Recirculating aquaculture systems (RAS) have been developed to reduce water usage through waste management, and so, making intensive fish production compatible with environmental sustainability (Piedrahita, 2003). However, these types of systems pose unique challenges in microbial community management, being extremely demanding to maintain a stable and healthy microbial community within the RAS environment (Schreier et al., 2010; Martins et al., 2013).

Microbiomes usually form specific communities in different physical and biological environments, with a dynamic and interactive nature crucial for the functioning and health of their hosts (Berg et al., 2020). Due to their dynamic nature, bacterial colonization in its host can be heavily influenced by diet and environmental conditions (Bledsoe et al., 2016; Wilkes et al., 2019). In fish, this translates, for example, in the role live feed plays in early development stages (Califano et al., 2017) as vectors for potential pathogenic bacteria of the genus Vibrio (Montanari et al., 1999; Olafsen, 2001). The gut microbiome has already been extensively studied due to its role in reinforcing the digestive and immune system of the fish (Talwar et al., 2018). The composition of the fish diet affects gut microbiome composition, thus different diets applied to the different stages of fish development are expected to influence gut microbial communities during its life cycle (Stephens et al., 2016). Because the live feed administered in the early stages of development is known for being relatively poor in nutrients, the richness of the fish diet is higher in later stages (with commercial feed), which is conflicting with the importance of early bacterial colonization (Yukgehnaish et al., 2020). Another factor contributing to an improvement in bacterial colonization is the transition to RAS systems, as the establishment of the fish microbiome can be affected by the change in environmental conditions, with fish developing different profiles after this period (Steiner et al., 2021).

There is a multiplicity of ecological processes in microbiomes that affect community assembly (Goldford et al., 2018), such as selective pressures and nutrient availability, which causes cross-feeding networks with microbes communicating and trading metabolites and services, especially relevant in anaerobic environments (Marx, 2009). On the other hand, competitive interactions may also play an important role in shaping host microbial communities (Coyte et al., 2015). In aquaculture, and RAS in particular, life cycle studies are still rare, although they are required to detect temporal changes of the microbiome along farming cycles to identify the core taxa for future modulation (Infante-Villamil et al., 2021).

As mentioned above, microbiome studies are important to better understand how pathogen outbreaks occur and identify dysbiosis events. The community in RAS, particularly in the biofilter (a sector for optimal but undifferentiated bacterial growth used for ammonia removal from the system), influences the farmed fish that is in constant contact with the water, with its own prokaryotic community (Laurent et al., 2000) that also provides continuity between different physical and biological environments (host and biofilter, for example). Therefore, in this complex and interactive environment, there is a risk that disruptions may cause pathogenic outbreaks by opportunist bacteria (Blancheton et al., 2013). Groups commonly associated with disease outbreaks in sole are the Tenacibaculum genus (Gourzioti et al., 2018), Vibrio (Austin, 2010) and Photobacterium (Toranzo et al., 2005). The first two have also been linked in a pathogenic dysbiosis event (Wynne et al., 2020). The family Mycobacteriaceae also includes a large number of pathogenic bacteria for a number of different fish species (Delghandi et al., 2020).

The prokaryotic community can also result in improved nutrition and effective disease control by inhibiting potential fish pathogens (Irianto and Austin, 2002). In aquaculture, several microbial species, mainly present in the fish gut and water, have already been identified as potentially probiotic with several health benefits such as improved fish productivity, resistance to diseases and increased immune functions (el-Saadony et al., 2021). Microbiome studies can then help to guide the best practices to promote the persistence of these agents (Borges et al., 2021). Some of the bacterial orders already identified as having potentially probiotic interest are Lactobacillales (Alonso et al., 2019) and Bifidobacteriales (Quigley, 2017). Additionally, the genera Bacillus (Kuebutornye et al., 2020), Roseobacter, Phaeobacter, Paenibacillus, Pseudoalteromonas, Alteromonas, Pseudomonas, Aeromonas, Arthrobacter, Clostridium (Ringø, 2020), Saccharomyces (Gaggìa et al., 2010), Streptomyces (Teng Hern et al., 2019), and Shewanella (García de la Banda et al., 2010)have also been linked to this activity.

Our goal in this paper is to start filling the gap on the microbiota analysis during fish life cycle in aquaculture. That is, to characterize the bacterial community along a farming cycle, accompanying a batch from egg to the pre-ongrowing stage. In this study we were able to evaluate the temporal microbiota progression across sole life cycle, providing a reference microbiota map for this species at different stages of development. In addition, we were able to determine how much of the sole microbiome is inherited from the egg stage, and how much is acquired in the different production stages. This work improved the background knowledge needed to develop future microbiome modulation in sole production. Additionally, the results presented here can have a direct impact in the production husbandry strategies.

The 16S rRNA gene sequencing dataset used had a minimum and maximum read counts per sample (after trimming) of 7,776 and 84,097, respectively. The mean read counts for all samples was 32,709, the complete list of read counts per sample is presented in Supplementary Table S2.

Recirculating aquaculture systems have a unique challenge in managing a stable and functional microbial community (Schreier et al., 2010; Martins et al., 2013), with communities that are crucial for the health of the host (Berg et al., 2020) that can be heavily influenced by diet and environmental condition (Bledsoe et al., 2016; Wilkes et al., 2019). To fill the gap in life cycle studies, crucial to improve microbiome managing strategies, we characterized the bacterial community along a farming cycle, from egg to pre-ongrowing juveniles, evaluating the temporal microbiota evolution.

We found that alpha-diversity indexes did not change throughout development, although previous studies on a different species (Atlantic cod), refer to a loss of bacterial species diversity when artificial feeding is introduced (Ringø et al., 2006). Besides the fish species difference, the mentioned study was also technically very different, since the bacterial diversity was explored through isolation (Ringø et al., 2006), possibly missing the difficult to cultivate members of the community. The lack of substantial changes in bacterial alpha-diversity observed over the course of our study, is worth underscoring since it suggests a crucial importance of the early stages of fish development in establishing microbial community diversity.

The most abundant phyla in this dataset, Proteobacteria, Bacteroidetes and Firmicutes are commonly found to be the most abundant in aquaculture systems (Bledsoe et al., 2016; Wilkes et al., 2019). At the genus level, it appears that there are no dominating genera across the bacterial community and there is some variability in the relative abundance of genera detected between duplicates of the same sample. This variability in the bacterial community composition may be a consequence of the formation of heterogenous physical and biological micro-environments within the fish host, with specific bacterial communities as described by other studies (Zhang et al., 2019; Sylvain et al., 2020).

The term “core microbiome” has become widely used in microbial ecology to describe the set of microbial taxa that characterize a host or environment of interest (Neu et al., 2021). In this work we used a shared core microbiome analysis to infer possible conserved ecological roles and found that it was composed of twelve genera, four of them (Allorhizobium-Neorhizobium-Pararhizobium-Rhizobium, Vibrio, Pseudoalteromonas and Tenacibaculum) present in all tissues analysed and in all growth stages. As mentioned before, two of them are potentially pathogenic (Vibrio and Tenacibaculum). One thing to keep in mind is that both the live feed and the border sole tissues collected are in permanent contact with the water, and when studying these frontier environments its complex to disentangle the host from the environment community. Indeed, three of these genera have already been identified in the water, tank biofilm and biofilter carriers in this aquaculture unit, Vibrio, Pseudoalteromonas and Tenacibaculum (Almeida et al., 2021).

Using Venn diagrams, we found that the inherited community had very few genera represented (2.2–2.9%), all of them included Tenacibaculum and Vibrio. However, to our knowledge, this is the first time this type of characterization is performed in an aquaculture setting. Studies that accompany the evolution of the microbiome are rare, although there are some studies that accomplished a similar characterization, but only in wild populations. In migratory wild salmon, there is a microbiota community destabilization in migratory phases of the life cycle (Llewellyn et al., 2016). Another study found that, although deep-sea anglerfish microbiomes are dominated by the same genera from larvae to adult, their characteristic bacterial bioluminescent symbionts were not present in the early stages and were acquired from the environment (Freed et al., 2019).

Two target groups were selected (potentially pathogenic and probiotic) as having the most impact during the sole life cycle in an aquaculture production batch. We found a sharp increase in the number of potentially probiotic genera when sole moved to the WE system (49 DAH), around 40 days after live feed was introduced in the diet. Commercial feed did not appear to substantially increase the number of potentially probiotic genera at 119 or 146 DAH. Despite carrying Pediococcus acidilactici in its formulation, the commercial feed B did not consistently increase the abundance of Pediococcus in the sole intestines fed with this diet (119 and 146 DAH). When considering the increased number of potentially probiotic genera in the fish tissues after day 49, it is worth noticing that most of these genera were not present in the feed itself (live or commercial). An explanation could be that components in these feeds may act as prebiotics, that is, nutrients that are not digested by the fish that may fortify certain components of the intestinal microbiota by stimulating the growth and the activity of particular bacteria (Ringø et al., 2010). Indeed, prebiotic supplementation has shown potential as a strategy to overcome chronic stress-induced disease susceptibility in farmed S. senegalensis (Azeredo et al., 2019). Although reaching its highest number at 119 DAH, the number of potentially probiotic genera drops abruptly at 146 DAH with no change in the feed, raising the question if it was a consequence of husbandry or an unsuccessful establishment of the potential probiotic community. We should note, also, that most of the prokaryotic diversity is found in the rare biosphere (Pascoal et al., 2020), a genetic pool mostly undetected with the sequencing depth applied in this study, and some rare taxa can remain rare while others may grow abundant when the conditions change. This seed bank can include low abundant pathogenic communities and its monitorization could be useful in early identification, but can also support host functions specific to the aquaculture environment (Pascoal et al., 2022). This shift from undetectable to detectable groups may happen when the production alters the diet (specially between feeds), as the nutrients available change, diversity of certain genera increases momentarily and then declines with the stabilizing environmental conditions, explaining the drop of potentially probiotic genera at 146 DAH. Much like in the human gut microbiome, a diverse diet provides a competitive advantage to low abundant taxa, and the more diverse the microbiome, the more adaptable it will be to perturbations (Heiman and Greenway, 2016). Studies in chinook salmon also found that the gut microbiome is shaped by the environment, both by water and by formulated feed (Steiner et al., 2021). However, high inter-individual variation suggests that the host physiology itself may affect the community structure as much as environmental conditions (Fossmark et al., 2021; Hossain et al., 2021). In our study, it has also been observed that some genera associated with nitrifying activity (e.g., Nitrosomonas and Nitrospira) increased their relative abundances when fish were introduced to RAS, during the pre-ongrowing stage (Supplementary Table S5). Other studies have also found colonization of this group in fish tissue under similar conditions (van Kessel et al., 2016). Most probably, this is a consequence of nitrifying groups circulating from biofilters to the different compartments of the RAS unit, where they were found to occur (Almeida et al., 2021).

For the potentially pathogenic genera, Tenacibaculum and Vibrio, they appear to be acquired at the egg stage, accompanying the sole microbiome through its development, from egg to 146 DAH. The other two, Photobacterium and Mycoplasma, appear to be detectably colonizing later in the life cycle. In this study they have been identified in the rotifers and brineshrimp and thus the live feed could be a potential vector as has been previously demonstrated (Hurtado et al., 2020). This early diet driven microbiome development can have a significant impact in the future fish microbiome (Wilkes et al., 2019). Differentiating which pathogenic genera are inherited from those that the fish acquires throughout production is paramount. By identifying where in the production, the fish is exposed to these groups, husbandry improvements can be implemented to control them. However, if these pathogens are inherited from a wild broodstock, it may be difficult to safely remove them in a sustainable way. However, it is important to have in mind that the genera included in this study are potentially pathogenic, but are not composed solely by pathogenic species. In fact, the genus Vibrio is an important ecological marker, as it is widely abundant in riverine, estuarine, and marine aquatic environment (Hurtado et al., 2020) and one of the most diverse marine bacterial genera (Gomez-Gil et al., 2014). In the case of Tenacibaculum, out of 28 total species (Parte et al., 2020), only seven are generally associated with disease outbreaks: T. maritimum, T. soleae, T. discolor, T. gallaicum, T. dicentrarchi, T. finnmarkense, T. ovolyticum (Fernández-Álvarez and Santos, 2018). There are technologies available that may help to increase the definition for pathogenic species identifications. For example, long read sequencing with Pacific Biosciences (PacBio) or Oxford Nanopore Technology (ONT). These technologies had recent advancements that now provide higher accuracy and can provide up to 60 kb reads (Hu et al., 2021), almost enabling the sequencing of the complete 16S rRNA gene. Additionally, after the detection of groups of interest by an overall sequencing approach, like the one implemented in this study, targeted approaches to detect pathogenic markers of a subset of species could be implemented in a production setting to provide confirmation of pathogenicity. However, this was out of the scope of this work, which aimed at understanding the microbial progression along a fishery life-cycle, and not the occurrence of disease/stress.

In the correlation matrix, we found that six genera with potential probiotic activity were significantly negatively correlated with Vibrio, two of them, Bacillus and Streptomyces have already been described as potential inhibitors of Vibrio pathogen species (Vaseeharan and Ramasamy, 2003; Teng Hern et al., 2019). Only one genus had a negative correlation with Tenacibaculum (Pseudomonas), two with Mycoplasma (Pseudoalteromonas, Streptomyces) and two with Photobacterium (Phaeobacter and Alteromonas). For this analysis, we must recognize the potential biases in NGS community correlation studies that may result in misleading positive correlations, derived from the fact that more taxa are detected in deeply sequenced samples and therefore taxa co-vary with sequencing depth (Faust et al., 2015). Attesting to this, the correlation matrix shows a positive interaction between Shewanella and Photobacterium, however it had been amply reported that the first increases resistance to the later (García de la Banda et al., 2010, 2012; Vidal et al., 2016). These detected correlations can be useful to unveil possible interactions, but more studies are needed to confirm or discard mechanistic hypotheses. Also, it is relevant to note that Photobacterium has a total of 14 positive correlations with potentially probiotic taxa, which might also be a consequence of the positive bias. A similar observation occurred between Phaeobacter and Tenacibaculum (Edward et al., 2022). With the limits of these techniques, it is unreliable to distinguish between positive bias from non-specific potentially probiotic activity with positive correlations between pathogenic and potentially probiotic taxa in our data (Phaeobacter and Vibrio) or cases like Bacillus that shows a negative correlation with Vibrio but a positive one with Photobacterium. The genera Streptococcus, Phaeobacter and Limosilactobacillus also have a similar behavior. Four genera had exclusive positive correlations with the potential pathogenic bacteria (Alteromonas, Pseudomonas, Gardnerella, Lactobacillus), therefore these might be the most promising for future empirical studies.

This work aimed to describe the sole microbiome development throughout the production cycle in a RAS. Through a description of the inherited and acquired community in the different tissues analysed at different production and life stages, we hope to promote the emergence of life cycle studies in aquaculture and to underscore its applicability. We found that the bacterial community was significantly altered throughout the Solea senegalensis early development. Two potentially probiotic genera were inherited from the egg stage (Bacillus and Enterococcus), but the main increase in potentially probiotic abundance and diversity occurred around 40 days after live feed was introduced in the diet (at the weaning stage). Notwithstanding this increase, the establishment of this community in the following development stages was not successful. Regarding potentially pathogenic genera, two appear to be inherited (Tenacibaculum and Vibrio), and two are suggested to be acquired during production (Photobacterium and Mycoplasma). These results are relevant, because acquired potentially pathogenic groups may be prophylactically treated with improvement in husbandry conditions, but those that are inherited from the egg stage may be difficult to safely eradicate.

Our study has conducted a comprehensive description of the bacterial community in different life cycle stages of the Solea senegalensis, to our knowledge, the first of its kind. By analyzing the composition of this community, particularly with the definition of key target groups and the definition of the inherited and acquired community in the production cycle, we have highlighted the importance of whole life cycle studies to understand the vulnerability of the stages of fish production with a direct impact in husbandry strategies. The shifts in the composition of key components of Solea senegalensis gut microbiome during its life cycle, open important questions related to the functional significance of the observed taxonomic changes in terms of potentially probiotic activity and pathogenic incidences in the life cycle of this fish that must be explored in future investigations.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: European Nucleotide Archive (ENA) repository, accession number PRJEB55703.

Ethical review and approval was not required for the animal study because the animals used in this work were not subjected to any experimental protocol and were a part of the routine procedures of a commercial hatchery facility. All animals were handled by the fish farm employees, the euthanasia method used was an anaesthetic overdose of the commercial anaesthetic Aquacen benzocaine 200 mg/ml (CENAVISA, S.L., Spain) following SEA EIGHT’s Veterinary Plan. According to the Portuguese legislation DL N° 113/2013, this work is exempted from the need for ethical approval. All methods are reported in accordance with ARRIVE guidelines.

DA, MS, CM, IB, and AM had substantial contributions in the conception and design of the work. DA and IB were responsible for the acquisition of the samples. DA and MS for the analysis. DA, MS, CM, and AM in the interpretation of the data. DA drafted the first manuscript. MS, CM, IB, and AM revised it critically. MS, CM, and AM provided final approval for publication of the content. All authors contributed to the article and approved the submitted version.

This work was funded by the project ATLANTIDA (NORTE-01-0145-FEDER-000040), supported by the North Portugal Regional Operational Program (NORTE2020), under the PORTUGAL 2020 Partnership Agreement and through the European Regional Development Fund (ERDF). DA was supported by the Ph.D. grant with the reference PD/BDE/135542/2018, and Safiestela Sustainable Aquafarming Investments, S.A. (part of the SEA EIGHT group). MS was supported by the project 39948_FeedMi for this work, supported by Portugal and the European Union through FEDER/ERDF, CRESC Algarve 2020 and NORTE 2020, in the framework of Portugal2020. AM and CM were supported by the Strategic Funding UIDB/04423/2020 and UIDP/04423/2020 through national funds provided by FCT and ERDF.

DA and IB were employed by Safiestela S.A., SEA EIGHT group at the time of the study.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.