Marine water samples were collected near the Naples coast during the autumn of 2022. The first sample (LN 40°49’30”–LE 14°18’48”) was used for the prey’s isolation, whilst the second one (LN 40°49’32”–LE 14°18’47”) was used for the predator isolation. Water samples were immediately transferred to the laboratory and analyzed within 1 h. pH and water conductivity were evaluated by a pH-meter (Model pH50 Lab).

Marine water was directly spread on Thiosulfate-Citrate-Bile Sucrose Agar (TCBS – Oxoid, Basingstoke, United Kingdom) plates and incubated overnight at 32 °C. At the same time, marine water was subjected to enrichment according to the protocol proposed by Huq et al. (2012). Briefly, marine water (50 mL) was added to 450 mL of Alkaline Peptone Water (APW) 10X (100 g/L peptone, 100 g/L NaCl, pH 8.6). After an overnight incubation at 32 °C, 100 μL of the top layer was spread plated onto TCBS agar plates. Colonies were randomly selected and, after repetitive streaking onto TCBS plates, cultures were used to inoculate ATCC medium n. 1,286 (Seawater complete), Tryptic Soy Broth (TSB), and Luria Bertani (LB) broths to point out the most suitable medium for cultivation. Specifically, both LB and TBS were supplemented with 20 and 30 g/L of NaCl, respectively.

A total of 10 prey isolates were identified by 16S rRNA sequencing using PCR conditions and primers - fD1 (forward, 5′-AGAGTTTGATCCTGGCTCAG-3′) and rP2 (reverse, 5′-ACGGCTACCTTGTTACGACTT-3′) - described by Weisburg et al. (1991). DNA was extracted through the protocol described by Wang et al. (1993). PCR amplicons were purified by QIAquick Gel Extraction Kit (Qiagen). The DNA sequences were determined by the dideoxy chain termination method (Sanger et al., 1977) by using the forward primers (fD1) described by Weisburg et al. (1991). Research for DNA similarity was performed with the National Centre of Biotechnology Information GenBank (Altschul et al., 1997).

Three potential prey were tested: V. mediterranei VM6 and Citrobacter (C.) portucalensis VM2 isolated during this study, plus one strain of Escherichia coli 32 isolated from meat during a previous survey. Overnight cultures of V. mediterranei and C. portucalensis in modified LB broth (20 g/L NaCl) and of E. coli 32 in LB (10 g/L NaCl) were centrifuged (6,500 rpm for 15 min), and the cell pellets were used for the predator isolation.

The enrichment protocol described by Jurkevitch (2006) was followed for the marine BALOs. Cell pellets of the prey prepared as described in paragraph 2.3 were resuspended in 100 mL of marine water to obtain a prey concentration around 10⁹/10¹⁰ cell/mL based on CFU counts. During incubation at 28 °C under constant stirring (Orbital shaker 300 rpm), cultures were daily monitored by spectrophotometry (600 nm) and microscope observation. Enrichments were then filtered (Minisart 0.45 μm) and, after decimal serial dilutions in sterile marine water, plated by the double-layer technique according to Jurkevitch (2006) in Pp medium (0.5 g/L tryptone, 0.5 g/L proteose-peptone, pH 7.7). Pp bottom (1.5% agar) and Pp top (1.95% agar) media were prepared in sterile marine water obtained by autoclaving water after a 0.45 μm filtration. Plates were sealed and incubated at 28 °C for 7 days. Lytic plaques were excised and resuspended in 500 μL of HM buffer [N-(2-Hydroxyethyl)-piperazine-N’-(2-ethanesulfonic acid) sodium salt] at pH 7.5. After agitation for a few minutes, the liquid containing the released BALOs was serially diluted in HM buffer up to 10^–4.

Plaque purification was carried out according to Jurkevitch (2006) with some modifications. Cell pellets of the prey were diluted in HM buffer up to a concentration of 5 × 10⁹ cells/mL. Pp bottom (1.5% agar) and Pp top (0.7% agar) media were prepared in artificial seawater (ASW) according to Ettensohn et al. (2004). A total of 400 μL of the prey was added to 5 mL of Pp top and 100 μL of each sample’s dilution. Plates were incubated at 28 °C and monitored daily.

Predator enrichments were obtained by transferring single plaques in tubes containing the prey resuspended in 20 mL of ASW. The prey without the plaque served as a control. After 24, 48, and 72 h of incubation at 28 °C under constant shaking, tubes were monitored by spectrophotometer and microscope observation. The prey population level was assessed by the drop method (Collins et al., 1989) on modified LB agar plates. After dilution (10¹ to 10⁶) in sterile Ringer’s solution (Oxoid), 12 μL aliquots were dropped onto agar plates using a pre-calibrated 20 μL micropipette. After incubation, individual colonies in drop areas were counted. The test allowed for obtaining indications about predatory efficiency.

For the identification of predators, enrichments were filtered (0.45 μm), and DNA was extracted by the protocol described by Wang et al. (1993). The PCR for Bacteriovoracaceae with primers Bac676F (forward, 5′- ATTTCGCATGTAGGGGTA-3′) and Bac1442R (reverse, 5′-GCCACGGCTTCAGGTAAG-3′) described by Davidov et al. (2006) was carried out according to the protocol detailed by the authors.

Results of CFU, PFU, and OD600 values were expressed as mean ± standard deviation. Significant differences among data were computed by using ANOVA and Tukey t-test (p < 0.05) (XLStat 2012.6.02 statistical pocket, Addinsoft Corp., Paris, France).

Based on the colonies’ appearance on TCBS agar plates seeded with marine water (pH 7.91 ± 0.01; conductivity −52 ± 0.01 mV) and from plates seeded with the enrichments, ten colonies in total were chosen. According to the results obtained by the 16S rRNA partial sequencing, one strain could be reported to V. mediterranei (99% similarity), two to Photobacterium spp. (98% similarity with Photobacterium sp. strain 7–11 KX806606.1 in both cases), two to C. portucalensis (99 and 100% similarity with C. portucalensis strain 68soilLBA LC717361.1), three to V. harveyi (98 or 99% with V. harveyi strain B8-1MK102617.1), and two to V. chagasii (99% with V. chagasii strain GCZ10 MH613265.1).

Upon enrichment with the three prey (V. mediterranei VM6, C. portucalensis VM2, and E. coli 32), the faster clarification was noticed only against the unique strain of V. mediterranei (Data not shown). At the microscope, small, speedy motile cells could be observed. Five lysis plaques were purified. None of the five strains - BV1, BV2, BV3, BV4, and BV5 – generated an amplicon by the family-specific PCR, so despite the morphological and physiological similarities with BALOs, strains isolated during this study cannot be considered members of this group.

Cultures BV1, BV2, BV3, BV4, and BV5 exhibited a rather variable predatory efficiency. Strain BV4 proved to induce a significant decrease in OD values at 24 (1.841 ± 0.002) and 48 h (1.149 ± 0.001); strain BV5 induced the highest clarification at 48 h (0.997 ± 0.001). Moreover, this strain produced a significant decrease in V. mediterranei CFU/mL at 48 h (from 9.00 ± 0.23 to 7.73 ± 0.09 CFU/mL) (Supplementary Figure 1). Based on results, strain BV5 was selected for further experiments.

First, the growth of V. mediterranei strain VM6 on TCBS was compared with that on WPCA and TSA media. The counts levels on TCBS and TSA were equal, whereas it was ascertained that the prey was unable to grow on WPCA, thus proving that counts on this medium could not be affected by the prey inoculum.

In the first experiment, mussels not previously subject to depuration were used. The prey inoculum was fixed at about 10² CFU/mL in trials A, B, E, and F. Vibrionaceae were monitored on TCBS. Still, specifically, only yellow colonies similar to those produced by V. mediterranei VM6 on this medium were selectively counted. Vibrio populations in mussels were in all trials (A–D) in the range of 10²–10³ CFU/mL, namely, the same adopted for the prey inoculation in this experiment (Figure 2A). However, by comparing TCBS counts in mussels collected by trials A and B, namely those inoculated with BV5 plus VM6 and strain VM6 alone, respectively, an interesting outcome may be pointed out: after 24 h, the decline in Vibrio populations in trial A, which included both prey and predator, was greater than one Log. Such difference disappeared after 48 h, likely as a result of the prey growth due to nutrients released by mussels (Figure 2A).

In trial C, the predator addition did not induce any changes, and this might be linked to the high specificity of the strain BV5 regarding the prey. As a general consideration, Vibrionaceae increased by more than one Log in the control (Trial D).

The monitoring of Vibrionaceae in the depuration water showed a rather different trend, and the difference between trials A and B became statistically significant only after 48 h of purification (Figure 2A). In the control (Trial D), vibrios grew exponentially, demonstrating that a transfer from mussels into water occurs during depuration. The monitoring of the prey (Trial E) and of prey plus predator (Trial F) in ASW without mussels confirmed the predatory efficiency of the strain BV5: Vibrio reduction after 48 h was higher than one Log (Figure 2B).

Predators were monitored by PFU counts in water collected from trials A, C, and F (Supplementary Figure 2). Populations increased by more than one Log after 48 h in trial F, namely when prey and predator were alone in sterile ASW, stayed constant in trial C (predator inoculated into mussel depuration water), and slightly decreased in trial A despite the prey presence.

Heterotrophic microflora was monitored in both water and mussels (Supplementary Table 1). Counts were almost stable in mussels, whilst a slight decrease characterized counts in water. The high contamination level in water, around 3 Log CFU/mL, could be due to the high bacterial release from mussels that were not washed and still with the socket. Despite not being depurated, Enterobacteriaceae on VRBGA, Coli-aerogenes group on VRBLA, E. coli on TBX, and enterococci on SB were undetectable throughout the monitoring (Supplementary Table 1).

The same set of experiments was repeated by using mussels already depurated and a higher level of inoculum for the prey (about 4 Logs CFU/mL of water). Despite depuration, the level of countable yellow colonies on TCBS was around 2 Log CFU/mL in trials C and D carried out without the prey addition (Figure 3A). However, in this case, the adoption of a higher level of inoculum for V. mediterranei allowed the discrimination of the prey from the naturally occurring vibrios in mussels. In fact, in trials A and B, after 5 h, the level of vibrios in mussels was almost one Log higher than trial C (only the predator added), and control trial D (Figure 3A).

Also in this depuration experiment, the maximum predation occurred after 24 h, and, exactly as for the first set of experiments, the BV5 inoculum in trial C did not affect the autochthonous Vibrio population (Figure 3A).

In the water, a significant drop in Vibrio population can be linked to predation: the difference in Vibrio loads between trials A and B at 24 h was higher than two Logs (Figure 3B). In trials E and F, carried out without mussels, the outcomes were even more remarkable: after 48 h, V. mediterranei counts dropped below the method’s detection limit. PFU monitoring confirmed the evidence collected during the first experiment: predators in water from trials in which the strain BV5 was added did not change in number (Supplementary Figure 3A). As expected, plaques were not retrieved in trial E, containing only the prey in ASW, whereas in trial B (prey and mussels), as well as in trial D – the control – autochthonous predators were found despite being undetectable at time 0 (Supplementary Figure 3A). The level reached was around 3 Log PFU/mL. Furthermore, in the unique case of trial A, predators were monitored in mussels as well: after 24 h, a noticeable predator concentration inside mussels was evidenced (Supplementary Figure 3B).

Water Plate Count agar counts were by two Logs higher in depurated mussels and stayed almost stable in both mussels and ASW in all trials, thus proving that the presence of strain BV5 does not affect the naturally occurring microflora in both environments. Despite being depurated, mussels hosted Enterobacteriaceae and coliforms, whereas enterococci were only seldom detected (Supplementary Table 2). In mussels from trial B and control sample D, blue colonies on TBX revealed a low occurrence of E. coli (Supplementary Table 2).

16S rRNA gene amplification and amplicon sequencing were successful for all samples. A total of 1,381,261 classified reads were obtained. The mean number of reads per sample was 38,368 ± 15,897. The phylum Campylobacterota was the most abundant in mussels from the experiment carried out with non-depurated mussels; conversely, in bivalves already subject to depuration, a higher biodiversity could be noticed (Supplementary Figure 4). In water samples, Proteobacteria (now Pseudomonadota) were prevalent, especially in the second set of the experiment. Subdominant phyla were Firmicutes (now Bacillota), Bacteroidota, Fusobacteriota, and Patescibacteria. The abundance of Bdellovibrionota was significant only in water samples from non-depurated mussels and not exclusively in predator-added trials (Supplementary Figure 4).

The phylum Campylobacterota was essentially represented by species within the family Arcobacteraceae in trials carried out with non-depurated mussels (Figure 4A); conversely, in depurated mussels, a higher variability could be recorded, above all in samples where strain BV5 was added (Trials A and C): after 48 h, a noticeable increase in the Bacteroidaceae, Lachnospiraceae, and Ruminococcaceae occurrence could be observed (Figure 4B).

Moreover, in the first set of trials, the presence of Vibrionaceae was almost identical in all trials (Figure 4A). As already evidenced by microbial count, the adjunct of 2 Log/mL of V. mediterranei VM6 was hidden by the occurrence of autochthonous vibrios at about the same population level. In mussels from trials carried out by using a higher inoculum level for the prey, the relative abundance of Vibrionaceae in mussels increased with time in both trials A and B (Figure 4B): in other terms, the predator’s adjunct does not seem to control the Vibrio accumulation due to mussels’ filtration (Supplementary Figure 3). Indeed, the abundance of Vibrionaceae in mussels increased over time in the control (Trial D) as well (Figure 4B).

Concerning water, the relative abundance of Vibrionaceae markedly decreased in trials A of both depuration experiments (Figures 4C, D). The family Bacteriovoracaceae was detected in all samples. The relative abundance increased by passing from 24 to 48 h in almost all trials, independently of the type of mussels used for the experiments.

Enterobacteriaceae appeared to be more represented in depurated mussels and, in trial A, after 24 h, this family dominated the microbiota (Figure 4B). Such an outcome corroborated results obtained by counting on selective media.

Arcobacteraceae were, as expected, dominant in water coming from trials with non-depurated mussels (Figure 4C), whereas, in the water of experiments with depurated mussels, Pseudoalteromonadaceae exhibited the highest relative abundance (Figure 4D).

By passing to genera, the percentage of reads that could not be reported to any taxa ranged from 2.04 to 25.93% and from 8.92% to 24.19% for mussels and waters, respectively (Figure 5). In non-depurated mussels, the genus Halarcobacter spp. dominated (Figure 5A). The genus Pseudoalteromonas spp. – was prevalent in depurated mussels (Figure 5B).

The genera Mycoplasma, Polaribacter, and Rubritalea spp. were detected in all mussel samples with relative abundances in the range 0.26–3.84, 0.15–2.63, and 0.02-2.17, respectively. Concerning the genus Vibrio spp., in the first set of experiments, the low level of the inoculum did not allow for highlighting any change. The relative abundance in trial A after 24 h is comparable with that of the control (Figure 5A). The second set of experiments, with a higher inoculum level, allows some considerations to be inferred. The relative abundance of vibrios increased by passing from 24 to 48 h in mussels from trial A with non-depurated mussels, allowing to confirm that a vibrios migration in bivalves takes place during depuration (Figure 5B). In waters, the relative abundance of the Vibrio genus was noticeably high only after 24 h in trial A, and, in all cases, the abundance of this genus decreased after 48 h (Figure 5C).

In waters, only one BALOs was detected and was reported as unidentified Bacteriovoracaceae by HTS. The relative abundance was quite low in water from both sets of experiments, regardless of the trial. Nevertheless, the relative abundance significantly increased by passing from 24 to 48 h in ASW from trial A, where the strain BV5 was inoculated together with the prey (Figure 5C).

The microbiome of depurated and non-depurated mussels was analyzed by HTS (number of reads 50,957 and 42,458, respectively). Distribution appeared to be rather different at both family and genus levels (Figures 6A, B). In mussels that were not previously depurated, Arcobacteraceae were dominant, followed by Fusobacteriaceae and Mycoplasmataceae. In depurated mussels, Mycoplasmataceae were almost equally represented as Arcobacteraceae.

At the genus level, apart from Halarcobacter and Mycoplasma - which were retrieved in mussels subject to depuration in the lab as well, it is remarkable the relative abundance of the Psychrilyobacter spp. This genus was instead poorly represented in all mussel samples coming from trials A, B, C, and D.

The alpha diversity analysis, especially the Simpson’s diversity index, revealed a higher biodiversity in depurated mussels (Figures 6C, D).

V. mediterranei VM6 was selected as prey since it is more susceptible to lysis than the C. portucalensis VM2 and E. coli 32, regardless of the predator. The lytic capabilities of bacterial predators against hosts may greatly vary, resulting in wildly disparate prey spectrum ranges. While some predators exhibit large ranges that cover numerous Gram-negative and even some Gram-positive bacteria, others have very narrow ranges, covering only a few species or strains (Najnine et al., 2020). The species V. mediterranei has been recognized as a pathogen of the razor clam - Sinonovacula constricta - (Fan et al., 2023), and of the noble pen shell - Pinna nobilis - a large bivalve on the brink of extinction in the Mediterranean (Andree et al., 2021). Recently, this species has been demonstrated to provoke significant acute immune responses and tissue-level reactions in M. galloprovincialis (Ter et al., 2024). Furthermore, the presence of a strain-specific pathogenicity island was established by the comparative investigation of 21 strains of V. mediterranei, underscoring the species’ pathogenicity toward bivalves (Zhang et al., 2025). From this angle, in addition to the hypothesized control of Vibrio populations during depuration, the one-log reduction of V. mediterranei VM6 ensured by the strain BV5 after 48 h could prevent larval vibriosis in the shellfish farming.

In all depuration experiments, Vibrionaceae naturally occurring in mussels were around 10²–10³ CFU/mL. Several authors have already reported a prevalence of Vibrio spp. in seafood samples collected in Italy (Parisi et al., 2004; Serracca et al., 2007; Sferlazzo et al., 2018). Moreover, during depuration, Vibrionaceae increased by more than one Log in the control (trial D), thus confirming that, as previously reported (Serratore et al., 2014; Sferlazzo et al., 2018), the purification treatment, worldwide utilized to purge bivalve mollusk from fecal contaminants, is unsatisfactory for seawater autochthonous Vibrio spp. These results might be related to the vibrios’ release dynamics by bivalves in depuration (Suffredini et al., 2014). At any rate, during the second set of trials carried out with a higher level of prey inoculum, the predator addition provided satisfactory outcomes. Indeed, prey concentration is well-known to influence the efficacy of BALOs’ predation. Specifically, predatory activity seems to be inhibited at prey levels below 10⁴ CFU/mL (Williams et al., 2016). Results were in agreement with those reported by Ottaviani et al. (2020) for the Halobacteriovorax sp. strain HBXCO1: the predator at 10³ PFU/mL was able to keep 10⁵ CFU/mL of V. parahaemolyticus to about 2 Logs lower than that of the control during mussels’ depuration.

During the experiment conducted with non-depurated mussels, the predators monitoring in the water revealed that populations increased only in trial F, namely when prey and predator were alone in sterile ASWF (Supplementary Figure 2). This discrepancy might be explained by the predator migration inside mussels. In fact, PFU monitoring validated the results when the experiment was conducted again using depurated mussels. After a day, the number of predators in trial A mussels increased, supporting the theory that predators migrate into mussels (Supplementary Figure 3B).

By HTS, the phylum Campylobacterota was the most abundant in non-depurated mussels (Supplementary Figure 4). Results did not match those reported for Australian M. galloprovincialis mussels. According to Odeyemi et al. (2019), the three major phyla in the mussel meat and pouch water of non-depurated and depurated mussels were Proteobacteria, Cyanobacteria, and Firmicutes. Proteobacteria, Tenericutes, and Bacteroidetes are indicated as dominant taxa at the phylum level in Greek mussels in both winter and summer months by Schoinas et al. (2023).

More info can be obtained by analyzing the families’ dynamics in the four trials and during the time (Figure 4). Arcobacteraceae dominated the microbiome of non-depurated mussels, whilst depurated mussels were characterized by a higher variability, above all in trials with the predator: Bacteroidaceae, Lachnospiraceae, and Ruminococcaceae increased along time (Figure 4B). The mussel microbiota appeared to include well-known animal microbial commensals involved in carbohydrate oxidation or fermentation and are likely able to influence the gastrointestinal metabolism of the host, as demonstrated in terrestrial animals. Indeed, the description of mussel microbiome on the tissue scale has revealed that the microbiota of each tissue is characterized by a specific pattern, with the digestive gland microbiota being dominated by Ruminococcaceae and Lachnospiraceae: bacteria able to ferment complex polysaccharides into short-chain fatty acids, and thus well matching the general assets of the animal gut microbiota (Musella et al., 2020).

Regardless of the trial, the abundance of Vibrionaceae in mussels increased over time. An expanding number of environmental studies have contributed to improving knowledge about the family Vibrionaceae, and some new species, such as V. crassostreae, V. breoganii, and V. celticus, are described as forming part of the molluscan microbiota (Romalde et al., 2014).

In non-depurated mussels, the genus Halarcobacter spp. dominated all samples (Figure 5). This genus is one of the six obtained by the split of Arcobacter spp. (Pérez-Cataluña et al., 2018). Natural inhabitants of environmental waters, including surface water, groundwater, rivers, lakes, and seawater, members of this genus have also been found in sewage and plankton (Zhang et al., 2019). The overall prevalence of Arcobacter spp. in bivalves has been previously reported (Laishram et al., 2016; Morejón et al., 2017; Rathlavath et al., 2017; Salas-Massó et al., 2016; Zhang et al., 2019). High percentages have been reported in Italy as well. Mottola et al. (2016) isolated Arcobacter spp. from shellfish samples in the Apulian region in Italy, while Fera et al. (2004) detected this genus in seawater and plankton samples collected from the Strait of Messina. The genus Arcobacter spp., previously known as the aero-tolerant Campylobacter, has gained clinical significance as an emerging diarrheagenic pathogen associated with water reservoirs in recent years. The complete clinical significance of Arcobacter remains rather speculative due to the virulence and antibiotic susceptibility of individual strains (Barel and Yildirim, 2023). On the other hand, the microbiome of depurated mussels appeared to be dominated by Pseudoalteromonas spp. This genus is widely distributed in various marine environments. Many Pseudoalteromonas species may induce the settlement of larvae of several invertebrates, including Mytilus coruscus (Wang et al., 2019).

The genera Mycoplasma, Polaribacter, and Rubritalea spp. detected in all mussel samples have been reported as dominant in the mussel microbiome by Schoinas et al. (2023). On the other hand, other genera reported as dominant by the authors, such as Anaplasma, Ruegeria, and Mariniblastus spp., were not detected in the present study.

The only BALOs family detected in waters by HTS was an unidentified Bacteriovoracaceae. The relative abundance in trials with both prey and strain BV5 increased over time (Figure 5C). Since genus-specific PCR did not provide amplification when tested on strain BV5, it is not reasonable to guess that the added predator was the unidentified Bacteriovoraceae detected by HTS. In general, the addition of the strain BV5 to depurated mussels, even in the absence of prey (Trial C), resulted in a disturbance of the microbiome at both the family and genus levels. The circumstance that mussels had previously undergone a depuration treatment may have caused a loss of the microbiome’s innate resilience, raising the risk that the bacterial communities would be altered.

The mussels used for the two decontamination experiments were provided by the same supplier, and they were non-depurated for the first set of trials and already depurated for the second. Since both types of mussels were analyzed immediately upon arrival, HTS might offer insights into how commercial depuration affects the mussel microbiome. A considerable decrease in the relative abundance of Psychrilyobacter spp. suggests that depuration may have a major effect on the mussel microbiome (Figure 6). Psychrilyobacter spp. is a globally distributed bacterial genus with an inhabiting preference for the gut of marine invertebrates such as the European abalone (Haliotis tuberculata), regardless of the season and feeding diet, but also of oysters, sea vases, Atlantic salmon, deep-sea snails, green-lipped mussels, Chilean mussels, and even deep-sea hydrothermal vent crabs (Liu et al., 2023). In the present survey, the relative abundance of this genus decreases along with the mussels’ permanence in water, and this evidence does not seem to prove its role as a mussel holobiont. As a matter of fact, in Mytilus chilensis, the genus Psychrilyobacter spp. appeared to be dominant in mussels living in natural conditions (Santibáñez et al., 2022), and its presence could be linked uniquely to the marine sediments: an ecosystem where the genus Psychrilyobacter is associated as an important protein and/or amino acid degrader (Pelikan et al., 2021). Additionally, in oysters, this genus increases in flesh during the moribund peak of the Pacific Oyster Mortality Syndrome caused by Ostreid Herpesvirus 1 infection (Richard et al., 2021).

The decrease in the relative abundance of both Psychrilyobacter and Halarcobacter genera goes along with the increase of Mycoplasma spp. (Figure 6). Indeed, the proliferation of subdominant phyla after a depuration of 15 h has already been reported for the mussel hepatopancreas bacteriome (Rubiolo et al., 2018) and the haemolymph (Vezzulli et al., 2018).