Contrasted Saharan dust events in LNLC environments: impact on nutrient dynamics and primary production

. The response of the phytoplanktonic community (primary production and algal biomass) to contrasted Saharan dust events (wet and dry deposition) was studied in the framework of the DUNE (“a DUst experiment in a low-Nutrient, low-chlorophyll Ecosystem”) project. We simulated realistic dust deposition events (10 g m − 2 ) into large mesocosms (52 m 3 ) . Three distinct dust addition experiments were conducted in June 2008 (DUNE-1-P: simulation of a wet deposition; DUNE-1-Q: simulation of a dry deposition) and 2010 (DUNE-2-R1 and DUNE-2-R2: simulation of two successive wet depositions) in the northwestern oligotrophic Mediterranean Sea. No changes in primary production (PP) and chlorophyll a concentrations (Chl a) were observed after a dry deposition event, while a wet deposition event resulted in a rapid (24 h after dust addition), strong (up to 2.4-fold) and long (at least a week in duration) increase in PP and Chl a . We show that, in addition to being a source of dissolved inorganic phosphorus (DIP), simulated wet deposition events were also a signiﬁcant source of nitrate (NO − 3 ) (net increases up to + 9.8 µM NO − 3 at 0.1 m in depth) to the nutrient-depleted surface waters, due to cloud processes and mixing with anthropogenic species such as HNO 3 . The dry deposition event was shown to be a negligible source of NO − 3 . By transiently increasing DIP and NO − 3 concentrations in N–P starved surface waters, wet deposition of Saharan dust was able to relieve the potential N or NP co-limitation of the phytoplanktonic activity. Due to the higher input of NO − 3 relative to DIP, and taking into account the stimulation of the biological activity, a wet deposition event resulted in a strong increase in the NO − 3 / DIP ratio, from initially less than 6, to over 150 at the end of the DUNE-2-R1 experiment, suggesting a switch from an initial N or NP co-limitation towards a severe P limitation. We also show that the contribution of new production to PP strongly increased after wet dust deposition events, from initially 15 % to 60–70 % 24 h after seeding, indicating a switch from a regenerated-production based system to a new-production based system. DUNE experiments show that wet and dry dust deposition events induce contrasting responses of the phytoplanktonic community due to differences in the atmospheric supply of bioavailable new nutrients. Our results from original mesocosm experiments demonstrate that atmospheric dust wet deposition


Introduction
The primary nutrients that limit marine phytoplankton productivity include nitrogen (N), phosphorus (P) and iron (Fe). Mineral dust deposition from arid regions provides a source of each of these nutrients to the open ocean surface waters, and can have important effects on marine biogeochemical cycles and potentially affect the global climate (e.g., Baker et al., 2003;Jickells et al., 2005;Mahowald et al., 2008). Researchers have suggested that Fe delivered by mineral dust could have stimulated oceanic productivity and may help explain the glacial-interglacial atmospheric pCO 2 transition (Martin, 1990;Falkowski, 1997;Ridgwell and Watson, 2002;Mahowald et al., 2006). While the impact of Fe on productivity has been recognized in high-nutrient low-chlorophyll (HNLC) oceanic regions (Boyd et al., 2000;de Baar et al., 2005, Boyd et al., 2007Blain et al., 2007), the ecological and biogeochemical effects of aeolian dust deposition of macro-(N, P) and micro-nutrients (Fe) in oligotrophic lownutrient low-chlorophyll (LNLC) environments remain less explored and poorly understood, although LNLC regions represent 60 % of the global ocean (Longhurst et al., 1995), and over 50 % of the global oceanic carbon export (Emerson et al., 1997). The mineral dust can be transported thousands of miles across the Atlantic and Pacific oceans and deliver large quantities of nutrients to the LNLC Atlantic and Pacific gyres, which in turn could play a significant role in stimulating primary production (Bishop et al., 2002;Johnson et al., 2003;Duarte et al., 2006;Maranon et al., 2010;Franchy et al., 2013), potentially increasing the efficiency of the biological pump in the sequestration of atmospheric CO 2 .
The Mediterranean Sea is a typical LNLC region, particularly well adapted to tackling the question of the planktonic response to atmospheric nutrient inputs. The Mediterranean Sea is an oligotrophic quasi-enclosed basin that receives a noticeable flux of dust, mainly derived from the Sahara Desert, in the form of strong pulses (Guerzoni et al. 1999;Guieu et al., 2010a). After the seasonal phytoplanktonic bloom during spring, the surface mixed layer is isolated from deeper waters by a strong thermal stratification, and becomes N and P depleted Pulido-Villena et al., 2010), leading to low primary productivity and phytoplanktonic biomass (Moutin and Raimbault, 2002;Bosc et al., 2004;Lopez-Sandoval et al., 2011). During the whole stratification period, atmospheric inputs are therefore the main source of allochtonous nutrients to the oligotrophic surface waters. By bringing new macro-nutrients (e.g., Herut et al, 1999;Ridame and Guieu, 2002;Pulido-Villena et al., 2010;Markaki et al., 2010) and Fe (e.g., Bonnet and Guieu, 2006;Wuttig et al., 2013) to the Mediterranean surface waters, Saharan dust deposition is strongly suspected of playing a key role in the control of primary production. During the BOUM (Biogeochemistry from the Oligotrophic to the Ultra-Oligotrophic Mediterranean) cruise in the summer of 2008, nutrient/dust additions in microcosms showed that primary production and chlorophyll a were N or NP co-limited in the western Mediterranean Sea (Tanaka et al., 2011), and that Saharan dust stimulated primary production (Ternon et al., 2011).
The goal of the DUNE project was to estimate the impact of Saharan dust events on an LNLC ecosystem like the Mediterranean Sea from virus to zooplankton over a period of one week, and to evaluate the biogeochemical implications associated with this forcing (Guieu et al., 2010b, Guieu et al., 2014a. Due to the logistical difficulties of investigating in situ natural dust events, and due to the inherent limitations of microcosm experiments, new experimental approaches have been developed in the framework of the DUNE project. This allowed for the first time the performance of realistic dust seeding in in situ large metalfree mesocosms (52 m 3 ), and the following of the impact on chemistry, biology and particle dynamics, taking into account the vertical dimension. In this context, the present study is focused on the response of the phytoplanktonic community to contrasted Saharan dust events (wet and dry deposition). Here, we quantified the changes in in situ concentrations of dissolved inorganic nitrogen, as well as the changes in the primary production and algal biomass after simulated dust deposition events. This study is complementary to the companion paper of Giovagnetti et al. (2013), focusing on changes in the structure and composition of the phytoplanktonic community as well as in the ecophysiological state of the phytoplanktonic cells after two successive simulated wet deposition events (DUNE-2-R).

Experimental design and dust characterization
Three distinct experimental dust additions to large mesocosms were conducted in June 2008 and from June to July 2010 in the northwestern Mediterranean Sea, in the framework of the DUNE ("a DUst experiment in a low-Nutrient, low-chlorophyll Ecosystem") project (http://www. obs-vlfr.fr/LOV/DUNE/index.html). More precisely, the experiments were realized in Elbo Bay, located in the Natural Preservation Area of Scandola (8.554 • E, 42.374°N), which is representative of the LNLC conditions of the open western Mediterranean Sea (Guieu et al., 2010b(Guieu et al., , 2014a. The mesocosm experiment design and the accuracy of the strategy are described fully in Guieu et al. (2010b). Briefly, six mesocosms (height 12.5 m, diameter 2.3 m, surface area 4.15 m 2 , and volume 52 m 3 ) entirely designed in plastic were deployed. The bags were made of polyethylene mixed with vinyl acetate, and the holding structure of PVC and polyethylene. The anchors were screwed into the sea floor at 25-30 m in depth. The mesocosms are closed systems, which prevent lateral advection and allow process studies in the vertical dimension over time. Sediment traps (a 250 mL HDPE bottle) were screwed to the bases of the mesocosms for the collection of the exported material, and were changed every day (DUNE-2-R) or every two days (DUNE-1-P, DUNE-1-P-Q). Three replicate mesocosms (D1, D2 and D3, hereafter referred to as "Dust-meso") were amended with 41.5 g of mineral dust, reproducing a high but environmentally realistic atmospheric dust deposition of 10 g m −2 (Guieu et al., 2010b). The dust used in the three seeding experiments resulted from the fine fraction (< 20 µm) of Saharan soils collected in southern Tunisia, which is a Saharan aerosol source area for the western Mediterranean Sea (details in Guieu et al., 2010b andDesboeufs, et al., 2014). To simulate a Saharan deposition event, we did not use collected rainwater or aerosols, but instead we used the fine fraction of soils as an analog to Saharan aerosols, in order to obtain enough quantity of the same material. The amount of dust per mesocosm required was 41.5 g, which resulted in a total of 125 g of dust for the three replicates for only one seeding experiment. Such a large amount of particles could not be collected from airborne dust in the vicinity of the experimental area. Moreover, Saharan dust events are sporadic, and collecting rainwater in the field at the time of the mesocosm experiment would have introduced large uncertainties into the feasibility of the project. Three other mesocosms served as controls (C1, C2 and C3, hereafter referred to as "Control-meso"). The sampling session took place every morning at the same time over the duration of the experiments. Each day, three different depths (0.1, 5 and 10 m in depth) were sampled in the six mesocosms using a trace metal clean system of permanent PVC tubing placed at the center of the bags and connected to a Teflon pump, as described in Guieu et al. (2010b). Additional depths (2.5 and 12.5 m) were sampled in two dust mesocosms during DUNE-2-R for nutrient determination. Every 48 h, seawater was also collected outside the mesocosms (hereafter referred to as 'out') at the same depths, in order to test the representativeness of the data between Control-meso and out.

June 2008 experiments
Two distinct seeding experiments were conducted with two types of dust. In the first one (10-18 June 2008, hereafter referred to as "DUNE-1-P"), the dust used was previously subjected to physico-chemical transformations through condensation/evaporation cycles that involved HNO 3 and H 2 SO 4 reproducing the photochemistry, the gradients in pH, and ionic strength during cloud processing of dust particles (Desboeufs et al, 2001, anddetails in Guieu et al., 2010b). Indeed, Saha-ran dust collected in the Mediterranean atmosphere is usually mixed with organic and inorganic material, such as sulfate and nitrate, due to cloud processing during atmospheric transport (Putaud et al., 2004;Crumeyrolle et al., 2008). This evapocondensed dust (hereafter referred to as "EC dust") contained on average 0.045 ± 0.015 % of P, 2.31 ± 0.04 % of Fe and 1.19 ± 0.05 % of N, in weight (Table 1). The seeding of the dust mesocosms in the DUNE-1-P experiment was performed with the EC dust mixed with 2 L of ultrapure water in order to mimic a wet deposition event, as (i) the mixing of dust with anthropogenic components is a process that occurs mostly during cloud processes, and (ii) dissolution processes occur mainly in rainwater in the case of a wet deposition event.
In the second experiment (20-27 June 2008, hereafter referred to as "DUNE-1-Q"), non-processed dust was used (hereafter referred to as "non-EC dust"), and contained on average 0.044 ± 0.009 % of P, 2.28 ± 0.19 % of Fe, and 0.11 ± 0.01 % of N (Table 1). The seeding of the dust mesocosms in the DUNE-1-Q experiment was conducted with the non-EC dust mixed with 2 L of unfiltered surface seawater in order to mimic a dry deposition event, as (i) the untreated dust is representative of dry-deposited dust in the Mediterranean Sea, and (ii) dissolution processes occur in seawater in the case of a dry-type deposition event. Indeed, the internal mixing between dust and polluted species is not systematically observed in the Mediterranean area (Marconi et al., 2014). Kandler et al. (2007) show that the mixing is below 2 % for the transported dust particles larger than 5 µm, and up to 5 % for particles larger than 1 µm, i.e., for the particles preferentially removed by dry deposition. The size distribution and the chemical composition of our untreated dust are consistent with the characteristics of longrange transported dust (Guieu et al., 2010b;Formenti et al., 2011;de Leeuw et al., 2014).

June-July 2010 experiments
Two successive seeding experiments (26 June-2 July 2010, hereafter referred to as "DUNE-2-R1", and 3-9 July 2010, hereafter referred to as "DUNE-2-R2"; see details in Guieu et al., 2014a) were performed with the same amount of EC dust and with the same deposition setup as in DUNE-1-P, simulating a wet deposition event. The dust was characterized by an average content of 0.055 ± 0.003 % P, 2.26 ± 0.03 % Fe and 1.36 ± 0.09 % N (Table 1; see details in Desboeufs et al., 2014). During the first seeding (DUNE-2-R1), the comparison of the data between the Control-meso and Dust-meso gives information on the effect of one wet deposition event (as for DUNE-1-P), while during DUNE-2-R2, the comparison between the Control-meso and Dust-meso gives information on the effect of the combination of two successive wet deposition events.  Ridame et al., 2013). Means that were not significantly different for a given chemical element between the different experiments (p > 0.05) are labeled with the same letter (in parentheses).

P, Fe, N contents of the dust
The particulate P and Fe contents of the EC dust (DUNE-1-P and DUNE-2-R) and non-EC dust (DUNE-1-Q) were similar (p > 0.05, Table 1). Due to the simulation of cloud water processes that involved HNO 3 , the N content of EC dust was about 10-fold higher in comparison with the non-EC dust (Table 1; Guieu et al., 2010bGuieu et al., , 2014a. Small differences of the N content in EC dust used in the 2008 and 2010 experiments (1.19 ± 0.05 % and 1.36 ± 0.09 %) were observed (details in Guieu et al., 2014a).

Primary production
All materials were previously acid washed (HCl Suprapur) following trace metal clean procedures. Before sampling, bottles were rinsed three times with the sampled seawater. One sample per depth of unfiltered seawater was collected using the trace metal clean sampling system, in the morning at two depths (0.1 and 5 m in depth) during DUNE-1-P and DUNE-1-Q, and at 5 m in depth during DUNE-2-R, for the determination of primary production (PP). Samples were collected in the six mesocosms and outside the mesocosms before and after dust seeding. CO 2 fixation rates were determined using the 13 C-tracer addition method (e.g., Slawyk et al., 1977). During DUNE-1-P and DUNE-1-Q, 2.5 mL of NaH 13 CO 3 (99 %, Eurisotop) were added to 4.5 L polycarbonate bottles for 13 C uptake determination, while during DUNE-2-R, 1.5 mL of NaH 13 CO 3 were added to 2.3 L polycarbonate bottles ( 15 N 2 tracer was also added in order to determine simultaneously N 2 fixation rates using the dual 13 C/ 15 N 2 isotopic labeling technique; see Ridame et al., 2013). Prior to DUNE-2-R, intercomparison of PP rates measured in both 2.3 and 4.5 L incubated volumes showed coefficients of variation (CV) lower than 15 % (unpublished data), which was on the same order of magnitude as those found between triplicate mesocosms (CV < 17 %). Immediately after sampling, 13 C tracer was added to obtain a final enrichment of about 9 atom % excess, and each bottle was shaken well. The atom % excess of the dissolved inor-ganic carbon (DIC) was calculated by using measured DIC concentrations at the LOCEAN laboratory (see the detailed protocol in Corbière et al., 2007). Then, the 13 C-amended bottles were incubated under in situ conditions on a mooring line outside the mesocosms for 24 h at the corresponding sampling depths (0.1 m and 5 m in depth). Incubations were terminated the following morning by filtration onto precombusted 25 mm GF/F filters (0.7 µM nominal porosity). Sample filters were stored at −20 • C and dried at 40 • C for 48 h before analysis. Concentrations of carbon (C) and nitrogen (N) in particulate matter and 13 C enrichment were quantified with an isotope ratio mass spectrometer (IRMS, Delta plus, ThermoFisher Scientific, Bremen, Germany) coupled with a C : N analyzer (Flash EA, ThermoFisher Scientific) via a type III interface. The standard deviations were 0.009 µmol L −1 and 0.004 µmol L −1 for particulate carbon and nitrogen, respectively, and 0.0002 atom % for 13 C enrichment. From these measurements, the C : N molar ratios in the particulate matter were calculated. For DUNE-1 (P and Q), PP measured at 0.1 and 5 m in depth was integrated over the mesocosm depth (12.5 m). We assume that the PP at 5 m in depth was similar to that at 12.5 m in depth, based on the similarity of the chlorophyll a concentrations (Chl a) measured at 5 and 10 m in depth in the Control-meso and Dust-meso over the experiments (p > 0.05). Besides, on some selected sampling days, PP was measured at the three depths (0.1, 5 and 10 m in depth). Depth-integrated PP calculated from measured data at the three depths was similar (−4 %) to that extrapolated from data measured at 0.1 and 5 m (assuming that the PP at 12.5 m in depth was similar to that measured at 5 m in depth).
For DUNE-2-R, PP measured at 5 m in depth was integrated assuming a uniform distribution over the entire mesocosm. Over DUNE-2-R1, Chl a was similar between the three sampling depths (p > 0.05). During DUNE-2-R2, Chl a at 5 and 10 m in depth was similar (p > 0.05), while Chl a in surface waters was slightly higher than deeper (+ ∼ 30 %, p < 0.05) at the three sampling times. Also, measurements of PP at the three depths at some selected sampling times showed that depth-integrated PP calculated from measured data at 0.1, 5 and 10 m in depth was not significantly different (+4 %) from that extrapolated from data measured at 5 m in depth (assuming a homogeneous distribution over the mesocosm depth).

Chlorophyll a
Two liters of seawater from 0.1, 5 and 10 m in depth were collected on a daily scale and then filtered onto 25 mm GF/F filters. After 24 h extraction in 90 % acetone, at 4 • C and in the dark, the fluorescence of Chl a was measured on a Turner Trilogy Laboratory fluorometer (Strickland and Parsons, 1972). The Chl a concentrations measured at the three sampling depths were integrated over the mesocosm depth (12.5 m).

Nutrient concentrations
Twenty milliliters of filtered seawater (< 0.2 µm, Sartobran cartridge filters) from 0.1, 5 and 10 m in depth were collected in acid-washed high-density polyethylene (HDPE) bottles and frozen until analysis. Nitrate (NO − 3 ) and nitrite (NO − 2 ) were analyzed according to classical methods using the automated colorimetric technique as described in Grasshoff et al. (1999), on a QuAAtro Continuous Flow Analyzer (SEAL Analytical). The precision of the measurements was ± 30 nM and ± 40 nM for NO − 3 and NO − 2 , respectively. The limits of detection, defined as three times the standard deviation of the blank, were 30 nM and 10 nM for NO − 3 and NO − 2 , respectively.
Dissolved inorganic phosphorus (DIP) concentrations presented in Pulido-Villena et al. (2010 were analyzed immediately after collection on 0.2 µm filtered seawater by spectrophotometry using a long waveguide capillary cell (LWCC); the detection limit was 2 nM (details in Pulido-Villena et al., 2010. The concentration of dissolved Fe (DFe) (< 0.2 µm) was measured by flow injection analysis with online preconcentration and chemiluminescence detection (FIA-CL); the detection limit was 10 pM (details in Wagener et al., 2010). These data are presented in Wagener et al. (2010) and Wuttig et al. (2013).

Dissolution experiments under abiotic conditions
Both types of Saharan dust (EC and non-EC dust) used during the DUNE-1-P and DUNE-1-Q experiments were studied to quantify under abiotic conditions the amount of inorganic nitrogen (nitrate + nitrite: NO x ; ammonium: NH + 4 ) potentially released in the Dust-meso after seedings. This work was complementary to the in vitro dissolution experiments performed for the quantification of DFe (Wagener et al., 2010) and DIP (Pulido-Villena et al., 2010) released from dust. All experiments were performed in a clean room under a laminar flow hood using trace metal clean techniques, and all materials were previously acid washed (5 % HCl).
Dissolution experiments were conducted in sterile artificial seawater (Chen et al., 1996). For each type of dust, a stock solution of 20 mg dust L −1 was made in artificial seawater, and quickly (less than 2 min to minimize the instantaneous dissolution processes) after homogenization, increasing amounts of this stock solution were added to artificial seawater to reach a range of nine concentrations from 0 to 20 mg dust L −1 : 0, 0.01, 0.1, 0.5, 1, 3, 5, 10 and 20 mg dust L −1 . Each condition was performed in triplicate in 250 mL Nalgene® polycarbonate bottles. After dust inoculation, bottles were placed on an agitation table in the dark for 3 and 24 h. At each time point, 125 mL of each sample were filtered on washed (0.5 % HCl and rinsed three times with ultrapure water) Nucleopore ® polycarbonate filters (0.2 µm porosity) under a 200 mbar vacuum. The filtration was performed to remove Saharan dust from the liquid phase and to stop dissolution processes. The filtrate from each sample was subsampled to measure the concentrations of NO x and NH + 4 . Samples for NO x were stored in Nalgene ® HDPE vials in the dark at 4 • C for 2 days, and NH + 4 samples were analyzed immediately. Briefly, the NO x concentration was analyzed with an AutoAnalyzer3 Digital Colorimeter (Bran Luebbe) according to the method described by Aminot and Kérouel (2007) (detection limit of 20 nmol L −1 ). For NH + 4 concentrations, 40 mL of filtrate were recovered in a Schott flask (Duran) and analyzed by the method described in Holmes et al. (1999) with a Turner Trilogy Laboratory fluorometer (detection limit of 11 nmol L −1 ). Due to salt contamination, the initial concentrations of NO x and NH + 4 in the artificial seawater were 0.19 ± 0.02 µmol L −1 (n = 12) and 1.46 ± 0.12 µmol L −1 (n = 7), respectively.

Statistical analysis
Means of Chl a, PP, and NO − 3 concentrations in the Dust-and Control-meso as well as the NO x concentrations in the dissolution experiments were compared using a repeated measure ANOVA and the Fisher least significant difference (LSD) means comparison test (α = 0.05). When assumptions for ANOVA were not respected, means were compared using a Kruskal-Wallis test and a post hoc Dunn test in XLstat software. Table 2 showed no significant differences between Controlmeso, Dust-meso and out before seeding for all experiments (Guieu et al., 2010b;2014a). Chl a (0.07-0.11 µg L −1 ) and PP (3.89-5.35 mg C m −3 d −1 ) were initially low, and were slightly higher in DUNE-1-P relative to DUNE-1-Q and DUNE-2-R (p < 0.05, Table 2). The picophytoplanktonic Table 2. Initial temperature and biological and chemical properties of seawater before seeding in experiments DUNE-1-P, DUNE-1-Q and DUNE-2-R (average in the Dust-meso, Control-meso and out). DIP: dissolved inorganic phosphorus; dl: detection limit (30 nM for NO − 3 ); nd: no data. Data for temperature, chlorophyll a, DIP, NO − 3 and DFe are the means at 0.1, 5 and 10 m in depth. Data for primary production and the C : N ratio are the means at 0.1 and 5 m in depth for the DUNE-1-P and DUNE-1-Q experiments, and at 5 m in depth for the DUNE-2-R experiment. Means that were not significantly different for a given parameter between the different experiments (p > 0.05) are labeled with the same letter (in parentheses). biomass (Chl a < 3 µm) accounted for about 70 % of the total phytoplanktonic biomass before the DUNE-1-Q and DUNE-2-R seedings (C. Brunet, personal communication, 2013; Giovagnetti et al., 2013) (no data for DUNE-1-P).

Statistical analysis of biological and chemical parameters in
Synechococcus was the most abundant phytoplanktonic organism in the < 3 µm size fraction before all the dust additions (unpublished data; Giovagnetti et al., 2013). The initial C : N molar ratios in the particulate fraction, ranging from 7.3 to 7.8, were slightly higher than the Redfield ratio of 6.6 (106 : 16). The water column was initially DIP depleted (2-5 nM), with concentrations lower in the DUNE-1-Q experiment (2 nM) than in DUNE-1-P and DUNE-2-R (5 nM; p < 0.05). The initial NO − 3 and NO − 2 concentrations were below the detection limit (< 30 nM) in the DUNE-R experiment. By assuming a maximum initial NO − 3 concentration of 30 nM in DUNE-2-R, it should lead to a maximum NO − 3 /DIP ratio of 6. Due to analytical problems, NO − 3 concentrations were not available for the DUNE-1-P and DUNE-1-Q experiments, but are strongly suspected to be below the detection limit before seeding, as shown in DUNE-2-R and in surface waters of the northwestern Mediterranean Sea during stratification Pujo-Pay et al., 2011). The initial DFe concentration was higher for DUNE-2-R (3.3 nM) relative to DUNE-1-P and DUNE-1-Q (∼ 2.3 nM, p < 0.05, Table 2).
The initial seawater temperatures over the water column were greater than 19 • C (Table 2). During DUNE-1-P, the temperature was lower than during the other ones (Guieu et al., 2014a). Over the duration of this experiment, the temperature in the water column of the mesocosms was homogenous and stable (mean T = 19.8 ± 0.5 • C, Guieu et al., 2010b), while over the course of DUNE-1-Q, temperatures increased rapidly (up to 26.0 • C at the surface), leading to a strongly marked thermal stratification typical of summer conditions (mean T 0-10 m = 3.6 • C between 21 and 27 June; Guieu et al., 2014a). Over DUNE-2-R1, changes in temperature were representative of the transition period between spring (low stratification) and summer (strong stratification) conditions. While at the beginning of the second seeding (DUNE-2-R2), stratification was well established, a destratification followed by a restratification was then observed (details in Guieu et al., 2014a). The highest temperature was recorded during DUNE-2-R2 (up to 27.3 • C at the surface). Photosynthetically available radiation (PAR) was measured at the subsurface of the mesocosms during DUNE-2-R (Giovagnetti et al., 2013;Guieu et al., 2014a). Over the 13 days of the experiment, PAR was high, with maximum values of ∼ 900 µmol photons m −2 s −1 at the subsurface. During DUNE-1-P, the percentage of sunshine duration over the whole daylight time was lower than during DUNE-1-Q, where the values were close to the maximum (Guieu et al., 2014a), so it is likely that PAR values increased between the P and Q experiments.

Changes in the nutrient concentrations after seeding
After the DUNE-2-R1 and DUNE-2-R2 seedings, strong increases in NO − 3 concentrations at the surface of the Dustmeso were observed relative to the Control-meso, where NO − 3 concentrations were lower than 30 nM (Fig. 1a). Ten hours after the first seeding, the NO − 3 concentrations in the surface waters reached 3.3 ± 0.3 µM, and then decreased to 0.5 ± 0.1 µM at the end of R1 (day 6). Five hours after the second seeding, an additional increase in the NO − 3 concentration (+9.8 µM relative to day 6 in the Dust-meso) was detected at the surface, leading to NO − 3 concentrations of 10.3 ± 2.4 µM. At the end of R2, the NO − 3 concentration in the Dust-meso decreased down to the detection limit. Lower increases in the NO − 3 concentration were recorded at 5 and 10 m in depth in the Dust-meso (maximum of ∼ 1 µM) (Supplement Fig. S1). NO − 3 and NO − 2 concentrations in the Control-meso were below the detection limit at all three depths over the 13-day duration of the experiment. No change in the NO − 2 concentration was observed in the Dustmeso after both seedings. Depth-integrated NO − 3 concentrations showed that the stock increased to 10.2 mmol NO − 3 m −2 a few hours after the R1 seeding, and then decreased to a value significantly higher relative to the Control-meso at the end of the R1 (6.3 mmol NO − 3 m −2 , day 6) (Fig. 1b). The second seeding led to an additional increase in the NO − 3 stock of about +10 mmol NO − 3 m −2 relative to day 6 in the Dustmeso. The NO − 3 stock was at a maximum a few hours after the second seeding, ∼ 16 mmol NO − 3 m −2 , representing the cumulated effect of the two seedings, and then dropped to undetectable value at the end of the experiment.
After the simulated wet deposition in DUNE-1-P and DUNE-2-R1, Pulido-Villena et al. (2010 showed transient increases in the DIP concentrations a few hours after seeding (up to +12 nM at the surface), resulting in net increases in the DIP stock (from + 21 to + 24 µmol DIP m −2 ) in the Dust-meso relative to the Control-meso. Twenty-four hours after these seedings, DIP stock dropped to initial concentrations (no significant difference compared to the controls), and remained constant until the end of the experiments. Over DUNE-2-R2, the dust addition led to a new increase in the DIP stock (+54-104 µmol DIP m −2 in the Dustmeso relative to the Control-meso) that was higher than after the first one. The DIP stock remained stable and significantly higher relative to the controls until the end of the experiment . No change in the DIP concentration was recorded in the Dust-meso after the DUNE-1-Q seeding simulating a dry deposition event (Pulido-Villena, personal communication, 2013).

Response of the phytoplanktonic community to dust seeding
Depth-integrated Chl a and PP showed no significant differences between out and the Control-meso during the whole DUNE-1-P and DUNE-1-Q experiments (Figs. 2 and 3). From day 4 in the DUNE-2-R experiment, integrated Chl a and PP increased in out relative to the Control-meso (Fig. 4) as bacterial abundance did . Over the duration of the three experiments, integrated Chl a and PP were homogeneous and remained stable in the Control-meso, as shown by the low coefficients of variation (CV < 20 %). The mean C : N molar ratio in the particulate matter did not change significantly (p > 0.05) in the Dust-meso after all the seedings (P, Q, R) relative to the Control-meso (Supplement Fig. S2). The average C : N ratios were 7.5 ± 0.4, 7.6 ± 0.6 and 7.8 ± 0.6, respectively, over the duration of the DUNE-1-P, DUNE-1-Q and DUNE-2-R experiments.

Response of the phytoplanktonic community to a wet deposition event
In the DUNE-1-P experiment, the EC dust addition led to a 1.6-fold increase (p < 0.05) in the integrated Chl a relative to the Control-meso 24 h after the seeding, and to an average 1.9-fold increase over the duration of the experiment (Fig. 2). The increase in Chl a was observed for the entire duration of the experiment, as shown by the 1.9-fold increase (p < 0.05) detected 7 days after the seeding. The increase in Chl a was observed at all the sampling depths (Supplement Fig. S3), with similar relative changes (p > 0.05). Like Chl a, integrated PP strongly increased after the seeding during 7 days (1.8-fold on average) compared to the unamended controls (p < 0.05, Fig. 2). Twenty-four hours after the seeding, PP in the Dust-meso was about twice higher in comparison with the Control-meso. Relative increases in PP at the surface and at 5 m in depth were similar (p > 0.05, Supplement Fig. S4).

Response of the phytoplanktonic community to a dry deposition event
Over the duration of the DUNE-1-Q experiment, integrated Chl a and PP did not significantly increase after the addition of non-EC dust throughout the water column (p > 0.05; Figs. 3, S5, and S6). A slight stimulation of PP (1.2-fold, p < 0.05) was detected in the Dust-meso at 5 m in depth, 48 h after dust addition (Supplement Fig. S6).

Response of the phytoplanktonic community to two successive wet deposition events
In the DUNE-2-R experiment, the first seeding (R1) led to averaged increases in integrated Chl a and PP of x1.9 and x2.3 relative to the Control-meso over the duration of DUNE-2-R1 (Fig. 4, Supplement Figs. S7 and S8). Six days after the first seeding, Chl a and PP in the Dust-meso were still at least twice as high (p < 0.05) as in the Control-meso. Twentyfour hours after the second seeding, an additional x1.8 stimulation of PP was observed. The combination of the two seedings induced a x2.4 increase in both integrated Chl a and PP relative to the Control-meso over the duration of DUNE-2-R2 (Fig. 4, Supplement Figs. S7 and S8). Over the 13-day duration of the DUNE-2-R experiment, the Chl a increase in the Dust-meso was observed over the entire mesocosm (Supplement Fig. S7). At the end of the experiment (day 13), Chl a and PP in the Dust-meso were twice as high (p < 0.05) as that in the unamended controls.

Nitrogen solubility associated with dust under abiotic conditions
Additions of both EC and non-EC dust did not significantly change (p > 0.05) the NH + 4 concentrations compared to the unamended artificial seawater (data not shown). Significant increases (p < 0.05) in the NO x concentrations were observed concomitantly with increasing EC dust concentrations for both contact times (Fig. 6a). No significant differences in the NO x concentrations (p > 0.05) released from EC dust have been found between 3 and 24 h contact times. NO x concentrations increased linearly (r 2 = 0.99) with the increasing EC dust concentration, up to 17 µM for 20 mg dust L −1 . The percentages of dissolution ( NO x × 100/N tot ) reached 100 % from EC dust concentrations higher than 0.5 mg L −1 , at both time points (Fig. 6b). Additions of non-EC dust did not significantly increase (p > 0.05) NO x concentrations relative to the unamended artificial seawater (Fig. 6a).

Contrasted responses of the phytoplanktonic community to dust deposition events
Wet and dry dust deposition events induced contrasted responses of the phytoplanktonic community. The dust seeding experiments mimicking wet deposition events (P, R) induced significant stimulation of both Chl a and PP, from 1.8to 2.4-fold (Fig. 5), while no changes were observed after the seeding simulating a dry deposition event (Q). The response of the phytoplankton to a wet deposition event was (1) fast, as shown by significant increases in PP and algal biomass 24 h after dust addition, (2) long, as depicted by increases recorded during at least a week after seeding, and (3) homogeneous over the water column, as illustrated by the stimulation of Chl a throughout the mesocosms. Despite the strong increases in biomass and PP following the wet deposition events, the values remained typical of oligotrophic systems (maxima of 0.23 µg Chl a L −1 and 13.1 mg C m −3 d −1 in the Dust-meso). PP and Chl a in the Dust-meso were higher over DUNE-2-R2 (combination of two seedings) than over DUNE-2-R1 (first seeding) (Fig. 4). The cumulated effect of the two seedings observed over DUNE-2-R2 led to a higher Chl a increase (R2, RC = 2.4) relative to the first one (R1, RC = 1.9, Fig. 5). During the first dust addition, 65 % of the increase in total phytoplanktonic biomass was indeed due to picophytoplankton (< 3 µm), while over DUNE-2-R2, the nanophytoplankton and microphytoplankton (> 3 µm) contributed to about 50 % of the Chl a increase (Giovagnetti et al., 2013). The rapid response of picophytoplankton after the R1 seeding fits their dominance and ability to outcompete bigger cells under LNLC and high light conditions, by having a higher efficiency in physiological processes in resource-limited habitats when compared to the larger ones (Raven et al., 2005). Surprisingly, no increase in Chl a was observed after the simulated dry deposition event (DUNE-1-Q experiment, non-EC dust). Previous results from bioassay experiments in the western Mediterranean Sea during stratification periods have also shown that the addition of the untreated Saharan dust analog (simulated dry deposition event of 2.5-12.5 g m −2 ) did not induce any significant Chl a change (Ridame 2001, Bonnet et al., 2005. Despite the lack of a detectable response of Chl a after the DUNE-1-Q seeding, the POC export was higher in the Dust-meso (8 ± 3 mg m −2 d −1 ) compared to the controls (2 ± 1 mg m −2 d −1 ) 24 h after addition Guieu et al., 2014b). This could be explained by potential top-down effects through grazing pressure, which could have regulated the phytoplanktonic biomass. Slight increases in PP (+20 % at 5 m in depth, p < 0.05, Supplement Fig. S6) and in Synechococcus abundance (+30 % at 5 m in depth, p < 0.05, unpublished data) were recorded after the DUNE-1-Q seeding. The abundance of copepods representing the main contributor to zooplankton also increased slightly in the Dust-meso at the end of the DUNE-1-Q experiment relative to the Controlmeso (L. Stemmann, unpublished data, 2013). The increase in the POC export without an associated increase in the biogenic stock could also be explained by the lithogenic ballast effect through aggregation and/or DOC sorption processes on the settling particles Bressac et al., 2014).

Impact of the pathway of deposition on the nutrient dynamics
Differences in the atmospheric supply of bioavailable new nutrients depending on the pathway of deposition (wet or dry) may explain the variability in the response of the phytoplanktonic community (Table 3). In the DUNE-1-P and DUNE-2-R experiments, the seeding of the mesocosms was performed with the EC dust mixed with ultrapure water in order to mimic a wet deposition event (and thus cloud processes, including the mixing of particles with compounds from anthropogenic activities such as HNO 3 ; , whereas in DUNE-1-Q, non-EC dust mixed with seawater was used to simulate a dry-type deposition event.

DIP
The total content of P was similar in EC and non-EC dust (Table 1). Through in vitro dissolution experiments, both types of dust have been evidenced as a source of DIP (Pulido-Villena, unpublished data; Pulido-Villena et al., 2010;Desboeufs, personal communication, 2013;Aghnatios et al., 2014). Nevertheless, the dissolution of P associated with Saharan dust has been shown to be about two-to five-fold higher, for an equivalent dust concentration, in ultrapure water (wet deposition) than in seawater (dry deposition), likely due to a lower pH in ultrapure water (Ridame and Guieu, 2002). Indeed, in situ measurements showed significant increases in the DIP stock after simulated wet deposition events (Pulido-Villena et al., 2010 and no changes after the simulated dry deposition event (Pulido-Villena, personal communication, 2013), while abiotic dissolution exper-iments in seawater have shown a release of DIP from non-EC dust (+3 nM for a particulate dust concentration of 1 mg L −1 after a contact time of 6 h; Pulido-Villena, unpublished data). The low amount of DIP released after the dry deposition event could have been used quickly by the biological activity, explaining why no increase in DIP concentrations was recorded after the DUNE-1-Q seeding. Indeed, the dry deposition event in DUNE-1-Q induced a strong stimulation of N 2 fixation  and bacterial respiration (Guieu et al., 2014b) (Table 3).
The N content of the EC dust was about ten-fold higher compared to the non-EC dust, resulting in significant changes in the atmospheric supply of dissolved inorganic nitrogen (DIN), depending on the pathway of deposition. Abiotic dissolution experiments showed that EC dust -mimicking Saharan dust wet deposition (DUNE-1-P, DUNE-2-R) -was a significant source of NO x , whereas non-EC dust -mimicking Saharan dust dry deposition (DUNE-1-Q) -was a negligible source of NO x . Due to the simulation of cloud water processes that involved HNO 3 , dissolution of EC dust released NO x as NO − 3 . The release of NO x by EC dust under in vitro conditions was confirmed by in situ increases in NO − 3 concentrations throughout the Dust-meso after the DUNE-2-R1 and DUNE-2-R2 wet deposition events (Fig. 1,  Table 3). Assuming a homogeneous distribution of the EC dust within the Dust-meso, this would lead to a particulate dust concentration of about 1 mg L −1 , resulting in a potential input of 0.9 µM NO − 3 (Fig. 6a). The vertical profiles of NO − 3 in the mesocosms indicate a maximum net increase in NO − 3 of +9.8 µM at the surface soon after the R2 seeding (relative to day 6 in the Dust-meso) (Fig. 1a), and reveal a non-homogeneous vertical distribution of the mineral particles in the Dust-meso. This was confirmed by optical measurements by Bressac et al. (2012Bressac et al. ( , 2014, and by the vertical distribution of crustal elements as aluminum . Based on the dissolution results (Fig. 6), a complete dissolution of NO − 3 associated with EC dust is expected within 3 hr after the additions, resulting in theoretical inputs of atmospheric new N of 8.5, 9.7 and 9.7 mmol NO − 3 m −2 in the Dust-meso, respectively, for DUNE-1-P, DUNE-2-R1 and DUNE-2-R2. Over DUNE-2-R, the successive net increases in the NO − 3 stock just after seedings (R1: +10.2 mmol NO − 3 m −2 and R2: +10 mmol NO − 3 m −2 ; Fig. 1b) are on the same order of magnitude as the input of atmospheric NO − 3 based on total dissolution of N associated with EC dust (9.7 mmol NO − 3 m −2 ). In addition to the atmospheric input of new DIN, N 2 fixation could be a significant source of new N in N-depleted waters, as strong stimulations of N 2 fixation (up to x5.3, Table 3) were detected after all the DUNE seedings . The increases in the N 2 fixation rates led to net inputs of new N of +39, +15 and 4794 C. Ridame et al.: Phytoplanktonic response to Saharan dust events Table 3. Summary of the biological and chemical changes observed after the DUNE seedings in the Dust-meso relative to the Control-meso. Results from: this study; Ridame et al. (2013);Pulido-Villena et al. (2010Pulido-Villena, personal communication (2013);Wagener et al. (2010); T. Wagener, personal communication, 2013;and Wuttig et al. (2013). +22 µmol N m −2 , respectively, after the DUNE-1-P, DUNE-2-R1 and DUNE-2-R2 seedings over the duration of the experiments , which are negligible compared to the estimated supply of new N from the wet deposition events (from 8.5 to 9.7 mmol NO − 3 m −2 ). The increase in NO − 3 concentrations observed after simulated wet deposition events could thus not be explained by the increase in N 2 fixation activity. Over DUNE-1-Q, N 2 fixation could be the main source of external N (x3.3-5.2 increases in N 2 fixation after dust addition, in Ridame et al., 2013), as dry deposition of non-EC dust was shown to be a negligible source of NO − 3 . The slight increase in PP 48 h after the DUNE-1-Q seeding (at 5 m in depth, +20 %), converted to N with a C : N ratio of 7.6 (measured during DUNE-1-Q), required an input of ∼ 8 nmol N L −1 d −1 , which was higher than the input from increased N 2 fixation (∼ 1-2 nmol N L −1 d −1 ).

DFe
EC and non-EC dust have a similar Fe content (Table 1), and represent a source of DFe to surface waters (Wagener et al., 2010;Desboeufs, personal communication, 2013;Aghnatios et al., 2014). Results of dissolution experiments of both types of dust in artificial rainwater did not show any significant differences in DFe concentrations released for equivalent dust concentrations and contact times (Desboeufs, personal communication, 2013;Aghnatios et al., 2014). Despite these similarities, unexpected changes in the DFe concentrations were observed after the DUNE seedings by Wagener et al. (2010) and Wuttig et al. (2013) (Table 3). DFe concentrations dropped to about 1.5-2 nM after the DUNE-1-P, DUNE-1-Q and DUNE-2-R1 seedings, due to DFe scavenging on settling dust particles in the Dust-meso (T. Wagener, personal communication, 2013;Wagener et al., 2010;Wuttig et al., 2013), while a transient increase in the DFe concentrations in surface waters of about 2 nM was observed after the DUNE-2-R2 seeding (Wuttig et al., 2013).
The simulated wet deposition events induced significant increases in the DIP and NO − 3 stocks in the mesocosms, as well as significant stimulations of PP and Chl a (Table 3). On the opposite end, after the simulated dry deposition event, no significant changes in DIP (and likely in NO − 3 concentrations) as in Chl a and PP were detected in the mesocosms ( Table 3). As a consequence, the pathway of atmospheric deposition clearly controls the response of the phytoplanktonic community (biomass and productivity) through the differences in the new nutrient supply. As wet deposition is the main pathway of Saharan dust deposition over the western Mediterranean Sea (e.g., Loÿe-Pilot and Martin 1996), atmospheric deposition is probably the main source of new nitrogen (NO − 3 ) (Markaki et al., 2010) and new phosphorus (DIP) to the open Mediterranean surface waters during stratification periods.

Nutrient factors controlling primary production and biogeochemical consequences Iron
Before seedings, DFe concentrations ranged from 2.3 to 3.3 nM in the tested waters (Table 2). After the DUNE-1-P, DUNE-1-Q and DUNE-2-R1 seedings, DFe concentrations dropped to about 1.5-2 nM, while primary production was either stimulated (DUNE-1-P, DUNE-2-R1) or unchanged (DUNE-1-Q) ( Table 3). Wagener et al. (2010) showed that enhanced biological activity cannot explain by itself the observed decrease in DFe after the wet deposition event simulated during DUNE-1-P. Instead, the sink of DFe was likely due to scavenging on settling dust particles and aggregates in the Dust-meso. Despite opposite changes in DFe concentration after the DUNE-2-R1 (decrease) and DUNE-2-R2 (increase) seedings (Table 3), primary production increased after both additions with a similar magnitude (Fig. 5). This is evidence that, during the DUNE experiments, the bioavailability of Fe was not a controlling factor of the growth and CO 2 fixing activity of the phytoplanktonic community. This is in good agreement with previous results of Bonnet et al. (2005) during summer in the northwestern Mediterranean Sea.

Nitrogen and phosphorus
The non-stimulation of PP after the dry deposition event suggests that the phytoplanktonic community may be limited by N and P, or be NP co-limited. In the DUNE-2-R experiment, the estimated NO − 3 /DIP ratio before seeding (< 6) was lower than the Redfield ratio (16/1), suggesting a N limitation of the phytoplanktonic activity. As the DIP concentration was also initially extremely low (5 nM, Table 2), PP could likely be co-limited by both N and P, as previously shown during summer in the northwestern Mediterranean Sea (Tanaka et al., 2011). By increasing DIP and NO − 3 concentrations in P-and N-depleted surface waters, simulated wet deposition events relieve the potential N or NP co-limitation of the phytoplanktonic activity. The wet deposition events induced significant changes in the biogeochemical conditions of the tested waters by modifying the NO − 3 and DIP dynamics and by altering the ambient NO − 3 /DIP ratio (Fig. 7). Due to the higher input of NO − 3 relative to DIP, wet deposition events (P, R1, R2) are characterized by a NO − 3 /DIP ratio much higher than the Redfield ratio, resulting in increases in the NO − 3 /DIP ratio in the Dust-meso just after seeding. Studies have previously shown that the atmospheric input to the Mediterranean Sea displays a high N : P ratio for dissolved inorganic forms Markaki et al., 2010).

DUNE-1-P and DUNE-2-R1
Both experiments showed similar patterns in the response of the phytoplanktonic activity (PP) to a wet event, as well as in the evolution of the DIP concentration (and likely of the NO − 3 concentration). As the NO − 3 stock in the Dust-meso was higher than in the controls at the end of R1, and as the DIP stock was similar to the controls, the NO − 3 /DIP ratio strongly increased from less than 6 (initially) to greater than 150 at the end of the R1 experiment (Fig. 7a), suggesting a switch from an initial N limitation or NP co-limitation of the phytoplanktonic activity towards a severe P limitation until the end of R1. The net increase in integrated PP (PP dust -PP control ) 24 h after seeding was converted to P using a C : P molar ratio of 245 : 1 determined in the particulate organic matter in surface waters of the northwestern Mediterranean Sea during stratification (Tanaka et al., 2011), and to N using a C : N ratio measured during DUNE. The phytoplanktonic P requirement (+25 and +18 µmol DIP m −2 for DUNE-1-P and DUNE-2-R1 at day 1) was on the same order of magnitude as the net increase in the DIP stock in the Dust-meso (+ 21 and + 24 µmol DIP m −2 for DUNE-1-P and DUNE-2-R1), while the phytoplanktonic N requirement was much lower than the increase in the NO − 3 stock. The stimulation of PP 24 h after the DUNE-1-P and DUNE-2-R1 seedings can explain the rapid DIP depletion (Fig. 7a). Bacterial respiration was also stimulated after these seedings (Pulido-Villena, 2010, suggesting a competition for the DIP uptake between heterotrophic bacteria and autotrophic phytoplankton, as already reported during microcosm experiments in the tropical Atlantic and the Mediterranean Sea (Thingstad et al., 2005;Maranon et al., 2010). As the ∼ 2-fold increase in PP and Chl a in the Dust-meso lasted for a week, it suggests a rapid remineralization of DIP by heterotrophic bacteria and/or the uptake of dissolved organic phosphorus (DOP) as a source of P through the alkaline phosphatase enzyme (Beardall et al., 2001). The PP integrated for the duration of the experiments and the estimated nutrient requirements showed that the atmospheric input of NO − 3 largely exceeds the phytoplanktonic N demand, which may explain the NO − 4796 C. Ridame et al.: Phytoplanktonic response to Saharan dust events

DUNE-2-R2
Over DUNE-2-R2, the temporal evolution of NO − 3 and DIP stocks was different relative to that observed over DUNE-2-R1 (Fig. 7). The second seeding (R2) permitted an additional supply of new nitrogen and phosphorus into the mesocosms (Fig. 7). Then, due to the biological activity, the NO − 3 stock in the Dust-meso decreased to undetectable values at the end of R2, while the DIP stock remained much higher than in the controls. This led to a NO − 3 /DIP ratio much lower than 16:1, indicating a potential N limitation of the phytoplanktonic activity at the end of the experiment. Over DUNE-2-R2, Giovagnetti et al. (2013) showed a change in the size structure of the phytoplanktonic community towards a larger dominance of big cells (> 3 µm). Larger-sized cells need further nutrient supply in order to be able to adjust their physiology and compete for resource acquisition and biomass increase (Raven et al., 2005;Irwin et al., 2006). After the DUNE-2-R2 seeding, the nutrient (NO − 3 , DIP) concentrations were indeed higher than after the first seeding, which can explain why the response of larger cells was mainly relevant over DUNE-2-R2. This change in the size structure of the phytoplanktonic community is consistent with a higher NO − 3 consumption (Giovagnetti et al., 2013), as large cells have higher half-saturation constants (Aksnes and Egge, 1991;Hein et al., 1995) and higher cellular metabolic requirements than small cells (Grover, 1991).
The DIP stock surprisingly remained quite stable and significantly much higher relative to the controls until the end of R2, despite twice as high rates of PP in the Dust-meso than in the Control-meso. This does not imply that the input of new DIP was not used by phytoplankton, but rather that the rate of DIP consumption through PP was compensated for by an equivalent rate of DIP production, thus maintaining a quite constant DIP stock. It is unlikely that the increase in PP was sustained by DOP, as the use of DOP by phytoplankton is more energy consuming, through the synthesis of the alkaline phosphatase enzyme. We thus hypothesize that the atmospheric DIP was assimilated by phytoplankton to sustain increases in biomass and PP, and that heterotrophic remineralization of organic phosphorus kept DIP stock quite stable through high rates of P recycling. This would imply that during DUNE-2-R2, heterotrophic bacteria did not compete with phytoplankton for the P resource, as is suspected for DUNE-2-R1. As suggested by Pulido-Villena et al. (2014), even if carbon appeared not to be the limiting nutrient, bacterial respiration could have been stimulated by labile phytoplankton-derived dissolved organic matter (DOM). Besides carbon, phytoplankton DOM also provides P, offering an alternate source of the limiting element to bacteria. This shift in the P source could explain the nonutilization of DIP by heterotrophic bacteria during DUNE-2-R2.

Consequences of dust deposition for new production
During the summer oligotrophic conditions in the western Mediterranean Sea, PP is mainly dominated by regenerated production, as about 85 % of PP is sustained by internal recycling of organic matter within the euphotic zone through remineralization processes Moutin and Raimbault, 2002;L'Helguen et al., 2002). As atmospheric deposition constitutes a source of external nutrients to the surface mixed layer, it induces by definition new production (NP). The increase in PP in the Dust-meso 24 h after seeding observed in the DUNE-1-P and DUNE-2-R experiments can thus be associated with NP. We consider that, after 24 h, the increase in PP could be mainly supported by regenerated nutrients. Based on the estimates of the NP in the Dust-meso 24 h after the DUNE-1-P and DUNE-2-R seedings from Ridame et al. (2013), the contribution of NP to PP increased in the Dust-meso from initially 15 % to 60-70 % 24 h after seeding. Wet dust deposition events induced a switch from a regenerated-production based system (NP/PP = 0.15) to a new-production based system (NP/PP = 0.60-0.70).

Summary and conclusions
Our results from original mesocosm experiments demonstrate that wet atmospheric dust wet deposition does greatly influence primary productivity and algal biomass in LNLC environments through the atmospheric supply of new nutrients. The response of the phytoplanktonic community to dust events was quantified: primary production and biomass increased ∼ 2-fold, and the stimulation can be observed for at least one week. Differences in the atmospheric input of bioavailable new nutrients depending on the pathway of dust deposition (wet or dry) led to contrasting responses of the phytoplanktonic community. Primary production increased significantly only after wet deposition events. In addition to being a source of DIP, wet deposition due to cloud processes and mixing with anthropogenic species such as HNO 3 also represent a significant source of NO − 3 , inducing a significant increase in the ambient NO − 3 /DIP ratio soon after deposition. As dry deposition was not a significant source of NO − 3 , primary production was likely N limited or co-limited by both N and P. By transiently increasing DIP and DIN concentrations in P-N starved surface waters of the Mediterranean Sea, wet deposition of Saharan dust relieves the potential N or NP co-limitation of the phytoplanktonic activity. This study underlines the importance of Saharan dust deposition on the phytoplankton dynamics in the Mediterranean Sea and potentially in all LNLC areas impacted by dust deposition such as the tropical Atlantic and Pacific oceans. Despite the significant stimulation of the phytoplanktonic production and the increase in the POC export after wet events Bressac et al., 2014), Guieu et al. (2014b) have shown that simulated wet deposition did not result in a simple fertilization effect, as the oligotrophic ecosystem keeps or reinforces its net heterotrophic character thanks to the high organic carbon remineralization due to microbial food web processes.
Modeling studies suggest that changes in climate and landuse practices in recent decades may have altered dust fluxes, and thus aeolian Fe, P and N inputs to the oceans (e.g., Mahowald and Luo, 2003). In addition, in the future, a warming atmosphere and surface waters could potentially increase the stratification of the surface waters in the Mediterranean Sea (Somot et al., 2008) and other oceanic areas such as subtropical gyres (Bopp et al., 2001). In response to this, the biogeochemical impact of the Saharan deposition events on the primary production could be more pronounced, in particular during stratification periods.
The Supplement related to this article is available online at doi:10.5194/bg-11-4783-2014-supplement.