Ambient nitrate switches the ammonium consumption pathway in the euphotic ocean

Phytoplankton assimilation and microbial oxidation of ammonium are two critical conversion pathways in the marine nitrogen cycle. The underlying regulatory mechanisms of these two competing processes remain unclear. Here we show that ambient nitrate acts as a key variable to bifurcate ammonium flow through assimilation or oxidation, and the depth of the nitracline represents a robust spatial boundary between ammonium assimilators and oxidizers in the stratified ocean. Profiles of ammonium utilization show that phytoplankton assemblages in nitrate-depleted regimes have higher ammonium affinity than nitrifiers. In nitrate replete conditions, by contrast, phytoplankton reduce their ammonium reliance and thus enhance the success of nitrifiers. This finding helps to explain existing discrepancies in the understanding of light inhibition of surface nitrification in the global ocean, and provides further insights into the spatial linkages between oceanic nitrification and new production. The underlying regulatory mechanisms of phytoplankton assimilation and microbial oxidation of ammonium in the surface ocean are unclear. Here, using isotope labeling experiments, the authors show that ambient nitrate is a key variable bifurcating ammonium flow through assimilation or oxidation.

A mmonium (NH 4 + ) is a central component in the marine nitrogen cycle. The relative flow of NH 4 + through assimilation by phytoplankton versus oxidation by microbial nitrifiers largely determines the composition of the upper ocean nitrogen pool, which plays a crucial role in controlling marine productivity and the export of fixed carbon in the ocean 1 . The pathway to organic nitrogen through assimilation represents regenerated production, while the pathway to nitrate (NO 3 − ) through nitrification fuels new production and produces the greenhouse gas nitrous oxide (N 2 O). The two competing pathways are known to show spatial separation in some oceans, where nitrification occurs deeper than ammonium assimilation, and this separation has long been assumed to be lightmodulated 2,3 . However, this view is challenged by recent evidence of active nitrification in the sunlit ocean [4][5][6][7][8] . Intriguingly, surveys on distributions of nitrification in the global sunlit ocean exhibit paradoxical results. In the subtropical North Atlantic, nitrification in the euphotic zone appears to be inactive during winter when light irradiance is low 9, 10 and becomes more active in late spring and summertime with higher light intensity 4,5 , resulting in a counterintuitive light effect on nitrification. In the North Pacific, nitrification remains undetectable in the entire euphotic zone of the subtropical gyre, but is prominent even at the surface in high latitude and equatorial areas where subsurface waters outcrop 6 . These inconsistencies indicate additional controls on nitrification are likely.
Nutrient concentrations are another fundamental environmental parameter influencing biogeochemical processes. They exert primary control on species composition, cell size structure, and associated nutrient uptake capacity of phytoplankton assemblages in the ocean [11][12][13][14][15] , with significant variation in uptake and growth kinetics of phytoplankton communities occurring under different nutrient conditions, including increasing halfsaturation concentrations from oligotrophic to nutrient replete regions 16 . Thus, we hypothesize the nutrient status to be an essential controlling factor on NH 4 + flows in upper ocean waters through regulating the relative competitiveness of phytoplankton and nitrifiers under limiting levels of NH 4 + . To explore controls of NH 4 + cycling by the interactive relationships between phytoplankton and nitrifiers, simultaneous rate measurement of NH 4 + assimilation and oxidation is necessary; surprisingly, such data remain sparse. To our knowledge, direct comparison of NH 4 + utilization capabilities between phytoplankton and nitrifiers is missing to date in both laboratory and field studies.
To address this gap and to unravel the biogeochemical implications of ammonium pathways in dynamic oceans, we measured simultaneously NH 4 + uptake, NO 3 − uptake, and nitrification rates, at sites spanning a wide range of nutrient conditions from coastal to open ocean waters in the South China Sea and the Northwestern Pacific Ocean (Supplementary Fig. 1). By using isotope labeling techniques and a refined rate calculation approach (see Methods), we obtained multiple rate profiles with high vertical sampling resolution, which allowed us to capture the sharp gradients in biological processes and associated physicalchemical conditions in the upper ocean.
Moreover, for the first time, we compared the competitive capabilities of nitrifiers and phytoplankton by testing their underlying ammonium utilization kinetics (specifically, the potential maximum reaction rate V max , the half saturation concentration K s , and the substrate affinity α) 12,17 , at varying light and nutrient levels. This kinetic information provides robust evidence regarding the competitive advantages of nitrifiers versus phytoplankton for NH 4 + , under different environmental conditions. We find that ambient NO 3 − level, which regulates phytoplankton's reliance and affinity toward NH 4 + , is a key variable to bifurcate NH 4 + flow through assimilation versus oxidation.

NH 4
+ uptake and nitrification rates. In the South China Sea, all three stations (X2, H3, and D1) displayed typical shallow stratification of the subtropical oligotrophic ocean in warm seasons. Based on a temperature threshold criterion with a difference of 0.8°C from the surface value 18 , we found that X2 held the deepest and strongest stratification with mixed layer depth of 50 m. Station H3 showed the second strongest stratification despite it held mixed layer depth of around 20 m similar to D1 ( Supplementary  Fig. 2). By defining the mid-point (average) of the depth range with the steepest gradient as the location (depth) of the nitracline 19 , we found the depths of the surface mixed layer and the nitracline co-varied. As the nitracline shoaled, the deep chlorophyll maximum (DCM) migrated upward accompanied by higher Chl-a (Fig. 1a-c) and higher particulate organic nitrogen (PON) (Supplementary Fig. 2). Corresponding to such migration, the light penetration depth was influenced by the biomass (see PON levels in Supplementary Fig. 2) with attenuation coefficients of 0.07 at D1, 0.06 at H3, and to 0.05 for X2. Here, the euphotic zone was defined by the 0.1% surface photosynthetically active radiation (sPAR) depth, because low levels of NH 4 + uptake were observed below the 1% sPAR level. The euphotic depth thus deepened from 95 m at D1 to 115 m at Station H3 and 140 m at X2, being positively correlated with the nitracline depth. NH 4 + concentrations were low (1-74 nmol L −1 ) without clear vertical maxima. By contrast, primary nitrite maxima (PNM, with maximal concentrations ranging from 185.5 to 254.0 nmol L −1 ) were observed slightly above the 1% sPAR depth and below the DCM for all stations. NO 3 − was depleted in the mixed layer for all stations (<0.2 μmol L −1 ) and remained low downward until the depth of the gradient-defined nitracline. Interestingly, below the DCM layer at all stations there were distinct NO 3 − /SiO 4 −4 maxima with peak values ranging from 1.1 to 1.2 ( Supplementary  Fig. 2), suggesting preferential remineralization and nitrification of N in sinking particle had occurred 20,21 . The paired light and dark incubations provide direct evidence of light effects on phytoplankton and nitrifiers. Nitrification rates were significantly higher in the dark than in the light at all sites (p < 0.05, Supplementary Table 1); the inhibition effect was most prominent in the upper euphotic zone and decreased downward with the attenuation of light ( Supplementary Fig. 3a-c). Nevertheless, detectable rates (higher than the detection limit of 0.04-0.16 nmol N L −1 d −1 , see Methods) remained in the dark incubations in the upper euphotic layer at most of our study sites, indicating recovery of ammonia oxidizers from photo damage. In other words, the efficiency of nitrification was substantially reduced but not completely inhibited within a diel cycle. Conversely, light enhanced NH 4 + uptake was low in the dark (Supplementary Fig. 3d-f), although the differences between light and dark rates were not as significant as for nitrification (Supplementary Table 1). Thus our results also reinforce the necessity of paired light and dark incubations to obtain precise reaction rates.
The most striking feature of the NH 4 + uptake and nitrification rate profiles was their inverse unimodal distributions, with assimilation above and nitrification below a remarkably strong transition boundary at the nitracline, regardless of its depth or irradiance level ( Fig. 1d-f, Supplementary Fig. 4). This implies ambient nitrate might play an important role in niche partitioning of NH 4 + assimilation versus oxidation. NH 4 + uptake was active above the nitracline with peak values ranging from 5.4 to 27.8 nmol N L −1 d −1 for the three stations. Dramatic downward decreases (reduction by 73−82% within 20 m) in NH 4 + uptake were found near the nitracline, and the rate remained detectable throughout the lower euphotic zone. By contrast, nitrification was low or undetectable in the mixed layer and increased rapidly downward around the nitracline, peaking (with maximum rates of 5.9−22.2 nmol N L −1 d −1 ) near the PNM and upper oxycline. Below the depth of the nitrification maximum, rates decreased gradually toward greater depth. The vertical distribution pattern of nitrification broadly agrees with current recognition that nitrification is fueled by remineralization of organic matter 22 and contributes a considerable amount of NO 2 − to form and maintain the PNM 23,24 . The values of maximum NH 4 + uptake and nitrification were comparable, suggesting both pathways actively contributed to the consumption of NH 4 + , but in separate domains in the sunlit oligotrophic ocean.
To further examine the applicability of the observed relationship between nitrate concentration and NH 4 + consumption pathways, we measured NH 4 + uptake and nitrification rates in the Northwest Pacific Ocean. Five stations across coastal, shelf, and open ocean conditions, with large gradients in surface nutrients and biological activities were selected ( Supplementary  Fig. 5). The coastal station showed significantly higher nutrient and biological activities (inferred from fluorescence and PON) while the open ocean stations all showed shallower nitraclines and lower biomass ( Supplementary Fig. 5). For the shelf (P5 and C5) and basin (K1 and X1) stations with shallow stratification, the transition boundaries from NH 4 + uptake dominance to oxidation dominance were observed in every case to be near the corresponding nitraclines (i.e. at the same depth within the site specific 5-10 m resolution of the sampling). At the eutrophic coastal station P1, where no nitracline was present, NH 4 + uptake and nitrification co-occurred throughout the water column ( Fig. 2), similar to results observed in Monterey Bay 7 . These results from a wide range of environmental conditions demonstrate a ubiquitous regulatory influence of nitrate on the competition for NH 4 + between phytoplankton and nitrifiers. We investigated the dependence of nitrification and NH 4 + uptake rates on substrate concentration by adding different amounts of isotopic-tracer-labeled substrate, providing a method that calibrates enhancement of rates by tracer enrichment 25 . The observed kinetic responses also enable us to compare the competitive capability of these organisms towards the limiting nutrient 12,17 . Values of V max for nitrification were significantly higher than NH 4 + uptake below the nitracline where NO 3 − was replete (p < 0.05) (Fig. 3a, Supplementary Fig. 6 and Supplementary Table 2). The K s for nitrification did not show a clear vertical trend, whereas the K s for NH 4 + uptake by phytoplankton exhibited values below the nitracline higher than above, although with strong variability (Fig. 3b). Accordingly, a clear difference between nitrifiers and phytoplankton in affinity (α) toward NH 4 + was observed under different NO 3 − concentrations, the α of phytoplankton was higher above the nitracline while the α of nitrifiers was higher below (p < 0.01) (Fig. 3c, Supplementary Table 2). These differences in substrate affinity above and below the nitracline obviously reflect the relative competitive advantages between phytoplankton and nitrifiers.

Discussion
Our results suggest light inhibition of nitrifiers is not the single determinant for the distribution of nitrification in the surface ocean. In spite of the inhibition of ammonia oxidizers by light via photo damage to the key enzyme ammonia monooxygenase, as well as other membrane components 26 or the toxicity of reactive oxygen species 27 , some taxa of marine AOA have been identified as having genetic capabilities to reduce oxidative stress and repair ultraviolet damage 28,29 and thus appear to be resistant to light in the upper ocean 30,31 . Consistent with these perspectives, our paired incubations showed that light-inhibited nitrifiers either recovered in the dark or some portion of their population was light resistant (Supplementary Fig. 3a-c). Consequently, detectable nitrification rates were observed even in the surface layer ( Supplementary Fig. 4). This is consistent with reported nitrification rates in the euphotic zones of a wider range of environments, which all exhibit resistance to light inhibition 6-8, 24, 25 .
The co-variation of light and NO 3 − in the stratified oligotrophic ocean often obscures differentiating their effects on regulating euphotic nitrification. Nevertheless, our measurements with high vertical resolution combined with the kinetic tests allowed us to distinguish the role of NO 3 − from light and thus to give a more comprehensive evaluation of the underlying mechanisms of NH 4 + consumption. Unexpectedly, we found that the nitrification rates have stronger correlations with NO 3 − concentrations than with light levels, from the surface to the depth of the rate maxima ( Supplementary Fig. 7). Moreover, we observed the depth transition from NH 4 + uptake to nitrification was highly consistent with the depth of the nitracline (Figs. 1, 2, Supplementary Fig. 4), even though light intensity at this depth varied among the stations. These results suggest a previously unrealized ecological control of NO 3 − in structuring the spatial domains of nitrifiers.
According to our kinetic tests, the nitrifier and phytoplankton communities had similar K s values for NH 4 + in layers above the nitracline (Fig. 3b), and these values were consistent with field and laboratory studies 25,32,33 . However, the V max and α values were higher for phytoplankton above the nitracline (Fig. 3a), resulting in a competitive advantage of phytoplankton toward NH 4 + . Light stimulates phytoplankton growth, and dark does not damage these cells, and accordingly the phytoplankton showed significantly higher V max than nitrifiers over the diel cycle. Phytoplankton thus outcompete ammonia oxidizers to be the main NH 4 + consumer in the surface mixed layer. Below the nitracline, by contrast, significantly higher values of V max and α for nitrification evidence a competitive advantage for nitrifiers over phytoplankton (p < 0.05, Supplementary Table 2). This was due to the benefit of reduced light inhibition on nitrifiers ( Supplementary Fig. 3) as well as the reduction in affinity of phytoplankton toward NH 4 + . In fact, our rate profiles demonstrate a downward shift of nitrogen source for phytoplankton from NH 4 + to NO 3 − . The NO 3 − uptake rates showed an increasing trend with depth and peaked around the nitracline, and the rates remained significantly higher than NH 4 + uptake rates below the nitracline at all sites (p < 0.05). Similar results of higher NO 3 − than NH 4 + uptake rates have also been observed at BATs during the deep mixed layer season 5 . Because NO 3 − utilization is more energetically costly, the low NH 4 + uptake rates were unlikely to be caused by light deficiency, an alternative possible cause is the high nutrient and low light conditions, which often lead to eukaryotic phytoplankton dominance. Previous studies have shown the interactions of NH 4 + and NO 3 − uptake by phytoplankton appear to be highly diverse and vary with changes in species, preconditioning (nutrient history) and light [34][35][36] . For example, incubation of diatoms under low light shows high NO 3 − uptake rates in spite of high NH 4 + enrichment 37 . Our results are also supported by short-term experiments that find phytoplankton are less reliant on NH 4 + as their nitrogen source, when pre-conditioned with enriched NO 3 − (compared to preconditioning with NH 4 + ) 34 , and with no inhibition of NO 3 − uptake by NH 4 + under low light, possibly due to acclimation to satisfy nitrogen requirements for cell growth 35 or because of enzymatic rather than energetic limitation of growth 36 . Analogous subsurface NO 3 − uptake maxima have also been observed in the North Atlantic subtropical gyre, associated with elevated abundances of picoeukaryotes, implying their active uptake of NO 3 −38 , which is also evidenced by the δ 15 N of sorted picoeukaryotes 39 . Moreover, our NO 3 − uptake rates showed significant positive correlation with fluorescence at all our study sites (p < 0.01, R 2 = 0.63, Supplementary Table 3), providing additional evidence that the phytoplankton communities below the nitracline were predominantly dependent on NO 3 − (Fig. 1a-c). Comprehensive understanding of the distinctive and variable kinetic responses of NH 4 + and NO 3 − uptake to varying light and nutrient conditions and histories remains unclear, but the spatially decoupled maximum V max of NH 4 + uptake (at 15 m) and NO 3 − uptake (around the nitracline) demonstrate strongly divergent nutrient utilization strategies for the phytoplankton communities living in different nutrient conditions. Collectively, these results suggest that the downward increase of ambient NO 3 − concentration has an important role in modulating the relative competitiveness of phytoplankton and nitrifiers for NH 4 + , likely via its influence on phytoplankton community structure.
This mechanism for reduced phytoplankton reliance on ammonium at depth as a result of structuring of the community by nutrient availability is consistent with broader ecological theory in which ambient nutrient conditions are regarded as the most important factor in determining the structure and mean cell size of phytoplankton communities in marine environments 14,15,40 , with oligotrophic regions dominated by phytoplankton with small cell size, and larger cells present in nutrient-replete regions. Consistent with this generality, transition from prokaryote dominance of the phytoplankton community in the mixed layer to eukaryote dominance below the nitracline was evidenced by pigment data in the SCS (for instance, in the H3 station of our cruise, prokaryotic phytoplankton account for 85.5 ± 0.1% of total biomass in the mixed layer and the ratio of eukaryotes increase to 67.8 ± 2.9% from DCM to the bottom of euphotic zone, Supplementary Fig. 8). Dominance of Prochlorococcus in the mixed layer and increase of eukaryotes near the nitracline has been previously observed at other locations in our study area 41,42 as well as other low latitude oligotrophic oceans 43,44 . Prochlorococcus preference on NH 4 + is consistent with short-term incubations of flow-cytometrically sorted Prochlorococcus populations, of which NO 3 − was found to account for only <10% of their nitrogen source 45 , despite their genetic capability to utilize NO 3 −46 . By contrast, eukaryotic phytoplankton often show higher nutrient demand and higher nutrient K s values due to their larger cell size 11,13,47,48 . This is consistent with our results showing that the K s of NH 4 + uptake increased from the NO 3 − -depleted surface waters downwards into the NO 3 − -replete depths (Fig. 3b), and indicates lower affinity on NH 4 + in the phytoplankton in the high NO 3 − environment. This perspective of ambient nutrient control of community structure is  6 ; BATs from ref. 10 ; HOTs from ref. 4 ; ETSP from ref. 67 ; ETNP from ref. 66 ; Southern Ocean from ref. 69 ; California Current from ref. 24 ; Hood Canal from ref. 25  substantially stimulate biomass production and push phytoplankton communities towards large diatom dominance [49][50][51] . Consequently, nitrification is likely to be enhanced due to alleviation of competition for NH 4 + from phytoplankton. However, to our knowledge, direct evidence of stimulation of nitrification by NO 3 − enrichment, via shifts in phytoplankton community structure, remains unexplored in the field and thus warrants more verification in future.
Our results also help to elucidate the causes of the discrepancies in the extent of surface nitrification in different oceanic regimes. By compiling reported nitrification rates in the upper 200 m of different ocean provinces, we found persistently low rates of nitrification above the nitracline, and that high rates were always associated with NO 3 − -replete conditions, in spite of the highly varying locations of the observations (Fig. 4). In the cases with deep nitraclines, nitrification rates were low throughout the euphotic zone, likely due to intense competition for NH 4 + by phytoplankton under NO 3 − -depleted conditions (Supplementary Note 1). Conversely, in nutrient-replete coastal or upwelling regions, active nitrification has been reported even in surface layers with high light intensity 7,8,52,53 , demonstrating a niche in high-nutrient environments for nitrifiers. This close relationship between nitrification and NO 3 − suggests an obligatory condition of NO 3 − for nitrification to occur in the sunlit ocean, which reconciles the discrepancies in the apparent light effects on nitrification for different ocean provinces.
Conventionally, nitrifiers and phytoplankton are thought to be separated by light, phytoplankton dominates NH 4 + consumption in high irradiance environments, and nitrification is driven by mineralization of PON and DON in low-light conditions, yet, whether there is a threshold light value for setting the niche boundary of nitrifiers has been unclear (Fig. 5a). Our evidence underscores that light is not the sole determinant and that the ambient NO 3 − concentration exerts an important influence. Based on our results, the depth of the transition boundary of NH 4 + uptake and nitrification was consistent with the nitracline at all stratified oligotrophic sites, suggesting the depth of the nitracline represents a reliable boundary separating the space domains of ammonium oxidizers from assimilators (Fig. 5b). This finding extends our current knowledge of regulatory mechanisms for nitrification in the stratified oligotrophic ocean.
Finally, euphotic nitrification has important implications regarding bottom-up NO 3 − supply versus oceanic new production 54 . The observed inter-relationships among NO 3 − concentration, phytoplankton community structure, and the optimal niche for nitrifers, bind nitrification tightly with new production inside the euphotic zone. Specifically, the NH 4 + assimilators dominated in the mixed layer, while NO 3 − assimilators and nitrifiers developed in response to the downward increase of NO 3 − (Supplementary Fig. 9a). Thereby, a predictable relationship between nitrification and new production is formed, the ratio of nitrification rate to new production rate increases from the surface to the bottom of the euphotic zone (Supplementary Fig. 9b). Notably, in our study the V max values of nitrification at the DCM and 1% PAR layers reached the same magnitude as NO 3 − uptake rates ( Supplementary Fig. 7). This suggests that equating nitrate uptake with new production is an over-simplification, and also that nitrifiers were substrate stressed and thus potentially capable of even greater dominance of ammonium recycling and nitrate supply.
The ongoing warming of the global ocean is expected to enhance ocean stratification, and thus may have significant impacts on the cycling of NH 4 + by influencing the supply of subsurface NO 3 − . For instance, global decline of phytoplankton and production accompanied by increases of niches for cyanobacteria Prochlorococcus and Synechococcus due to the climatedriven expansion of the stratified oligotrophic ocean 55, 56 would likely depress nitrification through less supply of NO 3 − to the euphotic zone and more intense competition of NH 4 + by the small size phytoplankton. This migration of nitraclines would also influence the spatial domains available to nitrifiers. Thus the future changes in ocean stratification may cause reallocation of spatial domains for both active NH 4 + assimilators and oxidizers. In view of the important role of NH 4 + in marine nitrogen and carbon cycling, our study helps to understand the nitrogen pathways in the sunlit ocean and improve our ability to predict how global change is likely to impact the nitrogen cycle and in consequence the ability of the ocean to absorb CO 2 .

Methods
On-deck incubations. Samples were collected from June to July 2014 in the South China Sea (SCS) and from April to May 2015 in the Subtropical Northwest Pacific (NWP) by R/V Dongfanghong II. Three stations spanning a range of hydrographic conditions and corresponding biological productivities from shelf to the basin in the SCS, and five stations spanning from coastal waters to the subtropical gyre of the NWP with even larger environmental gradients were chosen for on-deck incubations (Supplementary Fig. 1). Temperature, salinity, and chlorophyll concentrations were measured using a Seabird SBE 911 CTD sensor package equipped with a fluorometer. Photosynthetically active radiation was measured using a PAR sensor . Samples were collected using 12-L Niskin bottles mounted on the CTD rosette.
Depths for rate measurements and kinetic determinations were selected based primarily on PAR and location of the depth of the DCM. High vertical resolution profiles with depth intervals of 5-10 m (from DCM to the depth of 1% of surface PAR) and 10-20 m (from surface to DCM and from 1% of surface PAR to 200 m) were conducted. In total, 14-20 depths were sampled for each station in the SCS.
Samples for chemical, biological, and rate measurements were collected from the same cast. Triplicate 150 mL high-density polyethylene (HDPE) Nalgene bottles were used for nutrient collection; 250 mL polycarbonates (PC) Nalgene bottles and 250 mL HDPE Nalgene bottles were filled for NH 4 + oxidation incubations; 1 L PC Nalgene bottles were used for PON concentration measurements and for NH 4 + uptake and NO 3 − uptake incubations (carried out in duplicate for each treatment in the SCS cruise and in triplicate in the NWP cruise). All bottles and equipment were acid washed and rinsed with seawater three times prior to sample collection.
During the SCS cruise, a set of incubations including kinetics tests were carried out on board to estimate in situ rates for nitrification, NH 4 + uptake, and NO 3 − uptake rates. For the NWP cruise, incubations were carried out for nitrification and NH 4 + uptake rates. Different neutral density screens (Lee Filters) were used to simulate the in situ light intensity. Dark incubation was implemented in parallel to examine light effects and to derive daily rate estimates.
For nitrification rates, 1 mL of 15 N-NH 4 + tracer (98% of 15 N atom, Sigma-Aldrich) was injected into 250 mL samples to get a final tracer concentration of 20 nmol L −1 . Immediately after tracer injection, around 40 mL of sample was filtered through a 0.2 μm syringe filter to represent the initial condition. Paired incubations under simulated in situ light and dark conditions allowed us to properly integrate the daily nitrification rates, and were conducted on-board in on-deck circulating seawater incubators for 12 h (0700-1900 hours) and stopped by filtration. For NH 4 + uptake rates, 15 N−NH 4 + tracer (98% of 15 N atom, Sigma-Aldrich) was injected into 1 L samples to get a final concentration of 10 nmol L −1 . Two to three bottles were filtered (with pressure of~200 mmHg) onto pre-combusted (450°C, 4 h) 25 mm glass fiber filters (GF/F, Whatman) immediately after tracer injection to represent the initial conditions. The remaining bottles were incubated for 3 h (0700 to 1000 hours) on board under simulated in situ light and dark conditions in parallel. Similar incubations were conducted for NO 3 − uptake rates with different levels of tracer enrichment. For samples in the mixed layer, the final concentration of 15 N−NO 3 − (98% of 15 N atom, Sigma-Aldrich) was 10 nmol L −1 , and final tracer concentrations were 100 and 500 nmol L −1 for samples from the bottom of mixed layer to the DCM and below the DCM, respectively. Incubations were terminated by gentle filtration.
To investigate the dependence of nitrification and NH 4 + uptake rates on substrate concentrations, samples from selected depths were incubated at four different levels of 15 NH 4 + addition spanning from 10 to 2000 nmol L −1 during the SCS cruise. At each station, three depths (15 m, depth of DCM, and depth of 1% PAR) were sampled to conduct Michaelis−Menten kinetic tests. For each set, tracer was added separately into duplicate samples, and all the samples were simultaneously incubated using the same procedures described above.
Chemical measurements. Oxygen concentrations were measured using an SBE oxygen sensor, calibrated against discrete sample Winkler titrations. NH 4 + concentrations were measured on board using a fluorometric method with detection limit of 1.2 nmol L −1 and precision of ±3.5% 57 . Nutrient concentrations below the nitracline were measured using a Four-channel Continuous Flow Technicon AA3 Auto-Analyzer. The detection limits for NOx (NO 3 − + NO 2 − ), SRP (soluble reactive phosphate), and Si(OH) 4 were 0.03, 0.03, 0.05 μmol L −1 , respectively, with precision better than 1, 2, 2.8%, respectively 58 . For surface samples above the nitracline, NO 3 − and NO 2 − concentrations were determined by the standard colorimetric method coupled with a Flow Injection Analysis-Liquid Waveguide Capillary Cell system (World precision Instruments) 59 ; the detection limit was 5 nmol L −1 and precision was better than 3.1%.
Isotopic analyses of NO X − and PON. δ 15 N of NO X − (NO 2 − + NO 3 − ) were determined using the bacterial method 60, 61 with minor modifications, using a Thermo Finnigan Gasbench system with cryogenic extraction and purification system interfaced to a Delta V PLUS isotopic ratio mass spectrometer. Briefly, NO X − was quantitatively converted to N 2 O by using the bacterial strain Pseudomonas aureofaciens (ATTC no. 13985). The N 2 O was then introduced to the GC-IRMS through an online N 2 O cryogenic extraction and purification system. δ 15 N of NO X − values were calibrated against nitrate isotope standards USGS 34, IAEA N3 and USGS 32, which were run before, after, and at ten sample intervals. Accuracy (pooled standard deviation) was better than ±0.2‰ according to analyses of these standards at an injection level of 20 nmol N over the past 3 years. Quality control was also conducted by analyzing laboratory working reference material (3000 m deep sea water from the South China Sea). For samples with NO X − concentrations lower than 0.5 μmol L −1 , 1 mL of 5 μmol L −1 of in-house NO 3 − standard was added as carrier to 9 mL of samples, and the isotopic composition of the sample was than calculated from the measured composition of the mixture and the known in-house standard via mass conservation.
For the particulate samples, the wet digestion method was applied prior to the bacterial method according to previous studies 62 with slight modification. In brief, the PON (with filters) was oxidized to NO 3 − by using 1 mL of purified persulfate oxidizing reagent (POR) and 4 mL of deionized water (DIW) in a 12 mL 450°C precombusted boro-silicate glass tube. The persulfate (ACS-grade, Merck, German) was recrystallized at least three times and then made up to alkaline POR by dissolving 6 g K 2 S 2 O 8 and 6 g NaOH (ACS-grade, Merck, German) in DIW to final volume of 100 mL. Before the use of POR, the residual NO 3 − concentration in the initial POR (referred as POR blank) was measured and ensured to be sufficiently low (<2 μmol L −1 in digested solution, see below). Screw caps were closed tightly once the POR was injected into the samples, which were then autoclaved for 1 h under 120°C. At least five tubes went through the same procedure in parallel with solely POR to determine the δ 15 N of the post-digestion POR blank. In addition, ten unused filters were randomly selected after each cruise to estimate the concentration and δ 15 N of the background PON, i.e. the filter blanks. After the digested solutions cooled down to room temperature, pH was adjusted to neutral with 6 N HCl (ACS-grade, Merck).
The NO 3 − concentration after digestion was measured by chemiluminescene 63 . By injecting 10-1000 μL of sample into a 95°C acidic solution of vanadium (V (III)), a sufficient amount of NO 3 − was reduced to NO gas for detection. δ 15 N values of the PON-derived NO 3 − were determined using the bacterial method as described above. Calculation of the concentration and δ 15 N of the PON included separate correction for both the blank of the POR procedure and the filters. The PON blank in each tube was typically less than 1% of the total N and the blank of the filter was less than 5 nmol N (1.88 ± 0.85 nmol N after the SCS cruise, and 3.99 ± 1.49 nmol N after the NWP cruise), which accounted for less than 3% of the N content in our samples.
Rate calibration. Nitrification and uptake rates were primary determined by the accumulation of 15 N in the product pool relative to the initial conditions. The equations for the rate calculations have been previously reviewed and compared 64 . Equation 1 quantifies the transformation rate under bulk substrate concentration (ambient substrate + tracer). The 10% tracer enrichment principle to minimize tracer perturbation is impractical in oligotrophic oceans with extremely low NH 4 + concentrations 65 . To date, a wide range (10-750 nmol L −1 ) of tracer concentrations has been adopted in different studies 4-10, 24, 25, 30, 66, 67 , and given that NH 4 + concentrations are low across most of global ocean, has likely caused overestimation of measured rates to a certain degree. Similar to other studies, the tracer addition in this study (even though the amount was as low as 10-20 nmol L −1 ) represents a significant substrate enrichment relative to ambient concentrations. Thus, the tracer addition potentially stimulates the in situ rate (Eq. 1).
A few attempts to overcome the enhancement of in situ rate by tracer enrichment have been carried out in the field. By using a kinetic approach (Eq. 2), for instance, Horak et al. 25  + into samples at ambient concentrations of 5-650 nmol L −1 . This conclusion was drawn by assuming a constant K s over depth and among sites for NH 4 + oxidation; yet, K s may vary with environmental conditions such as temperature and pH. Although the kinetic approach has been suggested as an effective method with better accuracy in calibrating in situ rates, this method has been less commonly conducted due to its labor-consuming requirements.
Instead, because the reaction rate showed a near first-order response when substrate concentration was lower than K s , we applied a linear regression approach by using Eqs. 3-5 with the following assumptions to obtain the optimal in situ reaction rates: first, the 15 N dilution effect due to regeneration is insignificant due to relatively short-term incubations, and the isotopic fractionation during N transformation can be ignored because of the high 15 N enrichment in the substrate after tracer addition, and thus the at% 15 N of substrate remained constant during the incubation. This is also justified by a recent model study that showed the dilution effect is not significant within 12 h in oligotrophic conditions 68 ; second, the at% of 15 N of in situ NH 4 + was assumed to be 0.3663%.
In the above equations, R bulk is the bulk reaction rate for all substrates after tracer enrichment (nmol N L −1 h −1 ). R kinetic is the rate of kinetic response. R 15 , R 14 are the accumulation rates of atom 15 N and 14 N, respectively, in the product pool. R in situ is the in situ reaction rate calibrated by linear interpolation. C t and C 0 are product concentrations at the ending and beginning of incubation (nmol N L −1 ), respectively. f 15 is at% 15 N of substrate pool at the beginning of incubation; n t and n 0 are the at% 15 N of the product pool at the ending and beginning of incubation (%), respectively. t is the duration of incubation (h). n NH þ 4 is the at% 15 N of substrate after tracer injection. Ci NH þ 4 and Ct NH þ 4 are the initial substrate concentration and final tracer concentration, respectively.
In addition, instead of assuming a constant product concentration during incubation as in most of previous studies, all particle samples at different sampling times were measured directly. This concentration measurement was necessary for uptake rate quantification, because phytoplankton simultaneous uptake of different species of nitrogen may induce considerable increases in PON concentrations during incubation. For NH 4 + oxidation, the calculated rate was converted to the daily-integrated rate by assuming equal periods of day and night rates, i.e. by summing the dark and light hourly rates multiplied by 12 (R in situ of light incubation × 12 + R in situ of dark incubation × 12); daily-integrated NH 4 + and NO 3 − uptake rate was approximately estimated by multiplying the hourly rates by 16 according to previous tests at the Bermuda Atlantic Time Series station (BATS) 5 .
In order to test the validity of the linear regression approach presented here, we compared the R bulk , R in situ and R kinetic derived from the above equations ( Supplementary Fig. 10) by using the data obtained from kinetic experiments. The NH 4 + concentrations of 21-94 nmol L −1 (in situ NH 4 + + tracer) were lower than the derived K s values in this study, allowing the application of the proposed linear regression method for in situ rate calibration. Our results showed much higher values of R bulk relative to R kinetic , demonstrating significant overestimation of reaction rates by tracer addition. The reaction rates were reduced by 15.2-78.4% by using the kinetic correction. Larger deviations were found at depths with low in situ concentrations. By contrast, the R in situ derived from the linear regression was close to that of R kinetic , demonstrating the usefulness of linear interpolation for in situ rates determination. Further, given the persistent low NH 4 + concentration in the global oligotrophic ocean, this method offers a convenient approach to correct reaction rates derived from tracer incubations.
Detection limit of rate measurement. The detection limit depends on the concentration of the product pool and the fraction of the 15 N in the substrate pool during the incubation 24, 66, 67 . As mentioned, the accuracy of δ 15 -NO 3 − was better than ±0.2‰ in our lab, and we here use three times the standard deviation (0.6‰) as a reliable enrichment of 15 N in the product pool. Therefore, we calculated a detection limit of 0.04-0.16 nmol N L −1 d −1 for nitrification and 0.01-0.04 nmol N L −1 d −1 for NH 4 + uptake for our samples.
Data analysis. The kinetic parameters (V max and K s ) were estimated using Eq. 2 (and Grapher version 10; Golden Software). The α value, which represents a better indicator of advantage in substrate competition 12,17 , was calculated by using Eq. 6.
The comparisons of reaction rates between paired light and dark incubations, and the kinetic parameters were examined by using the Student's t test and SPSS (IBM, version 19).