Particulate organic matter and ballast fluxes measured using Time-Series and Settling Velocity sediment traps in the northwestern Mediterranean Sea lead to new ideas on particle export Cindy Lee 1, Michael L. Peterson 2, Stuart G. Wakeham 3, Robert A. Armstrong 1, J. Kirk Cochran 1, Juan Carlos Miquel 4, Scott W. Fowler 1,4, David Hirschberg 1, Aaron Beck 1, and Jianhong Xue 1 1 Stony Brook University, Stony Brook, NY 2 University of Washington, Seattle WA 3 Skidaway Institute of Oceanography, Savannah GA 4 International Atomic Energy Agency, Monaco Motivation and Introduction Observations made during the Joint Global Ocean Flux Study (JGOFS) indicate that 50-80% of the vertical flux of carbon through the mesopelagic zone and into the deep ocean occurs by gravitational sinking of particles. This redistribution determines the depth profile of dissolved CO 2, including its concentration in the surface mixed layer, and hence the rate at which the ocean can absorb CO 2 from the atmosphere and sequester it in the deep ocean. The depth-scale of OM remineralization also determines the depth profile of nutrient regeneration, which in turn determines the time scale for return of mineral nutrients to the photic zone. A quantitative and mechanistic understanding of water column POC remineralization is critical to predicting the response of the global carbon cycle to environmental change. Collection of sinking material in a manner that enables identification of underlying mechanistic controls on settling and decomposition should lead to better estimates of vertical fluxes and remineralization depth scales. Prompted by recent data analyses suggesting that the flux of particulate organic carbon sinking into deep waters is determined by fluxes of mineral ballasts, we undertook a study of the relationships among organic matter (OM), calcium carbonate, opal, lithogenic material, and excess aluminum fluxes at a site in the NW Mediterranean as part of the MedFlux project. We measured fluxes of particulate components during Spring and Summer of 2003 and 2005 using a swimmer-excluding sediment trap design capable of measuring fluxes both in a time-series (TS) mode and in a configuration for obtaining particle settling velocity (SV) profiles. Conclusions The distinct OM-ballast associations observed in particles sinking at a depth of ~200 m imply that the ratio of organic matter to ballast is set in the upper water column and the importance of different ballast types follows the seasonal succession of phytoplankton. There was no relation between mass flux and satellite Chl values. Dust appears to increase particle flux through its role in aggregation rather than in nutrient input that enhances productivity. Particles must be at least half organic matter before their settling velocity is affected by ballast concentration. This lack of change in ballast composition with SV in particles with <40% OM content suggests that particle SV reaches a maximum because of the increasing importance of inertial drag. Relative amounts of OM and opal decrease with depth due to decomposition and dissolution; carbonates and lithogenic material contribute about the same amount to total mass, or increase slightly, throughout the water column. The high proportion of excess Al cannot be explained by its incorporation into diatom opal or reverse weathering, so Al is most likely adsorbed to particulate oxides The MedFlux project was funded by the U.S. National Science Foundation with additional support from IAEA, Monaco and CNRS, France. Mass flux (A), POC flux (B), Satellite Chl and sea surface temperature (C), and percent OC (D) in 2003 TS sediment traps. Vertical bands show dust inputs, with darker bands being heavier inputs. Surface fluxes of bulk particulate matter (mass) and OC were highest (up to ~1000 mg m ‑ 2 d ‑ 1 for mass and 70 mg m ‑ 2 d ‑ 1 for OC) during the first two weeks of the first deployment period (P1) and decreased by about an order of magnitude thereafter. There was no obvious correlation between satellite Chl and flux. Surface mass flux continued to decrease further until the end of deployment 2 on June 30. At 771 m, mass and OC fluxes at the start of 2003 P1 were about 4-fold lower than at 238 m, but quickly increased till fluxes at 771 m were essentially the same as at 238 m. OC fluxes peaked at slightly different times at 238 and 771 m. After the initial two weeks of P1, mass fluxes at 238 and 771 m were generally similar. OC fluxes were slightly lower at 771 m than at 238 m. During P2, mass fluxes at 117 m were very low, and mass fluxes at 1918 m were often higher. OC fluxes at the two depths showed a pattern generally similar to mass flux. Mass flux (A), POC flux (B), satellite Chl and sea surface temperature (C), and percent OC (D) in 2005 TS sediment traps. Vertical bands show dust inputs, with darker bands being heavier inputs. The high point shown in Chl is a single point. Again, there was no obvious correlation between satellite Chl and flux. Mass fluxes at 313 and 924 m in 2005 were almost as high as surface fluxes in 2003, up to ~900 mg m ‑ 2 d ‑ 1 mass. However, OC fluxes never reached the higher 2003 levels, peaking at 25 mg m ‑ 2 d ‑ 1. Mass fluxes at 313 m were higher than at 924 m, but not greatly so, suggesting again that material from the surface phytoplankton bloom quickly reached 924 m, and that little particulate material was lost between these depths. However, OC fluxes at 313 m were higher on average than 924 m fluxes, again suggesting some loss of OC with depth. Contribution of measured organic matter, calcium carbonate, opal, and lithogenic material in 2005 TS sediment traps at 313 (A) and 924 m (B). In the near-surface trap, CaCO 3, opal, and lithogenic material were almost equal (~30%) in importance at the beginning of the deployment period. In the deeper trap, opal was much less important and lithogenic material was the dominant component. The lithogenic component decreased with time. Contribution of measured organic matter, calcium carbonate, opal, and lithogenic material in 2005 SV sediment traps at 313 (A), 524 (B and C) and 1918 m (D and E). Duplicates are shown at 524 and 1918 m. Blank spaces occur where not all four components were measured. There was no consistent pattern in composition vs settlling velocity in these samples. In 2003 samples with %OM>40%, composition did change with SV, becoming more OM-rich at lower SV. Relationship between lithogenic Al concentration (mg g -1 ) and mass flux (m ‑ 2 d ‑ 1 ) for 2003 P1 and P2 samples (m=0.021; r 2 =0.82). There was only a poor correlation between mass flux and biogenic Si (r 2 =0.05), lithogenic Si (r 2 =0.1), or Ca (r 2 =0.23) concentration. Explained variance between total [Al] and mass flux was Relation between the proportion of each ballast and percent organic matter. Explained variances (r 2 ) for the proportion of calcium carbonate ( ), opal (  ), lithogenic material ( ), and Al xs hydroxide (  ) in total ballast versus %OM are 0.31, 0.004, 0.50, and 0.36, respectively. Data are for all 2003 and 2005 samples from near-surface traps (~200m).