INTRODUCTION
The rearing of aquatic animals and plants for human consumption, i.e., aquaculture, is a rapidly growing industry due to the increasing demand for high-quality protein to feed the growing world population (
1,
2). In finfish rearing, animals undergo multiple developmental stages from eggs through larvae to juveniles and, finally, adult fish. During this process, fish are supplied with different food sources, which may include live feed such as algae, rotifers,
Artemia, and copepods. The live feed itself feeds on microalgae, and hence, multiple trophic layers are involved. In many cases, fish farming has become a high-throughput process, with individual companies producing thousands of tons of fish annually, and by 2030, fish production from aquaculture is expected to reach an annual output of 109 million tons (
2). Along with the sustainable development goals of the United Nations, there is an increasing focus on the sustainable production of food, ending hunger while protecting wild fish populations (
3).
One of the major bottlenecks in fish production is disease outbreaks, and approximately 55% of infections are caused by pathogenic bacteria (
4) that are typically introduced with supply water (
5), broodstock, humans, and plankton feed (
6). Particularly, vibrios such as
Vibrio splendidus,
Vibrio harveyi,
Vibrio vulnificus, and
Vibrio anguillarum are of major concern to mariculture facilities, as they cause severe fish diseases and mortalities (
7,
8). This is predominantly an issue related to marine fish larvae where several species are reared in nutrient-rich green water tanks, feeding on live feed (
6,
9,
10). Pathogenic
Vibrio spp. are naturally associated with zooplankton (
11–14), and they can also easily proliferate in cultures of phytoplankton used as feed for the live feed (
9). Thus, live feed organisms can act as vectors of opportunistic pathogenic vibrios.
Major crashes of fish larval populations are most likely due to detrimental interactions (dysbiosis) in the microbial communities associated with the fish larvae (
15). Microbial communities respond quickly to changes in environmental factors such as oxygen concentrations, nutrient levels, pH, salinity, or accumulated toxic compounds (
16–19). In the event of an imbalance in the system, e.g., rapid increase in nutrient levels and temperature, opportunistic pathogens proliferate. The pathogens have traditionally been controlled by disinfection of the rearing tanks (
20), sterilization of the rearing water (
21), deployment of antibiotics (
22), and, in recent years, by vaccination of the fish (
23). However, the latter is not effective at the larval stage due to their underdeveloped immune systems (
23). Thus, sustainable alternatives to antibiotics are sought (
22–24). One proposed alternative is the use of probiotics, which are live microorganisms that provide a health benefit to the host when administered in adequate amounts (
25). The potential application of probiotics in aquaculture as a prophylactic and acute treatment of disease outbreaks has been studied, primarily focusing on the gut microbiome of the farmed animal (
26). Currently, the majority of commercially available probiotics for aquaculture are based on mono- or mixed cultures of
Firmicutes (
3), which have been successful in humans and livestock, although not adapted to the diverse aquatic environments.
Proteobacteria such as
Shewanella spp. and tropodithietic acid (TDA)-producing members of the
Roseobacter group have been studied extensively for their bioactivity and probiotic potential (
3,
27–29) as alternative candidates of marine origin.
Pure TDA is a bactericidal antibiotic at high concentrations (
30), but it can also act as a signaling molecule (quorum sensing) affecting behavior such as motility, biofilm formation, and secondary metabolite production in TDA producers (
31). Its antibacterial effect is due to its ability to act as an antiporter (
32), and resistance does not develop easily (
30,
33).
Roseobacter organisms producing TDA are indigenous to microbiomes of marine eukaryotes, including micro- and macroalgae (
34,
35), zooplankton (
36), sponges (
37), and mollusks (
38,
39), where they are believed to control the abundance of pathogenic community members. However, increasing their abundance has only had subtle effects on the microbiomes associated with microalgae and oysters (
40). While some community members were unaffected,
Vibrio spp. and
Pseudoalteromonas spp. were shown to diminish in the presence of
Phaeobacter inhibens (
40), and the presence of TDA or TDA producers has also caused a decrease in the relative abundances of closely related members of the
Rhodobacteraceae family in microalgal microbiomes (
41,
42).
TDA-producing
Phaeobacter spp. of the
Roseobacter group have been isolated from multiple mariculture facilities rearing mollusks and various marine finfish (
38,
43,
44), and they have been able to antagonize fish pathogens in live feed cultures (
9,
44,
45) without affecting growth or survival of the feed organisms (
9,
10). Most importantly, addition of 10
6 to 10
7 CFU ml
−1 can decrease mortality of turbot and cod larvae when challenged with pathogenic vibrios (
9,
46,
47). The selective impact on host-associated microbiomes (
40), along with the lack of resistance development despite their global occurrence in microbiomes (
30,
33), including mariculture microbiomes, highlights the applicability of
Phaeobacter spp. as probiotics. However, perturbations with probiotic levels of
Phaeobacter spp. could potentially cause imbalance and thereby give rise to proliferation of other pathogens than vibrios. Furthermore, little is currently known about mariculture microbiomes, and hence, the purpose of this study was to investigate how the addition of TDA-producing
P. inhibens affects microbiomes related to host organisms present in the food webs found in mariculture systems.
DISCUSSION
TDA-producing roseobacters can function as probiotics in mariculture due to their efficient killing of common pathogens (
9,
44,
45) and protection of vibrio-challenged fish larvae (
9,
46,
47). Since the natural microbiomes in these systems are imperative for the health of, e.g., algae (
48), addition of probiotics should selectively remove pathogens while leaving the commensal majority of the population unaffected. Our results demonstrate that the impact of
P. inhibens on the microbiome is highly dependent on the commensal microbiome composition and the inoculation level, with a greater impact on the bacterial community structure at the lower levels of the food web.
Three eukaryotes, green microalgae, copepods, and turbot larvae, were selected to represent different important levels in the food web, including feed for live feed, live feed, and reared fish, found in mariculture. Several studies have been conducted on microalgal microbiomes and how roseobacters interact with these unicellular eukaryotes (
49–51). Despite
T. suecica being widely used and produced in hatcheries, the microbial community associated with this microalga is not well studied. Biondi et al. (
52) observed that the
T. suecica microbiome was dominated by
Proteobacteria, particularly members of the
Roseobacter group, but also
Rhizobiales and
Bacteroidetes (
Flavobacteriales). This is similar to our findings and also to findings for another mariculture-relevant microalgal genus,
Nannochloropsis (
53). The
A. tonsa microbiome was also dominated by
Proteobacteria, particularly
Gammaproteobacteria, in this study. This has previously been observed in copepods from the North Atlantic Ocean (
54). Cultivation-based methods have found that
Vibrio spp. was dominating (
14,
55); however, the order
Vibrionales was below the 2% relative abundance cutoff in our community composition analysis, indicating that the relative abundance of these bacteria is likely overestimated in cultivation-dependent studies. Moisander et al. (
54) also observed that
Rhodobacteraceae dominated the transient food microbiome and proposed that they contribute to copepod nutrition. Members of the
Rhodobacterales order were also found at high relative abundances in our copepod system, though
Alteromonadales and
Oceanospirillales bacteria exhibited the highest relative abundances. These differences in the composition of the copepod-associated community are likely due to differences in the composition of the bacterial community in the immediate environment (natural versus laboratory cultivation) and to different methodologies applied (cultivation dependent versus cultivation independent).
The culturable bacteria of turbot eggs and larvae have been studied for decades, and the isolates have been dominated by members of the
Vibrionales and
Aeromonadales orders (
56,
57). While we observed
Vibrionales in the egg microbial community, the
Aeromonadales were not abundant (below the 2% relative abundance cutoff) in any of the samples. In contrast, we observed high relative abundances of
Alteromonadales. However, most studies on aquaculture microbiomes are cultivation based, and poor correlations between culture-dependent and -independent microbiome investigations were also observed by Fjellheim et al. in cod larval microbiomes (
58).
In concordance with a previous study of the microalgal
Emiliania huxleyi microbiome (
40), we found that the effects of probiotic treatment did not cause major changes to the community across orders. Interestingly, the addition of
P. inhibens reduced the relative abundance of closely related taxa from the
Rhodobacterales order in the copepod and fish larvae microbiomes, while they were unaffected in the microbiome of
T. suecica. Effects on closely related
Rhodobacterales members have been observed in the haptophyte
E. huxleyi microbiome exposed to the same
P. inhibens strain (
40). Several genera of the
Rhodobacteraceae family, namely,
Sulfitobacter,
Phaeobacter,
Pelagicola, and
Loktanella, were reduced or absent in the presence of another TDA-producing strain of
P. inhibens (2.10 variant) in the diatom
Thalassiosira rotula microbiome (
42). In our study, among the 30 most abundant taxa in the copepod and turbot microbiomes, unclassified genera of the
Rhodobacteraceae family,
Ruegeria spp.,
Celeribacter spp., and
Pseudophaeobacter spp. decreased in relative abundance. Addition of pure TDA to cultures of
Nannochloropsis salina has also been shown to decrease the relative abundance of
Rhodobacteraceae at relatively low concentrations (31.25 to 500 nM) (
41), which could indicate that TDA is causing the observed decrease of
Rhodobacteraceae members in the copepod and turbot microbiomes. Potentially, production and/or sensing of TDA is involved in the interspecies competition within the
Roseobacter group in which specific species or strains are more susceptible than others.
Vibrio spp. and
Pseudoalteromonas spp. were reduced by the presence of
P. inhibens DSM 17395 in the microalgal
E. huxleyi microbiome (
40), though in the present study, the orders
Vibrionales and
Alteromonadales were unaffected compared to the controls. Majzoub et al. (
42) found that the
T. rotula microbiome exposed to a
P. inhibens 2.10 variant (NCV12a1), with reduced antagonistic effect, developed in the same way as microbiomes exposed to the original bioactive strain, and factors other than TDA could be responsible for our observations. Altogether, these results indicate that closely related roseobacters compete for the same niches and that the impact is dependent on the eukaryotic host as well as the abundance of the roseobacters present in the commensal microbiome. However, as the sampling of the coculture systems was different between the tested systems, planktonic bacteria could be washed away from the copepods and fish eggs/larvae during sampling and therefore not captured as highly abundant in the sequences, while more planktonic bacteria would remain in the pelleting of the microalgal culture. Further studies should reveal how specific interactions determine which species prevail.
All microbiomes had similar richness and diversity indices. Bakke et al. (
59) reported that richness and diversity varied throughout the life stages of cod larvae. While the turbot larvae in this study were younger, the alpha diversity measure was similar to the observations by Bakke et al. (
59). The richness and diversity of the rearing water (i.e., green water prepared with algae,
Nannochloropsis oculata, and paste) and live feed (copepod,
A. tonsa, and rotifers,
Brachionus ‘Nevada’) were much higher than observed in the larval microbiome (
59) and the live feed assessed in this study. However, this is most likely due to experimental differences; this study was conducted in laboratory, small-scale cultures, while the study by Bakke et al. (
59) was conducted in large-scale, aquaculture flowthrough systems. The addition of the probiotic treatment in this study had a slight yet significant effect on the alpha diversity of the live feed. Dittmann et al. (
40) observed that treatment with, and the inoculation density of,
P. inhibens DSM 17395 did not impact the richness and diversity of the low-complexity microalgal
E. huxleyi microbiome. In contrast, the more complex oyster microbiome increased in richness when
P. inhibens had been added to the system, though the diversity was unaffected. How the observed changes in alpha diversity of host-associated microbiomes translate to the overall health state of the host organism needs further scrutiny.
The microbial communities associated with the three microbiomes were generally dynamic and changed over time, which is in concordance with previous studies (
42,
59,
60). The addition of
P. inhibens had a significant impact on the microbiome structure of
T. suecica and
A. tonsa. In contrast, the microbiome associated with the turbot larvae was more affected by incubation time compared to probiotic treatment. All eggs hatched within the first 48 h of the experiment, and thereby, a sudden increase in nutrients has likely occurred. In contrast, no nutrients were added to the microalgal and copepod systems, and thus, nutrients from the medium and the eukaryotic hosts were slowly consumed, and competition likely increased. The minor impact of
P. inhibens addition to the turbot egg and larval microbiome would indicate that the addition of probiotics would not cause dysbiosis in a healthy larval microbiome and a subsequent population crash. To determine this fully, longer trials would have to be conducted, monitoring the overall health of the larvae. The minor impact on the larval microbiome might also mean that the probiotic is less efficient at this level.
Vibrio spp. are commonly reported as detrimental pathogens to fish larvae (
7,
8), while they are also part of the commensal microbiome (
56,
57). In this study, the high relative abundance of
Vibrionales in the turbot microbiome was due to relatively few ASVs belonging to the
Vibrio genus, and the abundance of these ASVs did not change regardless of treatment (from day 1 through day 4). We added
P. inhibens at concentrations at which a probiotic effect had been observed in previous challenge trials (
9,
47,
61). In those trials, vibrios were reduced in numbers if not kept at inoculation level (
9,
47,
61), depending on the initial concentration of
Vibrio spp. (
9). Combined, these results would suggest that the addition of
P. inhibens, or the presence of inherent, closely related taxa, can keep vibrios in the fish microbiome at a stable level; however, this does not necessarily eliminate potential pathogens from the system. The effect is likely dose dependent, which was observed in a previous study of the microalgal
E. huxleyi microbiome (
40). Altogether, these data emphasize the need for investigating the optimal addition of probiotic
P. inhibens in relation to dose and which part of the food web to add the probiotic treatment to in order to obtain the most efficient protection against opportunistic pathogens while keeping the effects on the commensal microbiome to a minimum. The addition of the probiotic is likely more efficient at the live feed stages such as microalgae or zooplankton where
P. inhibens establishes itself and changes the structure. However, results based on nested PCRs on low-biomass samples, such as those presented in this study, are prone to biases and vulnerable to contaminations. Hence, it is not possible to conclude yet whether any changes would be beneficial or detrimental to the microbiome or its function. Broader omics studies should elucidate this in the future.
In conclusion, the addition of TDA-producing P. inhibens caused significant changes to the microbiome structure of the live feed but had little effect on the order-level composition and various effects on diversity and richness of the microbial communities associated with microalgae and copepods. No effect was seen on the community structure associated with turbot larvae. Particularly, the relative abundances of closely related taxa from the Roseobacter group were reduced as a function of probiotic treatment, but only in the copepod and turbot larval microbiomes. Vibrio spp. were highly abundant in the turbot microbiome, and these were kept at a stable level, though not eliminated, which indicates that the probiotic effect toward vibrios is likely dose dependent. Hence, the effect of adding a probiotic bacterium such as P. inhibens to the microbiome of mariculture-related eukaryotes is not likely to cause major perturbations to the existing microbial communities.