Sample Collection. Samples were collected on the Antarctic continental shelf and slope west of the Antarctic Peninsula within the PAL-LTER sampling domain (http://pal.lternet.edu/) during summer (cruise dates 30 Dec 2017 through 12 Feb 2018; sampling dates 5 Jan to 4 Feb 2018) from the ARSV Laurence M Gould (LMG 1801, PAL-LTER cruise 26, DOI: 10.7284/907858). Sampling focused on three or 4 depths at each station chosen to represent the Antarctic Surface Water (ASW, 0 -34 m depth), the Winter Water (WW, the water column temperature minimum, generally between 35 and 174 m) the Circumpolar Deep Water (CDW, 175-1000 m) and slope water (SLOPE, >1000 m, generally ~10 m above the bottom at deep stations on the slope, 2500-3048m). Water samples were collected from Niskin bottles (General Oceanics Inc., Miami, FL, USA) into opaque 2 L HDPE plastic bottles or into aged, acid-washed, sample-rinsed 250 ml polycarbonate bottles (Nalge) completely filled (~270 mL) directly from Niskin bottles as soon as possible after the rosette was secured on deck. Subsequent processing took place in an adjacent laboratory.

Samples for DNA analysis were taken from the 2 L opaque HDPE bottles and were filtered under pressure through 0.22 um pore size Sterivex GVWP filters (EMD Millipore, Billerica, MA, USA) using a peristaltic pump. Residual seawater was expelled from the filter using a syringe filled with air, then ~1.8 ml of lysis buffer (0.75 M sucrose, 40 mM EDTA, 50 mM Tris, pH 8.3) was added to the filter capsule, which was capped and placed in a -20 °C freezer. The frozen samples were aggregated into Ziploc Freezer Bags and transferred to a -80 °C freezer for the remainder of the cruise and for shipping to the laboratory.

Two samples of the Sterivex filtrate (40 mL each into new 50 mL disposable centrifuge tubes, VWR, rinsed 3x with sample) were frozen immediately at -20 °C, then aggregated into Ziploc Freezer Bags and transferred to a -80 ° freezer for the remainder of the cruise and for shipping to the laboratory. These were used for subsequent determination of 1) urea concentration and 2) the natural abundance of ¹⁵N in the nitrite plus nitrate pools (¹⁵NOₓ hereinafter). An additional sample of the Sterivex filtrate was stored in a polycarbonate bottle at 4 °C for subsequent onboard determination of ammonia concentration by the Holmes et al (1999) o-phthaldialdehyde method and nitrite concentration by the diazo-coupling method (Strickland and Parsons 1972). Technical difficulties encountered during onboard analysis resulted in the loss of ammonium and nitrite data for some samples.

Samples for DNA and chemical analyses were shipped on dry ice from Punta Arenas, Chile to the Hollibaugh laboratory at the University of Georgia. Upon arrival they were stored in a -80 °C freezer until analyzed. Samples for ¹⁵N analysis were shipped on dry ice from Punta Arenas, Chile to the Popp laboratory at the University of Hawaii. Upon arrival they were stored in a -40 °C freezer until analyzed.

Nitrogen oxidation rates. Oxidation rates of N supplied as ammonium, nitrite, urea and putrescine (1,4 diaminobutane) were measured in ~48 h incubations using ¹⁵N-labeled substrates (>98 at% ¹⁵N, Cambridge Isotope Laboratories, Tewksbury,MA, USA) added within ~1 hr of sample collection to yield ~44 nM amendments (Santoro et al., 2010; Beman et al., 2012). Labeled substrates were added to duplicate bottles that were placed in cardboard boxes and incubated in the dark in a Percival incubator (Perry, IA, USA). Incubation temperature was recorded at 5-minute time steps with HOBO TidBit data loggers (Onset Computer Corp., Bourne MA, Figure 1) placed in bottles of filtered seawater incubated in cardboard boxes identical to those used for experiments (see the "Incubator_Temperature.xlsx" Supplemental File). Incubations were terminated after ~48 hr by decanting 40 mL subsamples from each bottle into new, sample rinsed, 50 mL polypropylene centrifuge tubes that were immediately frozen at -80 °C. Water in these tubes was used for subsequent analysis of ¹⁵NOₓ. The natural abundance of ¹⁵N in NOₓ was taken as the initial (time = 0) value for calculating the amount of ¹⁵N oxidized to nitrate or nitrite during the incubations.

Chemical analyses. Urea content was determined by the diacetyl monoxime method (Rahmatullah and Boyde 1980, Mulvenna and Savidge 1992). Subsamples from samples that were also used to determine oxidation of ¹⁵N supplied as putrescine were sent to Dr. X. Mou’s laboratory at Kent State University where they were analyzed to determine polyamine and DFAA content as described previously (Lu et al 2014).

¹⁵N in nitrite plus nitrate. The ¹⁵NOₓ in samples was measured using the ‘denitrifier method’ (Sigman et al., 2001) with Pseudomonas aureofaciens as described in Popp et al. (1995), Dore et al. (1998) and Beman et al. (2011). The nitrous oxide produced was analyzed using a Gas Bench II coupled to a MAT 252 mass spectrometer following the recommendations of Casciotti et al. (2002). Typically nineteen samples plus one sample duplicate was analyzed along with duplicate reference materials USGS 32, USGS 34 and USGS 35 (or NIST 3), which were used to normalize the measured d¹⁵N values to AIR. In addition, a laboratory reference solution made from analytical grade NaNO₃ with d¹⁵N value (-52.2‰) that was known through extensive characterization using NIST/USGS reference materials was also analyzed in duplicate with each batch of 19 samples.

We calculated oxidation rates from d¹⁵N enrichment of the NOₓ pool in the bottles at the ends of the incubations compared to the initial value in the unamended seawater sample ("natural abundance"). We assumed that the d¹⁵N of naturally occurring ammonia, urea and putrescine is the same as that of N in bulk organic matter, and that the d¹⁵N value of nitrite is our samples is -30 o/oo as reported by Smart et al. (2015). Samples with low or no activity sometimes yielded negative rates because the d¹⁵NOₓ "natural abundance" value for that sample was greater than the d¹⁵NOₓ value of amended sample. We analyzed control samples consisting of filtered seawater taken at the beginning of the cruise, amended with ¹⁵NO₂, then immediately frozen at -80 °C, or "time zero" samples from time course experiments performed throughout the cruise, to determine the contribution of autooxidation or isotope exchange to the apparent rate of nitrite oxidation. These samples were treated with sulfamic acid to remove unreacted ¹⁵NO₂ (Granger and Sigman 2009). The analysis indicated that about 7.9% of the ¹⁵N supplied as NO₂ had been converted to ¹⁵NO₃ by the time we analyzed the samples. We also performed an independent chemical analysis of nitrite and nitrate (Strickland and Parsons 1972) in the (nominally) 0.125 mM ¹⁵NO₂ working stock solution a few months after the cruise. This analysis indicated that about 14.3% of the of the nitrite plus nitrate in the stock was nitrate. Because this stock had been handled and shipped separately from the cruise samples, we used 7.9% as the best estimate of the amount of ¹⁵N label converted to nitrate. We have incorporated corrections for this reaction into rates calculated from field data.

We ran time-course incubations with samples from 2 or 3 depths at 34 stations to verify that oxidation rates did not change significantly during incubations, for example, due to substrate depletion or changes in the population of ammonia oxidizers. These experiments were set up in 250 ml polycarbonate bottles as above. Two bottles were sampled at each time point over time courses of 72 to 96 h. These data are presented in Table 1 (PDF; see Supplemental Files). AO rates determined from the slope of linear regressions of the data from a given sample were compared to rates determined from samples taken at the 48 hr time point. We performed analogous experiments to examine the effect on N oxidation rates of variation in incubation temperature and in substrate concentration.

Summary Statistics for d15NOx (mean, median, maximum, minimum, and standard error of d15NOx measurements grouped by the water mass depth ranges) are provided in the Supplemental File d15NOx_Summary_Statistics.JPG.

Precision. Analytical uncertainty in d¹⁵N values was determined from duplicate analyses of USGS reference materials, our laboratory reference solution and samples analyzed in duplicate and ranged from 0.36‰ to 0.56‰ (Table 1 (PDF) Supplemental File). All rate measurements were also performed in duplicate (biological replicates) and their uncertainty is also presented in Table 1. Accuracy was determined based on isotope analysis of the laboratory reference solution, which was not used to normalize the isotopic results of samples and was found to be 0.42‰ (at% ¹⁵N = 0.00019, n = 56).

Rate calculations. We integrated the data we collected to calculate oxidation rates of N supplied as ammonia and urea as described in Popp et al. (1995), Dore et al. (1998) and Beman et al. (2011). We used ammonium concentration data from shipboard analyses. Nitrite + nitrate concentrations were determined by PAL-LTER personnel and were obtained from their database. Urea concentrations were measured on samples shipped frozen to the University of Georgia (Hollibaugh lab). Chemical data needed for rate calculations were not available for some samples so we substituted water mass averages determined from other samples taken on the cruise.

We determined the limits of detection and precision of nutrient analyses as follows. The precision of nitrate plus nitrite analyses run by LTER personnel (https://oceaninformatics.ucsd.edu/datazoo/catalogs/pallter/datasets/27) was reported to be 100 nM. The precision and limit of detection of putrescine (polyamine) analysis is given in Lu et al. (2018) as 1 nM. We determined the precision of ammonium, urea and nitrite analyses as the mean standard deviation of replicate (2 or 3) analyses of a given sample. They are: ammonium, 65 nM; urea, 10 nM; and nitrite, 70 nM. The limits of detection were taken as 1.96 times the precision of the relevant measurements.

We ran Monte Carlo simulations to estimate the precision and the limits of detection of rate measurements. The models incorporated the estimates of precision given above and the means of the measured in situ concentrations of the reactants, the mean in situ concentration of NOₓ (from PAL-LTER data), the precision of the measured d¹⁵NOₓ in the experiments at the beginning (natural abundance) and end of the incubations. We ran 10,000 trials using random numbers generated with population means and standard deviations (assuming normally distributed variance and produced using an Excel spreadsheet in the EasyFit^® app) ) equivalent to the test values. The standard deviation of the 10,000 rates calculated in this manner was taken as an overall estimate of the precision of the rates we report. The results of these models are summarized in Supplemental Table 3. Our estimates of the precision of the rate measurements (oxidation of ¹⁵N supplied as ammonium urea, putrescine or nitrite to ¹⁵NOₓ, nmol L^-1 d^-1) are: ammonia, 2.18; urea 0.31; putrescine, 0.51; nitrite, 4.6, for relative standard deviations (RSD; ((standard deviation/mean)*100)) of: 15.3%; 11.3%; 8.2% and 32%, respectively, of the calculated rates. Model runs, which also tested the sensitivity of the calculated rates to the accuracy of estimates of some variables, are summarized in Supplemental Table 3 (.xlsx file), which also includes the equations used to calculate rates.