Between May 2022 and January 2023, during four sampling campaigns, a total of 45 marine water samples ( Supplementary Table S1 ; Figure 1 ) were collected aboard 4 military naval vessels: Amerigo Vespucci, Caio Duilio, Alliance, and Paolo Thaon di Revel. The routes crossed the Mediterranean Sea, the Indian Ocean (including Persian Gulf, the Gulf of Oman, the Gulf of Aden and the Red Sea), the Atlantic Ocean and the Arctic Ocean. Sampling procedures were adapted to the operational conditions of each vessel and to the prevailing weather and sea state. Whenever possible, the ship was brought to a complete stop and seawater was collected manually from the starboard side near the bow, after temporarily shutting off all nearby discharges to minimize contamination risk. Alternatively, when sea conditions permitted, a small boat (launch) was deployed to collect samples at a short distance from the main vessel. These precautions were taken to avoid any contamination from the ship’s structure or engine. At each site, 3–5 L of surface seawater were collected from approximately 30 cm depth using a dedicated bucket that was thoroughly sanitized after each use. To focus on the free-living microbial fraction and minimize membrane clogging, a two-step filtration procedure was performed on board using a vacuum filtration system. Seawater was first pre-filtered through a 5 µm mesh filter (mixed cellulose esters, Millipore) to remove larger particles. The resulting filtrate was then passed through a 0.22 µm sterile membrane filter (mixed cellulose esters, 47 mm diameter, Millipore) to retain microbial cells. Each 0.22 µm filter was placed in a sterile 2 mL tube and stored at −20 °C until DNA extraction.

Seawater temperatures, pH, salinity and chlorophyll were measured on site at each sampling location using a handheld multi-parameter water quality probe (IDRONAUT, Ocean Seven 316 Plus) equipped with dedicated sensors for each parameter.

Total DNA from the prokaryotic biomass retained on the membrane filters was extracted in the laboratory using the DNeasy PowerWater Kit (Qiagen) according to the manufacturer’s instruction. DNA purity and concentration were determined using a Qubit 4.0 Fluorometer with the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific), following the manufacturer’s instructions.

The V3–V4 region was amplified using the primer pair 341F (5′-CCTACGGGNGGCWGCAG-3′) and 785R (5′-GACTACHVGGGTATCTAATCC-3′), originally designed by Klindworth et al. (2013), which target a ~460 bp fragment of the 16S rRNA gene. The V3–V4 variable region was sequenced following the MiSeq rRNA Amplicon Sequencing protocol (Illumina, San Diego, CA) with some modifications. The detailed methodology for DNA extraction, PCR amplification, and sequencing, including reagent concentrations, cycling conditions, and sequencing protocol, is provided in the Supplementary Materials .

Sequencing was performed on an Illumina MiSeq platform. Bioinformatics analyses were performed in Quantitative Insights into Microbial Ecology v2022.2 (QIIME2) software suite and R packages, the DADA2 algorithm, and an in-house trained classifier for taxonomy assignment. The Silva database, release 138.1 was used to identify the bacterial flora in each sample and to compare them between groups.

The relative abundance rate of the bacteria was determined at both the phylum and genus level, focusing on taxa that accounted for at least 1% of the total microbial community in each water sample.

The alpha diversity index was calculated for each library, and the analysis refers to the variety and complexity of species in a community. Shannon index was used to describe population diversity in samples (Sarmiento et al., 2019).

Statistically significant differences in alpha diversity (Shannon) in pairwise comparisons between bacterial communities from different geographical areas were estimated using the Kruskal-Wallis pairwise test (QIIME2). The correlation between physico-chemical parameters and bacterial genera abundance was determined using Pearson’s correlation. Due to the compositional nature of relative abundance data, Pearson correlation analysis required appropriate data transformation to avoid spurious results. Following Gloor et al. (2017), relative abundance data were transformed using centered log-ratio (CLR) transformation implemented in the “compositions” package (version v.4.3.1).

Details on the sequence read summaries, including the total number of filtered reads, geographic group distributions, and amplicon sequence variant (ASV) clustering, are provided in the Supplementary Materials ( Supplementary Table S2 ). Sequences were submitted to the NCBI Sequence Read Archive (SRA) (link to be provided at proof stage, currently under review).

A total of 31 distinct phyla were identified. The most prevalent phyla, along with their occurrence percentages across the 45 sampling sites, were: Proteobacteria, Actinobacteriota, and Bacteroidota (100%), Cyanobacteria (95.5%), Verrucomicrobiota (93.3%), Marinimicrobia (82.2%), Campylobacterota (75.5%), and Firmicutes (15.5%). These phyla appeared in varying proportions across the samples. Proteobacteria were the most abundant at all sites, with relative frequencies ranging from 31.50% to 94.30%, followed by Bacteroidota (0.20%–48.11%), Cyanobacteria (0.19%-46.20%), Actinobacteriota (1.11%-18.59%), Firmicutes (0.09%–14.30%), Verrucomicrobiota (0.07%-3.84%), Campylobacterota (0.07%-2.80%), and Marinimicrobia (0.08 -1.77%).

Approximately 400 bacterial genera were identified within the 45 microbial communities examined. The heat map in Supplementary Figure S4 represents the relative abundance of bacterial genera identified in seawater samples collected from different geographical regions, as determined by 16S rRNA gene sequencing. Only genera with a relative abundance greater than 1% were included in this heat map to simplify visualization, highlight the most significant taxa, and minimize the influence of low-abundance taxa. A core group of 15 genera is shared by the Atlantic, Mediterranean and Indian Oceans, while the Arctic Ocean has a different composition, with a smaller number of genera.

The north-western (NW) Mediterranean shows a greater diversification of genus composition, as seen in the central part of the map. SAR11 clade Ia was the most widespread genus identified in this study, with higher relative abundances in Mediterranean and Atlantic water samples. Even at lower abundance levels, SAR11 clade II remained widely distributed. A high abundance of Alteromonas and Pseudoalteromonas was observed in some samples from the NW Mediterranean Sea and from the Atlantic Ocean, respectively. These two genera were predominant in number of ASVs (Alteromonas 28253 ASVs in the sample MED NW7; Pseudoalteromonas 19432 ASVs in ATL OC1). Except for the Arctic Ocean, where Synechococcus_CC9902 was detected sporadically and in trace amounts (below 1% relative abundance), the genus was present in all other geographic areas considered, with the highest relative abundance in Mediterranean Sea samples. Two other genera frequently encountered are AEGEAN-169_marine_group and SAR86.

The microbial community structures across different geographical areas are presented in Figures 2 – 6 . For each geographical area, the detected bacterial phyla and genera are presented in bar-coded form, with the results described individually. For clarity and consistency, taxonomic composition plots at the genus level ( Figures 2 - 6 , lower panels) display only those ASVs that could be classified at the genus level. ASVs assigned only at the family level or higher were excluded from these visualizations.

Figure 7 shows the Shannon alpha diversity index of microbial communities across the different geographic regions, highlighting variations in community richness and evenness. Statistically significant differences among oceanic regions are reported in Supplementary Table S3 (Kruskal–Wallis, p < 0.05).

The Mediterranean Sea exhibited the highest alpha diversity among all sampled areas. Both subregions—northwestern (MED NW) and southwestern (MED SW)—had significantly higher Shannon values than the Arctic Ocean and the Indian Ocean, with a particularly marked difference between MED NW and the Indian Ocean (p < 0.01), and a significant difference between MED SW and the Indian Ocean (p < 0.05). Among the two Mediterranean subregions, MED NW showed slightly higher median diversity than MED SW.

The Atlantic Ocean showed the broadest variability in Shannon values across sites, as reflected by the large interquartile range. While some samples had high diversity, the median value was comparable to MED SW but lower than MED NW.

The Arctic Ocean displayed the lowest alpha diversity, with minimal variation between samples. The Indian Ocean, which includes semi-enclosed basins like the Persian Gulf, Gulf of Oman, Gulf of Aden, and Red Sea, also showed significantly lower alpha diversity compared to both Mediterranean subregions.

The spatial arrangement of sampling sites, overlaid with Shannon diversity classes, is shown in Figure 1 , offering insight into the geographic structuring of microbial alpha diversity.

Detailed results for temperature, salinity, and chlorophyll in the seawater samples are provided in the Supplementary Materials . Supplementary Figure S1 illustrates seawater temperatures across the sampling sites, ranging from nearly 0°C in the Arctic Ocean samples to approximately 30°C in the Indian Ocean. Supplementary Figure S2 shows salinity values (PSU) across the sampling sites, ranging from approximately 30 PSU in Arctic Ocean samples to values approaching 40 PSU in tropical waters and in a few samples from the northwestern Mediterranean Sea. In the southwestern Mediterranean, salinity values were lower, reaching approximately 37.5 PSU. Supplementary Figure S3 illustrates chlorophyll concentrations (µg/L) across the sampling sites. The values show a high degree of variability, with peaks observed in certain samples from the North-Western Mediterranean and the Atlantic Ocean, reaching up to 17.8 µg/L, while most other sites exhibit lower concentrations.

The analysis of correlations between the relative abundance of microbial genera and environmental variables revealed distinct patterns associated with salinity, chlorophyll, and temperature, as shown in Figure 8 . Regarding salinity, many genera exhibited negative correlations, indicating higher abundance in lower salinity waters. Notably, Marine Group II, Candidatus Actinomarina, Thalassotalea, and Synechococcus CC9902 showed pronounced negative associations with salinity. Conversely, some taxa such as the SAR116 clade and KI89A clade were more abundant in environments characterized by higher salinity.

In relation to chlorophyll, only Leeuwenhoekiella and Mesonia were weekly positively correlated.

Correlations with temperature were more variable across taxa. Groups like the SAR86 clade, SAR116 clade, and SAR92 clade showed positive correlations with temperature, while others such as SAR11 clade I, and UBA10353 marine group correlated negatively, indicating differing temperature preferences. Although the main correlation analysis was conducted on the entire dataset (n = 45) to ensure adequate statistical power, additional correlation heatmaps stratified by geographic region are presented in Supplementary Figure S4 , allowing a more detailed inspection of area-specific patterns.

The bacterial community structure in the Arctic Ocean reflects the adaptations necessary to survive in one of the most extreme marine environments.

Key taxa such as SAR11, Polaribacter, Sulfitobacter, SAR 92, and Amylibacter drive essential biogeochemical processes, including carbon and sulfur cycling and nutrient fluxes, underscoring their significance in polar ecosystems.

Polaribacter, identified as a major degrader of polysaccharides (Avcı et al., 2020; Deming and Eric Collins, 2017), is particularly abundant in northern samples, emphasizing its role in processing organic matter derived from phytoplankton blooms. Similarly, Sulfitobacter, a key player in the sulfur cycle, interacts with marine microalgae to degrade algal-derived organic sulfur compounds (Zeng et al., 2020). SAR 92 bacteria are potential degraders of organosulfur molecules and sources of climate-active gases in the marine environment, especially in polar regions (He X.-Y. et al., 2023).

The consistent detection of Nitrincolaceae and Amylibacter highlights their contributions to nitrogen cycling and organic matter degradation, respectively. The exclusive presence of Amylibacter in Arctic samples, with relative abundances ranging from 1.3% (ARC OC 6-7) to over 14% (ARC OC 1), suggests its adaptation in cold environments (Wietz et al., 2021).

The detection of Fibrobacter, the only one genus comprised in the phylum Fibrobacterota traditionally associated with lignocellulose degradation in herbivore guts (Ransom-Jones et al., 2012), suggests a broader ecological role in Arctic waters, contributing to organic matter breakdown.

Rhodococcus, belonging to Actinobacteriota, previously identified in polar regions, underscores the unique metabolic versatility of Arctic microbial communities (Ferrera-Rodríguez et al., 2013). These bacteria are involved in the removal of hydrocarbons from Arctic seawater and coastal soils, and some authors suggest their potential application in bioaugmentation strategies for the remediation of oil-contaminated polar environments (Semenova et al., 2022).

As expected, the low abundance of Cyanobacteria, reflects their limited success in polar waters due to environmental constraints such as low temperatures, reduced light availability, and grazing pressure (Vincent, 2000; Zakhia et al., 2008). These factors inhibit their growth, making them less competitive compared to other bacterial taxa.

Among all the samples analysed, this area exhibits the highest presence of the Gram-positive phylum Firmicutes, especially in MED NW 8 (14.3%) and MED NW 12 (7.8%).

Although at low abundance, Campylobacterota phylum was detected only in this geographic area in 25% of the analysed samples. This group play an important role in the cycling of nitrogen and sulphur in the cold seeps of the deep ocean (Sun et al., 2023).

The genus Idiomarina, found only in this area and in the Atlantic Ocean, is associated with the breakdown of amino acids, hydrocarbons and the nitrogen cycle; its detection further highlights the role of these bacteria in maintaining ecosystem function (Hou et al., 2004; Rizzo et al., 2022).

Among the Flavobacteriaceae, Salegentibacter (McCammon and Bowman, 2000) has only been found in these sea waters, at sites not far from the coast. The occasional presence of Vibrio (3/16), known for its role in nutrient cycling and potential pathogenicity (Zhang et al., 2018), further adds to the functional diversity.

The high relative abundance of Acinetobacter in certain samples highlights its importance in bioremediation, particularly in degrading hydrocarbons and organic pollutants (Denaro et al., 2021). Its role in pollution mitigation underlines its ecological and practical relevance in particularly anthropized Mediterranean coastal waters.

Cyanobacteria, although less prevalent overall, showed significant variability, with high abundances in certain samples such as MED SW 4P. The unique microbial composition of MED SW 4P, with a lower contribution of Proteobacteria, a peak in Cyanobacteria and the highest detection of Verrucomicrobiota, suggests localised factors such as nutrient hotspots or water stratification. The role of Verrucomicrobiota in polysaccharide degradation further emphasizes the ecological relevance of this outlier (Martinez-Garcia et al., 2012; Sichert et al., 2020).

Bacteroidota, although less dominant than Proteobacteria and Cyanobacteria, were present in all samples from this area and showed a higher relative abundance when compared to samples from the northwestern Mediterranean.

Firmicutes were found less frequently in samples from the south-western Mediterranean than in the north-western area.

Idiomarina, already described above, had the highest measured abundance values in these samples.

Halomonas, a common genus in the Mediterranean and the Atlantic, was found in 75% of the sampled sites in this area. Members of the Halomonas genus are widespread in the biosphere, colonizing common to extreme environments. These bacteria display broad physiological plasticity and metabolic versatility and have developed specific adaptations that allow them to maintain or grow under extreme physical, chemical, and energetic conditions (Kim et al., 2013 ).

The observed pattern of Proteobacteria dominance and variable Cyanobacteria contribution is consistent with the findings of Quéméneur et al. (2020), who analysed prokaryotic communities in Tunisian coastal waters in the southwestern Mediterranean region. They identified Alphaproteobacteria (particularly SAR11), Gammaproteobacteria (including Alteromonas) and Cyanobacteria as major components of the microbial assemblage. They also found that spatial variability in community composition was associated with environmental gradients, such as salinity and nutrient availability. These observed community patterns may also be influenced by hydrodynamic processes and the inflow of Atlantic waters through the Strait of Gibraltar, which have been shown to affect microbial diversity and stratification in the Alboran Sea and across Mediterranean basins (Sebastián et al., 2021).

The bacterial communities in the Atlantic Ocean exhibit distinct spatial patterns driven by environmental gradients and proximity to coastal or open-ocean regions.

The presence of Marinimicrobia (SAR406) in samples closer to the American continent underscores its role in the degradation of complex organic matter and nutrient cycling in coastal or deeper waters (Wright et al., 2012). Its local enrichment may indicate distinct biogeochemical processes influenced by terrestrial runoff or proximity to the continental shelf.

In the open ocean samples (ATL OC 1-3), Pseudoalteromonas emerges as the dominant genus with relative abundances above 50%, but its presence decreases significantly as one moves closer to coastal areas. Conversely, Alteromonas displays a contrasting pattern: this genus becomes increasingly abundant in samples taken closer to the European coast (ATL OC 6–8). The dominance of Pseudoalteromonas in open ocean samples is consistent with its ecological role in nutrient cycling and organic matter remineralisation, while the abundance of Alteromonas in samples near the European coast (ATL OC 6-8) reflects its adaptation to the nutrient-rich conditions often found in coastal regions (Offret et al., 2016).

Synechococcus is notably prevalent in samples from the open ocean (except for ATL OC 2), whereas Prochlorococcus exhibits a higher relative abundance in samples collected near the European coast. While Prochlorococcus is typically associated with oligotrophic open-ocean environments (Flombaum et al., 2013), local environmental conditions—such as stratification or the presence of specific ecotypes—may explain its coastal enrichment in this region.

Finally, the genus Cobetia, which is involved in the marine sulphur cycle (Geng et al., 2024), was detected only in this area.

Microbial communities in the Indian Ocean exhibit remarkable diversity, with the highest number of phyla detected among all sampled regions. These include two archaeal phyla, Thermoplasmatota and Nanoarchaeota, albeit in low proportions, which may in part be attributable to methodological limitations previously discussed in this study. Within the Thermoplasmatota phylum, the genus Marine Group II was identified, while the genus Woesearchaeales was associated with the Nanoarchaeota phylum. Members of the order Nanoarchaeota are obligate symbionts, meaning that they require attachment to another archaeal host to grow. The detection of genera such as Marine Group II and Woesearchaeales highlights their potential role in methanogenesis and adaptability to diverse marine habitats, including the photic zone (Amils, 2011; St. John and Reysenbach, 2019).

Notably, SAR324 (Marine Group B), previously classified as Deltaproteobacteria, shows significant enrichment in sample IND OC 4. The metabolic versatility of this group, including genes for RuBisCO-mediated carbon fixation, highlights its ecological importance in carbon cycling in diverse marine habitats (Boeuf et al., 2021; Quero et al., 2020). Interestingly, this sample has the lowest representation of Proteobacteria.

Vibrio was found in 87% of the samples analysed, with the highest proportion in IND OC 6, highlighting its ecological flexibility and potential role in biogeochemical cycles (Zhang et al., 2018).

The differences in alpha diversity observed across oceanic regions reflect the influence of environmental gradients and ecological conditions that limit microbial community composition and richness. The low diversity in the Arctic Ocean is consistent with the harsh polar environment—characterized by low temperatures, limited nutrient availability, and strong seasonal light variation—which restricts microbial growth and ecosystem complexity.

In contrast, the high alpha diversity observed in the Mediterranean Sea may be explained by its dynamic hydrology, semi-enclosed structure, and biogeochemical variability (Sebastián et al., 2021). These features create a wide range of ecological niches that support diverse microbial communities.

The Atlantic and Pacific Oceans showed high spatial variability in alpha diversity, likely driven by large-scale oceanographic features such as currents, depth profiles, and upwelling zones. This variability highlights how local environmental drivers can modulate microbial diversity, sometimes overriding broader latitudinal patterns.

Reduced diversity in the Indian Ocean, particularly in semi-enclosed areas like the Persian Gulf and Red Sea, may result from strong physicochemical stressors and anthropogenic impacts such as high salinity, pollution, and eutrophication (Sheppard et al., 2010; Ismail and Al Shehhi, 2022; Qian et al., 2010). These factors likely favour a few stress-tolerant taxa, leading to lower overall evenness.

Together, these observations underline the importance of both regional conditions and large-scale gradients in shaping marine microbial communities and reinforce the value of geographically extensive surveys to understand global patterns of microbial diversity.

The observed correlations highlight temperature as a key structuring factor of bacterial communities, with clear distinctions between taxa associated with warm oligotrophic environments (the SAR86 clade, SAR116 clade, and SAR92) and those favoring colder conditions as SAR11 clade I, and UBA10353). Salinity-driven patterns were more variable, reflecting ecological differences between clades, including Marine Group II, often reported in high-salinity surface layers. The weak correlations with chlorophyll (Leeuwenhoekiella and Mesonia) suggest that primary production alone does not explain microbial community structure across broad oceanographic gradients, reinforcing the importance of including multiple physicochemical and biological parameters when interpreting bacterioplankton biogeography. Indeed, beyond the measured parameters, other factors may also influence the observed microbial distributions, including nutrient availability (e.g., phosphate, nitrate, silicate), the presence of dissolved organic matter, hydrodynamic processes (such as upwelling or stratification), interactions with viruses and grazers, and anthropogenic inputs, especially in semi-enclosed basins like the Mediterranean. These unmeasured or temporally variable drivers likely contribute to shaping bacterial communities in ways not fully captured by the environmental parameters assessed here.