4.1. Characterizing the gut mucosal microbiota of Diplodus vulgaris across a large geographical scale in the NW Mediterranean Sea
This study is the first to explore and to describe the gut microbiota of the two-banded sea bream (Diplodus vulgaris, Geoffroy Saint-Hilaire 1817) by using a large and comprehensive dataset including 129 individuals from different geographical regions. Generally, when investigated through the analysis of the large scale dataset, the gut mucosal microbiota of the three populations displayed similar levels of Shannon’s diversity (Fig. 3C) and inter-individual variability of the bacterial community composition (Fig. 3A). Inter-individual variability is claimed to be often a consequence of stochastic changes in the microbiota that may occur in case of host’s stress or diseases [15]; increasing levels of inter-individual variability of the gut microbiota composition have been observed in populations of butterflyfish (Chaetodon capistratus) dwelling degraded coral reefs [19] and of sharpbelly (Hemiculter leucisculus) from polluted sites along the Ba river (China) [91]. Therefore, the similar inter-individual variability observed between the three populations investigated in this study may be interpreted as a sign of local stability and low levels of stress.
The core microbiota was explored to define the most commonly occurring and abundant bacterial genera characterizing this fish species. Seven core genera known to have a mutualistic relationship with marine fishes were identified in the gut microbiota of D. vulgaris: Vibrio, Aliivibrio, Photobacterium, Enterovibrio, Endozoicomonas, Shewanella, Propionigenium (Fig. 3B). Vibrio and Aliivibrio alone represented almost half of the whole gut bacterial community of the three populations. In samples from BO, these two genera and the genus Photobacterium represented 64% of the community. Vibrio is a very common genus in the intestine of marine fishes [2]. This includes many strains, and although some are harmful for fish, others have been described to confer protection against fish pathogens such as Aeromonas salmonicida, Pasteurella piscicida and Listonella anguillarum [2, 92]. Moreover, this genus contributes to the fish’s metabolism of several dietary compounds by producing amylase, lipase and chitinase [93, 94]. Aliivibrio correspond to the Vibrio fischeri group and it is phylogenetically and phenotypically distinct from other members of the Vibrionaceae family [95]. The genus Aliivibrio contains only six strains, four of which are consistently found in association with different marine vertebrates and invertebrates (A. fischeri, A.logei, A.wodanis, A. sifiae and A. finisterris) [96–99]. A. fischeri, A. sifiae and A. logei have been described as bioluminescent [99], however the role of light-producing bacteria in the intestinal tract of fishes is not known yet. The genus Photobacterium was reported several times as a symbiont of carnivorous and omnivorous fishes, where they produce chitinases essential for the digestion of crustaceans in the diet [97].
In summary, our findings indicate that at the time of the investigation, D. vulgaris from the NW Mediterranean Sea held a stable gut bacterial community that was mainly composed of members of the family Vibrionaceae, especially Vibrio and Aliivibrio. Given the high abundance and occurrence of these bacterial genera, they can be considered resident of the gut microbiota of D. vulgaris. Therefore, future studies set in the NW-Mediterranean Sea pay attention to any major shift in the abundances, or the absence, of these bacterial genera in the gut mucosa microbiota of D. vulgaris.
Generally, the core microbiota is shaped by the selective pressure of the host intestinal system regardless of the environment [100]. However, to have a more complete overview of the core gut mucosal microbiota of D. vulgaris from the NW-Mediterraenan Sea, the microbial data presented in this study should be coupled with data from other D.vulgaris populations, and with data from the same D.vulgaris populations at the reproductive season (from October to February). Indeed, during this season fishes undergo different environmental pressures due to their increased movement activity [57] and physiological alteration linked to the increased production of sex hormones that may influence the gut microbiota [101].
4.2. The taxonomy and functionality of the gut mucosal microbiota of Diplodus vulgaris is spatially heterogeneous in the Mediterranean Sea
4.2.1. Geographic location influences the taxonomy and potential functionality of the gut microbiota of Diplodus vulgaris at a large geographic scale
At a large scale (i.e. the three regions BA, CR and BO), the taxonomical composition of the gut microbiota of D. vulgaris was observed to differ significantly at all taxonomic resolutions. Between the three populations, the one from BO showed the most unique bacterial community (Fig. 3A): 18 bacterial genera were differently abundant between the specimens from BO and those from at least one of the other two populations (Supplementary Fig. 2). Nevertheless, it is important to notice that most of these differently abundant genera were not representative of the bacterial community in BO as only 4 genera were included in the core microbiota of this population. Conversely, several genera reported to be differently abundant in BA and CR were recorded in the core microbiota of these two populations (Fig. 3B).
Analysis of the results obtained at large spatial scale can be useful to review the role that both the environment and the host’s genotype have on the development of the fish gut microbiota. The microbial colonization of the intestinal system of fish larvae occurs within the first 50 days of life. This microbiota colonizes the gut following the ingestion of suspended particles and eggs debris by the fish larvae [41], from the surrounding water and from the first feed [2]. Therefore, the free-living bacterial taxa that are most abundant in the local environment and along the larval dispersal routes during this developmental stage may contribute to the final structure of the fish gut microbiota. In the Mediterranean Sea, an uneven distribution of environmental bacterial taxa (i.e. biogeography) was described in [102] and was shown to be a consequence of the interplay between different environmental parameters such as oxygen concentration and salinity, the longitude and the latitude of the sampling points. However, the composition of the free-living bacterial community is also influenced by the microbes specifically associated with the different marine macro-organisms: the contribution of macro-organisms in the dispersal and in the geographic distribution of marine microorganisms was indeed recently revealed [103]. Besides being the most distant region among the three, that of Corsica (BO) displays also higher levels of salinity compared to the Lion Gulf (BA and CR) (MARC, Modélisation et Analyse pour la Recherche Côtière, (https://marc.ifremer.fr/) and it is characterized by distinct biodiversity conditions both in terms of fish assemblage and coralligenous benthic species [104, 105]. With this in mind, it is possible that the greater differences observed in the gut microbiota of the D. vulgaris from BO compared to BA and CR would be a consequence of the different environmental microbiota occurring in this region. In the future, the composition of the free-living bacterial communities living in the sediment and in the water column of the three regions investigated in this study should be further explored.
The contribution of the host genotype to the gut microbiota should also be considered further [7]. D. vulgaris is described to have a larval and juvenile dispersal range respectively, of 90 km and 165 km [106]. Given that the region of BO is more than 350 km away from CR and more than 450 km from BA, the possibility that the BO population is genetically separated from the other two is not unreasonable. However, this is currently unknown and therefore, a genetic analysis of the populations of D. vulgaris in the NW Mediterranean Sea will be required. In practical terms, this could be implemented by the combined analysis of mitochondrial DNA sequence markers and microsatellites to define the existing haplogroups for the three species in the Mediterranean Sea [107].
Despite the different microbiota structure found in the fishes from BO compared to those from BA and CR, the potential functionality of their microbiota was rather similar to that recorded for the fishes from BA. Conversely, the analysis of potential functions revealed that the bacterial communities from the CR region were significantly different from those from BA and BO for all the macro functional categories considered, except for the metabolism of terpenoids and polyketides, the energy metabolism, and the metabolism and biosynthesis of glycans (Supplementary Table 4).
The differences between the functional potential observed for the microbiota of the CR and BO populations are a consequence of their largely different microbiota taxonomic structure. Differently, those reported for the microbiota of the CR and BA populations – which were more similar taxonomically – may be caused by the differential abundance of a few functionally important bacterial taxa. The bacterial communities recorded in BA and CR were dissimilar only for the abundance of two bacterial genera: one unclassified genus of the Mycoplasmataceae family (more abundant in the samples from CR) and Romboutsia (more abundant in the samples from BA) (Supplementary Fig. 2). The genera included in the Mycoplasmataceae family are known for their important contribution to fish homeostasis and metabolism and among their several functions, they have been reported to support the metabolism of long chain polymers such as chitin and starch [18]. Indeed, the metabolism of starch was significantly more abundant in the samples from CR than those from BA (and BO) (Fig. 4A). Similarly, the genus Rombustia was described to be an important player in the metabolism of amino acids and of vitamins, except for vitamin B6 [108]. The higher abundance of this genus in the microbiota of individuals from BA, compared to those from CR, may explain why we inferred different abundances of genes linked to these metabolic pathways in the samples from the two regions (Fig. 4B).
Investigating the differential abundance of potential functions involved in metabolism and biodegradation of xenobiotics in the fish gut microbiota across a spatial range can be a starting point to detect compromised environments, as well as the distribution of specific pollutants [44, 109]. Even if CR is the most industrialized region among the three, being closely located to the area of Fos-Barre and to the commercial and touristic harbor of Marseille, a similar level of human impact was reported in the three regions (www.medtrix.fr in the IMPACT project,[61]). In light of this, it was not too surprising to find an overrepresentation of specific metabolic pathways linked to the degradation of xenobiotics in all three populations (Fig. 4C). Among all the metabolic pathways included in this category, the most discriminant one was that related to the degradation of drugs (i.e. pharmaceuticals), which was overrepresented in the gut microbiota of fish from the region of BO. Given this higher abundance in BO it would be worth investigating further the presence of these pollutants in the south western coast of Corsica. A recent study [110] set in the Western Mediterranean Sea reported that the highest concertations of the anti-inflammatory drug naproxen was detected in sites located around the Corsican island. Also the presence of seven other pharmaceuticals was detected at lower concentrations (diclofenac, ibuprofen, ketoprofen, paracetamol, caffeine, carbamazepine and sulfamethoxazole) [110]. Pharmaceuticals are an emerging source of pollution that reaches the sea through the river inputs as a main outflux of agricultural, urban and industrial runoff, through coastal wastewater treatment plants and through touristic coastal infrastructures; the distribution of pollution sources in the Mediterranean Sea needs a more intense monitoring [111]. To monitor the distribution of these pollutants in the Mediterranean Sea and more specifically its fauna, a possible approach would be to combine measurements of pollutant concentration in the fish tissues with the analysis of the fish gut microbiota for its biodegradation potential. Fishes such as D. vulgaris may be a good model for such an approach for two reasons: first, by being widespread in the Mediterranean Sea, they can be a good sentinel species; secondly, because they occupy a relatively high position in the trophic network (i.e. 3.5) [112] and therefore they are exposed to higher concentrations of pollutants through biomagnification [113].
4.2.2. Heterogeneity of the benthic habitats characterizes the taxonomy of the gut microbiota of Diplodus vulgaris at a small geographic scale
Exploring the gut microbiota of 50 individuals in seven sampling locations distant from at most 33.6 km along the coast of the BO region allowed to define the effect of geographic location on the taxonomy and functionality of the microbiota at small spatial scale (small-scale dataset).
Surprisingly, regardless of the proximity of the sampling stations, the structure and the diversity of the bacterial communities associated with D. vulgaris varied according to the fish catching site (Fig. 5A and 5B). Nevertheless, the extent of the variation was smaller than that observed at a large spatial scale, and the statistical difference was lost when both parameters were investigated at higher taxonomic ranks (i.e. phylum, class for the alpha diversity; phylum for beta diversity). The type of benthic habitat appeared to be the strongest determinant of the spatial variation of the gut microbiota. More specifically this was driven by the proportion of biocenosis of Posidonia oceanica meadows on the sea substratum (Table 1). In terrestrial animals, the type of habitat has been demonstrated to shape the structure of the gut microbiota in different classes. In the herbivorous howler monkeys (Alouatta pigra), the structure of the gut microbiota was observed to differ according to the type of vegetation present in the habitat (from evergreen to semi-deciduous forests)[50]. Similarly, blue tits (Cyanistes caeruleus) were found to harbor different gut microbiota communities depending whether they breed in dense deciduous forests or in meadows-like habitats [114]. In fishes, the gut microbiota structure was mostly described to change between freshwaters and marine habitat [39, 115]. However, in a recent multi-species study, the type of marine substratum (rocks, sand and detritic bottoms) was shown to be an important determinant of the diversity and structure of fish microbiota not only in the gut but also the skin and gills [42]. Both the bacterial communities living in the sediment and in the water column may influence the development of the gut microbiota in fishes [2]. In this regard, the presence of vegetation (i.e. seagrass) on the sea bottom has been shown to influence the composition of the sediment and free-living bacterial communities [116]: specifically, sediment and water samples collected in sampling sites characterized by unvegetated substratum displayed a distinct bacterial community composition compared to those collected in sampling sites with an increasing degree of vegetation (i.e. seagrass and algae, seagrass at low density, seagrass at high density). In this study, the sampling locations GDP-1 and GDP-7 were the only ones mostly surrounded by soft detritic bottom and completely lacking Posidonia oceanica meadows (Fig. 2). Thus, this difference in vegetation may have led to a distinct structure of the gut microbiota of the specimens collected in them (Fig. 5A), through a different composition of the bacterial communities living in the sediment and in the water column in this marine habitat.
The alpha-diversity of gut microbiota of the fish individuals correlated positively with the proportion of soft detritic bottoms, and negatively with the proportion of Posidonia oceanica meadows (Fig. 5B and C). This may suggest that the sandy substrata is more favorable for this fish species’ gut microbiota, as higher alpha diversity is associated with better health and homeostasis in fishes [7]. Indeed, D. vulgaris is recorded frequently in rocky-algal and sandy substrata and only occasionally in Posidonia oceanica meadows [117].
The spatial difference in the composition of the gut microbiota observed at a small scale was also supported by the different distance of the sampling locations from the fully protected areas (i.e. FPAs) (Table 1, Supplementary Table 8). However, the alpha-diversity (Shannon’s index) of the microbiota did not correlate with the distance from the FPA. Although the fully protected areas are described to increase fish biomass and promote ecosystem restoration in the Mediterranean Sea and in the oceans [46, 118], omnivorous species are generally less affected by habitat protection – or degradation – compared to specialist ones [119]. Therefore, the result obtained for D. vulgaris in this study is in accordance with the generalist foraging strategy of this species. In fact, individuals living closer to the FPA would benefit from a more intact habitat and trophic network without major changes in the nutritional intake of their generalist diet. However, they would also experience higher inter-specific competition due to higher fish biomass. Conversely, those living further away would benefit from reduced inter-specific competition and still be able to find suitable foraging resources. In both scenarios, this adaptable species would emerge victorious, and it would not experience major changes in the alpha diversity of its gut microbiota.
Although the composition of the diet of D. vulgaris varied across the sampling locations (Supplementary Fig. 5), this appeared not to be a determinant of the structure and diversity of the mucosal gut microbiota of D. vulgaris. As previously stated, the autochthonous gut mucosal bacterial community is largely driven by the environment and the first feeds during the development of fish [2]; they lead at the adult stage to a few highly abundant bacterial genera representing almost entirely this community [120]. The latter represent a limit for the colonization and proliferation of external taxa that reach the intestine through the food and the water ingested. Therefore, while the type of resources exploited by the fish throughout its life time may influence the composition of the autochthonous gut mucosal microbiota, the gut content, which reflects the food ingested in the last few days, is less likely a determinant of this community. Differently, the gut lumen transient bacterial community is generally influenced by the short term diet as it is the one actively metabolizing the diet input [34]. Lastly, it is also important to consider again the generalist behavior of this species: analyzing the intestinal content after catching a fish provides only a snapshot of the diet, that might not be very representative of the general diet of the individual. In future studies, the relationship between the long term diet – obtained through the analysis of stable isotopes – and the structure of the gut mucosal community could be further investigated to determine their relationship in D.vulgaris.
Although the taxonomic composition of the gut microbiota was observed to differ across locations at a small scale, its functional potential was mainly conserved. Only five KEGG metabolic pathways were significantly different between the sampling locations, which reflected differences in the metabolism of terpenoids and polyketides and in the metabolism of xenobiotics. In the latter category, it was interesting to observe the higher abundance of the functions involved in the degradation of aminobenzoate in the samples collected in GDP-8 compared to those from GDP-1. 4-aminobenzoate (or para-aminobenzoic acid, or PABA) is a compound showing high UV absorption properties which makes it a common component of sunscreen products [121]. Its dispersion in the marine waters linked to seaside tourism makes it a harmful product for the marine environment [122]. Specifically, this compound is reported to have estrogen activity and to induce feminization in fish juveniles [123]. The higher potential ability to degrade aminobenzoate by the gut microbiota of specimens from GDP-8 might be linked to a higher concentration of aminobenzoate in the water surrounding this sampling station. According to the maps available on www.medtrix.fr in the IMPACT project [61–124], the level of seaside tourism along the coast appears to be higher in front of the station GDP-8 than in front of the GDP-1 one (Supplementary Fig. 6). However, additional data on the concentration of 4-aminobenzoate in the area of this study are needed to confirm this speculation.
Although inferring the potential functions of D.vulgaris gut mucosal microbial community through PICRUST2 provided valuable insight into some specific metabolic pathways, it is important to stress the limits of such predictive methods [87]. Functional predictions are biased towards existing reference bacterial genomes, therefore the functions specific to an environment and performed by novel taxa are less likely to be identified. Secondly, the functions are predicted on short amplicon sequences (~ 254bp), therefore it is not possible to infer strain-specific functionality. In light of this, it is important that future studies investigate the functionality of the gut microbiota of D. vulgaris also through metagenomic and metatranscriptomic data: while the first would provide information about the potential functionality of the microbiota at the bacterial strain level, the second would inform on the functions effectively performed by the bacterial community.