Microbial communities in the nepheloid layers and hypoxic zones of the Canary Current upwelling system

Abstract Eastern boundary upwelling systems (EBUSs) are among the most productive marine environments in the world. The Canary Current upwelling system off the coast of Mauritania and Morocco is the second most productive of the four EBUS, where nutrient‐rich waters fuel perennial phytoplankton blooms, evident by high chlorophyll a concentrations off Cape Blanc, Mauritania. High primary production leads to eutrophic waters in the surface layers, whereas sinking phytoplankton debris and horizontally dispersed particles form nepheloid layers (NLs) and hypoxic waters at depth. We used Catalyzed Reporter Deposition Fluorescence In Situ Hybridization (CARD‐FISH) in combination with fatty acid (measured as methyl ester; FAME) profiles to investigate the bacterial and archaeal community composition along transects from neritic to pelagic waters within the “giant Cape Blanc filament” in two consecutive years (2010 and 2011), and to evaluate the usage of FAME data for microbial community studies. We also report the first fatty acid profile of Pelagibacterales strain HTCC7211 which was used as a reference profile for the SAR11 clade. Unexpectedly, the reference profile contained low concentrations of long chain fatty acids 18:1 cis11, 18:1 cis11 11methyl, and 19:0 cyclo11–12 fatty acids, the main compounds in other Alphaproteobacteria. Members of the free‐living SAR11 clade were found at increased relative abundance in the hypoxic waters in both years. In contrast, the depth profiles of Gammaproteobacteria (including Alteromonas and Pseudoalteromonas), Bacteroidetes, Roseobacter, and Synechococcus showed high abundances of these groups in layers where particle abundance was high, suggesting that particle attachment or association is an important mechanisms of dispersal for these groups. Collectively, our results highlight the influence of NLs, horizontal particle transport, and low oxygen on the structure and dispersal of microbial communities in upwelling systems.


| INTRODUC TI ON
Eastern boundary upwelling systems (EBUS) can be found off the coast of Peru, the USA, Namibia, and Morocco/Mauretania. EBUS are among the most productive marine environments in the world, accounting for ~10% of global ocean primary production (Behrenfeld & Falkowski, 1997). Due to the persistent upwelling of nutrient-rich waters, these upwelling systems harbor perennial phytoplankton blooms (Carr & Kearns, 2003;Carr et al., 2006;Gattuso, Frankignoulle, & Wollast, 1998). The high productivity results in extensive vertical and horizontal carbon transport (Arístegui et al., 2004), with importance for global carbon cycling. The Canary

Current upwelling system (CC) off the coast of Morocco and
Mauretania is the second most productive of the EBUS (Carr, 2001;Lachkar & Gruber, 2011). In the southern part of the CC off Cape Blanc (Mauretania) coastal upwelling of cold water sustains a year round phytoplankton bloom with highest productivity from January through June (Arístegui et al., 2009;Lathuilière, Echevin, & Lévy, 2008). Due to the off-shore Ekmann transport the coastal phytoplankton biomass production forms "giant Cape Blanc filament" that extend several hundred kilometers offshore, the largest filament in all EBUS (Van Camp, Nykjaer, Mittelstaedt, & Schlittenhardt, 1991).
The high primary production results in large vertical export of organic matter via settling of marine snow and fecal pellets thereby forming the "biological carbon pump".
Enhanced surface primary production also enhances secondary production by the bacterial and archaeal community in surface waters (Alonso-Sáez, Sánchez, & Gasol, 2012; Vaqué et al., 2014). When aggregates are formed and sink through the water column increased productivity may occur in deeper water layers (Baltar, Arístegui, Gasol, & Herndl, 2012). As zooplankton mainly feed in the surface layers, microbial activity becomes the dominant attenuation process of organic matter export at depth (Iversen et al., 2010;Stemmann, Jackson, & Gorsky, 2004). Previous studies have observed increased contribution from Bacteroidetes, Gammaproteobacteria, and Rhodobacteraceae (including the Roseobacter clade) to the microbial communities within the highly productive upwelling regions and during coastal phytoplankton blooms (Alonso-Sáez et al., 2007;Teeling et al., 2012Teeling et al., , 2016, pointing toward a connection between these groups and elevated primary productivity. In contrast, the open and oligotrophic Atlantic Ocean is often dominated by members of the SAR11 clade, whereas the abundance of Bacteroidetes is reduced (Alonso-Sáez et al., 2012;Schattenhofer et al., 2009;Thiele, Fuchs, Ramaiah, & Amann, 2012). However, most studies have so far focused on bacterial and archaeal diversity in surface waters, whereas microbial studies in deeper layers has focused on specific groups and found high abundances of SAR202, Crenarchaeota Group I, or Euryarchaeota Group II Varela, Van Aken, Sintes, & Herndl, 2008).
Here we report an investigation of the bacterial and archaeal community composition of depth profiles within two transects of the "giant Cape Blanc filament" conducted in two consecutive years.
We used CARD-FISH to analyze samples derived from depth profiles taken at six stations, and compared the results to profiles of fatty acid concentrations from the water column aimed to elucidate the bacterial community structure in the Canary Current upwelling system. In addition, we combined CARD-FISH-based community composition analyses with FAME-based taxonomy to combine benefits from both methods for the characterization of marine microbial communities. Even though the SAR11 clade is the most abundant bacterial group in marine systems, the fatty acid composition of members of this clade remains unknown. Therefore, we analyzed the fatty acid profile of Pelagibacterales strain HTCC7211, a representative of the globally abundant Ia.3 subgroup of the SAR11 clade (Stingl, Tripp, & Giovannoni, 2007), to enhance FAME analyses for Alphaproteobacteria in marine environments.

| Sampling stations
Samples used in this study were taken on an east west transect in the CC region off Cape Blanc during two cruises on RV Poseidon and RV Maria S. Merian POS 396 from 24/02/2010 to 08/03/2010 and MSM 18-1 from 17/04/2011 to 05/05/2017, as reported previously (Basse et al., 2014). Two well-described mooring stations CB (outer mooring) and CBi (inner mooring)   Sampling volumes were determined by the pump control software and via a mechanical flow meter. Samples for fatty acid analyses were collected on precombusted Whatman GF/F glass fiber filters with a diameter of 142 mm and a pore size of 0.7 μm, as previously described (Basse et al., 2014). In addition, for CARD-FISH analyses seawater was sampled at different depths from the surface to the bottom layer using a CTD rosette sampler from which 100-200 ml water samples were fixed with 1% formaldehyde final concentration. The fixed samples were serially filtered with 10 and 3 μm polycarbonate membranes (Millipore, Billerica, USA). The final filtration step was done in duplicate with 20 ml to 100 ml of the prefiltered seawater on 0.22 μm pore size polycarbonate membranes (Millipore, Billerica, USA). All samples were stored at −20°C until processing.

| Biogeochemical parameters
Water temperature, salinity, oxygen, turbidity, and chlorophyll fluorescence were measured with a self-contained SBE-19 CTD profiler equipped with a conductivity-temperature-depth probe plus oxygen sensor, a CHELSEA-fluorometer, and a WETLABS turbidity sensor.
The sensors were calibrated before the cruise, but the oxygen sensor could not be calibrated continuously during the POS396 cruise and consequently the measured values have to be interpreted as relative values.

| Lipid extraction and analyses
Filters were stored at −20°C and dried immediately prior to extraction. Four pieces were cut out of the dried GF/F filters with a broach (∅ = 12 mm) for determining particulate organic matter content (POM), assuming that the composition of cut-out filter pieces is representative of the entire filter. Lipids were extracted from the remaining filter parts as described previously (Basse et al., 2014). One subsample of a laboratory-internal sediment standard was extracted every 11 samples using the same methods. Total lipid extracts (TLE) were saponified with 300 μl of 0.1 M KOH in MeOH with 10% H 2 O at 80°C for 2 hr. After that, ~80% of the solvent was evaporated using dried N 2 , and the neutral lipids were repeatedly extracted into hexane five times. The acid fraction containing the fatty acids was recovered five times in DCM after acidifying the solution to pH = 1 with hydrochloric acid. The fatty acids were methylated by adding 3.5 ml MeOH and 174 μl 37% HCl, replacing the air in the vial with N 2 and reacting in the closed F I G U R E 1 Map of the sampling area with the stations CB, CBi, and Continental Margin vial at 50°C over night. After cooling down solvents were evaporated down to a small rest volume of a few μl and re-dissolved with ca. 200 μl DCM/Methanol (1:1). Serapure-H 2 O and hexane were added and fatty acid methyl esters (FAMEs) were extracted into hexane for five times. Hexane was evaporated and FAMES were eluted with DCM/hexane 2:1 over self-packed 6 mm diameter columns (4 cm of 1% H 2 O de-activated SiO 2 , 0.063-0.2 mm mesh size, and 0.5 cm of Na 2 SO 4 .
Culture purity was monitored by flow cytometry and further tested using a suite of three broths: ProAC, ProMM, and MPTB (Berube et al., 2015;Morris, Kirkegaard, Szul, Johnson, & Zinser, 2008;Saito, Moffett, Chisholm, & Waterbury, 2002). Cells were harvested by centrifugation, washed in phosphate buffered saline and stored at −20°C. Fatty acid methyl esters were prepared from cell pellets as described by Sasser (1990). In brief, samples underwent saponification with 15% NaOH in 50% methanol and acid methylation with 6 N HCl in 50% methanol. The FAME extracts were analyzed using a Hewlett-Packard model 6890 gas chromatograph equipped with a 5% phenyl methyl silicone capillary column and a flame ionization detector. The identity of fatty acids was verified by GC-MS with an Agilent model 7890A gas chromatograph equipped with a 5% phenyl methyl silicone capillary column and coupled with a model 5975C mass selective detector. The chromatographic conditions were used as described previously (Lipski & Altendorf, 1997). The positions of hydroxy, methyl, cyclopropane groups, and double bonds were determined from the carboxyl group of the fatty acid molecule according to the recommendations of the 1977 IUPAC-IUB Commission on Biochemical Nomenclature (CBN).

| CARD-FISH
CARD-FISH using probes specific for the investigated bacterial groups was done after a standard protocol according to Thiele (Thiele, Fuchs, & Amann, 2011) exactly as described in a previous study (Thiele, Fuchs, Amann, & Iversen, 2015). In brief, duplicate or triplicate samples from all depths were used to conduct CARD-FISH with specific probes (Supporting Information Table S1) and subsequent DAPI staining. The counting was done using an automated counting routine based on the MPISYS software and subsequent image processing using the software ACMEtool2 after manual quality control  and the relative abundance of the different probes was calculated based on DAPI counts. Posterior ANOVA tests were done using the software package R (R core team, 2014).

| Bacterial and archaeal distribution
We analyzed the bacterial and archaeal community composition using CARD-FISH (Supporting Information Table S2) and fatty acid composition (Bacteria only; Supporting Information Table S3) for similar transects sampled in consecutive years. The total detection rate of Bacteria, Thaumarchaea, and Marine Group I Euryarchaea ranged between 40% and 90%. Bacteria and Euryarchaea (measured only in 2010) abundance were highest in surface waters decreased mostly after 100 m depth, whereas the abundance of Thaumarchaea increased below 100 m and often reaching 25% of relative abundance (Figure 3a and b). This general decrease in bacterial abundance with depth is also reflected in the decreasing amount of total and bacterial specific fatty acids with depth.

| Alphaproteobacteria
The relative abundance of SAR11 bacteria was relatively low, particularly at station CB in 2010, where the relative abundance was only Roseobacter, the second group of Alphaproteobacteria investigated in the CC was found at a slightly higher relative abundance in 2010 than in 2011 and on the continental margin as compared to stations CB and CBi (Figure 3a and b). In 2010 the abundance at station CB ranged between 0.3% and 2.2%, whereas at stations CBi and the continental margin, the relative abundance was highest in the surface waters with ~4% and decreased to <1% at depth. In 2011, the relative abundance of Roseobacter decreased at all stations from the surface toward deeper waters. However, the highest relative abundances of Roseobacter were found to be 6.8 ± 3.5% at 2,600 m depth at station CBi and 6.2 ± 5.0% at 4,100 m at station CB (Figure 3b).
In addition, another peak with ~5% was found between 880 and 1,500 m at station CB (Figure 3b).
chain fatty acids showed a higher abundance at the lower depths and declined in the higher depths. The abundance of long straight chain fatty acids did not show any correlation with depth but were present in all depths for all profiles (Figure 4).

| Synechococcus
Depth profiles for the cyanobacterium Synechococcus were only taken in 2011. The relative abundance of this group increased with depth at all stations, however, the onset of the increase was found at a different depth between the stations. At the continental margin, the relative abundance of Synechococcus increased significantly from 2.3 ± 1.3% at 150 m to 5.0 ± 1.3% at 400 m (ANOVA; p < 0.01). A similar increase was observed at station CB from 2.7 ± 1.0% at 60 m to 6.0 ± 4.3% at 880 m (Figure 3b). Significantly higher numbers were found in the BL at 4,100 m with 8.7 ± 1.6% (ANOVA; p < 0.01;

Figure 3b). A significant increase in the relative abundance of
Synechococcus was found at station CBi from an average of 3.3% in the first 1,250 m of the water column to 5.6 ± 2.4% at 1,900 m and even 7.3 ± 1.0% at 2,600 m in the BL (ANOVA; p < 0.01; Figure 3b).
These lipid markers are characteristic for many bacteria, not solely Synechococcus, and showed a decrease for most profiles with depths.

| Gammaproteobacteria
Generally, slightly higher relative abundance of Gammaproteobacteria was found in 2011 as compared to 2010. The relative abundance of Gammaproteobacteria was, with the exception of station CB in 2011, highest in the surface waters and decreased immediately with depth (Figure 3b). At the continental margin, the relative abundance of Gammaproteobacteria was ~6% and decreased to ~2% below 45 m in both years (Figure 3a and b). On the contrary, the abundances of the subgroups Alteromonas and Pseudoalteromonas increased from <1% at the surface toward ~2% and ~2% with depth ( Figure 3b). In this year, two smaller increases were found within the INL (~7% between and 600 m) and the BL (11.4% between 2,150 m and 2,600 m) (Figure 3b). Similarly, Alteromonas and Pseudoalteromonas showed significantly higher abundance at 2,600 m depth with 14.0 ± 5.1% and 7.9 ± 3.8% (Figure 3b; p = 0.00162 and p = 0.0177). At station CB, Gammaproteobacteria showed a small peak of 6.4 ± 2.4% at 2,500 m, whereas in 2011, a relative abundance increased to 8.9 ± 4.6% at 1,000 m with a significant peak 18.2 ± 7.9% (ANOVA; p < 0.05) at 3,300 m ( Figure 3a and b). Similarly, peaks of Alteromonas and Pseudoalteromonas with 7.8 ± 3.2% and 7.4 ± 2.6% where found at 880 m and 6.1 ± 0.6% and 6.2 ± 0.7% at 3,300 m (Figure 3b). The fatty acid data showed the decrease in the short straight chain fatty acids, characteristic for most Gammaproteobacteria, with depth for all profiles (Figure 4) and therefore confirmed the findings from the CARD-FISH counts.

| Bacteroidetes
As for most groups tested, the relative abundance of Bacteroidetes The fatty acid profiles of Bacteroidetes are dominated by iso/ ante-iso fatty acids, such as 15:0 iso, 15:0 ante-iso, 16:0 iso, 17:0 iso, 17:0 ante-iso fatty acid (Alain, Tindall, Catala, Intertaglia, & Lebaron, 2010;Kwon et al., 2018;Thongphrom, Kim, & Kim, 2016). This is in contrast to the straight chain fatty acids found in most Alpha-and Gammaproteobacteria. The abundances of these branched fatty acids were low throughout both transects ( Figure 4) and resemble the low abundance of Bacteroidetes in the giant Cape Blanc filament.

| Eukaryotic algae
The fatty acid profiles provide also some information about the distribution of eukaryotic algae. In contrast to bacteria, algae produce long chain fatty acids including polyunsaturated fatty acids (Wood, 1988). In Figure 4 the distribution of long chain fatty acids 18:0, 18:1 cis9, and 18:1 cis9,12 (eukaryotic fatty acids) demonstrates a general relative increase in these lipid markers with depths for all profiles.

| D ISCUSS I ON
In this study, we report the structure and distribution of the bacterial and archaeal communities along a transect in the CC off Cape Blanc (Mauretania) for two consecutive years using CARD-FISH and FAME analyses. Here, the perennial phytoplankton bloom in the Cape Blanc filament of the CC shows highest intensities from January to June (Arístegui et al., 2009;Lathuilière et al., 2008). Turbidity and oxygen measurements showed that both, the NLs (INL, BL, and BL clouds) and the hypoxic waters were at the expected depths in both years (Fischer & Karakaş, 2009;Karakaş et al., 2006;Karstensen, Stramma, & Visbeck, 2008;Müller & Fischer, 2001), whereas the  (Ingalls, Huguet, & Truxal, 2012). However, high particle loads in the system results in clogging of the filter during the filtration process, which will reduce the pore size and allow a mostly representative recovery of the microbial community. In the same samples intact polar membrane lipids of archaea, which usually have small cells, were detected, indicating that small cells were captured (Basse et al., 2014). The bacterial and archaeal depth profiles of the transect, including the negative correlation of bacterial and archaeal abundance at depth, are comparable to depth profiles from other oceanographic regions (Karner, DeLong, & Karl, 2001;Schattenhofer et al., 2009).
The most abundant heterotrophic bacteria in most marine systems are Alphaproteobacteria, mainly due to the high and ubiquitous abundance of the SAR11 clade which account for 20%-50% of all marine bacteria cells and occur at all depths throughout the global ocean (Brown et al., 2012;Morris et al., 2002;Schattenhofer et al., 2009;Thiele et al., 2012). In contrast to the finding of higher abundances for many bacterial groups in the NLs, the SAR11 clade was more often found at elevated abundances in the hypoxic waters, especially at the most oceanic site station CB, where SAR11 abundance and oxygen concentrations correlated negatively (R² = −0.0424 (2010) and R² = −0.0806 (2011)). Here, the highest relative abundance of SAR11 was found either on the upper (2010) or lower (2011)  Synechococcus, the second most abundant phototrophic bacteria in the ocean is usually found in the surface layer (Flombaum et al., 2013). However, in the CC it showed lower relative abundance in the surface water potentially due to competition with the bloom forming diatoms. On the contrary, Synechococcus was found to increase in relative abundance with depth, which seems contradictory to their phototrophic lifestyle. Still, evidence exists that Synechococcus might be able to survive heterotrophically in darkness (Cottrell & Kirchman, 2009;Zubkov, Fuchs, Tarran, Burkill, & Amann, 2003;Zubkov & Tarran, 2005). As previously shown for this stations, Synechococcus abundance increases with depth which is the result of attachment to sinking particles and subsequent transport into dark waters (Thiele et al., 2015). This attachment to particles would explain the high abundance in the BL clouds and the BL, where Synechococcus might also be transported horizontally from the continental shelf.
In addition, peaks of Roseobacter, Gammaproteobacteria, and Bacteroidetes were found at depths, where particles of the INL, BL clouds, or BL were abundant. Members of all these clades are known to attach to particles (Bižić-Ionescu et al., 2014;DeLong et al., 1993).
However, the abundance of Alteromonas and Bacteroidetes on particles was found to decrease from surface to 400 m depth indicating that the attached living community was inherited from surface waters (Thiele et al., 2015). This is congruent with the finding of Bacteroidetes at the depth of the NLs, since particles from the continental shelf could function as vectors for horizontal dispersal within the INL, the BL, and the BL clouds Karakaş et al., 2006), which is supported by high relative abundances of Bacteroidetes (~25%) at the continental shelf in 2010 (data not shown). In addition, particles in the NL might accumulate bacteria by attachment or cell growth over time and distance from the shore, which would also reflect the increase in relative abundance from the continental margin to station CB in the corresponding INL and BL depths. The increase in Gammaproteobacteria abundance at 3,300 m at Station CB in 2011 could hence indicate a small BL cloud that was not detectable using the turbidity sensor of the CTD, but was indicated by increased total organic carbon levels and the influx of North Atlantic Deep Water at this depth (Basse et al., 2014;Iversen et al., 2010). Attachment to particles and subsequent vertical transport within the biological pump or horizontal Ekman transport would provide a vehicle for bacterial dispersal over large distances. However, the correlation of particle transport, bacterial attachment, and consequently bacterial dispersal within the NLs requires further research.

ACK N OWLED G M ENTS
We acknowledge the captains and crews of RV Poseidon and RV "Seasonal and regional food web interactions with the biological pump": VH-NG-1000.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA ACCE SS I B I LIT Y
CARD-FISH and FAME data were submitted to the PANGAEA database and are available online (https://doi.org/10.1594/ pangaea.889467).