Samples were provided from the UK GEOTRACES Titanium sampling rosette and were sampled and filtered into acid-cleaned 1L LDPE bottles by the onboard clean sampling team. water samples were filtered using Millipore 25 mm 0.4 μm polyethersulfone (PES) filters (>75 m depth) or 0.2 μm AcroPak Supor Pall PES capsules (<75 m depth or fish samples). Upon receipt of samples on shore, they were acidified to pH 2 using clean hydrochloric acid and left at this pH for at least 3 months before processing. Every sample was simultaneously extracted and purified for the analysis of d56Fe and [Fe] using published methods (Conway *et al*., 2013). Briefly, Fe, double-spike was added to the acidified samples and allowed to equilibrate for > 1 hour. Metals were then extracted from seawater onto an ethylenediaminetriacetic acid resin (Nobias PA-1; Hitachi) for 2 hours at pH 2. Metals were eluted off this resin using 3 N HNO3, and were then purified by anion exchange chromatography before being reconstituted in 0.1 N HNO3 for analysis.

Concentrations and stable isotope ratios were determined simultaneously by analysis on a Neptune multi-collector ICP-MS in high resolution mode. Every sample was analyzed twice, and the mean isotope ratios are presented here, expressed relative to the international isotope standard IRRM-014. Instrumental and procedural isotopic fractionation was corrected for using a double-spike data reduction scheme.

Concentrations were calculated using isotope-dilution to calculate ng of metal per sample, and then converted to nmol kg-1 or pmol kg-1 using the weight of each sample processed. Concentration numbers represent the mean of two ICP-MS analyses of each sample. The largest sources of error for concentration analyses are determining the weight of the sample and determining the volume of double-spike added to each sample. We approximate 2% error for each sample based on the combination of these two factors. Accuracy of this technique for concentrations were checked by analysis of S, D1 and D2 SAFe reference standards (for which we determined values identical to the consensus values; see Conway et al, 2013) and inter-comparison with two other groups measuring trace metal concentrations on US GEOTRACES GA03 samples (Middag *et al*., 2015).

Uncertainty for isotope ratios in the dataset is expressed as the 2 sigma uncertainty on the average isotope ratio, calculated from a combination of the internal standard error of each of two measurements and the bracketing isotope standards used, as described in Conway *et al*., (2013). We also estimate external precision for these samples (0.05‰), based on replicate analyses of S. Atlantic seawater samples (see Conway, John and Lacan, in review).

**References:**

Conway, T. M., John, S. G. and Lacan, F. (2016). Intercomparison of dissolved iron isotope profiles from reoccupation of three GEOTRACES stations in the Atlantic Ocean. Marine Chemistry. 183. 50-61. doi:10.1016/j.marchem.2016.04.007.

Conway, T.M., Rosenberg, A.D., Adkins, J.F., and S.G. John. (2013). A new method for precise determination of iron, zinc and cadmium stable isotope ratios in seawater by double-spike mass spectrometry. Analytica Chimica Acta, 793: 44-52.

Middag, R., Sefarian, R., Conway, T. M., John, S. G., Bruland, K. W. and de Baar, H. J. W. (2015). GEOTRACES Intercomparison of Dissolved Trace Elements at the Bermuda Atlantic Time Series Station. *Mar. Chem.* 177 (3). 476-479.