFLP analysis. A pilot cDNA-AFLP analysis was performed starting from RNA extracted from 1 individual and pools of 3, 7 and 9 sea bass larvae. A set of 8 primer combinations was performed. High quality expression patterns were obtained for all samples. However, to ensure reproducibility of the results, the use of RNA extracted from a pool of 10 sea bass larvae was preferred in order to avoid bias towards differences due to differences between individuals rather than differences due to treatment/condition.

Partner 1 has run 1 experiment which was repeated once. Each experiment had six treatments: xenic fish larvae, with an unknown microbial community, axenic fish larvae and conventionalized fish larvae. A mixture of five quorum sensing degrading bacteria was added to half of the axenic fish larvae to become the conventionalized fish larvae at stocking. Fish were either challenged or not. The survival of the fish larvae in both experiments was rather different for the xenic and axenic treatments. The survival of the conventionalized fish larvae was similar for both experiments. Samples were taken from both experiments for cDNA-AFLP analyses.

Two experiments on sea bass (ARC) and one on cod larvae (NTNU) have been performed.

For the cod experiment, duplicate samples for 0 dph; 0 + 18 hours dph; 4 dph and 8 dph and triplicate samples for 7 dph and 7 + 18 hours dph, described in experiment 1 (NTNU) were extracted and initially analysed on gel with a selected number of primer combinations. After our 42 month proaject meeting, it was decided to drop the xenic fish samples and to analyze samples at 3 timepoints in depth. Subsequently, 18 extra primer combinations were analysed on this subset of samples. However, there was aspecific amplification in 12 primer combinations. To overcome this problem, three different pre-amplification dilutions were tested and more selective primers were used to avoid aspecific amplification. Despite these efforts, aspecific amplification could not be avoided and no clear bands could be obtained on the gels.

For the seabass experiments, RNA extractions from xenic and axenic sea bass samples (control and conventionalised) of both experiments (ARC) were analysed as well as two Artemia samples that were used as feed. After running 8 primer combinations on gel, the intensity of about 630 individual bands was determined. From these preliminary cDNA-AFLP results and agreed after our project meeting, following the Prague conference, it seemed very interesting to analyse a subset of samples more into depth trying to cover the whole genome of sea bass. This subset of samples has been tested with all possible primer combinations. Processing of the 128 gel images and normalization for variations led to the identification of 8448 bands in total. Statistical analysis on the effect of bacteria on the gene expression in sea bass larvae at DAH 4 and 7, on the difference in gene expression of axenic sea bass larvae performing well or bad and on the baseline gene expression in sea bass larvae evolving in axenic conditions has been analyzed. After manual curation of the gels a selection of 178 fragments was excised from the gels and reamplified by PCR. These PCR fragments were sequenced directly and subsequently BLASTed. Of these 178 sequenced fragments, 66 ESTs/contigs could be identified and by searching sequence homologies against databases listed on the public sigenae contig browser sea bass, 47 ESTs/contigs could be annotated.

As these cDNA-AFLP on the sea bass samples revealed very interesting data it was suggested to divert the remaining cDNA-AFLP effort of the cod samples towards the seabass samples to obtain an in depth study with a robust statistical analysis as well as gathering as much information as possible from both sea bass experiments. Therefore, the effect of a defined microbial community on host gene expression when challenged by L. anguillarum was tested by analyzing duplicates samples of DAH11 – mix en DAH11 – mix/LA. The intensity of about 9578 individual bands was determined and a student t-test between subjects and fold changes were calculated. Hereby, we could identify 28 differentially expressed bands (P < 0.05; fold changes of > 1.1 and less than -1.1).

In WP3, from what we have learned about gut MC composition and functionality in the other WPs, and in previous experience, we attempted to steer it favourably, as a barrier against infection by enteropathogens, as a guarantor of gut integrity, and more generally, as a possible contributor to digestion, immunity and disease prevention for the host.

In cod larvae, a mixture of 4 candidate probiotic strains was added through feed and water (Task 3.1). Generally, the level of the probiotic candidate strains in the larvae appeared to be highest just after administration, but the amounts decreased to approximately background levels during the 11 days sampling period. These probiotics thus appeared to be transient irrespective to the time of treatment. Further candidate probiotics were screened, and documented for their ability to ferment potential prebiotics (Task 3.2). The most beneficial prebiotics were β-glucans due to high production of short chain fatty acids. Twenty candidate probiotics were selected, based on their ability to adhere to intestinal mucus, and on their antagonism against pathogens. Some of these strains were able to use prebiotics. A new method was based on cultivation in multiwell plates with membrane-fitted chambers, which provided an efficient way of screening strains with respect to antagonistic and synergistic properties, also including the effects of competition for substrates and the production of stimulatory compounds.

In sea bass larvae, the prebiotic-like Poly-β-hydroxybutyrate (PHB) was tested to steering the initial gut colonisation (Task 3.4.1). PHB-containing bacteria were not adequately taken up by sea bass larvae, but it was possible to enrich Artemia nauplii (Instar II stage) with PHB particles. The application of PHB to sea bass larvae 22 days after hatch is thus feasible, and it can still be considered as management of initial colonization. Alternatively, seabass larvae can be fed since mouth opening with compound diet that must contain protein hydrolysate for optimal dietary efficiency, while short peptides are also suitable substrates for many intestinal microbes. New protein hydrolysates from marine sources were tested in experimental diets, compared to a control diet containing a commercial fish protein hydrolysate (Task 3.4.2). The experimental diets increased or decreased growth, compared with the control group, and one diet was detrimental for survival, with upregulation of genes involved in inflammation and antioxidative responses. Enzymatic assays and microbiological data suggested a concomitant maturation of gut and associated microbiota. A high proportion of small peptides in the hydrolysate appeared as a requisite, but not sufficient criterion to assess its dietary value.

Methods for metabolite analyses were applied for metabolite "fingerprinting" in comparative studies of the gut microbiota of tilapia larvae reared in different water treatment systems (Task 3.6). Compounds occurring in significant different abundance in the various groups were identified by search in metabolite databases. The most significant differences between the groups were observed for phospholipid catabolism, but differences in protein degradation products were also detected. Significant differences between the same groups of compounds were also found in a later comparison of axenic and xenic cod larvae (Task 2.9) clearly indicating that the differences were caused by the microbiota.

In sea bass juveniles, a study evaluated the possible contribution of intestinal microbiota to the catabolic process of carbohydrate in the gut, and the subsequent effects on the intermediary metabolism, with possible prebiotic effects of resistant starch and non-starch polysaccharides (Task 3.3). The starchy diets stimulated glucose storage and lipogenesis in the liver, while lupine meal caused an increase in perivisceral fat. The diet influenced especially faecal, but also mucosal microbiota, though in a different way, likely due to the interaction with the host. Lupine meal seemed potentially interesting as a source of prebiotic polysaccharides, but starch, either resistant or not, was also partly metabolized by microbiota, resulting in an increased concentration of acetate in the faeces.

The effect of different diets was studied in juvenile tilapias, to create a wide range of C/N ratio in waste metabolites (Task 3.5). Faecal, branchiary and urinary losses are the main nutrient and microbial sources to rearing water, resulting in the formation of bacterial flocs and biofilms, which were qualified and partially quantified. The results of the experiment suggest that the microbial community in the gastrointestinal tract was not affected by different C/N ratio in water or in feed, while different C/N ratio had a significant effect on the composition of microbial community in the water.

Work Package 4 investigated the relation between microbial communities (MCs) in the system and culture animals in recirculating aquaculture systems (RAS). The focus was on tilapia culture, although some experiments used cod larvae.

The environmental MC has a crucial impact on many aspects of fish performance. MCs in the culture tank water in RAS are influenced by the microbiota present in feed, fish and the different RAS compartments. To determine which of these different microbiota has the strongest influence on the culture water MC, a first experiment (Task 4.1) was set up to analyze the relation between MCs in different compartments in RAS. For this purpose MCs in (1) rearing tank water, (2) feed, (3) fish gut, (4) biofilter water and (5) active suspension tank water were sampled over a 42 day culture period. The fish gut-MCs differed over time (P < 0.001). Gut microbiota were different from the microbiota in feed, rearing tank, biofilter and active suspension tank (P < 0.05). This difference existed from day 0 onwards and did not diminish over time. Contrary to this the MC composition in the rearing tank, biofilter and active suspension tank changed less over time than the gut microbiota and were not different (P > 0.05). In addition, microbiota in 2 turbot farms, each applying a different RAS concept were monitored for a full year. Although both farms stocked fingerlings from the same source, microbiota in gut and rearing water were different between farms.

The low reproducibility in marine juvenile production has in great part been attributed to harmful fish/microbe interactions. Therefore a set of experiments was carried out to investigate the impact of water-MC on gut-MC (Task 4.2 Subtasks 4.2.1 through 4.2.4). It was shown that a high share of opportunistic and fast-growing microbes negatively influenced fish performance. It was hypothesized that larvae, reared in water dominated by K-strategists (mature microbial communities), will perform better, because they are less likely to encounter opportunistic (r-selected) microbes and thus experience less detrimental host/microbe interactions. Rearing systems, which employ biological water treatment, are expected to have a higher microbial biomass and a more stable and mature MC. This is probably particularly true in RAS which provides an extended time for water maturation in the system. Furthermore, the food supply per bacteria is low in such environment. Such conditions should theoretically favor K-selection of the microbial environment and therewith improve fish performance by reducing the risk of negative host/microbe interactions through the proliferation of opportunistic bacteria. Against this background, subtask 2.2.1 investigated the effect of water source on gut-MC in larval rearing of cod. Flow-Through Systems (FS), where no microbial water treatment is employed, were compared with Microbial Matured Systems (MMS), which include a biofiltration unit at the inflow, and complete RAS, incorporating different biological treatment units. Throughout the life feed period of the experiment, a directional microbial community selection experiment was carried out to compare the development of the water-MC, live feed-MC and fish gut-MC in MMS and RAS to a FTS. The microbial composition of the incoming water differed significantly between treatments, and the microbial composition of the incoming water in the RAS and MMS was more stable over time than in the FTS. Generally, the larvae in the RAS seemed to include a higher diversity and a higher species richness of bacteria than larvae in the other treatments. The gap in microbial carrying capacity between the incoming water and the rearing water was reduced in the RAS compared to the FTS or MMS, as the incoming water of the RAS showed higher abundance of bacteria and higher levels of microbial activity. This is thought to favor K-selection of the MC. The measurements of total biomass and survival indicate that the larvae performed better when exposed to a K-selected microbial community, as shown by higher survival in RAS and MMS than in FS. It was concluded that community level K-selection of microbes by water treatment using biofilters in either MMS or RAS worked and may be a methodology applicable in aquaculture to enhance larval performance. However, more knowledge is needed on systems design and on the influence of K-selection on MC composition to provide more insight into the MC selection process.

Compared to natural systems, the density of bacteria in aquaculture production systems is high, and might influence fish performance . Therefore an experiment was set up to determine the effect of the density of the rearing tank-MC on gut-MC (subtask 4.2.2). Furthermore, the study aimed to evaluate the effect of feed microbial load to the gut and water microbial communities by doing a comparison between sterile and non-sterile feeding. Swim-up fry of tilapia were exposed to different MC densities in the culture medium. Fish were fed a commercial starter diet containing 68% fish meal, from the moment of first feeding for 42 days. To control the effect of feeding on the gut microbial community, half of the fishes were fed γ-radiated sterilized feed. Flow cytometry confirmed that the microbial density was different between treatments and stable over time. Denaturing Gradient Gel Electrophoresis (DGGE) analysis showed that diluting the water in the culture tanks did not alter rearing tank MC composition. Furthermore, the microbial density in the culture water did not affect gut-MC (P < 0.01) and feed sterilization had no effect on the gut microbial communities in the system (P < 0.01).

A subsequent experiment was developed to determine the effect of salt- and freshwater rearing conditions on rearing tank- and gut MC in tilapia culture (subtaks 4.2.3). Swim-up fry of tilapia were exposed to zero or 25 ppt salinity. Fish were fed a commercial starter diet containing 68% fishmeal, from the moment of first feeding for 42 days. To control the effect of feeding on the gut microbial community, half of the fishes were fed γ-radiated sterilized feed. Gut-MCs and rearing tank-MCs differed between fresh water and saline environments and changed over time (P < 0.001). Feed sterilization had no effect on the gut microbial communities in the system (P < 0.01).

Little is known about the initial routes of entry into the larval gastrointestinal tract by microbes during early development and the subsequent colonization of the GI-tract. Therefore an experiment was carried out to investigate the effect of bacterial load in water ingested with food on gut-MC (subtask 4.2.4). For this purpose four different treatments were executed on first feeding axenic tilapia larvae. At first, an activated suspension (AS) system was installed to produce microbiota to be used in a subsequent feeding experiment. To assure that the AS system was matured, it was run for six months before bioflocs were harvested and microbes isolated. Tilapia larvae were subjected to four different treatments, i.e. (i) microbes added to feed and rearing water, (ii) sterilized rearing water and microbe-laced feed (iii), microbe-laced rearing water and cleaned feed and (iv) no feeding and microbe-laced rearing water. The water-microbes were stained with red nucleic acid dye (Syto-62) and feed-microbes with green nucleic acid dye (Syto-9) to make them visible under a fluorescent stereomicroscope. As auto-fluorescence of some plant ingredients in the diet made it impossible to quantify whole-gut samples without bias, micro-coupes of the samples were created and similarly analyzed by stereomicroscopy. It was found that tilapia larvae take up microbes from water and feed. Although a precise quantitative determination of ingested bacteria was difficult, a trend was visible that microbes incorporated in the feed were taken up more easily in the gut of tilapia larvae than those originating from rearing water.

Algae are an important ingredient of larval diets. Microbes grow in microalgae cultures, but due to leakage of organic substances from the algal cells there are large variations in numbers and properties of the bacteria among different algal species and depending on the growth phases. Some algae cultures can have high proportions of opportunistic bacteria depending on their growth phase or leak nutrients that can select for harmful microbes. Therefore the physiological state of the microalgae used for green water is probably very important for the larval health. Contrary to live algae cultures, algae pastes, which are often used nowadays, are believed to be sterile. However, once added to the larval environment, also these are expected to influence the MCs in system and gut. Therefore the impact of live algae cultures as well as algal pastes on the water-MC and its cascading effect on gut-MC was studied (Task 4.3). In a first experiment the MCs in cultured algae and algal pastes currently used in marine juvenile production were analysed (subtask 4.3.1). Cultures of Pavlova and Nannochloropsis contained high numbers of bacteria, but no opportunistic species were observed. The pastes from these algae contained around 105 coliform units (CFU) per ml. The paste-MCs differed from those of living cultures of the same species. Furthermore, when used under green water conditions the MCs of the water-added pastes were found to change, whereas those of living cultures remained stable. No antagonistic activity against pathogens or opportunistic species could be measured in a soft agar test in algal pastes or living cultures.

To provide further insight on the effect of algal-MC on gut-MC a feeding experiment with cultured algae and algae paste was executed (subtask 4.3.2). Originally the experiment was to be carried out with first-feeding cod-larvae. However, as this experiment failed several times an alternative test was carried out with yolk sac larvae. The three tested microalgae species gave different mortality patterns when used in green water for yolk sac larvae. Algae cultivated at stationary phase induced mortality. Pastes of the three species had no negative effect on the survival.

The largest accumulation of live bacteria in RAS is found in decomposing sludge, which is mainly composed of metabolic feed wastes. Therefore, RAS sludge was evaluated for its potential to serve as a feed additive ((Task 4.4). Subtask 4.4.1. examined the composition of metabolic wastes from tilapia production, decomposing under different conditions of oxygen availability. Sludge was incubated for 22 days under aerobic, anaerobic and anoxic conditions and analysed for proximate-, MC- and volatile fatty acid composition. As expected, oxygen availability strongly influenced the microbial degradation process, resulting in significant differences between the different sludge types in all parameters measured. Particularly noteworthy was the ash content of incubated sludge which rose to 43-47%, being too high for practical use as a fish feed ingredient. Furthermore, yields after incubation were poor. Whereas fresh sludge represented 4.2 – 11.1 % of the feed input (DM basis), the sludge yield after incubation diminished to merely 0.08 and 1.5% of the feed input. Thus it would be preferable to use fresh sludge as feed additive or, instead, use culture systems which provide fish with direct access to sludge or bioflocs.

In a next experiment, the attractiveness of different sludge types for tilapia was monitored (Subtask 4.4.2). Fishes were exposed to aerobic, anaerobic and anoxic sludge offered as frozen pellets. By means of video recording, the approaching, eating and nibbling times were recorded. All types of bioflocs raised the fishes’ curiosity. Clearly, tilapia were most attracted to aerobic bioflocs compared to anaerobic and anoxic ones. This was shown by the shortest attraction-, longest eating- and nibbling time for the aerobic bioflocs.

Another sludge feeding experiment with tilapia was carried out to study the effect of ingested sludge microbiota on fish performance and body composition (Subtask 4.4.3). Groups of tilapia fry were fed diets incorporating either aerobic, anaerobic or anoxic sludge. A ‘pure’ diet without sludge supplements was fed to a control group. High mortality was occurred in all groups, ranging from 49 – 59%. The cause of mortality could not be determined, even after sending samples to a fish disease lab. Because mortality was uniform between treatments the experiment was continued. Fish of the control group showed the highest final body weight compared to the sludge incorporated diet treatments, but this difference was only significant for fish fed the anoxic sludge diet (P < 0.05). In terms of proximate composition, fish assigned to the four treatments differed predominantly in crude protein contents (CP). Tilapia fed the aerobic sludge diet showed the highest CP contents (807 g/kg DM CP) followed by anoxic, control and anaerobic-diet fed fish (736, 679 and 658 g CP/kg DM, respectively). Gut-MC composition did not differ between treatments. Hence, exposure to the different types of sludge through the diet had a minimum influence on growth.

Finally, the potential impact of microbial processes on environmental performance in aquaculture was reviewed (Task 4.5). First, the environmental performance of aquaculture production systems with focus on nitrogen and phosphorous losses was reviewed, secondly the role and place of microbial processes in RAS management was highlighted. It was concluded that the environmental impact from aquaculture can best be reduced by converting existing production systems from open to semi-closed or nearly zero-discharge systems through the gradual development of RAS technology. In RAS, the MC is crucial to water purification (e.g. nitrification, denitrification, fermentation) but also plays a decisive role for animal wellbeing. A balanced MC is important for maintaining probiotic organisms and reducing the risk of bacterial diseases.

Totally along the lines developed in this project, general recommendations can be made for further research. Three area of research can be identified and some specific examples for further research are given.

The first area of research is related to the quantification and the control of the amount of microbes reaching the larvae. A promising strategy that is still in need of further documentation is microbial maturation of the water and elimination of the gap in carrying capacity between component parts of the system being used. Better insight in the ongoing processes might give the fish farmer tools to gain microbial control.

The second area of research is related the control of the type of micro-organisms that are around in larval tanks. Community level K-selection of microbes by water treatment works in aquaculture systems can be considered as a strategic option, but more knowledge is needed on systems design and on the development of the microbial community composition under K-selection. The gut microbiota of marine fish larvae has a dynamic and transient nature. A strategy for the selection and the introduction of probiotic bacteria to larvae should ensure continuous supply. Special attention should be paid to the established live feed period, as this seems to be the most difficult phase to influence on the composition of the gut microbiota.

The third area of research is related to microbial metabolic impact on larvae. This relates to the vastly unexplored field of microbial metabolism of resident or transient microbes in the larval gut. It is very likely that microbes develop a different gene expression pattern once inside the gut, but it is also possible that life inside the gut (of resident micro-organisms) requires a change in the phenotypic status, a process that could be modulated by epigenetic control. Finally micro-organisms will have an influence on the larval metabolism, especially on the developing immune-system. Systematic research on that has been hampered by the stochastic control of the colonisation of the gut, making it difficult to reproduce experiments. In that respect a systematic investigation of the immediate and sustained effect of incoming or locally produced microbe-associated molecular patterns (MAMPs) is a necessity. Immunostimulation, by established and to be developed immunostimulants seems to be a method with potential, but still requiring further documentation on the effect of larval metabolism. Production scale verification will also be required.

Further studies on microbial management strategies should diversify methodologically and experimentally, preferentially in combination. This area has received considerable attention during the last two decades but hasn’t fully explored the potential presented in this report (Table 3 and 4). A relatively high fraction of studies has been on disinfection and probiotics. Some of these studies are very good, but there is a tendency of system specificity and documentation that makes generalization of limited value. Moreover, for e.g. studies on probiotics effect variables are recorded on a relatively crude level (survival and growth) and these global variables do not give insight into mechanisms of action.

Considering the crucial impact of microbes on larval development, further studies on microbial management should also consider the long-term impact of the treatments applied to fish larvae beyond metamorphosis.

Potential Impact:

Potential impact:

Scientific impact

PROMICROBE has established and/or refined the gnotobiotic larviculture of 3 important aquaculture fin fish species, namely: cod, sea bass and tilapia. These species represent carnivorous versus omnivorous fish as well as marine and fresh water fish.

In order to study the interactions between microorganisms and a host, gnotobiotic culture systems are indispensable. The protocols and set ups have been described in scientific publications that are open access (Dierckens et al., 2009; Forberg et al., 2011; Situmorang et al., submitted). This could help other research groups to start up or broaden research in this field or in other fields where bacteria may play a role (e.g. toxicology, nutrition, ontology, ecology, histology, immunology).

For all gnotobiotic model systems, a standard challenge test has been developed. This allows studying infection pathways, next to the (immune) reactions of the fish larvae (Rekecki et al., 2011).

Through the use of these gnotobiotic systems, the effect of different putative probiotic strains on survival of challenged and non-challenged sea bass and cod has been studied. There were ten putative probiotic strains available, five that were degrading quorum sensing molecules and five that did not have these characteristics. Each strain was tested solely and randomly composed mixtures with increasing strain diversity, from two strains to all ten strains, were tested to study the effect of probiotic diversity and probiotic mixture composition towards protection of challenged cod or sea bass larvae. It was clear from the results that increasing the diversity of the probiotic mixture did not lead to a better protection of sea bass or cod larvae within the tested range of species richness. In the cod experiments, the composition of the probiont mixtures seemed to be more important than the species richness. From the obtained results, it is advisable to continue further research towards development of probiotic mixtures that may be fish species dependant.

Socio-economic impact:

It is beyond any doubt that multi-resistant bacterial pathogens are a threat for human health. A lot of these pathogens emerge due to the (un)controlled use of antibiotics in the human food chain. This is also the case in fish production. The farmer cannot cope financially with the loss of too many fish, leading to the (prophylactic) use of antibiotics. There is more and more pressure to use sustainable ways of producing human food. However, in order to provide a good alternative, enabling fish producers to switch to more sustainable ways, there has to be a sufficient amount of knowledge, basic and applied. The PROMICROBE project has delivered a substantial amount of new insight in the effects/interactions bacteria have with 3 important European aquaculture fish species. This basic knowledge will trigger new applications but also direct field tests towards more founded goals. During PROMICROBE several ways to influence the already present microbial community were tested. This rather new approach of trying to join instead of trying to beat the present bacteria is a more feasible way, from a financial and a system/protocol point of view, for the industry. Diverting the microbial community can be achieved through feed compounds, like polyhydroxbutyrate, that selectively enhance certain bacterial genera lead to fish larvae with a microbial community that exerts a better protection against pathogens. Another way to steer the microbial community in a commercially feasible way, is through the microbial community of the water. The water microbial community can be controlled by the set-up of the culture e.g. recirculation culture (Attramadal, 2011) or the use of matured water (Skjermo et al., 1997). Methods that still are on the experimental stage are adding bacteria that interfere with the communication between pathogens, and thus may prevent disease outbreaks (Defoirdt et al., 2010).

By applying the above mentioned culture methods and feed additives, the industry can probably achieve higher survival and growth and improved robustness in fish larvae and juveniles.

The use of ‘in system production’ of bioflocs will lead to extra available proteins as feed for the cultured fish and at the same time results in a lower pressure on the water quality unit in a recirculation system.

The higher production will not lead to more pressure on the environment, as the results of the PROMICROBE project indicate that the use of recirculation systems and use of bioflocs lead to more stable and predictable production of fish juveniles. The use of closed systems where microbial community management can be fully obtained will lead to less waste water being released in the surroundings, but will also lead to less use of the available water sources. This is certainly important in the culture of fresh water fish.

These results have been published in ‘popularising’ journals.

As the production of fish juveniles will become more predictable, employment in this sector may be more secure and new job opportunities may be created for persons with different educational levels.

The next step could be to expand the novel protocols and set-ups and to integrate these into the food/feed chain. This would allow a significant improvement of the survival of larvae of commercially interesting fish species (sea bass, cod and tilapia) in different European countries.

The value driver for hatcheries to imply the approach to join rather than trying to beat the bacteria and convert the set ups increasingly into RAS systems will improve the economics of their activities. The attractiveness of such scenario can be illustrated with an example (i) the current annual production of marine fish fry in Mediterranean Europe amounts to 750 million bream and bass; (ii) the average market value per larva is 0.2 euro and the average production cost is estimated at 0.12 euro; (iii) the average survival rate is 20%; (iv) if our novel proprietary functional food can improve the survival rate to 40%, the impact can be calculated as follows: for every 1000 larvae, 400 will survive instead of 200 and this implies that you can produce twice the amount of larva with roughly the same production cost or you can produce the same volume of larvae at roughly half the production cost.

Extrapolation of this equation to the European market implies a yearly cost-saving of approximately 45 million Euro.


Attramadal, K. (2011): Recirculation, microbial control strategy for intensive marine larviculture. Global Aquaculture Advocate, July/August 2011, 63-65.

Defoirdt, T., Ruwandeepika, H., Karunasagar, I., Boon, N., Bossier, P., 2010. Quorum sensing negatively regulates chitinase in Vibrio harveyi. Environmental Microbiology Reports, 44-49.

Dierckens, K., Rekecki, A., Laureau, S., Sorgeloos, P., Boon, N., Van den Broeck, W., Bossier, P. (2009). Development of a bacterial challenge test for gnotobiotic sea bass (Dicentrarchus labrax) larvae. Environmental microbiology 11 (2):526-533.

Forberg T, Arukwe A, Vadstein O. 2011. Two strategies to unravel gene expression responses of host-microbe interactions in cod (Gadus morhua) larvae. Aquaculture Research 42: 664-676.

Forberg, T, Arukwe, A and Vadstein, O. 2011. A protocol and cultivation system for gnotobiotic Atlantic cod larvae (Gadus morhua L.) as a tool to study host microbe interactions. Aquaculture 315: 222–227.

Forberg, T, Vestrum, R.I. Arukwe, A and Vadstein, O. (2012). Bacterial composition and activity determines host gene-expression responses in gnotobiotic Atlantic cod (Gadus morhua). Veterinary microbiology 157 (3-4): 420–427.

Rekecki, A., Ringø, E., Olsen, R., Myklebust, R., Dierckens, K., Bergh, Ø., Laureau, S., Cornelissen, M., Ducatelle, R., Decostere, A., Bossier, P., Van den Broeck, W. (2011). Luminal uptake of Vibrio (Listonella) anguillarum by shed enterocytes – a novel finding of early defense strategy in larval fish. J. Fish Dis. 35 (4): 265-273. DOI: 10.1111/j.1365-2761.2011.01342.x

Skjermo, J.; Salvesen, I.; Oie, G.; Olsen, Y.; Vadstein, O. (1997). Microbially matured water: A technique for selection of a non-opportunistic bacterial flora in water that may improve performance of marine larvae. Aquaculture International 5 (1): 13-28. DOI: 10.1007/BF02764784

Situmorang, M., Dierckens, K., Bossier, P. Development of a bacterial challenge test for gnotobiotic Nile tilapia (Oreochromis niloticus) larvae. (Submitted to Diseases of Aquatic Organisms).

List of Websites:

• The address of the project public website, if applicable as well as relevant contact details.

Project public website:

Project logo:

Contact details: Contact Person:

Dr. Stamatis VARSAMOS (EC Scientific officer)

DG RTD, Unit E4: Agriculture, Forests, Fisheries, Aquaculture
SDME 8/92
Square de Meus, 8
B 1050 Brussels

Prof. Dr. ir P. Bossier (Project coordinator)

Lab for Aquaculture & Artemia Reference Center
Ghent University
Rozier 44, 9000 Ghent

Dr. F.-J. Gatesoupe
Department of Physiology of Marine Organisms (Brest)
ZI de la Pointe du Diable - CS 10070 - 29280 Plouzané

Prof. Dr. M. Verdegem
The Aquaculture and Fisheries group
Wageningen University

De Elst 1

6708 WD, Wageningen

The Netherlands

Prof. Dr. O. Vadstein
Department of Biotechnology
Norwegian University of Science and Technology
Sem Sælandsvei 6/8

7034 Trondheim


Dr. Inga Marie Aasen
Department of Biotechnology
SINTEF Materials and Chemistry
R. Birkelands vei 2B

Box 4760 Sluppen, 7465 Trondheim


Dr. J. Skjermø
Department of Marine Resources technology

SINTEF Fisheries and Aquaculture

Brattørkaia 17C
7010 Trondheim

Prof. Dr. F. Vanbreusegem

Department of Plant Systems Biology

Technologiepark 927, 9052 Ghent