Measurements of nitrite production in and around the primary nitrite maximum in the central California Current

Nitrite (NO−2 ) is a substrate for both oxidative and reductive microbial metabolism. NO −2 accumulates at the base of the euphotic zone in oxygenated, stratified openocean water columns, forming a feature known as the primary nitrite maximum (PNM). Potential pathways of NO −2 production include the oxidation of ammonia (NH 3) by ammonia-oxidizing bacteria and archaea as well as assimilatory nitrate (NO−3 ) reduction by phytoplankton and heterotrophic bacteria. Measurements of NH 3 oxidation and NO−3 reduction to NO − 2 were conducted at two stations in the central California Current in the eastern North Pacific to determine the relative contributions of these processes to NO −2 production in the PNM. Sensitive ( < 10 nmol L−1), precise measurements of [NH +4 ] and [NO − 2 ] indicated a persistent NH+4 maximum overlying the PNM at every station, with concentrations as high as 1.5 μmol L −1. Within and just below the PNM, NH3 oxidation was the dominant NO − 2 producing process, with rates of NH 3 oxidation to NO − 2 of up to 31 nmol L−1 d−1, coinciding with high abundances of ammonia-oxidizing archaea. Though little NO −2 production from NO−3 was detected, potentially nitrate-reducing phytoplankton (photosynthetic picoeukaryotes, Synechococcus , andProchlorococcus ) were present at the depth of the PNM. Rates of NO−2 production from NO − 3 were highest within the upper mixed layer (4.6 nmol L −1 d−1) but were either below detection limits or 10 times lower than NH 3 oxidation rates around the PNM. One-dimensional modeling of water column NO−2 production agreed with production determined from 15N bottle incubations within the PNM, but a modeled net biological sink for NO−2 just below the PNM was not captured in the incubations. Residence time estimates of NO−2 within the PNM ranged from 18 to 470 days at the mesotrophic station and was 40 days at the oligotrophic station. Our results suggest the PNM is a dynamic, rather than relict, feature with a source term dominated by ammonia oxidation.

Potential pathways of NO − 2 production in aerobic waters include the oxidation of ammonia (NH 3 ) by ammoniaoxidizing bacteria (AOB) and archaea (AOA) (Ward, 2002) as well as assimilatory NO − 3 reduction by phytoplankton (Kiefer et al., 1976) and heterotrophic bacteria (Allen et al., 2001). The relative roles of these two pathways in the formation of the PNM have been extensively debated (Brandhorst, 1959;Vaccaro and Ryther, 1960). Proposed mechanisms for the maintenance of the PNM usually invoke organismal responses to sunlight -either insufficient light energy for complete phytoplankton nitrogen assimilation (Lomas and Lipschultz, 2006) or light inhibition of nitrite oxidation (Olson, 1981b). Others have suggested that it is actually the availability of sufficient light and NO − 3 that allows PNM formation (Karl et al., 2008). Removal pathways for NO − 2 in aerobic environments include oxidation by nitrite-oxidizing bacteria and phytoplankton uptake. The unexplained accumulation of NO − 2 leaves fundamental questions unanswered about what limits primary productivity, nitrogen remineralization, or both, in the lower euphotic zone.
Direct measurements of NO − 3 assimilation into phytoplankton (Dugdale and Goering, 1967), nitrate reductase enzyme activity (Eppley et al., 1969), and more recent nucleic acid-based methods for detecting nitrate reductase-encoding genes (Jenkins et al., 2006;Paerl et al., 2008;Ward, 2008) and gene transcripts (Paerl et al., 2012) have all shown the importance and prevalence of nitrate-reducing activity in marine phytoplankton and the potential for NO − 2 production. None of these methods, however, address how much NO − 2 is released into the water column, where it is detected in the dissolved phase, versus how much remains in the cell. High-sensitivity, high-precision techniques now allow access to stable isotope ratios of both nitrogen and oxygen in NO − 2 (δ 15 N NO 2 and δ 18 O NO 2 ) (McIlvin and Altabet, 2005;. These new methods enable the use of natural abundance stable isotopes to trace the mechanisms of NO − 2 production and consumption pathways, although uncertainties remain Casciotti, 2010, 2013;Casciotti et al., 2010;Buchwald et al., 2012).
It has been recognized for some time that AOB are present at the depth of the PNM (Olson, 1981a;Ward et al., 1982). More recently, marine AOA abundance and AOA : AOB ratios have been correlated with [NO − 2 ] (Murray et al., 1999;Beman et al., 2010). Archaeal genes for ammonia oxidation (ammonia monooxygenase subunit A, amoA) have also been found in and around the PNM (Beman et al., 2010;Santoro et al., 2010;Newell et al., 2011). The photosynthetic community present near the PNM, at approximately the 1 % light depth, is composed predominantly of low-light Prochlorococcus ecotypes (Rocap et al., 2003) and picoeukaryotes. Both of these groups have been recently implicated in active NO − 3 utilization and thus are a potential source of NO − 2 in the PNM. Metagenomic data (Martiny et al., 2009), flow cytometry coupled with stable isotope tracers (Casey et al., 2007), and natural abundance stable isotope measurements (Fawcett et al., 2011) suggest that some Prochlorococcus ecotypes are capable of using both NO 3 and NO 2 . Picoeukaryotes also appear to actively assimilate nitrate (Fawcett et al., 2011) and can grow at rapid rates in low-nitrogen open-ocean environments (Cuvelier et al., 2010).
The purpose of this study was to determine the pathways of NO − 2 formation in the PNM of the eastern North Pacific Ocean by coupling high-precision, low-level measurements of NO − 2 and ammonium (NH + 4 ) concentrations with 15 N tracer experiments and enumeration of microbial groups. Ammonia oxidation (production of NO − 2 and NO − 2 + NO − 3 from NH 3 ) and NO − 3 reduction to NO − 2 were measured, as was abundance of potentially NO − 2 -producing organisms in and around the PNM (photosynthetic eukaryotes, Synechococcus, and Prochlorococcus, AOB, and AOA). These measurements allow a direct comparison of NO − 2 production from NO − 3 and from NH + 4 in and around the PNM in concert with quantification of relevant organisms, and precise NH + 4 and NO − 2 concentration measurements across a gradient in primary productivity.

Study site and sample collection
Samples were collected 1-12 October 2009 during cruise WFAD09 aboard the R/V Western Flyer. The cruise track was an ∼ 800 km transect extending from Moss Landing, California, USA, along California Cooperative Oceanic Fisheries Investigations (CalCOFI) Line 67. The transect extended across the central California Current (Collins et al., 2003) to station 67.155 (33.27 • N, 129.43 • W). Fifteen stations were sampled for nutrients, chlorophyll a, and flow cytometry, while two stations (67.70,36.13 • N,123.49 • W;and 67.155,Fig. 1a) were chosen for further microbial characterization and geochemical rate measurements.
Water samples were collected at discrete depths between 0 and 1000 m using a 12-bottle rosette sampler. The sampler was equipped with an SBE 9 (Sea-Bird Electronics) to measure conductivity, temperature, and pressure as well as an in situ ultraviolet spectrophotometer (ISUS; Johnson and Coletti, 2002) for real-time NO − 3 concentration profiling. Samples for standard nutrient analyses were collected in 27 mL HDPE scintillation vials and stored frozen until analysis (Sakamoto et al., 1990). Samples for onboard lowlevel nutrient analyses (described below) were collected in 60 or 125 mL polyethylene bottles that had been initially cleaned in Micro-90 cleaning solution (Andwin Scientific) and subsequently acid-washed in 1.2 N HCl. Phytoplankton samples for shore-based flow cytometry were collected from the rosette in 60 mL amber HDPE bottles, then transferred to triplicate 1 mL cryovials (Nunc) and preserved with 0.25 % (final concentration) TEM grade glutaraldehyde (Tousimis) and fixed in the dark at room temperature for 20 min before flash-freezing in liquid nitrogen. Samples for nucleic acid extraction were collected from the rosette in polycarbonate bottles. Cells were harvested by pressure filtration of 1-2 L of seawater onto 25 mm-diameter, 0.2 µm pore-size polyethersulfone membrane filters (Pall Supor 200) housed in Swinnex filter holders (Millipore) using a peristaltic pump (Cole Parmer Masterflex) and silicone tubing. Filters were flashfrozen in 2 mL gasketed bead-beating tubes (Fisher Scientific) using liquid nitrogen.

Low-level nutrient analyses
Measurements of low concentrations (< 1 µM) of NH + 4 , NO − 2 , and NO − 3 were made onboard the ship as soon as pos-sible after collection. Higher-resolution (12 depths in the upper 200 m) nutrient measurements were made on the rosette cast immediately preceding the cast for incubation water collection. Nutrient measurements were also made at the incubation depths for rate determinations (see below). The NH + 4 analyzer was based on the method described in Plant et al. (2009) but used a sequential injection platform. [NO − 2 ] and [NO − 3 ] were measured using standard colorimetric methods coupled with liquid waveguide capillary cells (LWCC) for detection (Zhang, 2000;Patey et al., 2008). The LWCC is composed of quartz capillary tubing (550 µm inner diameter) covered by an outer surface cladding of AF-2400 Teflon resulting in a refractive index of 1.29 (World Precision Instruments). This refractive index is smaller than seawater (1.34) so that light is internally reflected through the water-filled waveguide to achieve path lengths of 1 m (for [NO − 3 ] determination) and 2 m (for [NO − 2 ] determination). The chemistry (Cd reduction, followed by azo dye formation) and sample processing manifold used for the low-level NO − 3 and NO − 2 analyses were based on the rapid flow an-alyzer described in Sakamoto et al. (1990). Based on 3x the standard deviation of low nutrient blanks, the low-level NH + 4 , NO − 2 , and NO − 3 measurements had detection limits of 10 nmol L −1 , 2 nmol L −1 , and 4 nmol L −1 , respectively. Correlations among low-level nutrient concentrations and between nutrients and organism abundance were evaluated using Spearman non-parametric rank correlations implemented in MATLAB R2011b (Mathworks).

Chlorophyll a determination and flow cytometry
Chl a was measured by fluorescence determination on a Turner 10-005R fluorometer. Samples were vacuum-filtered onto 25 mm glass fiber filters (GF/F) and extracted for 24 h in 90 % acetone (Venrick and Hayward, 1984).
Fixed cell samples were analyzed on a flow cytometer equipped with a 200 mW 488 nM laser (InFlux, Becton Dickson). Autoclaved, 0.1 µm filtered 1x phosphate-buffered saline was used as sheath fluid. Fixed samples were thawed in a water bath in the dark, and fluorescent polystyrene beads (0.75 µm, Polysciences, Inc.) added, immediately prior to each run. Each sample was run for 2 min prior to data collection, and data were subsequently collected for 10 min at a flow rate of approximately 25 µL min −1 , measured by an inline flow meter (Sensirion SLG-1430 run with software designed by J. Swalwell, University of Washington). Sheath and sample pressure were adjusted as needed to maintain constant flow rate, to approximately 18.5 and 19.2 psi, respectively. Forward-angle light scatter (FALS), pulse width, side scatter (90 • angle; SSC), red (692 ± 40 nm band-pass filter) and orange (527 ± 27 nm band-pass filter) autofluorescence were recorded. Listmode files were analyzed in Winlist 6.0 (Verity Software House) to enumerate Prochlorococcus, Synechococcus, and small eukaryote populations that were defined based on natural autofluorescence and FALS characteristics.
While fixed samples were used for most analyses herein, unfixed samples were used to analyze oligotrophic surface Prochlorococcus populations (while at sea) that are better resolved when run live, and to verify that eukaryotic populations remained intact (numerically correct) with fixation and cryogenic storage. Live samples were run similarly to fixed samples, with slight modifications to accommodate the effect of the ship's motion. Specifically, because the ship's motion caused considerable movement of sheath fluid, the flow rate ranged from 12 to 30 µL min −1 in a time period of seconds. To minimize this effect, the instrument was run with at least 5 L of sheath fluid at all times and total event rates only recorded when the flow rate was 25 µL min −1 (the average rate for all runs). Flow rate and cell count comparisons of samples run live and fixed showed that sheath pressure variability did not lower accuracy (data not shown).

Ammonia oxidation and nitrate reduction rate incubations
Rate measurements were conducted using stable isotope tracer additions ( 15 N) for five depths at stations 67.70 and 67.155 targeting the middle of the euphotic zone, just above the PNM, at the PNM, just below the PNM, and a depth well below the photic zone (500 m). For each depth, six 500 mL bottle incubations were conducted: two ammonia oxidation rate bottles (with 15 NH + 4 added), two NO − 3 reduction rate bottles (with 15 NO − 3 added), and two filtered control bottles (one with 15 NH + 4 and one with 15 NO − 3 ). Water for the experimental bottles was collected directly from the rosette into 500 mL acid-cleaned polycarbonate bottles. Water for filtered controls was then 0.2 µm-pressure filtered into fresh acid-clean bottles. Each bottle was spiked with 100 µL of 1 mmol L −1 15 N-labeled substrate ( 15 NH + 4 or 15 NO − 3 , 99.5 atom percent 15 N; equivalent to a final label concentration of 200 nmol L −1 15 N; Cambridge Isotope Laboratories). 15 NO − 3 addition treatments from the 500 m depth at each station were instead spiked with 200 µL of 15 NO − 3 to achieve a final label concentration of 400 nM. These 15 N additions corresponded to 0.5-13x ambient NH + 4 and 0.02-0.23x ambient NO − 3 (see below). Bottles were incubated in an on-deck circulating seawater incubator in neutral density-screened bags calibrated to approximate the in situ light field at each depth using a photosynthetically active radiation sensor (Biospherical Instruments QSL-2200). Replicate 50 mL samples were removed from each bottle at 4 time points (0, 12, 24, and 36 h) for onboard determination of [NH + 4 ], [NO − 2 ], and [NO − 2 + NO − 3 ] as described above (unfiltered), shore-based determinations of [NO − 2 + NO − 3 ] and [PO 3− 4 ] (unfiltered, frozen), and δ 15 N NO 2 and δ 15 N NO x (0.2 µm syringe-filtered, frozen), as described below.

Stable isotope analyses
Analyses of δ 15 N of NO − 2 and NO x (NO − 2 + NO − 3 ) in time course samples from the incubations were performed with a Thermo-Finnigan Delta PLUS XP isotope ratio mass spectrometer outfitted with a custom purge-and-trap system. Samples were prepared from 5 nmol or 10 nmol of analyte using the azide (McIlvin and Altabet, 2005) and denitrifier methods (Sigman et al., 2001;McIlvin and Casciotti, 2011) for δ 15 N NO 2 and δ 15 N NO x determination, respectively. δ 15 N NO 2 values were calibrated against nitrite isotope reference materials N-23, N-7373, and N-10219  analyzed in parallel. δ 15 N NO x values were calibrated against NO − 3 isotope reference materials USGS 32, USGS 34, and USGS 35, analyzed in parallel. Duplicate δ 15 N NO 2 analyses were performed on each sample, while δ 15 N NO x measurements were performed once. Due to the necessary addition of carrier NO − 2 (see below), the δ 15 N NO 2 analyses had mean standard deviations (for repli- 2 ] in all incubation samples was too low (< 1 µmol L −1 ) to allow direct δ 15 N NO 2 measurements. Therefore, either 5 or 10 nmol of NO − 2 with known natural abundance isotopic composition was added (either reference material N-23, δ 15 N NO 2 = −3.7 ‰, or a laboratory NO − 2 stock, δ 15 N NO 2 = 1.1 ‰) to 5 mL of each incubation sample prior to preparation and analysis. To prevent potential 15 NO − 3 contamination of the laboratory and the mass spectrometer purge-and-trap system, 1 µmol of natural abundance NO − 3 was also added to each incubation sample that had been incubated with 15 NO − 3 prior to δ 15 N NO 2 analysis. There was no interference of the NO − 3 addition on δ 15 N NO 2 determination of reference materials, as demonstrated previously . The isotope ratio of the original sample was then calculated by subtracting the concentration-weighted isotope ratio of the standard addition.

Rate calculations
To calculate rates of NH 3 oxidation and NO − 3 reduction, we modeled the 15 N and 14 N contents of the receiving "product" pool (Table 1) as a box, with inputs from the labeled "substrate" pool and outputs through NO x or NO − 2 removal using Eq. (1): is the 15 N / 14 N ratio in the product pool at any time t, 15 N 0 and 14 N 0 are the initial 15 N and 14 N concentrations in the product pool, at in is the initial atom fraction of 15 N in the substrate pool, and k is the removal rate constant for NO − 2 or NO − x removal (with units h −1 ). For example, to calculate NO − 3 reduction rates, where NO − 3 is the substrate and NO − 2 is the product, the left-hand side of the equation represents the time course change in 15 N NO 2 / 14 N NO 2 in incubations with added 15 NO − 3 and at in refers to the initial atom fraction 15 NO − 3 . This approach has been described in detail previously (Santoro et al., 2010). at in was calculated by mass balance from the ambient nutrient concentrations and the 15 N tracer addition. Initial atom fraction 15 N in the NH + 4 additions ranged from 0.35 to 0.93. Initial atom fraction 15 N in the NO − 3 additions ranged from 0.02 to 0.19. Coefficients F in (the rate of NO − 2 or NO − x production, with units µmol L −1 h −1 ) and k were calculated using a non-linear least-squares curve-fitting routine, implemented in MATLAB R2011b with the Optimization Toolbox; results were then converted to units of nmol L −1 d −1 . Standard error in the fit coefficients was calculated by approximating the covariance matrix, and using the square root of the diagonal to calculate the standard error.
Our model does not prescribe a specific biogeochemical process for the NO − 2 or NO − x removal term; this removal process may be nitrite oxidation or assimilation. Fractionation factors (α) for NO − 2 or NO − x removal were approximated using values for NO − 2 and NO − x assimilation, 1.001 and 1.005, respectively. At the levels of 15 N enrichment used here, the model is not sensitive to changes in this parameter (and therefore the uptake mechanism) over a range of reasonable values (0.995-1.010, data not shown). Removal rates (nmol L −1 d −1 ) were calculated as k (converted to units of d −1 ) multiplied by the initial [NO − 2 ]. The detection limit for each rate measurement is dependent on the initial atom fraction 15 N enrichment in the substrate pool and the concentration of the product pool. High initial atom fraction enrichment in the substrate pool and low concentrations in the product pool will result in greater sensitivity. The theoretical detection limit, a rate which we can be reasonably certain is significantly different than zero, was calculated individually for each rate and depth as the NO − 2 production rate necessary to cause a 2 ‰ increase in δ 15 N NO 2 from the initial value (including the carrier), i.e., twice the precision of the measurement. Detection limits for ammonia oxidation at the mesotrophic station (67.70) ranged from 0.3 nmol L −1 d −1 to 1.7 nmol L −1 d −1 at 33 m, and it was 0.5 nmol L −1 at the PNM. At the oligotrophic station

Nucleic acid extraction
Nucleic acids (DNA) were extracted from frozen filters as described previously (Santoro et al., 2010), with modifications. Briefly, cells on the filters were lysed directly in the beadbeating tubes with sucrose-EDTA lysis buffer and 1 % SDS. Prior to mechanical lysis, filter samples were subject to three freeze-thaw cycles of 5 min in liquid nitrogen and 5 min in a 65 • C water bath. Tubes were then agitated in a FastPrep bead-beating machine (MP Biomedicals) for 1.5 min at speed 5.5, and proteinase K (Invitrogen) was added to a final concentration of 0.5 mg mL −1 . Filters were incubated at 55 • C for approximately 4 h and the resulting lysates were purified with the DNeasy kit (Qiagen) using a slightly modified protocol (Santoro et al., 2010). The purified DNA was eluted in 200 µL of DNase, RNase-free water (Gibco) and quantified using a fluorometer (Qubit and Quanti-T BR reagent, Invitrogen Molecular Probes).

Quantitative PCR
qPCR assays for total archaeal amoA (AOA) and betaproteobacterial amoA (AOB) were carried out in 25 µL reactions using SYBR Green chemistry on a StepOnePlus realtime PCR machine (PE Applied Biosystems). Unless noted otherwise below, each reaction contained 12.5 µL Failsafe Green Real-Time PCR PreMix E (Epicentre Biotechnologies), 400 nM each primer, 1.25 U Failsafe Real-Time Enzyme Blend (Epicentre Biotechnologies), and ROX passive reference dye at the concentration recommended by the manufacturer. AOA group-specific assays for "shallow" water column ecotype A (WCA) and "deep" water column ecotype B (WCB) (Mosier and Francis, 2011) used TaqMan Environmental Mastermix (Life Technologies) chemistry as described below. Detection limits for all SYBR assays were 10 copies mL −1 or better; detection limit for TaqMan assays was 1 copies mL −1 or better. All reactions were run in triplicate with a standard curve spanning approximately 10 1 -10 5 templates, run in duplicate. Plasmids containing cloned inserts of the target gene (TOPO pCR4 vector, Invitrogen) were used as standards as indicated below. Standards were linearized with the restriction enzyme NotI (New England Biolabs), purified (DNeasy, Qiagen), quantified by fluorometry (Quanti-T HS reagent, Invitrogen), and stored at −80 • C. Fresh standard dilutions were made from frozen stocks for each day of analysis. A minimum of three negative control qPCR reactions to which no DNA template was added were analyzed with every assay. A melting curve analysis was performed after each SYBR qPCR run with plate reads at a temperature increment of 0.3 • C. R 2 values for the standard curves (cycle threshold (Ct) vs. log 10 copy number) were 0.98 or better for all runs. Efficiency was calculated relative to a theoretical standard curve slope of 3.32.
The betaproteobacterial amoA qPCR assay used the amoA1F/2R primer set (Rotthauwe et al., 1997) and the following thermal profile: of 94 • C for 3 min followed by 35 cycles of 95 • C for 45 s, 56 • C for 30 s, 72 • C for 50 s, and a plate reading step at 82 • C for 10 s. The standard used for this assay was a marine Nitrosospira-like amoA gene generated using amoA genes PCR-amplified from Monterey Bay; average qPCR efficiency was 96 %. Total archaeal amoA genes were quantified using the primers Arch-amoAF/Arch-amoAR (Francis et al., 2005) with an additional 2 mM MgCl 2 added to the reaction chemistry described above, and the following thermal profile: 94 • C for 3 min, followed by 35 cycles of 94 • C for 30 s, 58 • C for 45 s, and 72 • C for 50 s, and a plate read at 80 • C for 10 s. The standard used for this assay was an archaeal amoA gene amplified from the California Current (Santoro and Casciotti, 2011); average qPCR efficiency was 94 %. Group-specific amoA assays used the WCA-amoA-F/R and WCB-amoA-F/R primer and probe sets (Mosier and Francis, 2011) and the following thermal profile: 95 • C for 10 min, followed by 40 cycles of 95 • C for 30 s and 55 • C for 30 s.

One-dimensional NO − 2 modeling
A one-dimensional, bulk mixed-layer model (Price, Weller, Pinkel (Price et al., 1986); PWP) was used to approximate physical processes in the mixed layer. The present version of the model was implemented in MATLAB using code developed by Glover et al. (2011) and modified by Martz et al. (2009). The PWP model was initially developed to model diurnal heating and wind-driven mixing over a few days. Subsequent work has extended the model to run times from months to years (Archer et al., 1993;Mathieu and Deyoung, 1995;Plueddemann et al., 1995;Babu et al., 2004;Vage et al., 2008). The model has 1 m vertical resolution down to 200 m, a time step of 15 s, and a vertical eddy diffusivity of 1.5 × 10 −4 m 2 s −1 . PWP is forced at the surface using NCEP/NCAR Reanalysis 1 surface flux data (Kalnay et al., 1996). The 4-times-daily data were provided by the National Oceanic and Atmospheric Administration ( Signorini et al. (2001). Climate data were extracted for the 2 months prior to the profile date at each station position. The model was initialized with the 12-depth station [NO − 2 ] profile data and run 48 h before being reinitialized with the station profile.

Location and magnitude of the PNM along the cruise transect
The cruise transect traversed productive, high-chlorophyll waters in Monterey Bay (Chl a = 5.2 µg L −1 ) to oligotrophic, open-ocean conditions (surface Chl a < 0.1 µg L −1 , Fig. 1d). As observed in previous occupations of this line, equatorward flow of the California Current causes upward tilting of isopycnal surfaces toward the coast (east) bringing relatively cold, saline, and high-NO − 3 waters to the surface near the coast (Fig. 1b and c). The low-salinity core of the California Current was west of 126 • W (Fig. 1b), relatively far offshore. NO − 3 in the surface waters decreased along the transect from a high of 7 µmol L −1 within Monterey Bay down to below detection (< 4 nmol L −1 ) at 67.155. A subsurface deep chlorophyll maximum (DCM) layer became apparent at 126.3 • W.
We observed a PNM at depths below the chlorophyll maximum at all stations. The PNM deepened offshore (Fig. 1f), concurrent with a deepening of the mixed layer and the appearance of the DCM. The magnitude of the PNM also decreased offshore, ranging from 0.53 µmol L −1 within Monterey Bay (31-40 m depth) to 0.17 µmol L −1 offshore (station 67.155 at 128 m depth). A distinct NH + 4 maximum was also observed near the base of the euphotic zone, below the DCM but above the PNM (Fig. 1e). The magnitude of the NH + 4 maximum ranged from 0.51 µmol L −1 near the coast to 0.014 µmol L −1 offshore, with a particularly high NH + 4 maximum observed during the second occupation of station 67.70 of 1.5 µmol L −1 . NH + 4 was below detection limits at all depths at 67-155. Depth-integrated (0-200 m) NO − 2 and NH + 4 were both correlated with depth-integrated Chl a (r = 0.86, p < 0.01; r = 0.63, p = 0.02, respectively).

Organism distribution in relation to the PNM
The shift from cool, nutrient-rich waters along the coast to oligotrophic offshore conditions was reflected in the phytoplankton community (Fig. 2). At the coast, Synechococcus and photosynthetic picoeukaryotes were abundant in surface waters, reaching abundances of 8.72 × 10 4 cells mL −1 and 2.66 × 10 4 cells mL −1 , respectively, within Monterey Bay where Prochlorococcus cells were undetectable. Both Synechococcus and picoeukaryotes were most abundant in surface waters and decreased in abundance with depth. At any one of these stations picoeukaryotes were present deeper in the wa- Table 2. Abundance of two ecotypes of ammonia-oxidizing archaea: water column A (WCA) and water column B (WCB) based on the abundance of archaeal amoA genes determined by quantitative PCR using ecotype-specific primers and probes (Mosier and Francis, 2011 ter column than Synechococcus. Synechococcus abundance was correlated with water temperature (r = 0.27, p = 0.01), but peak abundance was observed in 16 • C waters. Offshore, between 125.6 and 126.3 • W, there was a shift in the photosynthetic community coincident with a transition to ∼ 17 • C waters (Fig. 2 a,b). Prochlorococcus cells became increasingly abundant west of this transition, reaching high abundances within distinct subsurface peaks, with local maxima of 1.53 × 10 5 to 2.09 × 10 5 cells mL −1 . Prochlorococcus abundance was inversely related to [NO − 3 ] (r = −0.83, p < 0.001) and positively related to water temperature (r = 0.68, p < 0.001). Organisms from all three groups of phytoplankton were generally present at the depth of the PNM (Fig. 3d, h), though with local minima in cell abundance.
amoA genes were quantified to infer the abundance of ammonia-oxidizing organisms at stations 67.70 and 67.155. Archaeal amoA genes were detectable at both stations at nearly all depths. At 67.70 abundances ranged from 220 amoA copies mL −1 at the surface to a maximum of 1.4 × 10 4 copies mL −1 at 55 m depth (Fig. 3b), just below the PNM. At 500 m depth, there were 3.1 × 10 3 copies per mL −1 -less abundant than at the base of the euphotic zone, but more abundant than at the surface. At station 67.155, archaeal amoA abundance ranged from undetectable in the shallowest sample (50 m) to a maximum of 9.5 × 10 3 copies mL −1 at 128 m depth, coincident with the PNM (Fig. 3f). The abundance of water column group "A" amoA genotypes, thought to represent a shallow water ecotype (Francis et al., 2005;Hallam et al., 2006;Beman et al., 2008), accounted for 98 % or greater of the archaeal amoA genes at both stations at all depths except 500 m at station 67.155, where they were less than 5 % of the amoA ecotypes (Table 2).
Bacterial amoA gene abundance at 67.70 ranged from below detection limits in the two shallowest samples to 610 copies mL −1 at 55 m, coinciding with the peak in AOA abundance. Bacterial ammonia oxidizers were less abundant at the oligotrophic station, ranging from undetectable at 50, 150, and 500 m, to near the detection limit above the PNM.

NO − 2 production in and around the PNM
We determined rates of ammonia oxidation (NH 3 to NO − 2 and NH 3 to NO − 2 + NO − 3 ) at the same stations and depths that we quantified ammonia-oxidizing microbial abundances (Figs. 3, 4). [NO − 2 ] and [NO − 3 ] were constant in the incubation bottles during the 36 h of the incubation, varying only within the error of the measurements (data not shown).
[NH + 4 ] decreased in nearly all incubation bottles, between 30 and 120 nmol L −1 with the greatest decreases in the euphotic zone bottles. Due to the short duration of the cruise we were not able to control the time of day at which the incubations began, but we did not observe any obvious day-night differences in the rate of change in 15 N / 14 N in the product pools (e.g., Fig. 4). NO − 2 production from ammonia oxidation in individual bottles at station 67.70 ranged from below detection limits at 500 m to 31 ± 13 nmol L −1 d −1 at 55 m, just below the PNM (Fig. 3a). Combined NO − x production from ammonia in individual experiments ranged from below the detection limit at the surface and 500 m to 34 ± 12 nmol L −1 d −1 at 55 m. At a given depth, average rates of ammonia oxidation to NO − 2 and ammonia oxidation to NO − x were similar, except at 55 m where the average NO x production rate from ammonia (33 ± 14 nmol L −1 ) was greater than the average NO − 2 production rate (23 ± 15 nmol L −1 d −1 ).
Further offshore at the oligotrophic station (67.155), rates of NO − 2 production from ammonia oxidation ranged from 0.03 ± 0.03 nmol L −1 d −1 at the surface to 9 ± 22 nmol L −1 d −1 just below the PNM. The large uncertainties at this depth were due to very low [NO − 2 ] (Fig. 3e). Rates of NO − 2 + NO − 3 production showed a similar pattern, ranging from below detection limits at the surface to 8 ± 5 nmol L −1 d −1 just below the PNM. At station 67.70, NO − 2 production from NO − 3 was detectable at 20 m, with a mean rate of 4 ± 5 nmol L −1 d −1 (Fig. 3c). A similar rate, however, was detected in the filtered control at this depth. NO − 2 production from NO − 3 was also detected at 55 m with a magnitude of 2 ± 1 nmol L −1 d −1 , just below the PNM and coincident with the highest rates of nitrification. At station 67.155, NO − 2 production from NO − 3 above detection limits was only observed at 50 m, the shallowest depth sampled, at a mean rate of 0.8 ± 0.3 nmol L −1 d −1 (Fig. 3g). The  profiles are the result of both physical and biological processes. If the nutrient profiles are assumed to be in a steady state on the scale of days, and the physical changes can be accounted for, then the residual differences can be attributed to biological activity. Here, a onedimensional mixed-layer model (PWP) was used to simulate . At steady state, the rates of NO − 2 production needed to sustain the peak concentration against the rates of mixing, and the rates of NO − 2 consumption needed to prevent mixing from broadening the peak, can be computed (Fig. 5). Modeled rates were compared against net NO − 2 production within the PNM from the incubation experiments. Net NO − 2 production was calculated as the sum of NO − 2 production from ammonia oxidation and nitrate reduction, minus NO − 2 losses, calculated as the NO − 2 removal rate (k) times the in situ [NO − 2 ]. At station 67.70, the model predicts net NO − 2 production at the PNM of 9.8 mmol m −2 d −1 , greater than aerially integrated measured values (0.5 mmol m −2 d −1 ) though there is high uncertainty in the measured value due to uncertainties in the NO − 3 reduction rate. At station 67.155, modeled rates of net NO − 2 production within the PNM were 4.5 mmol m −2 d −1 and compared well with measured net NO − 2 production (also 4.5 mmol m −2 d −1 ).

Discussion
NH + 4 consistently showed a maximum overlying the primary NO − 2 maximum (Fig. 1e, f). Co-localization of these features has been observed in isolated profiles (Olson, 1981a;Lipschultz et al., 1996;Beman et al., 2012), and across the South Pacific Gyre (Raimbault et al., 2008). The localization of sequential maxima in Chl a, NH + 4 , and NO − 2 strongly suggest that the PNM is a remineralization feature, whereby the sequential products of nitrogen remineralization -NH + 4 and NO − 2 -accumulate one above the other and are correlated in magnitude. A nitrification source of NO − 2 in the California Current PNM is further supported by the high rates of NO − 2 production from NH + 4 relative to production from NO − 3 . The maximum rates of NO − 2 production from NH + 4 , however, were measured below and not within the peak of the PNM, consistent with previous reports from the North Pacific (Dore and Karl, 1996a). This rate distribution has previously led to the proposal that reduction of NO − 3 by phytoplankton 2 production or loss at the PNM are shown as blue diamonds, calculated as described in the text. Model-data comparisons are only shown for incubations at the PNM due to uncertainties in NO − 2 removal above and below the PNM. Error bars reflect total uncertainty in the net NO − 2 production calculation.
was the source of NO − 2 in the upper PNM. However, direct measurements of NO − 2 production via this pathway were not made. Although it may have contributed at rates below our detection limits, release of NO − 2 from NO − 3 was quite low at the depths of the PNM at our stations.
The potential for "bottle effects" causing either positive or negative impacts on the measured rates is always a consideration during at-sea incubation experiments. One such effect could be the potential for stimulation of the measured rates due to added substrate, particularly in the case of our 15 NH + 4 additions. Previous work in the open ocean (Olson, 1981a;Newell et al., 2013) suggests that environmental populations of ammonia oxidizers are saturated at < 100 nM NH + 4 with no rate stimulation beyond this concentration. Others have suggested large stimulation effects of NH + 4 on nitrification (Horak et al., 2013). Cultures of ammonia-oxidizing archaea (Martens-Habbena et al., 2009) suggests that the halfsaturation constant (K m ) for NH 3 + NH + 4 is very low (less than 140 nM). At the depths of the PNM in the California Current, [NH + 4 ] was generally less than 100 nM, so it is possible that our 100 nM 15 NH + 4 additions could have stimulated ammonia oxidation. However, as noted above, the concentration of NH + 4 decreased over the course of the experiment with no apparent decrease in the rate of 15 NH + 4 conversion to NO − 2 or NO − 2 + NO − 3 over time. Because of the relatively short duration of the cruise, we were unable to start experiments from the two stations at similar times of day. As light is a potential contributing factor to the rates of ammonia oxidation and nitrate reduction, we evaluated the influence of light period on the measured rates of 15 N / 14 N change in the intervals between each pair of time points. Changes in δ 15 N over the course of the experiments were nearly linear and did not appear to be linked to the day-night cycle. This suggests that differential light exposure did not affect the relative rates measured at the two stations; however, this cannot be completely ruled out as a factor affecting our results. A second potential issue with incubation experiments is saturation of 15 N / 14 N in the product pool (in this case NO − 2 ), which would cause an underestimation in the rates of NO − 2 production. With a pool of NO − 2 that is held at a constant steady-state concentration (inputs and outputs balanced), the time scale of NO − 2 turnover that is required to saturate our ability to accurately model the input over 36 hours is on the order of 1 d −1 . For example, a pool of NO − 2 that is 100 nM would show a plateau in the 15 N / 14 N after 36 h at rates of production and consumption of 100 nM d −1 . This guideline is relevant to any size NO − 2 pool that is turning over once per day and is insensitive to the substrate at % 15 N in the range used here for NH + 4 oxidation (35-93 %) and NO − 3 reduction (2-19 %). Rates this high were not observed in our experiments, and we never observed plateauing of δ 15 N NO 2 over time, or values approaching the initial at % 15 N of the substrate pool. Thus, we do not believe that the rates of NO − 2 production have been underestimated around the PNM. NO − 2 production could have been underestimated in the euphotic zone and at 500 m, where the lack of measurable pre-existing NO − 2 does not provide a trap for labeled NO − 2 . Our interest, however, in this manuscript was in the formation and maintenance of the PNM, for which the above arguments hold.
In contrast to data from Station ALOHA (Dore and Karl, 1996b), we did not observe a "double peaked" PNM, though we have observed this feature on previous occupations of LINE 67 (Santoro et al., 2010). Dore and Karl observed an upper PNM (UPNM) and lower PNM (LPNM) that they attributed to NO − 3 reduction and nitrification, respectively. They also detected NO − 2 below the depth of the LPNM; 2-5 nmol L −1 NO − 2 was present as deep as 800 m. We did not find evidence of deep NO − 2 , though measurements at 500 m and 1000 m were not made at every station. As suggested by Dore and Karl (1996b), this "tail" may be a result of episodic organic matter export events from the mixed layer not occurring during our cruise.
The patterns observed here in photosynthetic picoplankton distribution are consistent with previous observations from both the Atlantic (Zubkov et al., 2000;Cavender-Bares et al., 2001;Johnson et al., 2006) and Pacific. Zubkov et al. (2000) suggested that a high abundance of Synechococcus marked the transition between temperate and oligotrophic waters, consistent with the data presented here showing a high abundance of Synechococcus in 16 • C waters. Deeper distribution of picoeukaryotes relative to picocyanobacteria has been observed at Station ALOHA (Campbell and Vaulot, 1993;Campbell et al., 1997), consistent with observations presented here. Due to the strong inverse correlation between temperature and [NO − 3 ], the inverse correlation between Prochlorococcus abundance and [NO − 3 ] cannot be interpreted as a signal of NO − 3 uptake by Prochlorococcus, but instead likely reflects the strong influence of temperature on the distribution of these organisms (Zubkov et al., 2000;Johnson et al., 2006;Zinser et al., 2007). Previous work in the Sargasso Sea (Zinser et al., 2007) did not find a correlation between Synechococcus abundance and temperature as we report here.
We used flow cytometry as an alternative to qPCR to target potentially nitrate-assimilating taxa. Nitrate reductase (Nar), the enzyme encoding the first step of assimilatory nitrate reduction, has a broad phylogenetic distribution, and it would be impossible to capture the majority of nitrate-reducing taxa with two or perhaps even ten sets of qPCR primers. Capturing cyanobacterial narB diversity alone requires a nested PCR assay not suitable for qPCR (Paerl et al. 2008), and 7 sets of qPCR primers to cover different Synechococcus ecotypes of narB alone (Paerl et al. 2011). Paerl et al. (2011 found that their qPCR assay for narB gave about 40 % of the abundance estimate of flow cytometric Synechococcus counts. We also did not quantify heterotrophic organisms that may be responsible for NO − 3 reduction, and some previous studies have shown variable but significant contributions of bacterioplankton to total NO − 3 uptake in the upper ocean (4-14 %, Kirchman et al., 1994;5-60 %, Kirchman and Wheeler 1998). The extent to which these organisms contribute to NO − 2 production would have been captured by our NO − 3 reduction rate measurements, even though we did not attempt to quantify the organisms.
The abundance of AOA relative to AOB was greater at station 67.155 (AOA : AOB >2000) in the oligotrophic waters of the North Pacific Gyre than at the upwelling station 67.70 (AOA : AOB ∼ 10-400), supporting previous speculation that AOA in the ocean are oligotrophic specialists (Martens-Habbena et al., 2009;Horak et al., 2013). The high relative abundance of the water column A amoA ecotype suggests this ecotype is responsible for the majority of the ammonia-oxidizing activity detected in our incubations. At present the physiological characteristics differentiating the two ecotypes is unknown (Beman et al., 2008;Santoro et al., 2010).
Sharp gradients were observed in nitrification rates, with apparent maxima just below the PNM at both stations. At 67.70 the maximum nitrification rate was coincident with the maximum AOA abundance below the PNM, whereas at 67.155 the abundance of AOA peaked within the PNM. Beman et al. (2012) also observed peaks in nitrification rates just below the PNM in the eastern tropical North Pacific using high-resolution measurements, particularly within   Buchwald and Casciotti (2013) the Gulf of California. Accumulation of NO − 2 in the water column is not necessarily directly correlated to NO − 2 production, but rather reflects the balance of production and consumption of NO − 2 . Thus, a spatial offset between maximal rates of production and [NO − 2 ] is not surprising. Newell et al. (2013), however, did observe a correspondence between the PNM and maximum nitrification rates in the Sargasso Sea.
A one-dimensional model was used to estimate NO − 2 production rates from [NO − 2 ] profiles at station 67.70 and model the balance of biological production, consumption, and physical mixing (Fig. 5a, b). At station 67.70 (Fig. 5a), modeled NO − 2 production within the PNM is roughly a factor of 10 greater than net production rates calculated using 15 N rate measurements. Agreement between the model and measurements was better at station 67.155 (Fig. 5b). There are large uncertainties in the net NO − 2 production calculation, as errors from all three rate determinations are propagated through the calculation and the model output is highly dependent on the choice of eddy diffusivity (1.5 × 10 −4 m 2 s −1 used here). Above and below the PNM, removal rates are poorly constrained by our 15 N incubations because of large uncertainties in the removal rate constant (k) and very low . At these depths the 1-D model is probably a more accurate representation of net NO − 2 production. The mechanism that allows remineralization products (NH + 4 and NO − 2 ) to accumulate is still uncertain. Differential light inhibition of the two nitrifier groups is often proposed (Olson, 1981b), whereby NO − 2 oxidizers are inhibited at lower light levels (are more light sensitive) than ammonia oxidizers. Our data are consistent with this hypothesis; however, laboratory data suggest that AOA are also extremely light sensitive (Merbt et al., 2012), which argues against the differential light inhibition hypothesis. The extent to which these laboratory results reflect responses in nature is unclear, as extreme light sensitivity does not appear consistent with widespread presence of AOA amoA genes and transcripts in stratified Pacific surface waters (Church et al., 2010;Santoro et al., 2010).
Gradients in several other growth factors with "nutrientlike" depth distributions are seldom discussed, but equally plausible mechanisms for the accumulation of NO − 2 at the PNM. Iron (Fe), in particular, could be limiting at the depth of the PNM. Non-steady-state Fe limitation is associated with NO − 2 excretion in diatoms (Behrenfeld et al., 2006) and has been invoked to explain a potential phytoplankton source of NO − 2 within the PNM (Lomas and Lipschultz, 2006). Fe could also play an underexplored role in a nitrificationsourced PNM. Our results indicate that AOA are both the most abundant and active nitrifiers at the PNM, as suggested previously (Mincer et al., 2007;Santoro et al., 2010;Newell et al., 2011). Genome analysis suggests AOA rely on blue copper proteins for respiration and electron transport (Walker et al., 2010;Blainey et al., 2011) and may therefore have a uniquely low Fe requirement. The nitrite oxidoreductase required for NO − 2 oxidation, on the other hand, contains at least four Fe-S clusters, suggesting a very high iron demand for this process (Kirstein and Bock, 1993;Spieck et al., 1998). Genomic information has only recently become available for relevant open-ocean NO − 2 oxidizers, such as Nitrospina spp. (Mincer et al., 2007;Luecker et al., 2013), and is currently lacking for Nitrospira spp.
We measured low rates of NO − 2 production from NO − 3 at all stations and depths, arguing against a large role for NO − 2 excretion from phytoplankton in the formation of the PNM at these sites. Just below the PNM at 67.70, NO − 2 production from NO − 3 was 2 nmol L −1 d −1 compared with mean rates of nitrification to NO − 2 of 26 nmol L −1 d −1 and to NO − 2 + NO −