Inuence of the Depth and Season on Microbial Community Dynamics of the Black Sea

The Black Sea is a unique environment having a thin layer of oxic-zone above and anoxic-zone below. Seasonal, vertical, and horizontal microbial assemblages were studied in terms of diversity, abundance, community structure using NGS of the 16S rRNA gene. Total of 750 bacteria species from 23 different phyla were identied. The number of species richness increased from the surface to deeper zones. Although microbial community compositions between sampling stations were similar, microbial community compositions were signicantly different vertically between zones. Community compositions of the seawater and sediment were also signicantly different. Community composition at 5 meters in summer was signicantly different from other seasons, while remaining depths appeared similar. Species of nitrite-oxidizing, sulfate-reducing, thiosulfate reducing, Iron-reducing, Fe-Mn reducing and electricity-producing bacteria were reported for the rst time in the Black Sea. Proteobacteria dominated all the sampling depths. Proteobacteria, Cyanobacteria, Bacteroidetes, and Verrucomicrobia were present in the whole water column, while Nitrospinae, Chloroexi, and Kiritimatiellaeota were restricted, appearing abundant at 75 meters and deeper layers. Vertical microbial community composition variation is attributable to environmental factors and their adaptations to the various ecological niches. ANOVA performed using each sample sequence quality. For pre-processing and clustering, data containing sequence errors were removed. After removing the ambiguous base and chimeric sequence containing reads, clustering was performed based on sequence similarity. QIIME2 was used for chimera detection, diversity analysis, rarefaction, and taxonomy information. Amplicon Sequence Variant (ASV) was used with 100% similarity for the formation of taxonomic units. Taxonomy was assigned in QIIME 2 by RDP (Ribosomal Database Project) classier (Wang et al., 2007) against the RDP database.


Introduction
Marine microbial communities play a fundamental role in the biogeochemical cycles of elements (Ji et  The Black Sea is one of the largest enclosed and isolated inland water basins (Oguz, 2005) connected to the Mediterranean Sea by narrow straits. Large freshwater input from the rivers, more than 350 km 3 /year, produces a low-density surface layer, and salty Mediterranean water ow through straits forms a high-density bottom layer (Tomczak and Godfrey, 1994). A steep pycnocline layer between these two layers is the main physical barrier limiting vertical water circulation, causing strati cation (Murray et al., 1989).
Pelagic life is restricted by complete anoxic waters below approximately 200 m depths. Dissolved oxygen is depleted below180-200 m depth, resulting in a hydrogen sul de layer, the largest known anaerobic water body in the world. Despite these challenges, diverse microbial life inhabits different layers of the Black Sea, before the advent of next generation sequencing. The Black Sea includes a spectrum of hydrographic features, seasonal and vertical strati cation. According to the water's biochemical characteristics, the Black Sea is divided into ve major strata in terms of density. These depth strata are: i. Surface mixed layer (SML) is a homogeneous layer with little variation in temperature, density or other oceanographic conditions (Kara et al., 2000). The exception being the South Eastern Black Sea's mixed layer depth which ranged from 67 m to 15 m in December and June, respectively (Agirbas et al., 2015).
ii. Thermocline/Pycnocline layer (TL) corresponds to sigma theta 14.6 (Tamminen and Kuosa, 1998). iii. Mineralization/nitri cation zone (MNZ) depth ranges between sigma theta 14.6 to 15.4, where the majority of mineralization and nitri cation is known to take place (Tamminen and Kuosa, 1998) iv. Oxygen minimum zone (OMZ) corresponds to sigma theta 15.4 to 15.8. The OMZ depth varies both seasonally and spatially depending on the circulation and the eddies' intensity. However, several investigations have shown that the OMZ could be explained better by water density than depth (Saydam et al., 1993;Tugrul et al., 1992). v. Anoxic zone (AZ), which begins after OMZ, corresponds to the depth of sigma theta 16.2 (Besiktepe, 2001).
Only a few metabarcoding surveys have been conducted to describe the diversity and community composition of bacteria in the Black Sea. Previously, bacterial diversity of the suboxic and anoxic zone in the western-central gyre (42°30′N, 30°45′E) of the Black Sea was studied by constructing a 16S rRNA gene library (Fuchsman et al., 2011). Members of the SAR11, SAR324, and Microthrix groups were the suboxic zone's dominant bacterial groups. Members of the BS-GSO-2, Marine-Group-A, and SUP05 were found as the dominant bacteria groups of the anoxic zone. Bobrova et al. (2016) found that surface water samples collected from a single location in the Northeastern Black Sea had a high abundance of Proteobacteria (51.4%), Bacteroidetes (17.3%), and Cyanobacteria (11.1%). Similarly, Bobrova et al. (2016) revealed that Cyanobacteria, Proteobacteria, Bacteroidetes, and Actinobacteria dominated the bacteria assemblages in surface waters collected from three nearby estuaries in the Northwestern Black Sea. Zhang (2020) studied the microbial abundance, diversity, and community composition of the Black Sea's different water layers collected from 12 stations (6 in open waters and 6 in coastal waters). It revealed that the microbial communities were dominated by Proteobacteria, Actinobacteria, Bacteroidetes, and Cyanobacteria.
There is no study on the inter-seasonal variation in community compositions. Moreover, the driving force and mechanisms underlying the Black Sea ecosystem's microbial community assembly are still unclear and require further attention. This study aimed to reveal the seasonal variation in microbial communities in the Black Sea through comparative metabarcoding analysis at ve density depth strata and surface sediment from coastal and offshore sampling sites. We also attempted to nd the main environmental factors affecting the microbial community composition.
Normality and homoscedasticity of the environmental data were analyzed using the Shapiro-Wilk test and Levene's test, respectively.
Three-way ANOVA was performed to evaluate each environmental parameter based on three factors: depth, season, and station using SigmaPlot v12. 3 (San Jose, CA, USA).
2.3. DNA isolation, PCR, and sequencing DNA was extracted from the 80 water samples using DNeasy PowerWater Kit (Qiagen, Hilden, Germany) and eight sediment samples using DNeasy PowerSoil Kit (Qiagen), following the manufacturer's protocol. Same procedure was also applied for negative control samples. The DNA samples' were assessed using agarose gel electrophoresis and the Nanodrop spectrophotometer (Nano-200, Allsheng Inc. China) and stored at − 20°C before processing. Primers of 314F and 805R (Albertsen et al., 2015) were used to amplify the V3-V4 hypervariable region of the prokaryotes' 16S rRNA gene. PCR was carried out in a total volume of 25 µL reaction mixture containing 0.75 µL of each primer (0.3 µM each), 0.5 µL of Kapa HiFi HotStart DNA Polymerase (1 u/µL) (Kappa Biosystem, Cape Town, South Africa), 5 µL of 5X HiFi Buffer, 0.75 µL of dNTP Mix (0.3 mM of each), 2-3 µL of DNA template, and PCR-grade water. Nuclease free water as DNA templates was used for negative PCR controls. Applied ampli cation protocol was as follows: initial denaturation at 95°C for 3 min followed by 24 cycles of denaturation at 98°C for 20 s, annealing at 55°C for 18 s, extension at 72°C for 15 s, and a nal extension at 72°C for 10 min. PCR products were assessed on 1.5% agarose gel and puri ed with NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel, Dueren, Germany). Indexed adapters were added to the amplicon by using Nextera XT Index Kit V2 (Illumina, California, USA) with 8 PCR cycles. Then PCR products were cleaned-up with AMPure XP beads before DNA libraries quanti cation with Qubit and real-time PCR (Exicycler 96, Bioneer, Korea). According to the manufacturer's instructions, samples were pooled and loaded onto an Illumina MiSeq analyzer (Illumina, USA). Samples were sequenced in both library directions (2x300 bp) with Illumina Sequencing by Synthesis (Illumina). Sequencing was undertaken by Macrogen (Korea).

Taxonomic analysis
Adapter sequences were removed, and an error correction was performed for areas where the two reads overlap using the FastP program (Chen et al., 2018). A paired-end sequence was created by assembling via sequencing both directions of the library. Among the assembled reads, the FLASH (v.1.2.11) program was used to remove reads shorter than 400 bp (Magoč and Salzberg, 2011).
FastQC (Babraham Bioinformatics, UK) program was used to check each sample sequence quality. For pre-processing and clustering, data containing sequence errors were removed. After removing the ambiguous base and chimeric sequence containing reads, clustering was performed based on sequence similarity. QIIME2 was used for chimera detection, diversity analysis, rarefaction, and taxonomy information. Amplicon Sequence Variant (ASV) was used with 100% similarity for the formation of taxonomic units. Taxonomy was assigned in QIIME 2 by RDP (Ribosomal Database Project) classi er (Wang et al., 2007) against the RDP database. Rarefaction normalization test was used to be able to compare the different samples.

Microbial diversity
All statistical analysis were conducted using R Studio v.1.2.5033. Alpha diversity indexes: Number of Operational Taxonomic Unit (OTU), Species evenness, Chao1, Shannon, Simpson's, and Inverse Simpson's diversity indexes were calculated with the vegan package. Rarefaction curves were drawn based on each sample's OTU richness. The accumulation curves for the species richness were drawn and compared with the "specaccum" function using the vegan package. Indicator taxa were detected with "multipatt" function in Indicspecies package, based on point-biserial correlation index with 999 permutations to identify, which taxa had signi cantly different detection frequencies in each season, station, depth, and zone. Microbial indicator taxa were de ned as a single or group of family taxon representing a speci c community association.

Community structure
Phylum and family level OTU data matrices were Hellinger-transformed. Environmental parameters were normalized to z-scores. Nonmetric multidimensional scaling (nMDS) was performed using the "metaMDS" function in vegan based on Bray-Curtis similarity measure to visualize and compare community compositions' patterns. Furthermore, Permutational multivariate analysis of variance (PERMANOVA) and Kruskal-Wallis test was implemented to check the community composition differences between stations, seasons, and depths. PERMANOVA was applied in vegan using "adonis" based on the algorithm of Anderson (2001). Similarity percentage analysis (SIMPER) was performed in vegan using the "simper" function and PAST v4. to identify major phyla that contributed to the community difference between different stations, seasons, and depths with 999 permutations. The OTUs with the highest contribution to community differences and environmental parameters were visually represented on the nMDS ordination using the "env t" function in vegan. The environmental parameter's impact on the microbial community composition was assessed using partial redundancy analysis (pRDA) in vegan using the "rda" function.
Mantel test evaluates environmental drivers of microbial community composition based on taxonomic data. Euclidean distance was used for environmental data, while Bray-Curtis distance for taxonomic data. The Mantel test calculated the correlation between ecological and taxonomic data with 9999 permutations in ggcor based on calculated environmental and taxonomic distances. Pairwise-Pearson's correlation analysis was implemented to evaluate the signi cance of the relationship between environmental parameters in ggcor using the "corr.test" function.

Results And Discussion
A total of 80 seawater samples were collected seasonally at ve different depths: 5, 25, 75, 100, and 200 meters, which corresponds to 5 depth strata, from four stations. A total of 8 sediment samples were collected seasonally from two coastal stations (K1 and K3).

Environmental Parameters
Environmental parameters are presented in Table S1 and Figure S1. Clear vertical strati cation was detected in some environmental parameters with different gradients between zones. The surface mixed layer (SML, 5 m) and upper thermocline (TL, 25 m) was characterized by seasonal changes in temperature, whereas the deeper zone was relatively stable. Water temperature and salinity did not differ between sites or with increasing distance from shore. The temperature at SML and TL were warmer during the summer and spring, re ecting seasonal patterns. The temperatures at the mineralization/nitri cation zone (MNZ, 75 m) and below were within the same range in all seasons. Salinity and density increased with increasing depth in all seasons. Light dramatically declined between SML and TL. Depths below MNZ appeared to be aphotic. Drastic changes in Chlorophyll-a and dissolved oxygen were observed between TL and MNZ.
The highest silicate concentration was measured in MNZ, signi cantly different than SMZ. Seasonally, the lowest silicate concentration was measured in summer, and the highest concentration was measured in autumn. The availability of silica limits diatom production . Decreased silicate in summer may be related to the increased abundance of diatoms. The highest nitrate concentration was detected in winter, whereas the lowest in summer. Vertically, in terms of nitrate, SML was the poorest, and TL was the richest zone. Spatially and vertically, nitrate concentration signi cantly (P < 0.01) differed in coastal stations and OMZ. Seasonally, nitrate concentration in autumn and winter was signi cantly different from spring and summer. Comparatively higher nitrate and lower nitrite in this study indicated the rapid conversion of ammonia to nitrate. The highest phosphate concentration was measured in autumn, which signi cantly differed from other seasons. Phosphate concentration increased with depth, with the highest concentration in AZ. The lowest ammonia concentration was calculated in autumn, which showed a signi cant seasonal difference. Vertically highest ammonia concentration was calculated for OMZ, whereas the lowest for TL. Dissolved oxygen concentrations decreased dramatically after TL in all seasons, and hydrogen sul de concentrations increased exponentially after Oxygen minimum zone (OMZ) and remained relatively constant in all seasons. The highest total organic carbon (TOC) concentrations were measured in surface waters during the summer months, attributed to increased water temperature and phytoplankton growth. Although water quality parameters measured at stations close to the shore (K1 and K3) were considerably higher in most of the cases compared to the offshore stations, differences were insigni cant (p<0.05).

Metabarcoding and Microbial Diversity
The MiSeq paired-end sequencing of the libraries yielded a total of 2,358,752 reads. After assembly of paired-reads, quality ltering, and chimera removal, a total of 2,019,141 valid reads were retrieved from 88 samples. The total number of sequence reads ranged between 157,998 (Summer-K3-25 m) and 285,952 (Spring-K1-SED), with an average of 207,960. The number of raw reads for each library was relatively uniform. Average Phred quality score for Q30 was 98.66.
Metabarcoding successfully detected 23 phyla (Table S2), 45 classes, 102 orders, 208 families (Table S3), 527 genera, and 750 species (Table S4) from a total of 88 samples collected seasonally from 4 different stations at ve different depths in each station in the Black Sea. Rarefaction curves of sampling sites were similar, and most reached a plateau, excluding sediment samples, indicating that sampling effort was su cient to detect microorganisms in the water community ( Figure S2). The rarefaction curves of the sediment samples did not form plateau which might indicate that sediment samples is not a good representation of the microbial community. Thus, the accumulation curves failed to reach a plateau ( Figure S3). Sampling different layers of sediment would give a better representation of the sediment microbial community. Alpha diversity indices are presented in Table S5 and Figure S4. There are a few metabarcoding studies performed in the Black Sea. Bobrova (Bobrova et al., 2016) detected the presence of 8 bacterial phyla from the surface waters of three estuaries. Zhang et al. (2020) conducted the most comprehensive metabarcoding study in which water samples were collected at four vertical layers from 12 stations for once and detected 141 prokaryotic species. Our nding differed from previous studies in which comparatively more diverse microbial communities were found, with higher-level at deeper stations. This could also be due to methodological differences especially in the bioinformatics. Also, the previous study did not have sediment samples which would increase diversity. In this study, prokaryotic reads were assigned to 750 bacteria species. The percentage of taxonomically classi ed reads was high, which is the expected outcome of seasonal sampling from different water layers. The present study's annotation results re ected the Black Sea's microbial communities' composition based on the current knowledge and information contained in reference databases. Our ndings indicated that the number of observed OTUs and species richness increased from the surface to deeper zones re ecting diversi ed species adapted to a broader range of ecological niches. An increase in species richness from the surface to deeper layers was also reported previously (Pommier et  At the phylum level, almost half of the phyla (9 out of 25) represented in each sample (Figure 2a). Overall, the majority of the reads and OTUs belonged to Proteobacteria (40% reads and 58.4% OTUs), Cyanobacteria (19.5% reads and 2.8% OTUs), Planctomycetes (12.2% reads and 2.5 OTUs), Bacteroidetes (10.9% reads and 13.8% OTUs), Verrucomicrobia (9.6 reads and 2.9 OTUs), and Actinobacteria (2.3% reads and 5.6% OTUs) (Table 1, Figure 2b and 2c). The major phyla composition, which represented the highest number of reads and OTUs, was uniform among seasons ( Figure 2d) and stations (Figure 2e). Meanwhile, the composition of most phyla differed among depths. Almost 20% of lineages (5 out of 25) were dominant in SML and TL, 48% lineages (12 of 25) in MNZ and OMZ, and 32% lineages (8 of 25) were more abundant in AZ (Figure 2f).
After excluding OTUs with low reads at the family level, 29 out of 76 OTUs were presented in each station (Figure 3a). Overall, the bacterial community was dominated by Prochloraceae, a family of Cyanobacteria (16.9% reads) represented with a single species, Prochlorococcus marinus. The remaining majority of the reads and OTUs belonged to Rhodobacteraceae (11.6% reads and 11% OTUs), Planctomycetaceae (10% reads and 0.9% OTUs), and Flavobacteriaceae (7.5% reads and 13.2% OTUs), which is the most diverse family represented with 60 OTUs (Figure 3b and 3c). The composition of the major families represented with the highest number of reads and OTUs was uniform among seasons ( Figure 3d) and stations (Figure 3e). Most of the community composition differed with depth at the family level (Figure 3f). Some of the lineages' distribution was non-uniform among different depths with asymmetrical allocations of reads ( Figure S5-S10).
A total of 12 families were identi ed as indicator taxa for the seasonal changes in microbial communities. A detailed list of indicator family taxa for stations, seasons, depths, and zones is tabulated in Table S6. In particular, Cyanobacteria families of Chamaesiphonaceae and Spirulinaceae were among the strongest indicator taxa in the Autumn and Spring season. Anaerolineaceae and Spirulinaceae families were recovered as indicator taxa for station groups of K1 and K1-K3, respectively. However, not all sampling sites belonging to these groups are included in these taxa. We observed zonation in the microbial communities belonging to Desulfobacteraceae, Vicinamibacteraceae, Ectothiorhodospiraceae, Dehalogenimonas (Genus), Desulfuromonadaceae, and Methyloligella (Genus) being commonly recovered from water samples below MNZ (Sewell et al., 2017; Kato et al., 2019; He et al., 2020), while Lentimicrobiaceae and Desulfobulbaceae were identi ed as indicator taxa for AZ. The absence or low recovery of Caldilineaceae appeared as an indicator for surface SML. Families of Rhodothermaceae and Mycobacteriaceae were identi ed as indicator taxa for SML, TL, MNZ, and OMZ, respectively, with low restriction to these zones. On the other hand, four taxa were assigned as an indicator for an anoxic zone. The strongest indicator taxon for AZ was OTUs assigned to Anaerolineaceae. The presence of OTUs belonging to the families, Desulfobacteraceae, Caldilineaceae, and Vicinamibacteraceae were strongly associated with aphotic and anoxic zones (Table S6).

Variation in microbial community composition
Microbial community composition difference between sampling sites was insigni cant (p>0.05). K3 station was chosen to assess whether the copper mining discharge system in that speci c region at a depth of 180 meters affected the microbial community structure and composition. However, there was no signi cant difference between sampling stations (PERMANOVA, F=0.927, R 2 =0.031, p=0.538). Similar communities could be found between long distances as long as no oceanographic front was crossed.  , 2020). Seasonally, only winter samples were signi cantly different from spring samples when all the depth groups were considered whole for each station (p = 0.0201). Speci cally, community composition in SML in summer was signi cantly different than in other seasons. Temperature appeared to be a leading cause of the difference between shallow depths. Community composition at remaining zones was insigni cant seasonally (p>0.05). Sampling fast-changing unstable surface layer can be problematic. Besides uctuations in environmental parameters, sampling at different sites and seasons may result in microbial community differences. The grouping of samples (based on Bray-Curtis dissimilarity) displayed on nMDS plots showed similarity in microbial community composition between the four sampling sites (stress: 0.1161) ( Figure 4A). In contrast, dissimilarities were observed between seasons and depths ( Figure 4B and C). There was a strong in uence of sampling depth on microbial community composition, where samples from SML-TL, MNZ-OMZ, and AZ were separately clustered ( Figure 4D). Clustering samples from SML and TL were expected since both depths were within the mixed layer. Samples from 75 and 100 m did not join the surface cluster. Moreover, these samples were more similar to the deeper than the surface samples. The grouping of microbial communities at the various season, depths, and zone in each sampling station are presented in Figure S11. Seasonally, microbial community composition in autumn was less tightly clustered within seasonal clustering. SIMPER analysis revealed that the AZ was differentiated from the other zones by an increase in abundance of taxa belonging to the family Helicobacteraceae. In contrast, decreased abundance of Rhodobacteraceae, Prochloraceae, Verrucomicrobiaceae, and increased abundance of Planctomycetaceae and Nitrospinaceae in the MNZ and OMZ were the major contributor to the dissimilarity between the communities of SML-TL and MNZ-OMZ. The major contributor phyla and families to the dissimilarities between seasons, stations, depth, and zones are tabulated ( Figure 4, Table S7-S10).
Among the most abundant phyla, Proteobacteria, Cyanobacteria, Bacteroidetes, and Verrucomicrobia were present in the whole water column. Conversely, Nitrospinae, Chloro exi, and Kiritimatiellaeota were more restricted, appearing abundantly at 75 meters and deeper layers. While the abundance of Planctomycetes, Acidobacteria, and Firmicutes increased with depth, the abundance of Cyanobacteria, Bacteroidetes, and Balneolaeota decreased with depth. Ecologically distinct microbial taxa occupy different niches. Nevertheless, information about most of the bacterial taxa in the marine environment is limited. The Proteobacteria members' adaptation (Planctomycetes, Bacteroidetes, Verrucomicrobia, and Acidobacteria group) to most water depths spanning the entire chemical gradient could indicate an adaptation to a wide range of chemical stressors and an e cient survival strategy. SML and TL were dominated by Cyanobacteria that perform photosynthesis and nitrogen xation with light and oxygen. Cyanobacteria were found in all water column and sediment, but their abundance decreased in the deeper layers. A comparatively lower but still high abundance of Cyanobacteria in AZ might be due to the precipitation of dying bacteria from the water column into the sediment. The abundance of Nitrospinae, Chloro exi, Kiritimatiellaeota, and Planctomycetes were high in MNZ and below, inconsistent with the previous report on the abundance of the Chloro exi taxa below the chlorophyll maximum zone (Morris et al., 2002;Qian et al., 2011). Iron-reducer (Geoalkalibacter ferrihydriticus) and Iron-Manganese reducer (G. subterraneus) of the genus Geoalkalibacter (Greene et al., 2009) were found at 75 m and below, including sediment. Geobacter sulfurreducens was mostly found in the water column where H 2 S is present. Since G. sulfurreducens produce electricity (Poddar and Khurana, 2011), environmentally friendly microbial fuel cells can be produced (Bond and Lovley, 2003) to create electricity by the degradation of waste products using G. sulfurreducens (Poddar and Khurana, 2011).
Phylum Nitrospinae is represented with ve species in the Black Sea. Nitrococcus mobilis was detected in high abundance in the aphotic and anoxic zones, whereas N. gracilis was only detected in the sediment. N. mobilis and N. gracilis are obligate chemoautotrophic nitrite-oxidizing bacterial species. Enrichment of N. mobilis indicates prevalent nitrite cycling in the aphotic and anoxic zones (Watson and Waterbury, 1971). Although the other three obligately anaerobic, thermophilic, sulfate-reducing bacterial species (Thermodesulfovibrio hydrogeniphilus, T. aggregans, and T. thiophilusare) were previously isolated from thermophilic environments (Haouari et al., 2008;Sekiguchi et al., 2008), their presence was detected in AZ with water temperature around 8°C.
Sulfate-, sulfur-and sul te-reducing bacteria produce hydrogen sul de by reducing different inorganic and organic materials. Sulfuroxidizing and anoxic bacteria groups were enriched in AZ due to the high H 2 S and low O 2 content in these deep waters in the Black Sea. Sulfate-reducing bacteria use sulfate as an electron acceptor during the breakdown of organic material to produce hydrogen sul de (Rückert, 2016). Sulfate-reducing bacteria families (Helicobacteraceae, Granulosicoccaceae, Desulfobacteraceae, and Desulfobulbaceae) of the phylum Proteobacteria dominated the AZ. Enrichment of these groups is a clear indication of prevalent sulfur cycling. Shewanella hafniensis and Shewanella vesiculosa that produce H 2 S from thiosulfate were found in all water column and sediment. Metabolic activities of identi ed sul de reducing bacteria taxa distributed abundantly in sediment and 100-200 meters might be the source of H 2 S in the Black Sea. The presence of strict anaerobic bacterial groups in high abundances, such as the Anaerolineaceae family of the phylum Chloro exi in AZ, re ects the anoxic properties of these layers. Members of Fusobacteria, Gemmatimonadetes, and Lentisphaerae phyla were absent in the sediment. Distribution and presence of bacterial phyla in AZ were similar, except for Lentisphaerae, found only in waters, and Fibrobacteres found only in sediment.
Based on abundance, the most common ve species found in each depth and sediment of the Black Sea were Prochlorococcus marinus, Gimesia maris, Kineobactrum sediminis, Azotobacter beijerinckii, and Planctopirus limnophila. Genus of Prochlorococcus has a single species called Prochlorococcus marinus, common in oligotrophic oceans (Partensky et al., 1999). P. marinus contains both chlorophyll-a and chlorophyll-b, which supports its growth in deeper waters with low light (Ralf and Repeta, 1992). Chlorophyll-b absorbs blue light and carries out photosynthesis even at 200 m depth as long as blue light penetrates (Zinser et al., 2007). This might be the main reason behind the distribution of photosynthetic cyanobacteria, P. marinus, abundantly in all of the Black Sea's sampled depths or environmental DNA of the dead individuals might be another reason behind the presence of photosynthetic cyanobacteria in anoxic zone and aphotic zone. Although P. marinus adapt in low phosphate waters by replacing phospholipids membranes to sulfolipids membrane (Van Mooy et al., 2006), phosphate concentrations were not scarce, ranging between 0.02 µM in summer at surface water and 2.5 µM in fall in AZ.
Members of the Planctomycetes are known to colonize both oxic and anoxic layers (Dedysh and Ivanova, 2019). Planctomycetes are an important phylum for carbon and nitrogen cycles in the environment found in fresh, brackish, and marine water (Van Niftrik et al., 2004). Gimesia maris, previously known as Planctomyces maris, a species of the phylum Planctomycetes, was rst isolated from the shallow waters at USA (Ferreira et al., 2016). Planctopirus limnophila is a member of the Planctomycetaceae family that do not contain peptidoglycan in their cell wall (Jeske et al., 2015). G. maris and P. limnophila were recorded in each sampling depth abundantly, yet the research on these species' function is limited. Kineobactrum sediminis was rst described and isolated from marine sediment (Chang et al., 2019). K. sediminis was found in each sampling depth and sediment in the Black Sea. The Genus of Azotobacter is mostly a soil bacterium that play a role in the nitrogen cycle by binding atmospheric nitrogen and converting it to ammonium. Azotobacter beijerinckii is found in soils and water associated plants (Tejera et al., 2005). It was the fourth most abundant bacterial species in the Black Sea. Although Azotobacter xes atmospheric nitrogen in the soil, its function in the water column is unknown. All the ve most abundant species were also found in the sediment of AZ. Since we used molecular techniques to identify microorganisms, both the live and debris of dead organisms could be identi ed. Thus, their actual presence in the sediment cannot be ascertained.

In uence of Environmental Parameters on Microbial Community Structure
When the entire water column was considered, the permutation test indicated that temperature, salinity, sigma-t, conductivity, light, chlorophyll-a, O 2 NO 3 , NH 3 , PO 4 , Si, and H 2 S appeared to have a signi cant effect on the vertical bacterial community richness (p<0.05). In contrast, TOC and NO 2 did not have any signi cant impact (p>0.05) on vertical bacterial community compositions following Bonferroni type correction (Table S11). Among the signi cant environmental factors, temperature (r 2 = 0.605), PO 4 (r 2 = 0.634), and O 2 (r 2 = 0.33) were found to have the most in uence on the overall bacterial community composition. Correlation between each environmental parameter with the microbial community composition at each zone is presented in Figure S12. Correlation between environmental parameters and vertical distribution of bacterial families are presented in Figure 5. Varied correlations were detected between environmental variables and the community structures of samples. Partial redundancy analysis ( Figure S13) was performed to assess further the relationship between environmental variables, which revealed a similar set of environmental variables that statistically in uenced the microbial community. Environmental parameters are represented by arrows that point toward the direction of variation.
Correlation between environmental parameters ( Figure S13) and microbial communities appeared to be different in each zone. In SML and TL, the temperature was positively correlated with conductivity (Pearson's test R 2 >0.95, p<0.001), and salinity negatively correlated with TOC. Mantel test indicated that temperature and conductivity were strongly related to microbial composition (Mantel's R>0.5, p<0.01) ( Figure 6A). In MNZ and OMZ, conductivity was positively correlated with salinity and dissolved oxygen (Pearson's test R 2 >0.95, p<0.001). Conversely, oxygen was negatively correlated with conductivity and H 2 S. Salinity and H 2 S were positively correlated with conductivity. Mantel test indicated that NO 2 , NO 3 , PO 4 , and Si had an in uence on community composition (Mantel's R>0. 25, p<0.01) ( Figure 6B). In AZ, NO 3 was negatively correlated with temperature, salinity, conductivity, and NH 3, whereas conductivity was positively correlated with temperature and salinity (Pearson's test R 2 >0.95, p<0.001). Mantel test indicated that PO 4 was strongly related to microbial composition (Mantel's R>0.5, p<0.01) ( Figure 6C).
The Black Sea is an inland water body with permanent vertical strati cation. In terms of water density, there are ve well-de ned strata in the Black Sea. In this study, seasonal uctuations and heterogeneity of environmental parameters between the zones led to differences in the microbial community. Microbial diversity and density showed trends consistent with vertical strati cation and environmental parameters for some of the taxa. High temperature, light, and dissolved oxygen in the upper zones facilitated the growth of photosynthetic bacteria in high density. However, lack of dissolved oxygen and light and high concentrations of hydrogen sul de in the anoxic zone facilitated heterotrophic and anaerobic microorganisms in great abundance. OMZ and MNZ appeared to be the most diverse zones. Numerous studies addressed the environmental factors (

Conclusions
Marine microorganisms in uence environmental dynamics and play a key role in marine productivity and biogeochemical processes by the combined effect of countless interactions at a single-cell level. Microbial communities in the Black Sea are under-studied. In this study, 88 metabarcoding libraries were generated to assess the seasonal, vertical, and horizontal variation of microbial community compositions in the Northeastern Black Sea. Our results revealed that microbial communities differ with depth and uctuate seasonally at upper layers. This study is the rst to assess microbial communities seasonally and vertically in the Black Sea and provide insight into microbial ecology in one of the world's unique marine ecosystems. Contrary to popular belief, extreme conditions in the deep sea, such as the absence of light, low dissolved oxygen level, and low temperature, do not limit microbial diversity, con rmed by the comparatively higher diversity in deeper zones.  Figure 1 Sampling stations in the Black Sea    Two-dimensional Nonmetric multidimensional scaling (NMDS) ordination of microbial communities detected at different depths.
Black lines represent environmental parameters responsible for causing the major differences. Strong predictors have longer lines than weak predictors. TOC and NO2 did not exhibit signi cant relationships with bacterial community composition following the Bonferroni type correction and excluded.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.