Sample processing:
Preserved samples were analyzed for nitrate, nitrite, and nitrous oxide isotopes via mass spectrometry using a Thermo Finnigan DELTA V in continuous flow mode. For nitrate isotopic analysis, filtered and frozen samples were analyzed for NOx concentration with a discrete analyzer system (Westco), which uses a cadmium column to convert NO3- to NO2- and colorimetric methods to quantify the concentration of NO2-. Following nitrate concentration analysis, filtered and frozen samples were prepared for nitrate isotopic analysis using the denitrifier method (Sigman et al., 2001; Casciotti et al., 2002), with updates from McIlvin and Casciotti (2011). Samples with any detectable nitrite ([NO2-] < 0.076 µM) were treated with sulfamic acid to convert the nitrite present to sulfuric acid + N2 gas (Granger and Sigman, 2009), then prepared similarly with the denitrifier method. Three reference materials with a range of δ15N and δ18O values as well as a process blank were prepared alongside each run of samples (USGS 32, USGS 34, and USGS 35; Table S1 from Kelly et al., 2020). USGS reference materials were also treated with sulfamic acid for sulfamic-treated sample runs.

Samples prepared with the denitrifier method were analyzed via a purge-and-trap system coupled to a Thermo Finnigan DELTA V isotope ratio mass spectrometer in continuous flow (McIlvin and Casciotti, 2011). Likewise, samples preserved shipboard with the azide method (McIlvin and Altabet, 2005) were analyzed for the δ15N and δ18O of nitrite on the same IRMS system, following the injection of 100 uL antifoam emulsion (Sigma-Aldrich) into each vial. Samples with nitrate < 0.1 µM or nitrite < 0.25 µM were excluded from the nitrate or nitrite δ15N and δ18O datasets, respectively, due to the large error associated with these low concentrations.

For nitrous oxide isotope analysis, each 160 mL bottle was analyzed on a custom-built purge-and-trap system coupled to a Finnigan DELTA V isotope ratio mass spectrometer (McIlvin and Casciotti, 2010) against a tank of pure N2O calibrated by S. Toyoda (Tokyo Tech). Molecular masses 44, 45, and 46 were measured for sample and reference gases, as well as fragment ion masses 30 and 31. These values, along with a set of coefficients to account for “scrambling” at the ion source, were used to solve for the δ15N2O, δ15N2Oβ, and δ18O-N2O of each sample. See Kelly et al. (2020) for a full description of the scrambling calculation. N2O concentrations were obtained from the N2O peak area, known instrument sensitivity (conversion of mass 44 peak area to nmols N2O), and sample weights pre- and post-analysis (McIlvin and Casciotti, 2010).

Data processing:
Samples prepared via the denitrifier method were corrected for drift and any offset/blank as described in McIlvin and Casciotti (2011). For samples prepared for nitrite isotopic analysis via the azide method, first, a correction was applied to account for instrument drift over the course of the run, using the run numbers and raw, measured δ15N and δ18O of the high-concentration reference materials. Subsequently, a size correction was calculated from two different sizes of standard curves (5 nmol and 10 nmol) and their respective slopes and intercepts; Finally, a size-corrected standard curve was used to calculate the actual isotope ratios from the drift-corrected values.

The isotope ratios of N2O and that of the fragment ion NO — namely, 45/44, 46/44, and 31/30 — were first corrected via comparison to direct injections of a common N2O reference gas. Following this correction, a linearity relation was applied to these isotope ratios, in order to correct for variations in measured isotope ratios due to peak size. To obtain from these three isotope ratios the individual isotopomers of N2O, it was necessary to correct for “scrambling” at the ion source, or the phenomenon by which the NO fragment ion actually contains the outer, beta N atom from the linear N2O molecule, instead of the inner, alpha N atom (Toyoda and Yoshida, 1999). Finally, a two-point scale decompression (Mohn et al., 2014) was applied to correct for a consistent offset between measured and inter-calibrated standard values. See Kelly et al. (2020) for a full discussion of the N2O isotopomer data corrections.

Problem Report:
Issues were occasionally encountered during analysis of nitrous oxide stable isotopes (clogged lines, etc.). Samples for which issues occurred are flagged in the data accordingly: no flag=good, 3=questionable, 4=bad.

BCO-DMO processing:
- Timestamp data from all cruises were joined to isotopomer dataset
- Timestamp converted to ISO DateTime UTC format in additional column
- Parameter names adjusted to comply with database requirements
- Added a conventional header with dataset name, PI name, version date
- Units added to parameter description metadata section
- Missing data identifier of 'nd' (no data) used