We report here the elemental composition of sinking particles in sediment traps and in the water column following four artificial dust seeding experiments (each representing a flux of 10 g m−2). Dry or wet dust deposition were simulated during two large mesocosms field campaigns that took place in the coastal water of Corsica (NW Mediterranean Sea) representative of oligotrophic conditions. The dust additions were carried out with fresh or artificially aged dust (i.e., enriched in nitrate and sulfate by mimicking cloud processing) for various biogeochemical conditions, enabling us to test the effect of these parameters on the chemical composition and settling of dust after deposition. The rates and mechanisms of total mass, particulate organic carbon (POC) and chemical elements (Al, Ba, Ca, Co, Cu, Fe, K, Li, Mg, Mn, Mo, N, Nd, P, S, Sr and Ti) transfer from the mesocosm surface to the sediment traps installed at the base of the mesocosms after dust deposition show that (1) 15% of the initial dust mass was dissolved in the water column in the first 24 h after seeding. Except for Ca, S and N, the elemental composition of dust particles was constant during their settling, showing the relevance of using interelemental ratios, such as Ti/Al as proxy of lithogenic fluxes. (2) Whatever the type of seeding (using fresh dust to simulate dry deposition or artificially aged dust to simulate wet deposition), the particulate phase both in the water column and in the sediment traps was dominated by dust particles. (3) Due to the high Ba content in dust, Ba/Al cannot be used as productivity proxy in the case of high dust input in the sediment traps. Instead, our data suggests that the ratio Co/Al could be a good productivity proxy in this case. (4) After 7 days, between 30 and 68% of added dust was still in suspension in the mesocosms. This difference in the dust settling was directly associated with a difference in POC export, since POC fluxes were highly correlated to dust lithogenic fluxes signifying a ballast effect of dust. The highest fraction of remaining dust in suspension in the mesocosm at the end of the experiment was found inversely correlated to Chl a increase. This suggests that the fertilizing effect of dust on autotrophs organisms, the ballast effect, and POC fluxes are strongly correlated. (5) Our data emphasize a typical mass ratio Lithogenic/POC fluxes around 30 which could be used as reference to estimate the POC export triggered by wet dust deposition event.

The deposition of atmospheric dust is the primary process supplying trace elements abundant in crustal rocks (e.g. Al, Mn and Fe) to the surface ocean. Upon deposition, the residence time in surface waters for each of these elements differs according to their chemical speciation and biological utilization. Presently, however, the chemical and physical processes occurring after atmospheric deposition are poorly constrained, principally because of the difficulty in following natural dust events in situ. In the present work we examined the temporal changes in the biogeochemistry of crustal metals (in particular Al, Mn and Fe) after an artificial dust deposition event. The experiment was contained inside trace metal clean mesocosms (0–12.5 m depths) deployed in the surface waters of the northwestern Mediterranean, close to the coast of Corsica within the frame of the DUNE project (a DUst experiment in a low Nutrient, low chlorophyll Ecosystem). Two consecutive artificial dust deposition events, each mimicking a wet deposition of 10 g m−2 of dust, were performed during the course of this DUNE-2 experiment. The changes in dissolved manganese (Mn), iron (Fe) and aluminum (Al) concentrations were followed immediately after the seeding with dust and over the following week. The Mn, Fe and Al inventories and loss or dissolution rates were determined. The evolution of the inventories after the two consecutive additions of dust showed distinct behaviors for dissolved Mn, Al and Fe. Even though the mixing conditions differed from one seeding to the other, Mn and Al showed clear increases directly after both seedings due to dissolution processes. Three days after the dust additions, Al concentrations decreased as a consequence of scavenging on sinking particles. Al appeared to be highly affected by the concentrations of biogenic particles, with an order of magnitude difference in its loss rates related to the increase of biomass after the addition of dust. In the case of dissolved Fe, it appears that the first dust addition resulted in a decrease as it was scavenged by sinking dust particles, whereas the second seeding induced dissolution of Fe from the dust particles due to the excess Fe binding ligand concentrations present at that time. This difference, which might be related to a change in Fe binding ligand concentration in the mesocosms, highlights the complex processes that control the solubility of Fe. Based on the inventories at the mesocosm scale, the estimations of the fractional solubility of metals from dust particles in seawater were 1.44 ± 0.19% and 0.91 ± 0.83% for Al and 41 ± 9% and 27 ± 19% for Mn for the first and the second dust addition. These values are in good agreement with laboratory-based estimates. For Fe no fractional solubility was obtained after the first seeding, but 0.12 ± 0.03% was estimated after the second seeding. Overall, the trace metal dataset presented here makes a significant contribution to enhancing our knowledge on the processes influencing trace metal release from Saharan dust and the subsequent processes of bio-uptake and scavenging in a low nutrient, low chlorophyll area.

Intense Saharan dust deposition occurs over large oligotrophic areas in the Mediterranean Sea and in the Tropical Atlantic, and its impact on the biogeochemical functioning of such oligotrophic ecosystems needs to be understood. However, due to the logistical difficulties of investigating in situ natural dust events, and due to the inherent limitations of microcosm laboratory experiments, new experimental approaches need to be developed. In this paper, we present a new experimental setup based on large, clean mesocoms deployed in the frame of the DUNE (a DUst experiment in a low-Nutrient, low-chlorophyll Ecosystem) project. We demonstrate that these tools are highly relevant and provide a powerful new strategy to in situ studies of the response of an oligotrophic ecosystem to chemical forcing by atmospheric deposition of African dust. First, we describe how to cope with the large amount of dust aerosol needed to conduct the seeding experiments by producing an analogue from soil collected in a source area and by performing subsequent appropriate physico-chemical treatments in the laboratory, including an eventual processing by simulated cloud water. The comparison of the physico-chemical characteristics of produced dust analogues with the literature confirms that our experimental simulations are representative of dust, aging during atmospheric transport, and subsequent deposition to the Mediterranean. Second, we demonstrate the feasibility in coastal areas to installing, in situ, a series of large (6 x 52 m(3)) mesocosms without perturbing the local ecosystem. The setup, containing no metallic parts and with the least possible induced perturbation during the sampling sequence, provides an approach for working with the required conditions for biogeochemical studies in oligotrophic environments, where nutrient and micronutrients are at nano-or subnano-molar levels. Two, distinct "seeding experiments" were conducted by deploying three mesocosms serving as controls (CONTROLS-Meso = no addition) and three mesocosms seeded with the same amount of Saharan dust (DUST-Meso = 10 gm(-2) of sprayed dust). A large panel of biogeochemical parameters was measured at 0.1 m, at 5 m and 10 m in all of the mesocosms and at a selected site outside the mesocosms before seeding and at regular intervals afterward. Statistical analyses of the results show that data from three mesocosms that received the same treatment are highly reproducible (variability < 30%) and that there is no significant difference between data obtained from CONTROLS-Meso and data obtained outside the mesocosms. This paper demonstrates that the methodology developed in the DUNE project is suitable to quantifying and parameterizing the impact of atmospheric chemical forcing in a low-nutrient, low-chlorophyll (LNLC) ecosystem. Such large mesocosms can be considered as 1-D ecosystems so that the parameterization obtained from these experiments can be integrated into ecosystem models.