Dataset: Sediment-trap particle flux and composition measurements before, during, and after the passage of Hurricane Fabian (2003) and Hurricane Igor (2010) that resuspended large amounts of sediment from the Bermuda Platform

Particle Fluxes
The OFP mooring and sample collection methods are provided in Conte et al. (2001). The OFP mooring uses conical Parflux sediment traps (McLane Research Laboratories, Falmouth MA, USA) having a 0.5 square meters (m2) sampling area. Traps are deployed at 500, 1,500 and 3,200 m depths and continuously collect the sinking particle flux at an approximate biweekly resolution. Trap cups are filled with deep seawater brine (41 parts per thousand (ppt)) poisoned with ultra‐trace metal purity HgCl2 (200 milligrams per liter (mg L-1)) to prevent organic matter degradation. Before deployment, trap cups are filled in a laminar flow hood with a trace metal clean brine (41 ppt), prepared from seawater collected at 3,000 m depth using trace‐metal clean Go‐Flo bottles, poisoned with ultra‐purity mercuric chloride (200 mg L-1) to arrest bacterial activity. Process and deployment blanks are collected during each deployment to assess potential contamination.

Analytical Methods
OFP sample processing: Sample processing protocols are described in Conte et al. (2001, 2003, 2019). Prior to quantitative sample splitting, >1,000 micrometer (μm) -sized material is transferred to a pre‐weighed Petri dish for photog- raphy, removal of swimmers, and dried at 55 degrees Celsius (°C) for mass determination. The remaining, <1000 µm material is split using a McLane rotary splitter (McLane Research Laboratories, Falmouth, MA, USA). Three subsamples are designated for organic analysis and one for trace elemental analysis. The remaining subsamples (60%) are recombined and fractionated into 500–1,000 μm, 125–500 μm, and <125 μm size fractions. For Hurricanes Fabian (25 Aug‐8Sep 2003) and Igor (13-28 Sep, 28 Sep - 12 Oct 2010) samples, the <125 μm fraction was divided into additional size fractions to better characterize the hurricane sediment plumes. The 63-125 μm ("fine sand") and 37-63 μm ("coarse silt") fractions were separated using stainless‐steel sieves. The 4-37 μm ("medium‐fine silt") fraction was concentrated by centrifuging 7 minutes at 1,000 rpm, and the supernatant containing the <4 μm ("clay") fraction was concentrated by centrifuging 10 minutes at 3,000 rpm (modified from Pedrosa‐Pàmies et al., 2013). We note that for these detrital carbonate sediments the standard nomenclature commonly used for these fractions is an operational definition only. The larger size fractions (>125 μm) were quantitatively photographed (described below), dried at 55°C and weighed to the nearest 0.01 mg. <125 μm size fraction was freeze‐dried and weighed. Mass flux was calculated from combined weights of all size fractions.

Carbonate analyses: Carbonate analyses were performed using a Coulometrics model 5011 coulometer (UIC Inc.) equipped with a System 140 module for inorganic carbon determination. Analytical uncertainty is <1.8% based on repeated measurements of flux material working standards. Carbonate δ13C and δ18O were analyzed using a Finnigan MAT252 mass spectrometer following the procedure of Ostermann and Curry (2000). Analytical precision was ±0.04 for δ18O and ±0.05 for δ13C based on the reproducibility of the internal WHOI Atlantis II coral standard. Bulk and isotopic analyses are made on the <125 μm size fraction and converted to total flux by assuming that the total mass composition approximates that of the <125 μm fraction which comprises most of the mass. Analytical precision was ±0.04 for δ18O and ±0.05 for δ13C based on the reproducibility of the internal WHOI Atlantis II coral standard. Bulk and isotopic analyses are made on the <125 μm size fraction and converted to total flux by assuming that the total mass composition approximates that of the <125 μm fraction which comprises most of the mass.

Organic carbon analyses: Particulate organic carbon (POC) and nitrogen (N) concentrations and stable isotopic composition were analyzed using a Europa 20‐20 CF‐IRMS interfaced with the Europa ANCA‐SL elemental analyzer. Before analysis, carbonates were removed by pre‐treatment with 4% sulfurous acid using a modified Verardo et al. (1990) method. Analytical uncertainty is <0.18% based on repeated measurements of flux material working standards.

Elemental analysis: Elemental analyses were made on the total <1,000 μm material using a fusion‐Inductively Coupled Plasma Mass Spectrometry (ICPMS) method developed for multi‐elemental analysis of flux material (Huang et al., 2007). Briefly, the dried sample (4-6 mg) is fused with high purity lithium metaborate (LiBO2) flux at 1,000°C in a dedicated combustion furnace, using a sample to LiBO2 flux ratio of 1:2.5. The fused sample bead was dissolved in 1M HNO3 for ICPMS analysis. Samples were analyzed on a Finnigan Element 2 ICPMS at the Woods Hole Oceanographic ICPMS Facility. Lithogenic concentration was estimated from Si and Al concentrations, assuming that the Al flux was carried mainly by lithogenic particles whose composition approximates that of pelagic clay sediments (25% Si and 8.4% Al, Li and Schoonmaker, 2003): [Lithogenic] = [Al] /0.084. Biogenic Si was estimated by subtracting the lithogenic Si from the total Si and converted to opal assuming an opal opal water content of SiO2·0.4H2O (Mortlock & Froelich, 1989). To assess analytical reproducibility and uncertainty over the analysis period and to allow for data intercalibration, we ran the certified standard PACS-2 (National Research Council of Canada) with each fusion group, and also periodically ran well-characterized working standards of OFP sediment trap material.