The supplementation of a prebiotic improves the microbial community in the gut and the skin of Atlantic salmon (Salmo salar)

Abstract

Aquaculture growth is hindered by an increasing number of challenges, primarily infectious diseases and inappropriate or unsustainable fish nutrition. Hence it is critical to develop novel prevention strategies to minimise infectious diseases and pharmaceutical interventions. Nutritional challenges and the health of the fish could be improved by managing their microbial communities. Microbiomes can play a crucial role in fish physiology, particularly in digestion, by metabolizing largely indigestible feed components for the host or synthesis essential micronutrients. Beyond their nutritional role, microbiomes are considered the first line of defence against pathogens. In this study, a novel prebiotic mix (Selectovit), composed of 1,3/1,6-beta glucans, mannan-oligosaccharides, nucleic acids, nucleotides, medium chain fatty acids and single chain fatty acids, was tested at different inclusion levels (0.0; 0.5; 1.0; 2.0 g/kg) in the diet of Atlantic salmon (Salmo salar). Using experimental feed trials and 16 S rRNA microbiome profiling, the impact of the prebiotic blend on fish growth and microbial community within both the gastrointestinal tract and the skin was assessed. Overall, the supplement showed no significant impact on growth. However, we clearly demonstrate that the prebiotic can significantly manipulate the microbial community of the distal intestine and the skin. Several potential beneficial bacteria such as Bacillus and Mycoplasma spp. were significantly more abundant in the prebiotic-fed groups compared to the control. In contrast, putative pathogenic bacteria were less abundant in the salmon fed the prebiotic blend. Interestingly, the supplement induced more changes in the skin than in the gut. There is growing evidence in fish for highly complex interactions between the microbial communities of the digestive system and external mucosa, and with the host immune system. Further research in this field could lead to the creation of novel bacterial biomarkers and new non-invasive strategies for fish digestive health monitoring.

Highlights

• •Improving mucosal health and commensal microbial communities in the intestine of fish.
• •Prebiotics influence non-target tissues including the skin – a barrier to pathogens.
• •Using 16 S rRNA profiling, we demonstrate positive effects on beneficial bacterial taxa.
• •Prebiotic additive reduced abundance of putative pathogenic genera in the gut and skin.
• •Strong interaction between the skin microbiome and the diet of Atlantic Salmon.

1. Introduction

In the last decade, aquaculture production has been the fastest-growing animal food production sector (FAO, 2020). Despite steady growth, future expansion is hindered by an increasing number of challenges, primarily: (1) infectious diseases and (2) inappropriate or unsustainable fish nutrition (Shinn et al., 2018). Infectious diseases are considered the greatest threat to sustainable aquaculture production development. High mortalities and losses in production volume, resulting from disease outbreaks, are a substantial risk for global food security and represent an increasing challenge for maintaining animal welfare (Bank, 2014, Assefa and Abunna, 2018). Hence, it is critical to develop prevention strategies to minimise infectious diseases, in addition to treating diseased fish. Vaccinations have been one of the most efficient strategies for some pathogens (Ma et al., 2019). However, due to the involved costs, the application of vaccines is limited in many countries. In addition, vaccines are not available yet for all diseases and/or fish species (Adams, 2019). Therefore, alternative preventative strategies are urgently needed to promote high animal welfare standards on farms.

The second major challenge to the aquaculture industry is sustainable nutrition (Hasan and Halwart, 2009). Feeds typically represent the single greatest production cost to fish farms (Asche and Oglend, 2016). Due to the rapid growth of aquaculture production, and widely exploited wild fish stocks, marine-derived fishmeal and fish oils are limited and increasingly expensive feed ingredients (Hodar et al., 2019). Plant-based feed components are often used as substitutes, but their use is constrained due to differing nutritional profiles to fish-based ingredients (Francis, Makkar and Becker, 2001). Fish have specific nutritional requirements, and as such inappropriate nutrition and plant-derived anti-nutritional factors can negatively impact growth performance, health, and welfare of the fish (Refstie et al., 2000, Pratoomyot et al., 2010, Al-Thobaiti et al., 2018).

Nutritional challenges and the health of the fish could be improved by managing their microbial communities. Microbial communities including bacteria, archaea and fungi – “microbiomes” – which colonise all mucosal surfaces of fish that are in contact with the environment, including the skin, gills and digestive tract (Merrifield and Rodiles, 2015). Genetic origin, environmental factors and diet are the main drivers of the microbial composition in fish (de Bruijn et al., 2018). Microbiomes can play a crucial role in fish physiology, particularly in digestion. For example, biosynthesis of exogenous enzymes by microbiota in the gastrointestinal tract increases digestion efficiency (Ray, Ghosh and Ringø, 2012). Moreover, certain bacteria and other microbes can metabolise plant starches and cellulose, that are largely indigestible to the host especially for carnivorous fish (Ni et al., 2014). In addition, bacteria can synthesise essential vitamins and other compounds, such as eicosapentaenoic acid (EPA), that contribute to the overall health of the host (Dailey et al., 2016, Yang et al., 2020). Therefore, there is rapidly growing interest in how fish microbiomes can be managed and enhanced for the benefit of aquaculture productivity (Kuebutornye et al., 2020).

Beyond their nutritional role, microbiomes contribute to host immune responses and disease resistance. Epithelial mucus, rich in microbes, is considered the first line of defence against invading pathogens. In fish and other vertebrates, the host immune system (both innate and adaptive pathways) can regulate the abundance of commensal and pathogenic bacteria and actively shape the microbiota composition (Merrifield and Rodiles, 2015, Chen et al., 2019, Zheng et al., 2020). Conversely, metabolites produced by bacteria (e.g., Polysaccharide A (PSA), Short Chain Fatty Acids SCFAs) can travel through the circulatory system of the host, act on immune cells and subsequently induce tissue specific immune responses, which may in turn contribute to resistance against invading pathogens (Merrifield and Rodiles, 2015, Xiong et al., 2019, Zheng et al., 2020). In addition, commensal microbes can compete with pathogens through niche exclusion, competition of essential resources, and/or the production of antimicrobial compounds also known as antibiosis (Kamada et al., 2013, de Bruijn et al., 2018). Healthy fish are typically characterized by a stable and highly diverse microbial community (Li et al., 2017). In contrast, diseased fish often show a loss of diversity - known as dysbiosis - and a high abundance of pathogenic bacteria (Stecher et al., 2013, de Bruijn et al., 2018). As our understanding of the crucial roles of microbiomes for fish increases, there is a growing interest to maintain and beneficially manipulate fish microbiomes to improve productivity, health and disease resistance in aquaculture.

Recent strategies to manage the microbial community include the alteration of diet composition (e.g., lipid and protein sources), and the supplementation of probiotics and prebiotic in diets (Egerton et al., 2018). Probiotics involve the application of live organisms, such as Bacillus or Lactococcus spp., to improve the balance between beneficial and harmful bacteria (Perumal, Thirunavukkarasu and Pachiappan, 2015). Their inclusion in diet has been shown to improve welfare (e.g., improved digestion, upregulation of immune genes, increased disease resistance) and growth performance in various fish species (El-Haroun, Goda and Kabir Chowdhury, 2006). In contrast, prebiotics comprise specific feed materials or feed additives that enhance the growth of specific microorganisms (Merrifield et al., 2010). A broad range of prebiotics have been tested in fish, including yeast cell wall derived components (e.g., 1,3/1,6-beta-glucans, mannan-oligosaccharides), yeast cytoplasm derived components (nucleic acids and nucleotides) or short- and medium-chain fatty acids.

1,3/1,6-beta-glucans and mannan-oligosaccharides have been widely tested in cultured fish species with supplementation typically improving growth performance, feed utilisation and immune activity (Abid et al., 2013, Selim and Reda, 2015, Jami, 2019; Staykov et al., 2007; Jami et al. 2019). In addition, mannan-oligosaccharides may reduce the multiplication of specific bacteria in the intestinal tract (Momeni-Moghaddam et al., 2015, Teng and Kim, 2018, Gainza and Romero, 2020). Crucially, fish fed with 1,3/1,6-beta-glucans and mannan oligosaccharides showed higher survival when challenged with Yersinia ruckeri (Selim and Reda, 2015). Nucleic acids and nucleotides are another promising group of micronutrients for enhancing fish health and improving digestive capacities. These have been shown to increase lysozyme activity, suggesting raised non-specific as well as specific immune response (Akhter et al., 2015, Xu et al., 2015). In rainbow trout, nucleotide supplementation resulted in higher survival when exposed to Vibrio anguillarum (Burrells, Williams and Forno, 2001). Other prebiotics such as the supplementation of medium-chain fatty acids and single chain fatty acids actively shaped the microbial community by increasing the abundance of several beneficial and health-promoting bacteria (Rimoldi et al., 2018). Most research studies focus on the interaction between supplements and the gut microbial community, and few studies have analysed the impact of feed prebiotics on microbiota outside the gastrointestinal tract. These non-gut mucosal microbiomes – such as on the skin and gills – can play a critical role in fish disease resistance (Boutin et al., 2013, Zhang et al., 2018). Moreover, many existing and emerging disease issues in aquaculture result in skin and/or gill pathologies (Schmidt, Thompson and Padrós, 2018). Therefore, novel management strategies to enhance these mucosal microbial communities may play an important role in aquaculture health.

The relative success of prebiotics varies and depends on a range of factors including dose, time of supplementation, production methods, culture conditions and fish species, but they hold promise as strategies to take control of the microbial community of cultured fish (Merrifield et al., 2010, Akhter et al., 2015). However, studies of their effects in combination, and/or the impact on microbial communities outside the digestive systems remains understudied. In the following study, a novel prebiotic mix, composed of 1,3/1,6-beta glucans, mannan-oligosaccharides, nucleic acids, nucleotides, medium chain fatty acids and single chain fatty acids, is tested at different inclusion levels in Atlantic salmon (Salmo salar). The main goal of the study was to assess, using experimental feed trials and 16 S rRNA microbiome profiling, how the prebiotic blend impacts fish performance and the overall microbial community of the host, within both the gastrointestinal tract and the skin.

2. Methods

2.1. Feeding trials and sampling procedures

Nutrition fish trials were carried out at the Pontus research facility (Aberdare, Wales), using Atlantic salmon parr (Salmo salar), supplied by Cooke Aquaculture (average weight 32 g). 384 salmon were acclimatised for two weeks in a recirculating aquaculture system (RAS) prior to start of feeding experiment. The system consisted of 12 × 200 L tanks, with flow rates 400 L/hr per tank, and 10 % water changes to maintain water chemistry parameters (see below). The RAS was equipped with a drum filter and filter bag for mechanical filtering, a biological moving bed filter, an ultraviolet disinfection unit, as well as a degassing tower, with oxygen supplementation via an oxygen cone. Animal handling procedures were approved by Pontus animal ethics committee.

At the start of trials, fish were randomly distributed into 12 tanks (200 L), with triplicate tanks per treatment group. Fish were raised for 12 weeks on one of four experimental diets with different inclusion levels (0.0, 0.5, 1.0, 2.0 g/kg) of a blend of potential health stimulants and prebiotics. The feed was formulated and produced in cooperation with Sparos and Chemoforma AG (Augst, Switzerland), and composed of yeast derived mannans and,1,3/1,6-beta glucans, nucleic acids and nucleotides, medium chain fatty acids as well as single chain fatty acids. The soy protein concentrate (SPC) and wheat were replaced as the inclusion of the prebiotic blend increases (Table 1). All diets were formulated to meet commercial standards and the principle nutritional requirements of Atlantic salmon parr (Table 1). The quality and nutritional profile of the feed was assessed by a proximate composition analysis (Stirling University). Fish were fed to satiation by hand, four times a day. Feed intake was monitored daily. Throughout the trial, tanks were exposed to an 8:16 h light: dark regime. Water parameters in the RAS system were maintained at 12.0 °C ( ± 1 °C), > 80 % oxygen saturation, pH 7.25 ( ± 0.15), < 0.03 mg/l ammonia, < 0.6 mg/l nitrite and < 75 mg/l nitrate, in accordance with optimal welfare conditions for salmon.

Table 1.

Feed formulation based on nutritional requirement of the parr stage of Atlantic salmon [SPC=Soy Protein Concentrate, MCP=Monocalcium phosphate].

Diet	1 [Control]	2 [0.5 g/kg]	3 [1 g/kg]	4 [2 g/kg]
Fishmeal [g/kg]	300.0	300.0	300.0	300.0
Maize gluten [g/kg]	200.0	200.0	200.0	200.0
SPC [g/kg]	133.0	133.5	133.5	133.5
Sunflower meal [g/kg]	100.0	100.0	100.0	100.0
Wheat [g/kg]	68.8	67.8	67.3	66.3
Fish oil [g/kg]	159.5	159.5	159.5	159.5
Rapeseed oil [g/kg]	16.0	16.0	16.0	16.0
Experimental Blend [g/kg]	0	0.5	1	2
Premix [g/kg]	10.0	10.0	10.0	10.0
MCP [g/kg]	12.7	12.7	12.7	12.7
Total	1000.0	1000.0	1000.0	1000

For the assessment of growth performance, batch weights (total biomass per tank) were taken at week 0, 6 and 12 of the trials. Fish were starved for 24 h prior to weighing. At the end of trial, ten random fish from each tank were individually weighed and measured for standard length. Growth performance was assessed using specific growth rate (SGR); percentage body weight gain per day. In addition, feed intake (FI) was calculated as percentage of body weight per day. Moreover, feed conversion ratio (FCR) was calculated as the ratio of feed intake to weight gain. Mortalities were recorded and summarized as overall survival ( %). For microbiome analysis, five random fish from every tank were sampled (15 per treatment group). Fish were anaesthetized with MS222 (Methansulfonate, 200 mg/L) and killed by the destruction of the brain. Microbiome swab samples were taken from the skin and from the distal intestine. Skin swabs were run along the entire lateral body surface six times. For intestine samples, the whole intestine was removed using a sterile dissection kit. A 1 cm long piece of the distal intestine was cut, opened, and faecal residues removed using sterile distilled water, followed by rubbing the mucosal surface with a swab. All swab samples were immediately frozen and stored at - 80 °C until DNA extraction.

2.2. DNA extraction, 16 S rRNA amplification and sequencing

Total DNA was extracted from each skin and intestine sample using the Qiamp DNA mini kit, following manufacturer instructions. Extracted DNA was stored immediately at − 20 °C. A subset of the samples was quantified using the Qubit BR DNA assays to verify successful DNA extraction. PCR amplification and library preparation were performed by 2-step PCR targeting of the V1-V2 region of the 16 S rRNA gene. First round of PCR amplification used 27 F (5’-AGAGTTTGATCCTGGCTCAG-3’) and 338 R (5’- TCTGCTGCCTCCCGTAGGAGT −3’) primers with addition of universal tails (Bohmann et al., 2021) and were performed in triplicates for each extraction sample. The PCR reaction volume was 25 ul, including 12.5 ul PCR mix (NEB Q5 Hotstart High fidelity PCR master mix), 0.5 ul of each primer (10 µM), 10.5 ul H₂O and 1 ul of DNA. The cycling protocol was as follows: 98 °C for 30 s, 35 cycles of 98 °C for 10 s, 55 °C for 30 s, 72 °C for 30 s and final elongation for 10 min. PCR products were visualised by agarose gel electrophoresis to ensure successful amplification. Negative controls for DNA extractions and PCRs, and a mock community as a positive control, were included for sequencing. PCR round 1 triplicates were pooled and cleaned using Agencourt AMPure XP beads according to manufacturer instructions. The second round of PCR introduced Illumina adapter sequences and unique indexes for samples identification (Bohmann et al., 2021). PCR round 2 conditions were as above using 15 cycles. Final PCR products were individually bead cleaned as above and quantified using Qubit BR DNA assays. Cleaned libraries were pooled equimolarly and sequenced using an Illumina MiSeq v2 2 × 250 bp run at Bangor University Centre for Environmental Biotechnology. Raw sequence data are available at the NCBI Short Read Archive (SRA) under accession PRJNA800661.

2.3. Bioinformatics & statistics

Paired end demultiplexed sequencing reads were imported into Quantitative Insights Into Microbial Ecology 2 (QIIME2). Sequences were then quality filtered, trimmed, dereplicated, chimeras rejected, and pair-end reads merged in QIIME2 using DADA2 with standard settings. Reads were clustered by 99 % identity using the de-novo function. Classification of Amplicon Sequence Variants (ASVs) was performed using a scikit-learn naive Bayes machine-learning classifier trained using sequences representing the bacterial V1 – V2 rRNA region available from the SILVA database (https://www.arb-silva.de/download/archive/qiime;Silva_132), and taxonomic classifications were based on the q2-feature classifier in QIIME2. The classifier then assigned taxonomic information to representative sequences of each ASV. The QIIME2 output was further processed in RStudio (Version 4.0.3) with the package “phyloseq”(McMurdie and Holmes, 2013). Rarefaction analysis was used to determine sufficient read depth, samples with less than 10 000 sequences were excluded. Subsequent filtering excluded taxa with less than 100 reads and taxa found in only one sample. R was used to analyse significant differences in alpha (Pairwise Wilcoxon signed-rank test) and beta (Pairwise Adonis) diversity measures. Significant differential abundance of ASVs between fish fed the prebiotic blend and the control (no prebiotic) was determined using DESeq2 (FDR-corrected p < 0.05). In addition, significant differential abundances of all taxonomic levels (non-parametric Wilcoxon test, FDR-corrected p < 0.05) were assigned and visualised using the Microbiome Analyst heat trees(Chong et al., 2020). Growth performance indicators were tested in R for normality (Shapiro-Wilk test) and homogeneity of variance (Levene’s test). If normality and homogeneity was confirmed, significant (p < 0.05) differences between treatment groups were determined using a Tukey pairwise pos hoc analysis of the Anova results.

3. Results

Over the 12-week feeding trial period, survival was greater than 95 % in all treatments with no significant difference among groups (1 = 95 %, 2 = 98 %, 3 = 97 %, 4 = 98 %; Table 2). Specific growth rate (SGR) was marginally lower in the two highest inclusion level diet treatments; however, these differences were non-significant (Table 2). Feed intake and food conversion ratios were also non-significant among treatments.

Table 2.

Summary of average growth performance indicators of salmon parr on experimental diets. SGR = Specific Growth Rate, FI=Feed Intake, FCR=Feed Conversion Ratio, Standard deviation added.

Diet	1 [Control]	2 [0.5 g/kg]	3 [1 g/kg]	4 [ 2 g/kg]
SGR ( % bw d-1)	0.92 ± 0.04	0.90 ± 0.07	0.83 ± 0.05	0.87 ± 0.03
FI ( % bw d-1)	0.77 ± 0.01	0.76 ± 0.03	0.74 ± 0.03	0.73 ± 0.03
FCR (kg feed/kg gain)	0.83 ± 0.01	0.86 ± 0.06	0.91 ± 0.02	0.86 ± 0.07
End weight (g)	98.95 ± 1.07	95.04 ± 5.95	91.37 ± 1.66	91.70 ± 2.11
Survival ( %)	95.83 ± 1.80	98.96 ± 1.47	97.92 ± 1.47	98.96 ± 1.47

A total of 9.5 million raw read pairs were produced from the 115 sequenced samples. After filtering and data pre-processing, a total of 7 million reads (average reads per sample 65 371, range = 12 820 – 119 551) were retained. Rarefaction curves determined samples with less than 10 000 reads should be removed due to insufficient read depth (all samples passed this criteria). For diversity tests, skin samples were rarefied to 12 820 reads and gut samples were rarefied to 22 247 reads per sample. In total, 266 ASVs for the gut and 330 ASVs for the skin were retained for further analysis.

In the skin, Shannon diversity of the third treatment (inclusion 1 g/kg) was significantly lower compared to all other experimental groups, (Fig. 1 A). No significant differences were observed between the other three groups. No differences could be determined for Chao 1 richness. In gut samples, we found no significant difference in either diversity measures, (Fig. 1 B).

Fig. 1.

Diversity of the microbial community in skin and the gut under prebiotic inclusion levels (green = 0.0 [control], orange = 0.5 g/kg, purple = 1.0 g/kg, blue = 2.0 g/kg), Alpha diversity measured by Chao1 and Shannon indices in the A) skin and B) gut. NMDS ordination of Beta diversity of C) skin and D) gut communities. Ellipses indicate 95 % confidence.

Test of beta diversity of the distal intestine samples indicated significant group segregation (p < 0.05). This was largely driven by differences in the 1 g/kg inclusion group, (Fig. 1 D). The communities of this group were significantly different to the control group (p = 0.003) and the lowest inclusion level of the prebiotic (p = 0.003). Similar to the gut, within the skin samples the third treatment group was the biggest driver of group differences, although not significant (Fig. 1 C).

Proteobacteria were found to be the dominant phylum of microbial communities in both the gut and the skin (Fig. 2). At genus level, the skin microbial community is largely dominated by Acinetobacter (38 %), followed by Pseudomonas (15 %) and Alkanindiges (5 %, Fig. 2 B), whereas the gut has a more evenly distributed community (Fig. 2 A), with Pseudomonas (22 %), Cupriavidus (9 %) & Janthinobacterium (8 %) the most abundant taxa.

Fig. 2.

Relative abundance of the top 20 genera of the microbial community in the A) gut and B) skin, colour shades separate taxa at Phylum level.

In comparison to the control group, the feed supplement induced more differences in the microbial community for the skin compared to the gut using both differential abundance tests (gut: 27 ASVs differentially abundant vs. skin: 44 ASVs, Figs. 3 & 4). In the skin samples, Proteobacteria and Bacteroidetes were significantly less abundant in fish fed the prebiotic blend, while Actinobacteria and Firmicutes were more abundant (Fig. 3 D-F). In contrast, no significant differences were found at phylum level in gut samples (Fig. 3 A-C). At genus level, Bacillus, Mycetocola and Salinocola were consistently more abundant in the gut of all prebiotic treatment groups compared to controls, whilst Mobiluncus and Paraperlucidibaca were less abundant in prebiotic-fed fish. The intermediate inclusion level (1 g/kg) showed the greatest impact on abundance of microbial taxa compared to the control group in the distal intestine, with genera such including Corynebacterium and Paraperlucidibaca less abundant compared to the control (Fig. 3 A-C). Similar to gut samples, Bacillus is among the genera that were increased in the skin of fish fed the prebiotic blend. Towards a higher inclusion level, the number of ASVs in the skin from the phylum of Bacteroidetes rise, including Chryseobacterium, Flectobacillus and Mesonia (Fig. 3 D-F).

Fig. 3.

Differential abundance analysis with Deseq2 for the gut (A, B, C) and the skin (D, E, F). Significant different (p ¡ 0.05) ASVs summarized to Genus level, differences in colours distinguish in Phylum levels.

Fig. 4.

Heat tree analysis Control vs third treatment (1 g/kg) for the A) gut and B) skin, significant different features are printed, yellow indicates higher abundance in the control, while green indicates greater abundance in the treatment group.

4. Discussion

The present study compared the effects of different inclusion levels of Selectovit, a prebiotic blend (mannans and,1,3/1,6-beta glucans, nucleic acids and nucleotides, medium-chain fatty acids, single chain fatty acids), on growth performance indicators and the microbial composition of the skin and the distal intestine of Atlantic salmon. We demonstrate that the prebiotic supplementation, although has negligible effects on growth, induces significant changes in the mucosal microbiomes. Interestingly, we observed greater changes in the skin microbiomes, illustrating potentially important interactions between the use of nutritional prebiotic supplements and wider host health. Further interrogation of these interactions could lead to the development of specific bacterial biomarkers that reveal new insights into the overall health status of the fish and could result in novel strategies of health monitoring (O'neill et al., 2016, Zheng et al., 2020, De Pessemier et al., 2021).

Supplementation of Atlantic salmon diet with Selectovit resulted in no significant effects on any growth performance indicators (SGR, FI, FCR, End weight and survival, Table 2). To our knowledge, the individual components of Selectovit have not been tested so far as a combined blend. However, the single components of the prebiotic blend (mannans and,1,3/1,6-beta glucans, nucleic acids and nucleotides, medium-chain fatty acids, single chain fatty acids) have been used widely in different cultured fish species (Selim and Reda, 2015, Torrecillas et al., 2015, Hisano et al., 2018). Interestingly, mannans and 1,3/1,6-beta-glucans have been shown to have variable impacts on growth performance in salmon and trout. Whilst some studies indicate reduced growth performance (Grisdale-Helland et al., 2008, Refstie et al., 2010), others show no effect (Grisdale-Helland et al., 2008, Refstie et al., 2010, Dimitroglou et al., 2011) or improved growth (Staykov et al., 2007, Jami, 2019). Similarly varying growth results were found using nucleotides as a single prebiotic supplement in feeds (Xu et al., 2015, Bowyer et al., 2019). Recent findings suggest that the optimal dosage for best growth results using nucleotides stands at around 0.1 % and 0.2 % inclusion, however, the optimal dosage for mannans and glucans is highly variable and needs further investigation (Ringø et al., 2011, Akhter et al., 2015). The variation in growth performance could be correlated to differences in product composition. Mannans, 1,3/1,6-beta-glucans and nucleotides are typically derived from yeasts, that may vary in their properties and/or quality due to differences in processing, culture conditions or the genetic origin of the yeast (Klis et al., 2002, Kwiatkowski and Edgar, 2012, Orlean, 2012, Liu, 2021).

Although prebiotic supplementation regimes may not always affect growth, several studies have revealed their positive impacts on fish gut morphology (Ringø et al., 2011). Nucleotides, mannans and 1,3/1,6-beta-glucans have all been shown to induce significant increases in microvilli length, thus increased the absorptive surface in the gut (Dimitroglou et al., 2011, Selim and Reda, 2015, Guo et al., 2017). As immunostimulants, prebiotics can enhance the activity of non-specific defence mechanisms and improve resistance to infectious diseases (Burrells et al., 2001, Akhter et al., 2015). For example, Pattern recognition receptors (PRRs) such as 1,3/1,6-beta-glucan receptors expressed by macrophages, directly interact with 1,3/1,6-beta-glucan supplements, resulting in their immunomodulatory activity (Akhter et al., 2015, Zheng et al., 2020)^. Prebiotic-driven alterations in intestine morphology and/or host immune interactions can subsequently affect the microbial composition in the gut (Xu et al., 2015, Guo et al., 2017, Yang et al., 2020, Hu et al., 2021). The microbial composition of mucosal surfaces plays a vital role fish health, and thus is considered an important factor to ensure successful aquaculture production (de Bruijn et al., 2018, Wang et al., 2018, Bozzi et al., 2021).

In the present study, the prebiotic blend demonstrated no clear trend of overall microbial diversity in the gut and the skin (Fig. 1). The overall structure of abundant Phyla in the gut and the skin are similar to findings observed in recent studies of farmed Atlantic salmon, indicating Proteobacteria and Actinobacteria as the predominant groups (Wang et al., 2018, Wang et al., 2020, Uren Webster et al., 2021). However, prebiotic supplementation resulted in significant alterations in the relative abundance of certain bacteria taxa in both tissues (Figs. 3 & 4).

At genus level, Bacillus ASVs were consistently more abundant in the microbial community of the skin and the gut in fish fed the prebiotic blend at all inclusion levels (Fig. 3). Various species of this genus are considered being beneficial for the digestion and the health of the host. Bacillus spp. can synthesize various enzymes such as cellulase, amylase, protease and chitinase and thus can increase the overall digestibility of feeds. Moreover, cellulase and chitinase can break down feed components that are largely indigestible for carnivorous species (Ray et al., 2012, Hodar et al., 2019). This can be beneficial for the aquaculture industry that is facing nutritional challenges induced by the increasing substitution of a marine-based product by alternative plant or insect-based sources (Refstie et al., 2000, Pratoomyot et al., 2010, Al-Thobaiti et al., 2018). As a health benefit, Bacillus spp. produce antimicrobial compounds that inhibit the growth of pathogenic bacteria and reduce the risk of disease outbreaks and mortalities. For example, increased survival and growth has been demonstrated in fish fed Bacillus spp. as a feed supplement (e.g. Bacillus subtilis and Bacillus licheniformis), including rainbow trout, challenged with Yersinia ruckeri, a common fish pathogen (Raida et al., 2003). In addition to the health and growth benefits, Bacillus spp. are spore-forming and can colonize the fish gut that enable its use as a probiotic. Due to their beneficial properties, Bacillus spp. are becoming widely used probiotic additives in the aquaculture industry (Rahman et al., 2021). In addition, Mycoplasma was more prevalent in the distal intestine of the treatment groups, and significantly more abundant in individuals fed 0.5 g/kg of the prebiotic blend (Fig. 2 A & 3A). Mycoplasma is typically found in the gut of healthy and faster-growing individuals of farmed Atlantic salmon and therefore considered a potential biomarker for health assessment (Bozzi et al., 2021, Cheaib et al., 2020).

In contrast, species of the genus Chryseobacterium are considered emerging fish pathogens in aquaculture production. For example, Chryseobacterium pisciola has caused mass mortalities in farmed rainbow trout and Atlantic salmon (Zamora et al., 2012). In this study, Chryseobacterium ASVs were significantly less abundant on the skin at higher inclusion levels of the prebiotic (Fig. 3 E & F), suggesting a positive effect of the prebiotic blend. Reduced growth of pathogenic bacteria can be explained by various pathways. Bacteria growth may be suppressed by antimicrobial compounds or niche exclusion by other commensal bacteria such as Bacillus, or improved host immune response (Kamada et al., 2013, de Bruijn et al., 2018, Zheng et al., 2020). Indeed, a number of components of the tested prebiotic (e.g., 1,3/1,6-beta-glucans, mannans) are known immunomodulators (Abid et al., 2013, Selim and Reda, 2015, Jami, 2019). Future studies to investigate the relative contribution of enhanced anti-pathogenic microbiome community members and/or beneficial immune stimulation via Selectovit supplementation are needed.

Interestingly, we found that the dietary supplementation induced greater changes in the microbial community in the skin than in the intestine (Figs. 3 & 4). This suggests a link between the skin microbiome and dietary changes. In zebrafish, a dietary probiotic containing Lactobacillus rhamnosus induced greater changes in the skin microbiome along with enhanced immune functions showing improved lysozyme activity, immunoglobulin concentrations and peroxidase activity (Hu et al., 2021). A previous study in Yellowtail Kingfish on the effect of gut enteritis on the gut and skin microbiota found diseased fish exhibited more microbial dysbiosis in the skin compared to the gut (Legrand et al., 2020). Further, multiple metabolic and immune pathways were up-regulated mostly in the skin, suggesting the high sensitivity of the skin mucosal surface to the host health (Legrand et al., 2020). In humans, the interaction between the gut and the skin is far more understood. There is evidence, that metabolic products such as short-chain fatty acids (SCFAs) from fibre fermentation in the gut have a crucial role in determining the microbial composition, thus subsequently influencing cutaneous immune defence mechanisms (Egawa et al., 2017, Salem et al., 2018). Metabolites produced by pathogenic bacteria are now considered biomarkers of human intestinal dysbiosis (O'neill et al., 2016, Salem et al., 2018). These gastrointestinal-derived metabolites can circulate the host via the blood and subsequently accumulate in the skin, resulting in impaired epidermal differentiation and skin barrier integrity in humans (O'neill et al., 2016, De Pessemier et al., 2021). In addition, host pattern recognition receptors (e.g., toll-like receptors, TLRs) can sense gut derived microbial signals and induce skin-specific host immune response against pathogens but also shape the skin microbial composition (Akhter et al., 2015, Zheng et al., 2020). Further investigation of the interactions between the gut health status and skin microbiota in fish could be highly beneficial to health monitoring in aquaculture. Specific changes in the skin microbiome – which can be assessed non-destructively - could be used as biomarkers to diagnose diseases and health issues in the gut (Xiong, Nie and Chen, 2019).

5. Conclusion

Overall, the inclusion of Selectovit as a prebiotic supplement for Atlantic salmon showed no significant impact on growth. However, we clearly demonstrate that the prebiotic can significantly manipulate the microbial community of the distal intestine and the skin. Several potential beneficial bacteria such as Bacillus and Mycoplasma spp. were significantly more abundant in the prebiotic-fed groups compared to the control. In contrast, pathogenic bacteria were less abundant in the fish fed the prebiotic blend. This suggests a positive effect of the prebiotic blend on the microbial composition along with several potential crucial health benefits that can be provided to the host. Findings support the use of prebiotics as a promising method to take control of the microbial communities in fish. Interestingly, the supplement induced more changes in the skin than in the gut. There is growing evidence in fish for highly complex interactions between the microbial communities of the digestive system and external mucosa, and with the host immune system. Further research in this field could lead to the creation of novel bacterial biomarkers and new non-invasive strategies for fish digestive health monitoring. In the future, it will be beneficial to investigate host gene expression, to assess the effect of prebiotics and the microbial community on host metabolic and immune pathways.

CRediT authorship contribution statement

Simon Baumgärtner: Methodology, Validation, Formal analysis, Investigation, Writing – original draft, Project administration. Jack James: Conceptualization, Supervision, Resources, Project administration. Amy Ellison: Supervision, Resources, Writing – review & editing, Project administration.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.