1 Diatom frustule cleaning protocol 2 Previously published methods for determining diatom bound 15N values required 3 adaptation due to the high organic carbon concentrations in Guaymas Basin sediments [Robinson 4 et al., 2004; Robinson and Sigman, 2008]. A more universal cleaning method for use with all 5 sediment, culture and water column samples was developed to allow more direct data 6 comparisons. Specific details for cleaning culture and water column diatoms differ slightly from 7 sediments and will be reported elsewhere [Horn et al., 2011].The most significant change 8 incorporated here was the use of potassium permanganate oxidation in place of hydrogen 9 peroxide to reduce variability of 15N values of culture and water column samples [Horn et al., 10 11 2011]. Several adaptations to the persulfate-denitrifier method were tested to determine how best 12 to remove excess organic carbon without affecting the 15N value of nitrogen contained within 13 diatom frustules. We added an initial potassium permanganate oxidation prior to any cleaning 14 methods described by Robsinson et al. [2004] and Robinson and Sigman [2008], replaced the 15 hydrogen peroxide (H2O2) oxidation with a potassium persulfate oxidation, and added additional 16 perchloric oxidations prior to dissolution of the frustules. Described below is a detailed 17 description of the adapted cleaning procedures. 18 An initial oxidation was necessary to prevent particle aggregating during the physical 19 separation of more dense aluminosilicates and metal oxides from biogenic silica. The use of 20 H2O2 and potassium permanganate as oxidizing agents was compared. The permanganate 21 oxidation followed the procedure described by Hasle and Fryxell [1970]. Briefly, 50% 22 concentration H2SO4 was added to the samples and then 3 mL of saturated potassium 23 permanganate was added in 0.5 mL increments. In order to reduce the excess potassium 1 24 permanganate, saturated oxalic acid was added slowly until the solution was clear. The H2O2 25 oxidation involved the addition of 10 mL of cold 10% H2O2 to the dry sample with brief 26 sonication. After 20 minutes at room temperature, 10 mL of 2 M HCl was added to the sediment- 27 H2O2 mixture, and sonication repeated. The samples reacted for an additional 30 minutes before 28 rinsing. After this initial oxidation, the physical separation and reductive cleaning of the opal 29 fraction, as described in Robinson et al. [2004] were used for all samples. In the published 30 method, the oxidation after the reductive cleaning used H2O2. Here, we replaced the H2O2 31 oxidation with a second potassium permanganate oxidation [Hasle and Fryxell, 1970] after 32 which the sediments were further oxidized using perchloric acid at 100˚C [Robinson and Sigman, 33 2008]. The samples that were initially oxidized using potassium permanganate removed more 34 organic nitrogen than the H2O2 oxidation with far less variability, 16.8 ± 0.8 µmol N per gram of 35 opal compared to 26.7 ± 8.0 µmol N per gram of opal (Figure S2). For the diatom bound 15N 36 method to be effective, the 15N value should measure the 15N value of the nitrogen pool 37 contained within the frustule, not nitrogen from organic matter outside of or adhering to the 38 frustule. The lower standard deviation of the 15N values associated with the potassium 39 permanganate oxidation suggests that this method consistently removed similar quantities of 40 external organic matter. 41 Perchloric acid was previously used to oxidize external organic matter from the frustules 42 [Robinson and Sigman, 2008]. Data from cleaning diatoms grown in culture indicate that 43 additional perchloric acid oxidation steps remove more organic matter and improve the precision 44 of the 15N measurement (Horn and Morales, unpublished data). A comparison of results where 45 the samples were subjected to 4, 5, or 6 perchloric acid treatments showed that the precision of 46 the 15N values was the best when 4 perchloric acid treatments were used. The standard 2 47 deviation of 6 samples with 4 perchloric acid oxidations was 0.35‰, 5 oxidations was 0.63‰ 48 and 6 perchloric acid oxidations was 0.57‰ (Figure S3). We interpret the higher standard 49 deviation associated with 5 or more perchloric acid oxidation steps as resulting from possible 50 alteration of the diatom-bound organic N. Each additional cleaning subjects the samples to 51 additional heating thus a higher likelihood for opal dissolution. For future work using 52 sedimentary diatom bound 15N measurements, it is suggested that the number of perchloric acid 53 oxidations be determined for each specific sedimentary environment. 54 55 56 57 58 59 60 61 62 63 64 65 66 67 References Hasle, G. R., and G. A. Fryxell (1970), Diatoms: cleaning and mounting for light and electron microscopy, Transactions of the American Microscopical Society, 89, 469-474. Horn, M., et al. (2011), Nitrogen isotopic relationship between diatom-bound and bulk organic matter of cultured polar diatoms, Paleoceanography, doi:10.1029/2010PA002080, in press. Robinson, R. S., et al. (2004), Revisiting nutrient utilization in the glacial Antarctic: Evidence from a new method for diatom-bound N isotopic analysis, Paleoceanography, 19(3), PA3001, doi:3010.1029/2003PA000996. Robinson, R. S., and D. M. Sigman (2008), Nitrogen isotopic evidence for a poleward decrease in surface nitrate within the ice age Antarctic, Quaternary Science Reviews, 27(9-10), 10761090. 3