The Arctic summer microbiome across Fram Strait: Depth, longitude, and substrate concentrations structure microbial diversity in the euphotic zone

The long-term dynamics of microbial communities across geographic, hydrographic, and biogeochemical gradients in the


INTRODUCTION
Recurrent observations provide a deeper understanding of ecosystem functioning over geographic, biogeochemical, and hydrographic gradients; enabling to record changes and predict future states (Buttigieg et al., 2018;Lannuzel et al., 2020).Since the Arctic warms at a fourfold faster rate than the global average (Rantanen et al., 2022), multi-year observations are key for characterizing associated shifts in biological and physicochemical regimes.Remote sensing technologies provide an avenue for tracking large-scale patterns, such as sea-ice reduction and primary productivity (Frey et al., 2023;Horvat et al., 2017;Lewis et al., 2020).However, to measure the biological responses of pelagic communities, in situ observations are necessary (Grebmeier et al., 2019;Solan et al., 2020).For instance, continuous in situ studies demonstrated that plankton diversity scales with sea-ice extent and water temperature (Lin et al., 2021).Decadal records of sinking particles in the Arctic Ocean revealed long-lasting effects of a warmwater anomaly; stimulating small phytoplankton while larger diatoms decreased in abundance, coincident with shifting bacterial composition (Cardozo-Mino et al., 2023).These dynamics have major consequences for carbon export and benthopelagic coupling (Jacquemot et al., 2022;Kohlbach et al., 2023;Salter et al., 2023).Future ocean scenarios predict substantial ecosystem shifts in the Arctic, supported by a changing microbiome structure at higher temperatures (Ahme et al., 2023).One important aspect is the impact of northward expanding Atlantic waters, termed Atlantification, on microbial diversity and the biological carbon pump (Carter-Gates et al., 2020;Oldenburg et al., 2023;Oziel et al., 2020).
A multiannual inventory of microbiological and biogeochemical patterns-comparing Arctic-versus Atlantic-influenced ecosystem states across the euphotic zone-is yet missing, and can help to predict how climate change will impact the regional microbial loop.Here, we portray microbial, biogeochemical, and oceanographic dynamics over five summers, covering $10 stations from western to eastern Fram Strait in the euphotic zone (5-100 m depth).Through ampliconsequencing and the quantification of biogeochemically important substrates, we illuminate fundamental dynamics and drivers of Fram Strait bacteria and archaea.We hypothesized that surface waters carry a signature reflecting phytoplankton productivity at different bloom stages, especially under Atlantic influence, compared to more uniform patterns in the lower photic zone.This microbial-biogeochemical inventory establishes a benchmark to assess future ecosystem shifts.

DNA extraction and sequencing
Per sampling event, 2-4 L of seawater were filtered onto 0.22 μm Sterivex cartridges (Millipore, Burlington, MA).Filters were stored at À20 C until DNA extraction using the PowerWater kit (QIAGEN, Germany) according to the manufacturer's instructions.The V4-V5 region of 16S rRNA genes was amplified using primers 515F (GTGYCAGCMGCCGCGGTAA) and 926R (CCGYCAATTYMTTTRAGTTT), providing high coverage of both bacteria and archaea (Parada et al., 2016).Amplicon libraries were prepared following the 16S Metagenomic Sequencing Library Preparation protocol (Illumina, San Diego, CA) and sequenced using MiSeq technology in 2 Â 300 bp paired-end runs.

Cell numbers
Seawater samples were fixed with glutardialdehyde (2% v/v final concentration) and frozen at À80 C until further analysis.Cells were stained using SYBR Green I (Thermo Fisher Scientific, Waltham, MA) and counted on a FACSCalibur flow cytometer (BD, Franklin Lakes, NJ) using Cell Quest v3.3 (detection limit 2000 events s À1 ) after calibration with TruCount beads (BD).

Satellite data
Sea-ice and chl-a concentrations, derived from the AMSR-2 and Sentinel 3A OLCI satellites, were downloaded from seaice.uni-bremen.deand data.marine.copernicus.eu,respectively, considering grid points within 15 km around stations.

RESULTS AND DISCUSSION
We characterized microbial, biogeochemical, and hydrographic patterns in the summers of 2015-2019 across the western (0-6 W) and eastern (0-12 E) Fram Strait (Figure 1 and Table S1).By analysing $200 seawater samples from the surface (average depth 10 m), chl-max (average 24 m), below the chlmax (average 46 m) and the lower photic zone (100 m), we determined regional and vertical patterns in microbial diversity, cell numbers and substrate concentrations across Arctic-and Atlantic-influenced ecosystem states.

Environmental parameters, cell numbers, and substrate regimes
Seawater temperatures showed marked regional differences, varying from À1.6 C to 8 C between western and eastern Fram Strait (Figure 2A).Per site, temperatures across the photic zone were similar, with a maximal difference of $2 C between surface and 100 m depth.Lower salinities in the upper western Fram Strait (Figure 2A) illustrate the influence of polar surface water, sea ice-derived meltwater or a combination of both.
The western Fram Strait harboured 1 Â 10 5 cells ml À1 throughout the upper 100 m.In eastern Fram Strait, the lower photic zone harboured similar numbers, compared to an order of magnitude more cells in the upper 25 m (Wilcoxon rank-sum test, p < 0.001; (Figure 2A).These numbers agree with cell abundances reported near Svalbard (Cardozo-Mino et al., 2021).
The concentrations of neutral carbohydrates (CHO), amino acids (AA), sugar acids, and amines peaked in the chl-max of the eastern Fram Strait (Wilcoxon ranksum test, p < 0.05), corresponding to elevated primary production (Nöthig et al., 2015) and availability of labile substrates (Piontek et al., 2014).Bulk DOC concentrations were 12% higher in the western Fram Strait (Wilcoxon rank-sum test, p < 0.05; Figure 2B).Presumably, this relates to higher concentrations of terrestrialand ice-derived DOC, which can constitute up to 30% of organic matter in Arctic waters (Nguyen et al., 2022;Opsahl et al., 1999).These patterns underscore that specific organic compounds prevail under Arctic versus Atlantic influence (Engel et al., 2019;Priest, Vidal-Melgosa, et al., 2023;von Jackowski et al., 2020).Consequently, bacterial communities in polar waters are enriched in genes targeting terrestrial compounds, compared to genes targeting phytoplankton-derived compounds under Atlantic influence (Priest, von Appen, et al., 2023).Over vertical scales, concentrations of CHO, AA, sugar acids, amines, chl-a, and DOC peaked in surface and chl-max depths, independent of the sampling site (Wilcoxon rank-sum test, p < 0.001; Figures 2B and S1).TDN concentrations were consistently higher in the lower photic zone (Wilcoxon ranksum test, p < 0.001), similar to Svalbard fjords (Osterholz et al., 2014).

Broad community patterns in the environmental context
Microbial community composition varied most by longitude and depth, explaining 15% and 16% of variability, respectively (Figure 3; PERMANOVA, p < 0.001).To a lesser extent, composition varied with Julian day, which explained 6% of the variability (PERMANOVA; p < 0.001).Separate PERMANOVA for bacterial and archaeal ASVs showed that depth was the strongest determinant for archaeal composition (R 2 = 0.17), in line with their common preference for aphotic waters.In contrast, archaea were less influenced by latitude (R 2 = 0.07) and Julian day (R 2 = 0.01) than bacteria (R 2 = 0.16 and 0.06), but all factors tested were significant (PERMANOVA; p < 0.01).
In the western Fram Strait, alpha diversity significantly increased with depth (Figure S2; Kruskal-Wallis test, p < 0.001).This subsurface peak suggests marked separation of water layers through lower salinities at the surface (Figure 2A).Alphaproteobacteria, Gammaproteobacteria, Bacteroidetes, and Verrucomicrobia represented $90% of the community at the surface and chl-max, whereas the abundance of other bacterial classes increased to $50% in the lower photic zone (Figure 4A).However, class-level proportions varied between western and eastern Fram Strait, with shifting dominance from Gamma-to Alphaproteobacteria, respectively (Figure 4A).At the order level, this shift corresponded to decreasing abundances of Dadabacteriales and Cytophagales and increasing abundances of OM182 and Puniceispirillales from the western to the eastern Strait.Some lower-photic taxa, for example, Nitrospinales, had similar abundances across all subsurface samples (Figure 4B).ASV-level dynamics indicated the presence of both "cosmopolitan" and locally confined genotypes.For example, ASVs affiliated with Pseudohongiella were restricted to the chl-max (Figure S3).Pseudohongiella has been reported from the Barents Sea and Svalbard fjords, and linked to hydrocarbon degradation (Kampouris et al., 2023;Peng et al., 2020).Overall, western subsurface waters comprised the largest number of unique ASVs (Figure S4), illustrating that earlier observations (Fadeev et al., 2018) are interannually consistent.The greatest regional overlap among subsurface ASVs indicates that lowerphotic waters are more uniform, in line with lower regional variability in water temperatures and cell numbers at 100 m depth (Figure 2A).Nonetheless, $280 ASVs were detected with a minimum 0.001% relative abundance in all samples (Figure S4).Hence, regional and vertical differences in microbiome structure not only relate to presence-absence of specific taxa but also variability in the relative abundances of shared ASVs.
Correlation analyses underlined how environmental variability influences ASV distribution.ASVs predominantly correlated with DOC, amines, and sugar acids in the western Fram Strait compared to temperature and CHO in the eastern Fram Strait, with ASVs being associated with different bacterial families (Figure 5).Such trends are probably connected with phytoplankton distribution, which varies between eastern and western Fram Strait (Nöthig et al., 2015).Accordingly, analysis of chl-a concentrations in different size fractions indicated the prevalence of smaller phytoplankton (0.4-3 μm fraction) in eastern Fram Strait (Figure S5), in agreement with previous findings (Kilias et al., 2014;Metfies et al., 2016).

Effect of sampling time and seasonality
Although depth and longitude were the strongest drivers of microbial composition (Figure 3), the influence of Julian day (i.e., the time of sampling) was pronounced for samples from surface and chl-max depths (Figure 6A).These patterns likely corresponded to varying mixed layer depth and productivity between June and September (Oldenburg et al., 2023;Wietz et al., 2021).In parallel, substrate regimes shift to more refractory compounds once phytoplankton blooms collapse (von Jackowski et al., 2022).Accordingly, Julian day explained 9% of the variability in substrate concentrations, being comparable to depth (12%) (PERMANOVA; p < 0.001).
The 2016 and 2018 samplings occurred in June/July, when diatom abundances are typically highest (von Jackowski et al., 2022, Wietz et al., 2021).Hence, these samplings likely occurred during the peak phytoplankton bloom, supported by maximal CHO concentrations and Flavobacteriales abundances (Figure 6B).These patterns mirror the ecological relationships between flavobacteria, phytoplankton, and algal substrates in temperate seas (Teeling et al., 2012).In contrast, the 2019 sampling occurred in August/September, with high SAR11 abundances and low CHO concentrations signifying the transition to oligotrophic conditions in autumn (von Jackowski et al., 2022).Nonetheless, flavobacteria remain translationally active during such periods, even at lower abundances (Priest, Vidal-Melgosa, et al., 2023).Interannual differences were possibly magnified by contrasting sea-ice conditions, with the ice edge varying by several degrees of latitude and longitude between years (Figure S6).Such variability can markedly influence microbial composition and function, especially in the marginal ice zone (Priest, von Appen,  et al., 2023;von Appen et al., 2021).Furthermore, local variability in biological and physical parameters can have a direct effect on microbial patterns.For instance, Arctic submesoscale filaments can harbour distinct microbial communities and substrate regimes, with twofold higher organic matter export than in the surrounding waters (Fadeev, Wietz, et al., 2021).

Signature populations
Sparse partial least square regression (sPLS) established a refined picture of community composition and its environmental drivers.sPLS revealed that approximately half of ASVs were associated with distinct environmental parameters, corresponding to 10 clusters representing defined ecosystem states (Figure 7A and Table S2).On average, ASVs from signature populations constituted a relative abundance of 66%, with a maximum of 78% in the lower-photic western Strait (Table S2).The maximum in subsurface Arctic waters supports the notion of a stable community in "true" polar conditions, which might be affected by progressing Atlantification (Priest, von Appen, et al., 2023).Each cluster displayed a specific taxonomic composition (Figure 7B).The higher fraction of unclassified genera among subsurface/polar signature taxa (Table S2)-together with higher alpha-diversity and more unique ASVs (Figures S2 and S4)-underlines the presence of uncharted microbial diversity in polar waters (Fadeev et al., 2018).In the following, we discuss the potential ecology of selected signature taxa.
Cluster C2-Surface (eastern Fram Strait) C2 signature taxa (75 ASVs; Table S2) predominated in Atlantic-influenced surface waters, illustrated by strong positive correlations with temperature and longitude.Prior studies suggest associations with phytoplankton or their metabolites.For instance, SAR116 commonly encodes genes to degrade DMSP (Choi et al., 2015), an algal compound mediating interactions with bacteria (Kuhlisch et al., 2023).The predominance of Thiotrichaceae has been linked to the genetic capacity to oxidize methanethiol (Priest, von Appen, et al., 2023), a compound resulting from DMSP demethylation.The C2 taxon OCS116 occurs in areas of high primary productivity (Morris et al., 2012).The surface signature was underlined by higher proportions of SAR11 clade Ia and SAR86.Detection of Roseibacillus and Saprospiraceae indicates connectivity with nearby Svalbard fjords (Delpech et al., 2021;Park et al., 2022).
Cluster C5-Lower photic zone (central Fram Strait towards EGC) This cluster comprises 211 ASVs (Table S2) associated with subsurface Arctic conditions.The SAR324 clade, common in deep waters of the Arctic (Cardozo-Mino et al., 2021) and worldwide, encodes a versatile metabolism including alkane oxidation (Sheik et al., 2014).Lentibacter has been observed in both coastal and deep Arctic waters, especially during lownutrient conditions (Angelova et al., 2021).This combination possibly favoured establishment in polar waters, which presumably harbour a more refractory substrate pool (Priest, von Appen, et al., 2023).
Cluster C6-Julian day (western Fram Strait) This narrow cluster (15 ASVs; Table S2) predominated during the summer/autumn transition in the western Fram Strait, constituting a fourfold higher abundance in 2019 samples.This period probably features maximal seeding of ice-derived substrates into the underlying seawater (Underwood et al., 2019).Alteromonas might rely on proteolytic activities (Park et al., 2014), potentially related to ice-derived substrates as observed in related Alteromonadaceae (Underwood et al., 2019).
Cluster C8-Surface "freshwater bloom" (western Fram Strait) This cluster of 30 ASVs (Table S2) is associated with productive fresher waters of the eastern Fram Strait (positive correlation with sea-ice cover and chl-a; negative correlation with salinity).Hence, lower salinities probably stimulate primary production and the microbial web (Lester et al., 2021).SAR11 clade III has a wide salinity tolerance (Lanclos et al., 2023) and predominates in fresher Arctic waters (Kraemer et al., 2019).Polaribacter, Colwelliaceae, and Paraglaciecola are often associated with sea ice (Bowman, 2014;Deming & Eric Collins, 2017).The presence of Nitrincolaceae and Methylophagaceae suggests methylotrophic metabolism and possible methane oxidation (Gründger et al., 2021).

CONCLUSIONS
Microbial communities and substrate regimes in the Arctic Fram Strait showed marked regional and vertical gradients across five summers.Revisiting the same stations over consecutive years established a robust inventory of microbial and biogeochemical dynamics, expanding upon regional and vertical gradients reported from single expeditions (Fadeev et al., 2018;Cardozo-Mino et al., 2021).The majority of populations were coupled to sea-ice cover, temperature, depth, sampling time, and substrate regimes.The predominance of phytoplankton-associated taxa, CHOs, and AAs in Atlantic-influenced surface waters contrasted with a predominance of nitrogenous substrates in the more uniform subsurface waters.Further climate warming will presumably alter polar signatures, affecting vertical diversity gradients and the resident polar microbiome.Patterns attributed to the timing of the annual expedition highlight that time CAZyme family Longitudinal and vertical patterns in environmental parameters across Fram Strait between 2015 and 2019, averaged over longitudinal ranges (western Fram Strait: 0-6 W; eastern Fram Strait: 0-12 E). (A) Sea-ice cover, seawater temperature, salinity, chlorophyll concentrations, and bacterial cell numbers.(B) Concentrations of carbohydrates (CHO; μM), amino acids (AA; μM), sugar acids (μM), amines (μM), total dissolved nitrogen (TDN; μM), and dissolved organic carbon (DOC; mM).The full data including standard deviations are shown in Figure S1.U R E 3 Regional and vertical structuring of bacterial and archaeal communities.RDA of community structure (Bray-Curtis dissimilarity), separating samples by depth and longitude (western Fram Strait: 0-6 W; eastern Fram Strait: 0-12 E).Arrows indicate the relative influence of associated environmental parameters, determined via the envfit function.
Longitudinal shifts in bacterial and archaeal composition.(A): Relative abundances of bacterial classes, averaged between 2015 and 2019 in surface, chl-max, and lower-photic zone.(B) Underlying abundance shifts on order level, displayed as Hellinger-transformed relative abundances of the 100 ASVs with the strongest depth and/or longitude correlations.
Bacteria-environment linkages.(A) Total number of correlations between ASVs and environmental parameters in western versus eastern Fram Strait, focusing on surface and chl-max samples due to their higher substrate concentrations (Figure2B).(B) Order-level assignment of ASVs that correlate with environmental parameters.Numbers indicate the total number of correlations per order and variable.
Interannual differences in microbial and substrate regimes.(A) RDA of bacterial community structure (top) and PCA of environmental parameters (bottom), separating surface and chl-max samples by year.(B) Carbohydrate concentrations (top) and relative abundances of bacterial classes (bottom) in western and eastern Fram Strait by year.Numbers in parentheses indicate the average Julian day of sampling by year.DHAA: Total hydrolyzable amino acids; CHO: carbohydrates.
Signature populations for distinct ecosystem states.(A) ASV clusters with significant environmental correlations, based on sparse partial least squares regression (sPLS).(B) Major bacterial genera per sPLS cluster, displaying the most prominent cluster affiliation per genus as Hellinger-transformed relative abundances.
series studies need to consider the prevailing ecosystem state.Overall, our evidence establishes a benchmark to quantify persistence versus change in the Fram Strait and identifies potential consequences for ecosystem functioning.(INSPIRES).Anabel von Jackowski was supported by the Helmholtz Association and the MicroARC project (03F0802A) within the Changing Arctic Ocean program, jointly funded by the UKRI Natural Environment o b a c t e r a l e s T h i o m i c r o s p i r a l e s T h i o t r i c h a l e s N i t r o s o c o c c a l e s P u n i c e i s p i r i l l a l e s R h o d o s p i r i l l a l e s F l a v o b a c t e r i a l e s A r c t i c 9 n i c e l l a l e s T h a l a s s o b a c u l a l e s S t e r o i d o b a c t e r a l Predicted functional capacities of surface and lower-photic bacterial communities, based on MAGs with 100% 16S rRNA gene identity with signature ASVs.(A) Presence/absence of CAZyme genes encoding glycoside hydrolases (GH) and polysaccharide lyases (PL).(B) Fraction of GH-and PL-encoding genes by bacterial orders.