Introduction

Nitrogen has a central role in biogeochemical cycles in the ocean because it generally limits marine biological production. New production, which is defined as production based on nitrogenous nutrients newly delivered from outside the productive layer (Dugdale and Goering, 1967), is balanced by the sinking particle flux in a steady-state system (Eppley and Peterson, 1979) and thus is used to evaluate the capacity of the biological pump.

The nitrate assimilation rate has been examined to determine new production because nitrate is considered the major allochthonous source of nitrogen, supplied to the euphotic zone from deep water (Falkowski et al., 2003). Nitrification has been ignored because it is inhibited by sunlight (Olson, 1981; Guerrero and Jones, 1996). However, nitrification in the euphotic zone has been evaluated at the same time and has been found to be detectable (Ward et al., 1989; Yool et al., 2007; Clark et al., 2008; Raimbault and Garcia, 2008; Beman et al., 2012). These findings suggested that nitrate-based new production would be overestimated because of the occurrence of nitrification in the euphotic zone. Such surface nitrification is posing a challenge to our current understanding of the surface nitrogen cycle in the ocean. The environmental factors controlling nitrification are still poorly understood; hence, there is large uncertainty in the estimation of nitrification and new production.

Nitrification has been considered to be a two-step process consisting of ammonia oxidation (ammonium to nitrite) followed by nitrite oxidation (nitrite to nitrate), and each step is carried out by different microbes. (Although very recent studies show that some bacteria belonging to genus Nitrospira perform both steps (Daims et al., 2015; van Kessel et al., 2015), the present study does not consider this process. The marker gene of the complete ammonia oxidizer is not found in the ocean in public databases (Daims et al., 2015)). The first step of nitrification is the rate-limiting process, and it was formerly thought to be performed almost entirely by ammonia-oxidizing bacteria (AOB). Therefore, the limiting factors of nitrification in the ocean were thought to be tightly coupled with the physiological characteristics of AOB (Ward, 2002). However, ammonia-oxidizing archaea (AOA) belonging to the phylum Thaumarchaeota were discovered in the 2000s (Könneke et al., 2005; Brochier-Armanet et al., 2008), and subsequent studies found that AOA outnumber AOB in various regions, indicating that AOA could be a major ammonia-oxidizing organism (AOO) in the ocean (Wuchter et al., 2006; Beman et al., 2008, 2012; Church et al., 2010). As an explanation for the abundant AOA, Martens-Habbena et al. (2009) suggested that the higher ammonium affinity of AOA compared with AOB could provide an advantage to AOA because ammonium concentration is typically low throughout most of the ocean. Few studies have investigated the transcript abundance of AOA in the ocean (Church et al., 2010; Horak et al., 2013; Urakawa et al., 2014; Smith et al., 2014), and little is known about the factors limiting their ammonia oxidation activities. A genome analysis (Walker et al., 2010) demonstrated that AOA has different metabolic systems than AOB, suggesting that limiting factors for AOA could be different from those for AOB.

The regions studied in this research covered several biogeochemical provinces, from the equatorial Pacific to the Arctic Ocean; therefore, environmental conditions were expected to differ greatly among the provinces. We examined the influence of nitrification on the nitrogen cycle in the euphotic zone and the factors controlling ammonia oxidation via large-scale observations. We investigated ammonia oxidation and the nitrate assimilation rate simultaneously. Furthermore, we examined the kinetics of ammonium utilization for ammonia oxidation in each oceanic region and determined the ammonium concentration at the nanomolar level using supersensitive colorimetric methods onboard a vessel (Hashihama et al., 2015; Kodama et al., 2015). We also determined the ammonia monooxygenase subunit A (amoA) gene and the transcript abundances of AOB and shallow and deep clades of AOA, for which the affinity to ammonium and the distribution are considered to be different.

Materials and methods

Sampling was performed onboard the R/V Hakuho-Maru (KH-14-3) from 23 June to 11 August 2014 in the North Pacific Ocean and the Arctic Ocean (Figure 1). A depth profile of light intensity was determined just prior to each water sampling using a Hyper Profiler (Satlantic LP, Halifax, NS, Canada) at all stations except Sts.1, 2 and 3; at these stations, observations were carried out only at night due to time constraints and thus the light profile was assumed to be the same as that estimated in the same region and during the same season during the KH-12-3 cruise. Temperature, salinity and dissolved oxygen profiles were measured using a SBE 911 plus CTD system (Sea-Bird Electronics, Inc., Washington, DC, USA). Water samples were collected by an acid-cleaned bucket from the surface and by acid-cleaned Teflon-coated 12 l Niskin-X bottles from other depths. Samples for nutrients and chlorophyll a were collected from depths corresponding to 100%, 25%, 10%, 1%, 0.3% and 0.1% of the surface light intensity and from the depth of 200 m. In addition, samples for chlorophyll a were taken from depths of 5, 10, 20, 30, 50, 75, 100, 125, 150, 250 and 300 m. Samples for experiments of primary production and nitrate assimilation were collected from the five light depths corresponding to 100%, 25%, 10%, 1% and 0.1% of the surface light intensity. Samples for the experiments of ammonia oxidation and DNA analyses were collected from the depths corresponding to 100%, 10%, 1%, 0.3% and 0.1% of the surface light intensity and from the depth of 200 m. Samples for RNA analyses were taken only from the 0.1% light depth. As primary production generally increases near the surface, samples for the primary production and nitrate assimilation experiment were mainly collected from shallow depths. In contrast, maximum ammonia oxidation occurs around the bottom of the euphotic zone (Beman et al., 2008, 2012), and thus, samples for ammonia oxidation experiment were mainly collected from deeper water.

Figure 1
figure 1

Sampling stations in the North Pacific Ocean and Arctic Ocean during the KH-14-3 cruise. The background contour represents satellite-derived chlorophyll a during the study period. Black lines denote 200-m isobaths. Gray areas in the ocean indicate no data.

Nutrients and chlorophyll a

Samples for nutrients analysis were collected in single acid-cleaned 30-ml polypropylene tubes and in duplicate 10-ml acrylic tubes. The samples in the 10-ml acrylic tubes were kept frozen until analysis onshore. At Sts.1–6, which correspond with the leg 1 cruise of KH-14-3, the concentrations of nitrate, nitrite, ammonium and phosphate were determined immediately onboard at the nanomolar level using supersensitive colorimetric systems (Hashihama et al., 2009, 2015). The detection limits were 3, 2, 4 and 3 nm, respectively. At Sts. 7–15, which correspond with the leg 2 cruise of KH-14-3, only ammonium concentration was determined onboard using a highly sensitive and large concentration range colorimetric analysis (Kodama et al., 2015). The detection limit of this analysis was 6 nm. The concentrations of nitrate, nitrite and phosphate at Sts. 7–15 were determined at the nanomolar level on land using the supersensitive colorimetric systems (Hashihama et al., 2009). When the concentrations of nitrate, nitrite and phosphate were higher than 1 μm, they were determined using an AACSII auto-analyzer.

Samples for chlorophyll a were filtered using 25 mm Whatman GF/F filters, and the chlorophyll a concentrations were determined onboard using a Turner Design 10-AU fluorometer after extraction with N, N-dimethylformamide.

Ammonia oxidation

Samples for the ammonia oxidation experiment were collected in duplicate in acid-cleaned 0.3 l polycarbonate (PC) bottles. 15N-labeled ammonium sulfate (99 atom% 15N; SI Science) was spiked to give a final tracer concentration of 31 nm. In addition, to examine the kinetics of ammonia oxidation, the 15N-labeled ammonium sulfate was spiked to the samples collected from the 0.1% light depth to adjust the final tracer concentrations to 31, 62, 234 and 1560 nm. The atom% 15N and ammonia oxidation rate in each incubation bottle are listed in Supplementary Table S1. The samples from 200 m were covered with black screen and those from the other depths were covered with neutral-density screens to adjust the light levels (10%, 1%, 0.3% and 0.1%). Those bottles were incubated in a thermostatic incubator (CN-25B, MEE) whose temperature was adjusted to the sea temperature of the 0.3% light depth under 150 μmol photons m−2 s−1 light intensity. The light and dark cycle was adjusted to the cycle at each station. Samples collected from the surface water were incubated without a screen in the on-deck incubator. After 24 h incubation, the filtrate passing through a 0.2 μm pore size cellulose acetate in-line filter (DISMIC, ADVANTEC MFS, Inc., Tokyo, Japan) was collected in 50-ml polypropylene bottles and in 10-ml acrylic tubes and was kept frozen until analysis. Analysis of the δ15N of nitrate+nitrite in the filtrate was performed using the denitrifier method (Sigman et al., 2001; Isobe et al., 2011). In brief, 10-ml filtrate was dispensed into a 50-ml vial, purged for 10 min with ultrapure He, and inoculated with 2 ml concentrated medium of nitrate-starved Pseudomonas chlororaphis subsp. aureofaciens ATCC13985, which was also purged for 2 h with ultrapure He. The reaction was terminated by adding 1 ml 8 m NaOH after 24 h incubation to allow complete conversion of nitrate+nitrite into N2O. The N2O that was produced was injected into a GasBench II equipped with a cold trap system for concentration and purification and then introduced into a DELTAplus XP isotope ratio mass spectrometer. The s.d.s of the δ15N values measured for the 1000 and 100 nm nitrate standards were 0.49‰ and 2.12‰, respectively. The concentration of nitrate+nitrite in the filtrate was determined by colorimetry using an AACSII auto-analyzer (Bran+Luebbe, Norderstedt, Germany). Ammonia oxidation rates were calculated using the equation given by Beman et al. (2012).

We defined detectable activity as when the difference between δ15N-nitrate in the incubated and initial samples was more than threefold the s.d. of the nitrate standard. Hence, when the nitrate concentration was >1000 nm, we considered that ammonia oxidation was detectable when the difference was >1.47‰. When the nitrate concentration was 100–1000 nm, the difference was set >6.36‰. Furthermore, we regarded ammonia oxidation to be below the detection limit when the nitrate concentration was <100 nm. At St. 15, the ambient ammonium concentration at the 0.1% light depth was too high (5540 nm) to detect ammonia oxidation at the final tracer concentrations of 31 and 62 nm. Meanwhile, the ammonia oxidation rates for 234 and 1560 nm were 5.01 and 5.16 nmolN l−1 d−1, respectively, and the ammonia oxidation at that depth was taken as the mean of these values.

Primary production and nitrate assimilation rates

Samples for the primary production experiment were collected in duplicate in acid-cleaned 4.5 l PC bottles. Primary production was measured with nitrogen fixation using a dual (15N and 13C) isotopic technique (nitrogen fixation data are not shown here). 13C-labeled sodium bicarbonate (99 atom% 13C; Cambridge Isotope Laboratories, Inc., Andover, MA, USA) was added to each bottle at a final tracer concentration of 200 μmol l−1. After 15N2-dissolved water was added to the bottle, it was sealed with a thermoplastic elastomer cap. Samples for the nitrate assimilation experiment were collected in acid-cleaned 2 l PC bottles. At Sts. 2–8, 14 and 15, nitrate assimilation was estimated by the Michaelis–Menten kinetics approach to correct the overestimation caused by the excessive use of the 15N tracer (Kanda et al., 2003; Shiozaki et al., 2009); 15N-labeled nitrate (99 atom% 15N; SI Science) was added to each bottle to adjust the final tracer concentrations to 10, 22, 109 and 2170 nm. The enrichment of 15N-labeled nitrate at the other stations was: 109 nm at St.1, 217 nm at St. 9–11, 1090 nm at St.12 and 435 nm at St. 13. The atom% 15N and nitrate assimilation rates in each incubation bottle are listed in Supplementary Table S2. The bottles for the primary production and the nitrate assimilation experiments were covered with neutral-density screen to adjust the light levels (100%, 25%, 10%, 1% and 0.1%) and then incubated in an on-deck incubator cooled by flowing surface seawater. Where the bottles for primary production were incubated for 24 h, those for nitrate assimilation were incubated for 2–3 h during daytime. At Sts. 3, 8 and 11, samples for nitrate assimilation were collected from the water column during both daytime and nighttime and were immediately incubated with 15N-labeled nitrate. At the other stations, samples for night-time incubation were collected from the surface, incubated without 15N tracer in the on-deck incubator until midnight, and then 15N-labeled nitrate was added. The night-time incubations were also performed for 2–3 h. Although ammonia oxidation is less susceptible to temperature (Horak et al., 2013; Baer et al., 2014), primary production and nitrate assimilation are significantly influenced by temperature (Harrison et al., 1996; Behrenfeld and Falkowski, 1997). Although these properties are not linearly related to temperature and the deviation from the actual value is difficult to estimate, there is the possibility of bias in these rates measured using the on-deck incubations. Incubation was terminated by gentle vacuum filtration of the seawater samples through precombusted GF/F filters. Analyses and calculations were performed as described previously (Shiozaki et al., 2009, 2011). The s.d. of the δ15N values of a working standard (l-alanine) was <0.3‰.

DNA and RNA sampling and extraction

Samples (2 l) for RNA and DNA analyses were filtered onto Sterivex-GP pressure filter units with a 0.22 μm pore size (Millipore, Billerica, MA, USA). RNA samples were filtered within 30 min of the water sampling and then added to RNAlater Stabilization Solution (Life Technologies, Carlsbad, CA, USA). The Sterivex filter units were frozen at −80 °C until onshore analyses. Total DNA was extracted using a ChargeSwitch Forensic DNA Purification Kit (Invitrogen, Carlsbad, CA, USA). For RNA extraction, a mirVana miRNA Isolation Kit (Life Technologies) was used after the RNAlater solution in the Sterivex filter units was removed. Then the extracted RNA was treated with the Turbo DNA Free Kit (Ambion, Austin, MD, USA) to remove contaminating DNA. The concentration of the purified RNA was measured using Nano Drop 2000 (Thermo Scientific, Waltham, MA, USA). Complementary DNA synthesis was performed using Transcriptor First Strand cDNA Synthesis Kit (Roche Applied Science, Mannheim, Germany) with random hexamer as the primer. Sample treatment using the kits followed the manufacturer’s instructions.

Quantitative polymerase chain reaction analyses

AOA have been classified into three major groups in the ocean based on their amoA gene sequences: the Nitrosopumilus maritimus-like cluster, water column cluster A and water column cluster B (Francis et al., 2005; Beman et al., 2008). As the water column cluster A is dominant in surface water, this cluster is also called the shallow clade (Hallam et al., 2006). Based on the AmoA sequences, the Nitrosopumilus maritimus-like cluster is also included in the shallow clade (Hallam et al., 2006; Beman et al., 2008). In contrast, water column cluster B is found only in deep water, and hence, is called the deep clade (Hallam et al., 2006). The primer set Arch-amoAFA & Arch-amoAR that we used in this study covers both water column cluster A and the Nitrosopumilus maritimus-like cluster, namely the shallow clade (Beman et al., 2008). Further, the primer set Arch-amoAFB & Arch-amoAR covers water column cluster B, namely the deep clade (Beman et al., 2008). To avoid confusion of terminology, we denoted the quantified AOA using Arch-amoAFA & Arch-amoAR and Arch-amoAFB & Arch-amoAR as the shallow and deep clades, respectively. The primer sets that we used did not target the water column cluster C and soil assemblages, which are minor groups in the marine water column (Francis et al., 2005). Therefore, we might miss these contributions if these groups dominated in the samples.

AOB have been classified into two groups: Betaproteobacteria and Gammaproteobacteria. We used the primer sets amoA-1F & amoA-r-NEW for Betaproteobacteria (hereafter βAOB) (Hornek et al., 2006) and amoA-3F & amoA-4R for Gammaproteobacteria (Purkhold et al., 2000). When we applied quantitative PCR assay to gammaproteobacterial amoA gene, non-specific bands were detected from most of the PCR products. Therefore, we only quantified βAOB.

The quantitative PCR assays were performed on a LightCycler 480 Real-Time PCR System (Roche Applied Science). The reaction mixtures (20 μl) contained 10 μl SYBR Premix Ex Taq (TaKaRa, Shiga, Japan), 0.2 μm each primer and 1 μl DNA. The assays were run in triplicate under the following cycling conditions: 95 °C for 30 s, followed by 45 cycles of 95 °C for 15 s; 56 °C (shallow clade AOA), 55 °C (deep clade AOA) or 55 °C (βAOB) for 30 s; and 72 °C for 30 s; with a detection step at the end of each cycle. The primer (and relevant references), quantitative PCR standards, standard curve correlation coefficients, PCR efficiencies and detection limits are listed in Supplementary Table S3.

Results

Biogeochemical provinces and environmental conditions

We classified oceanic regions according to temperature, salinity, surface chlorophyll a concentration and geographical location (Longhurst, 2007) (Figures 2a–c and Supplementary Figure S1). High surface chlorophyll a at St. 1 indicated that this station was located in the Pacific Equatorial Divergence Province (PEQD). Surface salinity at St. 2 was lower than at surrounding stations, suggesting that this station was in the North Pacific Equatorial Countercurrent Province (PNEC). The North Pacific Tropical Gyre Province (NPTG) extended from St. 3 to St. 8. Surface nitrate was depleted from St. 2 to St. 8 (Figure 2d and Table 1). Temperature and salinity at the surface changed markedly from St. 8 to St. 10, and the temperature and salinity at St. 10 were typical of values in the Pacific Subarctic Gyres Province (PSAG) in summer (Longhurst, 2007). Therefore, St. 9 was located in the North Pacific Subtropical and Polar Front Provinces (NPST/NPPF). The surface nitrate concentration at St. 9 (305 nm) was somewhat low compared with those in the PSAG (>1000 nm; Harrison et al., 2004). Although the temperature-salinity diagram indicated that St. 11 was close to St. 12 (Supplementary Figure S1), St. 11 was classified as PSAG from its geographical location. The surface nitrate concentration at Sts. 10 and 11 was very high (9950 and 7570 nm, respectively). To the north of St. 12, stations were divided into the North Pacific Epicontinental Sea Province (BERS; Sts. 12–14) and the Boreal Polar Province (BPLR; St. 15) according to the geographical location. The surface nitrate concentrations at Sts. 13 and 14 were almost depleted (<3 and 5 nm, respectively), and those at Sts. 12 and 15 were 435 and 211 nm, respectively.

Figure 2
figure 2

Vertical profiles of (a) temperature (°C), (b) salinity, (c) chlorophyll a (μg l−1), (d) nitrate (nm), (e) nitrite (nm) and (f) ammonium (nm) along the transect from the equatorial Pacific to the Arctic Ocean.

Table 1 Environmental variables at the surface (SST, chlorophyll a, and nitrate), depth-integrated biological activities (primary production, nitrate [NO3] assimilation and ammonia [NH3] oxidation) and contribution of nitrification to nitrate assimilation at each station

Nitrite had a clear subsurface maximum throughout the study regions, except at Sts. 1 and 14 (Figure 2e). The primary nitrite maximum was emerged in deep water in the subtropical region but tended to shoal to the north (Supplementary Figure S2a). The concentration at the nitrite maximum varied between 65 and 554 nm.

The subsurface ammonium maximum was not very significant from the PEQD to the PSAG (Figure 2f). In contrast, in the BERS and BPLR, the ammonium concentration clearly increased at the subsurface or near the bottom, and the maximum concentrations at Sts. 12, 13, 14 and 15 were 1380 nm (31 m), 1650 nm (38 m), 950 nm (26 m) and 5740 nm (44 m), respectively.

The subsurface chlorophyll a maximum (SCM) was developed in the study region, except in the PSAG and at St. 12 in the BERS. Although the depth of the SCM deepened in the NPTG and became shallow to the north, as for the nitrite maximum, the SCM tended to be shallower than the nitrite maximum (Supplementary Fig. S2a). At Sts. 14 and 15 in the BERS and BPLR, the SCM occurred in the deepest water, near the bottom. The surface chlorophyll a concentrations at Sts. 10 and 11 (0.25 and 0.96 μg l−1, respectively) were lower than that at St. 12 (1.28 μg l−1), whereas the surface nitrate concentrations at Sts. 10 and 11 (9950 and 7570 nm, respectively) were an order of magnitude higher than that at St. 12 (435 nm) (Table 1), suggesting that Sts. 10 and 11 were located in a high-nutrient low-chlorophyll region.

Ammonia oxidation, primary production and nitrate assimilation

Ammonia oxidation occurred at all stations except St. 14, and the peak in the water column was below the 1% light depth (Figure 3a). Ammonia oxidation at the 100% light depth was detected at only St. 1, and that at the 10% light depth was found only at Sts. 1, 9, 12 and 15. The peak of ammonia oxidation was below the SCM, except at St. 15 (Supplementary Figure S2a), and it varied from 2.87 to 67.1 nmolN l−1 d−1. The depth-integrated rate to the 0.1% light depth varied from 0.02 to 1.90 mmolN m−2 d−1 (Table 1). The depth-integrated rates in the PNEC and NPTG (0.11–0.53 mmolN m−2 d−1) were lower than those in the surrounding regions (0.79 mmolN m−2 d−1 in the PEQD and 1.71 mmolN m−2 d−1 in the NPST/NPPF). North of the NPST/NPPF, the depth-integrated value at St.10 (0.35 mmolN m−2 d−1) was lower than the value at the adjacent stations (1.08 mmolN m−2 d−1 at St. 11). The highest ammonia oxidation value (67.1 nmolN l−1 d−1) occurred at St. 12 in the BERS. Ammonia oxidation was low at Sts. 13, 14 and 15 (<5.08 nmolN l−1 d−1). The kinetics experiment at the 0.1% light depth demonstrated that substantial elevation of ammonia oxidation with addition of 15N-labeled ammonium was observed only at St. 12 among Sts. 1–12 (Figure 4a). At St. 8, ammonia oxidation was slightly elevated with the addition of the 15N tracer.

Figure 3
figure 3

Vertical profiles of (a) ammonia oxidation (nmolN l−1 d−1), (b) primary production (nmolC l−1 d−1) and (c) nitrate assimilation (nmolN l−1 d−1) along the transect from the equatorial Pacific to the Arctic Ocean. Dashed lines indicate the 1% light depth (green), the 0.1% light depth (blue) and the nitracline depth (red), respectively.

Figure 4
figure 4

Selected results of the kinetics experiments for (a) ammonia oxidation (Sts. 1, 2, 3, 8, 9, 10 and 12) and (b) nitrate assimilation (Sts. 2, 3, 8 and 14). Each station was a representative in each region. The horizontal axes indicate total substrate concentration (15N tracer addition+ambient substrate concentration).

The maximum primary production generally occurred at the surface, except at Sts. 3, 14 and 15, and ranged between 255 and 8270 nmolC l−1 d−1. Primary production was detected but was low at the 0.1% light depth (3–46 nmolC l−1 d−1), except at St. 14 (505 nmolC l−1 d−1) (Figure 3b). The euphotic zone is conventionally defined from the surface down to the 1% light depth, below which no appreciable photosynthesis can occur (Ryther, 1956). In this study, according to the results of primary production, the euphotic zone was defined down to the 0.1% light depth. Therefore, unless otherwise noted, productivity per area is integrated from the surface to the 0.1% light depth.

Nitrate assimilation was always stimulated by the addition of nitrate, ranging from 10 to 2170 nm in the depths where ambient nitrate concentration was almost depleted (31 nm; Figure 4b). Nitrate assimilation in the study regions showed a maximum at the surface in the PEQD, NPST/NPPF, PSAG, and at Sts. 12 and 13 in the BERS. (Figure 3c). In the PNEC and NPTG, the maximum nitrate assimilation occurred at the 0.1% light depth, which was always placed below the nitracline. The maximum values and depth-integrated values varied as 5.30–2280 nmolN l−1d−1 and 0.20–26.1 mmolN m−2 d−1 (Table 1), respectively. At St. 14, although ambient nitrate concentration was low (<500 nm) throughout the entire water column, the maximum nitrate assimilation (1980 nmolN l−1 d−1) was the highest value in the study regions, suggesting that supplied nitrate was rapidly consumed by microbes at this site. The depth-integrated nitrate assimilation was positively related with nitrate concentration, except for the data collected at St. 14 (r=0.72, P<0.05) (Figure 5a), and was positively correlated with primary production throughout the whole region (r=0.91, P<0.05) (Figure 5b).

Figure 5
figure 5

Relationships between (a) depth-integrated nitrate and nitrate assimilation, (b) depth-integrated primary production and nitrate assimilation and (c) depth-integrated primary production and ammonia oxidation. Open circles denote data excluded from the regression analysis (see text).

The ratio of ammonia oxidation to nitrate assimilation was significantly higher in the PNEC and NPTG (average 55.6%) than in the other regions (Table 1). The highest ratio (87.4%) was observed at St. 4.

amoA abundance and expression

Shallow clade AOA were detected at all stations except St. 14 and had a subsurface maximum within the depth of 200 m (Figure 6a and Supplementary Figure S2b). At St. 7, the maximum occurred at 200 m. Shallow clade AOA were not observed in the surface water at Sts. 5–14. The abundance of shallow clade AOA varied from 590 to 33 × 106 copies l−1 at all stations, and at Sts. 13–15, the values (from below detection limit to 0.0043 × 106 copies l−1) tended to be lower than those in the other regions. The maximum in the PNEC and NPTG (average 7.95 (±4.57) × 106 copies l−1) was lower than those in the neighboring regions: the PEQD (12.5 × 106 copies l−1) and the NPST/NPPF (16.5 × 106 copies l−1). The depth of the subsurface maximum did not correspond with that of ammonia oxidation at most stations (Supplementary Figure S2b).

Figure 6
figure 6

Vertical profiles of amoA abundance (Log10(copies l−1)) of (a) shallow clade AOA, (b) deep clade AOA, (c) βAOB and (d) the ratio of shallow and deep clade AOA to βAOB (AOA:βAOB).

Deep clade AOA were detected in the study region, except at Sts. 13–15, and occurred only below the 1% light depth (Figure 6b). The maximum abundance of deep clade AOA was always placed at 200 m at the stations (Supplementary Figure S2b), ranged between 0.36 × 106 and 12.8 × 106 copies l−1, and there was no apparent difference among the study regions. The abundance of deep clade AOA exceeded that of shallow clade AOA only at the 200 m depth of Sts. 1–3.

βAOB were detected at the stations, except for Sts. 2–5 and 14 (Figure 6c). They were observed at the surface only at Sts. 8 and 10. The vertical distribution of βAOB showed a subsurface maximum at the each station, and at Sts. 7 and 8, the maximum occurred at 200 m (Supplementary Figure S2b). The maximum βAOB abundance ranged between 0.0024 × 106 and 1.31 × 106 copies l−1. The maximum depth did not match the depth of maximum ammonia oxidation at most stations. North of the NPST/NPPF, the abundance of βAOB was higher than that of deep clade AOA at the depth of the βAOB maximum. The abundance of βAOB was generally one order of magnitude lower than that of shallow clade AOA, except at St. 13 where their abundances were comparable and the ratio of shallow clade AOA to βAOB at the depth of the βAOB maximum was 1.43. The ratio of AOA to βAOB tended to be low in ammonium-rich water (Figure 6d).

Except at St. 14, amoA expression was detected at all stations, and that of shallow clade AOA always exceeded that of deep clade AOA and βAOB (Figure 7a). The amoA expression of shallow clade AOA varied from 0.004 × 106 to 0.33 × 106 transcripts l−1 at Sts. 1–7, and it increased from St. 8 (1.05 × 106 transcripts l−1) to St. 12 (3.94 × 106 transcripts l−1), except at St. 10 where it decreased to 0.22 × 106 transcripts l−1. At Sts. 13 and 15, the amoA expression of shallow clade AOA was low at 0.099 × 106 and 0.013 × 106 transcripts l−1, respectively. Although the amoA gene of deep clade AOA was detected at Sts. 1–11, expression was found only at Sts. 1, 6, 7, 9 and 11, with a value of 0.018–0.13 × 106 transcripts l−1. βAOB amoA expression was observed only at Sts. 9, 11, 12 and 13, where it varied from 0.0065 × 106 to 0.087 × 106 transcripts l−1. At St. 13, the amoA expression of βAOB (0.087 × 106 transcripts l−1) was comparable to that of shallow clade AOA (0.099 × 106 transcripts l−1).

Figure 7
figure 7

(a) Latitudinal variation in amoA expression (Log10 (transcripts l−1)) of shallow clade AOA, deep clade AOA, and βAOB at the 0.1% light depth (see dashed line in Figure 3b). Error bars indicate s.d.s of triplicate qPCR analyses. (b, c) Relationships at the 0.1% light depth between (b) amoA abundance of shallow clade AOA, deep clade AOA, and βAOB and ammonia oxidation and (c) amoA expression of shallow clade AOA, deep clade AOA and βAOB and ammonia oxidation.

At the 0.1% light depth, the amoA gene abundance of shallow clade AOA was positively but weakly related with ammonia oxidation (ρ=0.63, P<0.05, Spearman rank correlation) (Figure 7b). In addition, the abundance of βAOB was positively related with ammonia oxidation (ρ=0.56, P<0.05) (Figure 7b). There was no relationship between the abundance of deep clade AOA and ammonia oxidation at that depth (ρ=0.11, P>0.05) (Figure 7b). The amoA expression of shallow clade AOA was positively and strongly related with ammonia oxidation (ρ=0.79, P<0.001; Figure 7c). The amoA expressions of deep clade AOA and βAOB were not correlated with ammonia oxidation (ρ=0.19 and ρ=0.29, respectively, P>0.05) (Figure 7c).

Discussion

Distribution of ammonia oxidation and AOO and influence of nitrification on nitrate assimilation

Although studies on ammonia oxidation in the open ocean remain limited, our measurements obtained from various environments are in accordance with those of previous studies. In the PEQD, Raimbault et al. (1999) reported that the maximum ammonia oxidation ranged between ca. 20 and 40 nmolN l−1 d−1, and our estimate (24.5 nmolN l−1 d−1) was within this range. Ammonia oxidation in the subtropical oligotrophic region of the Atlantic Ocean was found to vary from 1 to 10 nmolN l−1 d−1 (Clark et al., 2008; Newell et al. 2013), which is similar to the values we found (1.53–24.4 nmolN l−1 d−1). Ammonia oxidation determined by the 15N tracer method has not been reported in the subarctic open ocean or the BERS. In the Chukchi Sea, ammonia oxidation is recognized to be low in summer (0.2–0.4 nmolN l−1 h−1 (=4.8–9.6 nmolN l−1 d−1)) (Baer et al., 2014). We also observed low rates of ammonia oxidation in the BPLR (5.08 nmolN l−1 d−1).

In 15N tracer experiments for ammonia oxidation in the open ocean, 15N-labeled ammonium is recognized to substantially increase the substrate concentration and may enhance ammonia oxidation (Clark et al., 2008; Horak et al., 2013; Newell et al., 2013). Kinetics experiments performed in previous studies demonstrated that ammonia oxidation was stimulated with increasing concentration of 15N-labeled ammonium (Horak et al., 2013; Newell et al., 2013). In our study, at Sts. 1–12, the substrate concentrations increased from 1.9 to 5.4 times by the addition of 31 nm 15N-labeled ammonium. However, in the kinetic experiment, significant enhancement of ammonia oxidation was observed only at St. 12. These results suggest that ammonia oxidation reached a plateau below the minimum substrate concentration (31 nm added 15N-labeled ammonium+ambient ammonium). Newell et al. (2013) reported a substantial increase in ammonia oxidation after the addition of 1.8–12 nm 15N tracer. Hence, except at St. 12, AOO communities could have a high affinity for ammonia. In addition, the absence of a difference in ammonia oxidation after the addition of an excess amount of 15N-labeled ammonium suggests that other factors could have influenced the AOO activity.

The maximum depth of ammonia oxidation generally occurred between the 1% and the 0.1% light depth and was placed below the SCM except at Sts. 13–15. Similar vertical profiles have been observed in previous studies of the South Pacific Ocean (Raimbault and Garcia, 2008; Raimbault et al., 2008). AOA and AOB are both regulated in their activity by light (Olson, 1981; Guerrero and Jones, 1996; Merbt et al., 2012). In contrast, phytoplankton has an extremely high ammonium affinity in surface water (Harrison et al., 1996), and the ammonium assimilation activity decreases with light degradation (MacIsaac and Dugdale, 1972). Therefore, the difference between the depth of the SCM and the depth of the ammonia oxidation maximum is ascribable to the light environment and ammonium resource competition (c.f. Ward, 1985). In our study, the ammonia oxidation maximum occurred around the nitrite maximum, which corresponds with the results of previous reports (Raimbault and Garcia, 2008; Raimbault et al., 2008; Beman et al., 2008, 2012; Newell et al., 2013; Santoro et al., 2013). Although the formation of the nitrite maximum is considered to be due to ammonia oxidation (Santoro et al., 2013) and/or the activity of phytoplankton (Lomas and Lipschultz, 2006), the data set obtained did not clarify which factor was essential.

The maximum abundance of shallow clade AOA was located above the depth of the maximum abundance of deep clade AOA throughout the study regions, the same trend reported in previous studies (Beman et al., 2008; Smith et al., 2014). At all stations, the amoA gene abundance of shallow clade AOA dominated in the AOO to the depth of 200 m, and the expression of shallow clade AOA at the 0.1% light depth exceeded those of deep clade AOA and βAOB, suggesting that shallow clade AOA was the major AOO in this study region. This inference was also supported by the fact that the amoA gene abundance and expression of shallow clade AOA were both positively correlated with ammonia oxidation. This relationship was also observed in a previous study (Smith et al., 2014). It should be noted that variation in amoA expression of shallow clade AOA was not necessarily linked with gene abundance. For example, amoA gene abundance of shallow clade AOA at the 0.1% light depth at St.10 was comparable to that at St. 9, but the gene expression at St. 10 was about one-tenth that at St. 9. Furthermore, although the amoA gene of deep clade AOA (βAOB) was detected at the 0.1% light depth at Sts. 2–5, 8, 10 and 12 (Sts. 1, 7, 8, 10 and 15), it was not expressed there. Although these results were obtained at one sampling depth, the deviation between amoA gene abundance and gene expression could cause the difference in maximum depth between ammonia oxidation and AOO abundance.

The nitrate assimilation rates that we obtained were comparable to those of previous studies in each region (Supplementary Table S4). The concentrations of chlorophyll a and nitrate indicated that Sts. 10 and 11 were in the high-nutrient low-chlorophyll region, suggesting that biological production was limited by trace metals (particularly iron) (Boyd et al., 2004; Moore et al., 2013). The surface nitrate concentration at St. 10 was higher than that at St. 11. On the other hand, the chlorophyll a concentration and the nitrate assimilation at St. 10 were less than half those at St. 11. This suggests that biological production at St. 10 was likely more limited by iron than biological production at St. 11. St. 11 was located near the Aleutian Islands. Iron concentration near the Aleutian Islands (>0.4 nm) is considerably higher than in the subarctic gyre (<0.01 nm) (Suzuki et al., 2002), and thus iron delivered from coastal water may have contributed to the production. Satellite-derived chlorophyll a indicated that active production occurred near the islands (Figure 1), which supports our inference. At St. 12, in the off-shelf area of the BERS, the surface nitrate assimilation (586 nmolN l−1 d−1) was about six times higher than that at St. 9 (104 nmolN l−1 d−1), whereas the surface nitrate concentration at St. 12 (435 nm) was similar to that at St. 9 (305 nm). St. 12 was located at the shelf edge and hence was probably in the Green Belt, which is defined as a highly productive region along the edge of the continental shelf of the Bering Sea (Springer et al., 1996). Iron input to the Green Belt is high and enhances primary production (Aguilar-Islas et al., 2007). Therefore, the elevated nitrate assimilation at St. 12 was ascribable to a higher iron supply than at St. 9. Nitrate assimilation was strongly related with primary production, suggesting that the biological production could be influenced by the supply of not only nitrate but also micronutrients.

When we assumed that ammonia and nitrite oxidations were tightly coupled, ammonia oxidation had a significant impact on nitrate assimilation, particularly in the subtropical region. A previous study in the South Pacific (Raimbault and Garcia, 2008) and a modeling study (Yool et al., 2007) came to the same conclusion as we have presented. Raimbault and Garcia (2008) reported that the ratio of ammonia oxidation to nitrate assimilation was 80–100% in the subtropical South Pacific Ocean. Although this ratio in the subtropical region (PNEC and NPTG) was ca.80% at some stations, the average of 55.6% was lower than that in the subtropical South Pacific. The surface environment at Sts. 13–15 was also under oligotrophic conditions. However, the ammonia oxidation rates were low and the majority of nitrate assimilation was new production, suggesting that ammonia oxidation does not necessarily significantly influence the estimate of nitrate-based new production in all oligotrophic regions.

Environmental controls on the distribution of AOO and ammonia oxidation

Although βAOB was not detected0 or barely detected in the PNEC and NPTG, it was abundant from the NPST/NPPF to St. 13. The ammonium affinity of AOB is lower than that of AOA (Martens-Habbena et al., 2009); thus, it would be difficult for AOB to survive in the oligotrophic unproductive region. The variation in βAOB abundance was likely due to the supply of ammonium. In fact, the relative abundance of βAOB in AOO has been reported to increase with ammonium concentration (Bouskill et al., 2012; Urakawa et al., 2014), and we had a similar result here. Distribution of microbial communities is influenced by physical structure and is described as stratified with depth (Giovannoni and Vergin, 2012). The mixed layer depth at Sts. 2–4 was deeper relative to that at the adjacent stations, and there was no significant regional difference at the other stations (Supplementary Figure S3a). Thus, the vertical distribution of AOO was not consistent with the mixed layer depth; rather, it was correlated with the euphotic depth, suggesting that light environment was an important factor in determining the vertical distribution of AOO. Some studies reported that oxygen concentration is also important for determining AOO distribution and their activities (Ward, 2002; Beman et al., 2008; Bouskill et al., 2012) and the AOO abundances and activities were reported to be high in the oxygen minimum zone (Beman et al., 2008; Bouskill et al., 2012). In the present study, although low-oxygen water was observed below 100 m at Sts. 2, 3 and 11 (Supplementary Figure S3b), AOO abundances and activities were not markedly high in these waters. Oxygen concentration above 100 m in the north of the PSAG increased, and it was not notably correlated with AOO abundance or activities. AOO abundance and ammonia oxidation were not detected at St. 14 and were low at St. 15. In the Arctic Ocean, nitrification activity was reported to be low in summer (Christman et al., 2011; Baer et al., 2014). Christman et al. (2011) and Baer et al. (2014) showed that chlorophyll a was high and ammonium concentration was low in summer and suggested that phytoplankton could outcompete AOO for ammonium resources. In our study, AOO abundance and activity were low in the region, but a high ammonium concentration was observed near the bottom. The existence of ammonium-rich water indicated that the low nitrification activity was not due simply to ammonium resource competition with phytoplankton. In the shelf region of BERS and BPLR, ammonium-rich water generally occurs near the bottom during summer as a result of active ammonia regeneration (Saino et al., 1983; Whitledge et al., 1986; Nishino et al., 2005). As Christman et al. (2011) and Baer et al. (2014) collected their samples from 2–6.5 m, they missed sampling the ammonium-rich water. Such ammonium-rich (>1 μm) water masses are rarely observed in the ocean (Brzezinski, 1988; Rees et al., 2006; Zhang et al., 2007). This might also be influenced by the low ammonia oxidation and not only be due to the active ammonia regeneration.

The question of why AOO abundance and activity were low at Sts. 13–15 requires consideration. One of the possible factors is the light environment. Light is known to generally inhibit the activity of AOO (Olson, 1981; Guerrero and Jones, 1996; Merbt et al., 2012). At. Sts. 14 and 15, the SCM occurred near the bottom, suggesting that the light environment was sufficient for the growth of phytoplankton. Therefore, it could make it difficult for AOO to survive, at least at Sts. 14 and 15. Furthermore, competition with other organisms for resources except ammonium and the existence of nitrification inhibitors could also influence the activities. This should be clarified in future studies.

When we excluded the data at Sts. 14 and 15, where nitrification was clearly suppressed by various factors, the depth-integrated ammonia oxidation rates were positively correlated with primary production (r=0.63, P<0.05) (Figure 5c). Substrate supply is generally considered to be essential for ammonia oxidation (Ward, 2002). This significant correlation suggests that ammonia oxidation was likely enhanced by high ammonium supply through decomposition of organic matter in highly productive water. However, our kinetics experiment demonstrated that ammonia oxidation did not increase significantly with additional ammonium of 31–1560 nm, except at St. 12. Although we performed the kinetics experiment at a single depth and results from other depths might differ, the obtained result implies that other factor(s), in addition to the difference in ammonium supply, could contribute to ammonia oxidation. As mentioned previously, the primary production and nitrate assimilation revealed in this study would have been influenced by the supply of micronutrients. The shallow clade AOA was the major AOO in the study regions, and previous genome analyses have indicated that copper is probably important for ammonia oxidation by AOA (Walker et al., 2010; Hollibaugh et al., 2011). During our cruise, dissolved copper concentration at the SCM was low (1.8 nm) at Sts. 1–10 and significantly increased in the BERS (2.0 nm) (S. Takeda, pers. comm.). Previous studies also reported that dissolved copper concentration to the 200-m depth was <~2 nm in the subarctic North Pacific (Fujishima et al., 2004; Takano et al., 2014), <~1 nm in the subtropical North Pacific (Takano et al., 2014), and 2–10 nm in the Bering Sea (Cid et al., 2011). This distribution pattern is similar to that of iron (Fujishima et al., 2001; Cid et al., 2011), the supply of which likely limited primary production and nitrate assimilation. Considering this pattern, the lower ammonia oxidation at St. 10 in the subarctic ocean can be explained: it was attributable to a shortage of copper. In contrast, at St. 12, in the Green Belt, the copper supply would be high enough. Thus, the ammonia oxidation increased significantly with the addition of ammonium.

Conclusion

Although the contribution of nitrification to nitrate assimilation was significant in the subtropical oligotrophic region, it was low (<20%) in most of the other regions. The amoA gene abundance and expression of shallow clade AOA were the highest among the AOO from the equatorial Pacific to Arctic Ocean, indicating that shallow clade AOA was the major AOO in the study regions. The ratio of AOA to βAOB changed significantly with region, and it was related to ammonium concentration. Although AOO was detected even in the surface water, ammonia oxidation occurred mainly below the 1% light depth throughout the study region, indicating that it would be susceptible to light environment. Light was likely one of the reasons for very low ammonia oxidation in the shallow Bering and Chukchi Sea shelves. Ammonium concentration seemed to be important for determining the AOO community; however, it did not explain the distribution of ammonia oxidation. The kinetic experiments and the significant positive relationship between depth-integrated primary production and ammonia oxidation suggested that ammonia oxidation might be controlled by micronutrient (probably copper) availability. This was likely related with the fact that AOA, which require copper for ammonia oxidation (Walker et al., 2010; Hollibaugh et al., 2011), were dominant in the study regions.