Characterising the surface microlayer in the Mediterranean Sea: trace metals concentration and microbial plankton abundance

Characterising the surface microlayer in the Mediterranean Sea: trace metals concentration and microbial plankton abundance Antonio Tovar-Sánchez1, Araceli Rodríguez-Romero1, Anja Engel2, Birthe Zäncker2, Franck Fu3, Emilio Marañón4, María Pérez-Lorenzo4, Matthieu Bressac5,6, Thibaut Wagener7, Karine Desboeuf3, Sylvain 5 Triquet3, Guillaume Siour3, Cécile Guieu7 1Department of Ecology and Coastal Management, Institute of Marine Sciences of Andalusia (ICMAN-CSIC), Puerto Real, 07190, Spain 2Leibniz Institute of Marine Sciences (IFM-GEOMAR), Kiel, Germany 3Laboratoire Interuniversitaire des Systèmes Atmosphériques (LISA), CNRS UMR 7583, Université de Paris, Université Paris10 Est-Créteil, Institut Pierre Simon Laplace (IPSL), Créteil, 94000, France 4Departamento de Ecología y Biología Animal, Universidad de Vigo, 36310 Vigo, Spain 5Institute for Marine and Antarctic Studies, University of Tasmania, Hobart, Tasmania, Australia 6Sorbonne Université, CNRS, Laboratoire d’Océanographie de Villefranche, LOV, F-06230, Villefranche-sur-mer, France 7Aix Marseille Univ., CNRS, IRD, Université de Toulon, MIO UM 110, 13288, Marseille, France 15


Introduction
The Mediterranean Sea (MS) is enriched in many trace metals relative to similar nutrient-depleted waters in the open ocean (e.g. Cd, Cr, Co, Cu, Ni, Fe, Zn) (Bonnet et al., 2013;Boyle et al., 1985;Sarthou and Jeandel, 2001;Sherrell and Boyle, 1988). The enrichment of metals in surface water has been 5 associated to different sources including atmospheric deposition, river inflows, groundwaters, anthropogenic sources and the Atlantic Ocean inflow through the Gibraltar Strait (Boyle et al., 1985;Elbaz-Poulichet et al., 2001;Migon, 2005;Trezzi et al., 2016). The MS has one of the highest rates of aeolian deposition in the world with strong pulses of mineral dust from Africa, in addition to consistent anthropogenic aerosol inputs from Europe. Therefore, atmospheric deposition, dry and wet, is the 10 dominant pathway for large scale transport of trace metals to the water column and sediments in MS (Guieu et al., 2002(Guieu et al., , 2010Jordi et al., 2012;Ternon et al., 2010;Tovar-Sánchez et al., 2010, 2014. Many of these metals play an important role in biogeochemical processes of this sea. For example, it has been hypothesized that the high Co concentrations in the MS stimulate "de novo" synthesis of vitamin B12 as Co is the central metal ion in the B12 molecule (Bonnet et al., 2013). Although present in higher 15 concentration than in other oceans, Fe has been considered as an important factor controlling phytoplankton growth (Sarthou and Jeandel, 2001). Copper from aerosol depositions has been demonstrated to have toxic effects on marine phytoplankton (Jordi et al., 2012;Paytan et al., 2009) while Ni and Zn have been considered as good geochemical tracers of aerosols impact in Posidonia oceanica (Tovar-Sánchez et al., 2010). 20 Studying the Sea-Surface Microlayer (SML), especially in a region dominated by aeolian deposition, is crucial for understanding trace metal dust solubility, ocean distribution, and the processes influencing the primary production and the vertical particle fluxes in the water column. The SML is considered the skin of the ocean as it serves as a boundary layer between the atmosphere and the ocean.
With a thickness of 1-1000 µm, it is a prevalent feature of the surface ocean that shows distinct physical, 25 chemical, and biological properties than the rest of the water column. This sea-air interface plays a key role in regulating the exchanges of gases, solutes and energy between water and atmosphere and is central to a wide range of global biogeochemical and climate regulation processes (Cunliffe et al., 2013). 4

Material and Methods
Samples from SML, SSW and aerosols were collected during the cruise PEACETIME (ProcEss studies at the Air-sEa Interface after dust deposition in the MEditerranean sea) on board the French R/V 'Pourquoi Pas?' in the MS, from May 10 th to June 11 th , 2017. Twelve stations were sampled (Figure 1). 5 Three of these stations were sampled twice (TYR 1-2, ION 1-2) or five times  at different days, counting for a total of 17 groups of samples (Table 1).

Aerosol sampling and analysis
The PEGASUS container was installed aboard the R/V Pourquoi Pas?, this container is a mobile 10 platform equipped with a set of instruments optimized to collect and analyze in real time, gaseous compounds and particles in the atmospheric boundary layer (Formenti et al., 2019). Atmospheric sampling was performed using isokinetic and wind-oriented aerosol multi-samplers with a total sampled flow rate around 400 L min -1 per inlet. This inlet was developed for sampling both fine and coarse particles, with particles of aerodynamic diameter of about 40 µm (Rajot et al., 2008). This total flow was 15 subdivided to various transmission lines which served the majority of the instrumentation. The aerosol size distribution from 10 nm to 30 µm was measured by a combination of standard optical and electrical mobility analyzers. The total mass concentration was obtained by an on-line Tapering Element Oscillating Microbalance (TEOM, model 1400a, Rupprecht and Patashnick).
One of the sampling lines was equipped with filtration unit to collect the aerosols on 47-mm 20 polycarbonate membranes of 0.4 µm pore size (Whatman Nuclepore TM ). The volume flow rate was set at 20 L.min -1 . All the filters were previously cleaned by immersion in ultrapure HCl (2%) during 2 hours and rinsing with ultrapure waters. A sampling strategy was made to avoid the contamination by the cruise smoking, firstly when the vessel was in station, the R/V was systematically positioned such as inlets were facing the wind (PEGASUS container and boat's chimney are on the opposite side of the deck). On the 25 route, contamination-free sampling was operated when the relative wind direction was not in the direction of chimney exhaust. In total, 36 series of filters were collected which 17 filters during the stations and 5 blanks of filters were also prepared. The sampling locations for each filter is presented in Figure 1.

5
Aerosols filters were first analyzed by X-ray fluorescence spectrometry (SFX, spectrometer PW-2404, Panalytical™) for measuring chemical markers of particles origin sources (as Al and Ca). Filters were then leached by ultrapure water in order to determine the soluble fraction of metals. Finally, the filters were mineralized using an acid digestion protocol adapted from (Heimburger et al., 2013) et al., 2012;Tovar-Sánchez et al., 2019) which had been previously cleaned with acid overnight and rinsed thoroughly with ultrapure water (MQ-water). The 39 x 25 cm silicate glass plate had an effective sampling surface area of 1950 cm 2 considering both sides. In order to check for procedural contamination, we collected SML blanks in some stations on board of the pneumatic boat by rinsing the glass plate with 20 ultra-pure water and collecting 0.5 L using the glass plate system. The surface microlayer thickness was calculated following the formula of Wurl (2009) (Wurl, 2009). Total fraction of SML (i.e. T-SML) were directly collected from the glass plate system without filtration in a 0.5 L acid cleaned LDPE bottles, while that the dissolved fraction in the SML (i.e. D-SML) was rapidly filtered on board the pneumatic boat through an acid-cleaned polypropylene cartridge filter (0.22µm; MSI, Calyx®). SSW were collected 25 using an acid-washed Teflon tubing connected to a peristaltic pump and directly filtered on the same cartridge to collect the dissolved fraction (D-SSW). All samples were acidified on board to pH< 2 with Ultrapure-grade HCl in a class-100 HEPA laminar flow hood. Metals (i.e. Cd,Co,Cu,Fe,Ni,Mo,V,Zn 6 and Pb) were pre-concentrated using an organic extraction method (Bruland et al., 1979) and quantified by ICP-MS (Perkin Elmer ELAN DRC-e). Prior to the preconcentration and for the breakdown of metalorganic complexes and the removal of organic matter (Achterberg et al., 2001;Milne et al., 2010), total fraction samples (i.e. T-SML) were digested using an UV system consisting in one UV (80 W  Microorganism inhabiting the SML are collectively referred to as the "neuston" (Engel et al., 2017). At the same time than trace metal samples collection, microorganism in the SML were sampled also using a glass plate system (50 x 26 cm silicate glass plate with an effective sampling surface area of 2600 cm 2 considering both sides). The water from the SSW was manually collected in acid clean 25 borosilicate bottles at around 20 cm depth. Bacterial numbers were determined using flow cytometry from a 4 mL sample that was fixed with 200 mL glutaraldehyde (GDA, 1% final concentration). Samples were stored at -20 ºC for at most 2.5 months until analysis and were stained with SYBR Green I (Molecular 7 Probes) prior to quantification using a flow cytometer equipped with a 488 nm laser (Becton & Dickinson FACScalibur). A plot of side scatter (SSC) vs. green fluorescence (FL1) was used to detect the unique signature of the bacterial cells. The internal standard consisted of yellow-green latex beads (Polysciences, 0.5 mm). Abundance and area of Transparent Exopolymer Particles (TEP) were measured microscopically following a previously described method (Engel, A., 2009). 5

Phytoplankton and Primary Production
Chl-a concentration and primary production were measured in the SSW at 5 m depth. Primary production was measured with the 14 C-uptake technique. Seawater samples, collected from Niskin bottles at dawn, were dispensed into four (3 light and 1 dark) polystyrene bottles of 70 mL in volume, which 10 were amended with 15 µCi of NaH 14 CO3 and incubated for 24 h inside a deck incubator refrigerated with surface seawater from the continuous water supply. The incubator was covered with a neutral density filter that provided an irradiance level of 70% of incident PAR. After incubation, samples were filtered, using low vacuum pressure, through 0.2-µm polycarbonate filters, which were exposed to HCl fumes overnight to remove non-fixed, inorganic 14 C. After adding 5 mL of liquid scintillation cocktail to the 15 filters, the radioactivity on each sample was determined on-board with a liquid scintillation counter. To compute the rates of carbon fixation, the dark-bottle DPM value was subtracted from the light-bottle DPM value and a value of 26,000 µgC L -1 was used for the concentration of dissolved inorganic carbon. Chla concentratios were measured by HPLC (HPLC Agilent Technologies 1200) following the method described by Ras et al. (2008) (Ras et al., 2008.

Statistical analyses
Spearman rank correlation coefficient (rs) was used to determine significant relationships (p< 0.05) between the parameters measured in the different compartments (air, SML and SSW) and parameters. Coefficient of determination (R 2 ) between the selected parameters were also calculated in 25 order to determine how well correlations fit with a linear regression relationship. Statistical analysis was performed with the aid of the statistical software package SPSS 25. 8

Aerosols deposition
Metal aerosols composition is shown in Table 1. Average concentrations in our study were in the same order of magnitude, than previous measurements collected in the same region and season (Becagli et al., 2012;Calzolai et al., 2015;Tovar-Sánchez et al., 2014), and hence were consistent with Western Ni and V in the collected aerosols all along the cruise, suggest a common source associated to heavy oil combustion; i.e. marine ship traffic (Becagli et al., 2012). Some rains occurred during the cruise, but only 15 one was measured when the vessel was in station, June 5 th from 2:36 am to 3:04 am between Fast 3 and Fast 4 samples. However, all the zone around the Fast station was rainy from the 3 rd of June ( Figure S3).
As the collected rain composition was typical to dust wet deposition with high particulate concentrations of Al, Fe and Ca (Fu et al., in preparation), we suppose the rain-out of dust in the atmospheric column around this station occurred between the 3 rd to the 5 th of June.  (Table 1).

Trace metals in the SML
Trace metals concentrations of T-SML (Table 1)  Dissolved concentrations of Co, Zn, Pb, Cu and Ni showed a decreasing trend from the SML to the SSW, with concentrations 10.4 ± 0.7; 9.3 ± 5.5; 4.2 ± 1.8; 3.1 ± 1.5; and 1.2 ± 0.1 times higher in the SML than in the SSW, respectively. Vanadium (1.2 ± 0.42) and Fe (1.3 ± 1.5) varied lightly between 10 SML and underlayer water, and Mo (1.0 ± 0.1) did not showed any differences (Table 1). Only Cd concentrations were consistently lower in the SML compared to the underlayer water (0.8 ± 0.2 times lower). Such depletion of dissolved metals in the SML compared to the underlayer water has been previously observed in areas with no significant aerosols inputs (Ebling and Landing, 2015, 2017).
Although not fully understood, some mechanisms such as dominance of removal mechanisms versus 15 diffusion, or higher influence of underlaying metal sources have been previously suggested to explain this metal depletion (Ebling and Landing, 2017;Hunter, 1980). Spatial distribution of Co and Ni concentrations in the D-SML were well correlated with those measured in the D-SSW (Spearman's correlation coefficient (rs): 0.87 for Co and 0.91 for Ni; p<0.01, Table 2), indicating for these elements an efficient diffusive mixing between these two compartments. 20 These elements were also well correlated with the surface salinity distribution (rs: 0.62 for Co and 0.93 for Ni; p<0.01, Table 2), and presented an eastward trend of increasing concentration, which is consistent with the characteristic distribution of metals on the surface of the MS (see section 3.2.4. below).
Variations in concentrations for the rest of the elements (i.e. Cd, Cu, Fe, Pb, V and Zn) in the D-SML were not correlated either with the underlayer water or salinity gradient. Multiple physical, chemical and 25 biological processes taking place in the SML could be affecting and controlling the mobility and diffusion of these elements between compartments. However, the concentrations of Cu, Fe and Zn in the T-SML showed an opposite trend with a longitudinal gradient inversely correlated with the salinity (rs: -0.59 for Cu; -0.69 for Fe, and -0.61 for Zn; p<0.01, Table 2). Since aerosols metal concentrations did not show any longitudinal trend and no other natural or anthropogenic sources were identified in the region, gradient concentration of these reactive trace elements in the T-SML must be influenced by other factors different than sources inputs, water exchange or dilution with Atlantic waters.  (Table 3). Residence time of particulate metals (T-SML) ranged from 12 min for Co and Fe to 7.6 h for Cu. Since Mo and Cd are not enriched in the SML they were not considered in this calculation. Although variable among stations, residence time of Cu (3.0 residence time. Since, such fast transfer of these particle metals to the underlayer water (in the order of 1-3 min) is unlikely (mainly due their affinities to organic ligands), and dissolution is not immediately reflected in an increase of concentration in the dissolved fraction (i.e. D-SML), other parameters (linked to dynamic or biological activity) would be affecting the residence time of these elements in the SML. In the case of Co, wind seems to partly explain this short residence time. Wind speed showed high and 5 significant negative correlation with the residence time of Co (rs: -0.67, p<0.01), Ni (rs: -0.76 p< p<0.01) and V (rs: -0.80 p< p<0.01) in the SML, suggesting an influence of wind on the diffusion of these elements to the underlayer water (Table 2).  Table 1. Bacterial abundance in the SML ranged from 2x10 5 to 1x10 6 cell mL -1 (average: 5.1x10 5 ± 2.2x10 5 cell mL -1 ) that is of the same order of magnitude 15 than abundance measured in the SML of other regions (e.g. in the Peruvian Coast with average of 8.9x10 5 ± 4.3x10 5 cell mL -1 ) (Zäncker et al., 2018). Bacterial community was dominated by low nucleid acidcontent bacteria (LNA) with an average concentration of 2.8 x10 5 ± 1.0x10 5 cell mL -1 . In general, and with the exception of phytoplankton middle and CBL-small, microbial abundance was higher in the SML than in the SSW with abundances ranging from 1 to 6 times higher for bacteria and CBL-middle-large, 20 respectively (Table 1).

Neuston composition
A microbial abundance decreases from west to east related to the increasing oligotrophy (explained though an increased P limitation) of the surface Mediterranean waters has been described (Pulido-Villena E. et al., 2012). In this study, microbial abundance in the SML and T-SML reactive elements (i.e. Cu, Fe, and Zn) showed the same longitudinal gradients with decreasing eastward  (Table 1). No general relationship between concentrations of metals and TEPs (high molecular weight polymers released by phytoplankton and bacteria and with high metal binding capacity (Passow, 2002)) were found in the SML, although both showed strongly increased values after a dust deposition event at FAST 3 (Table 4). We 5 therefore assume that metal assimilation by microbial communities would also explain the higher residence time of Cu and Zn (in the order of hours) in the SML. However, in the case of Fe with an estimated residence time of a few minutes, other processes different than wind speed and neuston uptake, should be contributing to facilitate the transfer from the SML to the underlayer water. For example, photochemical reactions drive by intense solar radiation exposure in the SML could play an important 10 role in the dissolution processes of this metal (Boyd et al., 2010). On the other hand, Ni was strongly and negatively correlated with bacteria abundance in the D-SML (rs = -0.93, p<0.01; R 2 = 0.74, p<0.01) suggesting, contrarily to Cu, Fe and Zn, a possible inhibiting role on the microbiology growth (Table 4 and Figure 2) (see next section for more discussion).

Subsurface water
The D-SSW concentrations of Cd, Co, Cu, Ni, Mo and Zn showed a longitudinal gradient of concentrations increasing from west to east, with strong significant positive correlations with longitude for Cd, Co and Ni ( Figure 3). This trend is consistent with previous studies where the increased eastward concentration along the southern coast of the MS is indicated to be due to factors such as: more intense 20 Saharan deposition on the eastern MS (Guieu et al., 2002); more rapid exchange of water masses and margin inputs in the Western part (Yoon et al., 1999) or, as suggested for Co, the regeneration of biogenic particulate eastward that yields a westward decreased of the dissolved Co in surface (Dulaquais et al., 2017). Since surface salinity showed the same eastward increase and was close correlated with those metals (rs ranged from 0.51 p<0.05 for Mo to 0.97 p<0.01 for Ni; Table 2), the exchange with the surface 25 Atlantic Ocean waters seems to be the main cause of this gradient of concentrations in our study, although higher aerosol inputs in the western MS could also contribute to this gradient. Other metals (i.e. Fe, Pb and V) did not show any clear geographical trend and variations in surface concentrations could be 13 influenced by several factors other than dilution or exchange, such as vertical diffusive fluxes or, specific metal sources, as it is in the case of Fe and Pb that has been suggested to be more affected by atmospheric inputs (Nicolas et al., 1994;Yoon et al., 1999). In fact, Pb was the only element that showed significant positive correlation with latitude (rs: 0.88 p<0.01, Table 2) suggesting an influence of the northern region of the MS. bioaccumulation and/or toxicity of many trace elements. Even if a general decreasing trend from west to east of microbial abundance due to the increasing oligotrophy has been described, it is interesting to mention that primary production and Chl-a concentration (measured a 5 m depth), did not show significant 20 correlations with Ni (Table 4). We therefore assume that toxicity of Ni is mainly affecting the bacterial community and/or on the top meter of the surface ocean. Nickel, as other transition metals, is an essential cofactor of several enzymes, however, it becomes toxic when homeostasis fails. Multiple potential mechanisms of Ni toxicity to aquatic organisms, and in particular to bacteria, have been identified  (Brix et al., 2017), which are abundant in the SML. It appears that Nickel-dependent toxicity involving ROS may be likely mechanism of oxidative stress in marine microbial organism of the surface ocean. There is many information on the effect of Ni on insolate laboratory microalgae experiments, however its toxicity role in oceans have been 5 poorly explored. Therefore, additional studies on Ni diffusion from SML, solubility, speciation, and the effects on phytoplankton at the species level are required to fully understand the magnitude of this finding.

Conclusions
Our results show that the SML in the MS is enriched in trace metals relative to the SSW even under low      or nmol.m -3 for aerosols and nM or pM in water compartments. Table 2. Spearman's rank correlation coefficients for selected parameters. Significant correlations at p<0.05 and p<0.01 are marked with one asterisk (orange numbers) and two asterisks (red numbers), respectively.    Table 4. Spearman's rank correlation coefficients for selected parameters. HNA: High nucleid acid-content bacteria;: HNA; LNA: Low nucleid acid-content bacteria; CBL-small: Small cyanobacteria like cells; CBL-middle-large: middle-high cyanobacteria like cells.
Significant correlations at p<0.05 and p<0.01 are marked with one asterisk (orange numbers) and two asterisks (red numbers), respectively.