Nitrogen fixation and the diazotroph community in the 1 temperate coastal region of the northwestern North Pacific 2 3

Abstract. Nitrogen fixation in temperate oceans is a potentially important, but poorly understood process that may influence the marine nitrogen budget. This study determined seasonal variations in nitrogen fixation and the diazotroph community within the euphotic zone in the temperate coastal region of the northwestern North Pacific. Nitrogen fixation as high as 13.6 nmol N L−1 d−1 was measured from early summer to fall when the surface temperature exceeded 14.2 °C (but was lower than 24.3 °C) and the surface nitrate concentration was low (≤ 0.30 μM), although we also detected nitrogen fixation in subsurface layers (42–62 m) where nitrate concentrations were high (> 1 μM). Clone library analysis results indicated that nifH gene sequences were omnipresent throughout the investigation period. During the period when nitrogen fixation was detected (early summer to fall), the genes affiliated with UCYN-A, Trichodesmium, and γ-proteobacterial phylotype γ-24774A11 were frequently recovered. In contrast, when nitrogen fixation was undetectable (winter to spring), many sequences affiliated with Cluster III diazotrophs (putative anaerobic bacteria) were recovered. Quantitative PCR analysis revealed that UCYN-A was relatively abundant from early to late summer compared with Trichodesmium and γ-24774A11, whereas Trichodesmium abundance was the highest among the three groups during fall.


Introduction
The amount of bioavailable nitrogen introduced into the global ocean via nitrogen fixation is considered to be roughly balanced at the large spatiotemporal scale by nitrogen loss through denitrification, as indicated by the sedimentary nitrogen isotope record during the Holocene epoch (Brandes and Devol, 2002;Deutsch et al., 2004).However, rate measurement data have revealed that denitrification far exceeds nitrogen fixation (Codispoti, 2007).This discrepancy in the nitrogen balance has raised the possibility that the current estimate of marine nitrogen fixation, which is primarily based on data collected in tropical and subtropical oceans where large cyanobacterial diazotrophs (e.g., Trichodesmium spp.and Richelia intracellularis) are considered to be primarily responsible for nitrogen fixation (e.g., Capone et al., 1997), might be too low (Codispoti, 2007).This is supported by the results of recent studies using molecular approaches that have increasingly revealed that marine diazotrophs are more diverse and widespread than previously thought (Riemann et al., 2010;Zehr, 2011).Recently discovered marine diazotrophic taxa, including those belonging to unicellular cyanobacteria and heterotrophic bacteria, are abundant in oceanic regions where large cyanobacterial diazotrophs are scarce (Needoba et al., 2007;Moisander et al., 2010;Halm et al., 2012;Bonnet et al., 2013;Rahav et al., 2013;Shiozaki et al., 2014a), suggesting that a failure to account for nitrogen fixation mediated by these diazotrophs might result in underestimation of marine nitrogen fixation.
The temperate coastal ocean is one of the regions where nitrogen fixation rates have been understudied and potentially Published by Copernicus Publications on behalf of the European Geosciences Union.
T. Shiozaki et al.: Nitrogen fixation in the temperate coastal region underestimated.Conventionally, nitrogen fixation in temperate oceans has been assumed to be low because of the relatively low temperatures (< ∼ 20 • C), which generally inhibit the growth of large cyanobacterial diazotrophs (Breitbarth et al., 2007), and development of high dissolved inorganic nitrogen (DIN) concentrations (> 1 µM).High DIN concentrations are generally understood to inhibit nitrogen fixation (Falkowski, 1983), especially during mixing periods.However, recent studies have indicated that nitrogen fixation, presumably mediated by unicellular cyanobacteria and heterotrophic bacteria, is detectable even in the relatively cold (< 10 • C) and DIN-rich waters (> 1 µM) of the Atlantic coast (Mulholland et al., 2012) and the Baltic Sea estuaries (Bentzon-Tilia et al., 2015).These results highlight the necessity of re-evaluating the extent, variation, and control mechanisms of nitrogen fixation in temperate oceans, with recognition of the widespread occurrence of diverse diazotrophic microbes.
This study examined the seasonal variation in nitrogen fixation along two onshore-offshore transects in the interfrontal zone of the northwestern North Pacific.In this temperate region, physical, chemical, and biological properties vary widely between seasons (Shiozaki et al., 2014b) due to the confluence of three currents: the Kuroshio (warm current), the Tsugaru Warm Current, and Oyashio (cold current).Data on nitrogen fixation rates in the temperate Pacific are limited (Needoba et al., 2007), and to the best of our knowledge, the present study is the first to examine diazotrophy during all seasons in the temperate ocean.This study was conducted as part of a project to monitor the dynamics of the coastal ecosystem and the recovery thereof after the 2011 Tohoku-Oki Tsunami, which struck the region on 11 March 2011.
Temperature, salinity, and dissolved oxygen profiles of regions near the bottom floor were measured using a SBE 911plus conductivity-temperature-pressure (CTD) system (Sea- bird Electronics, Bellevue, WA, USA).Water samples were collected in an acid-cleaned bucket and Niskin-X bottles.At offshore stations, samples for nutrient analysis were collected from 7-15 different depths in the upper 200 m, while at shallower (< 200 m) bay stations, samples were collected from 4-9 different depths in the entire water column, except at Station (Stn.)OT1 where only surface water samples were collected.Samples for DNA analysis and incubation experiments were collected from the surface at almost every station, and from depths corresponding to 10 and 1 % of the surface light intensities at Stns.OT4 and ON5.Light attenuation was determined using a submersible PAR sensor.

Nutrients
Samples for nutrient analysis were stored in 10 mL acrylic tubes and kept frozen until onshore analyses.Nitrate, nitrite, ammonium, and phosphate concentrations were determined using an AACSII auto-analyzer (Bran+Luebbe, Norderstedt, Germany).The detection limits of nitrate, nitrite, ammonium, and phosphate were in the ranges of 0.01-0.04,0.01-0.02,0.01-0.03,and 0.01-0.02µM, respectively.The nitra-cline was defined as the depth where nitrate concentrations increased above 1 µM.

Nitrogen fixation activity and mannitol enrichment experiment
Nitrogen fixation was determined by the 15 N 2 gas bubble method (hereafter, the bubble method; Montoya et al., 1996).Samples for incubation were collected in duplicate acidcleaned 2 L polycarbonate (PC) bottles.The time-zero samples (n = 1) were immediately filtered onto precombusted GF/F filters.Two milliliters of 15 N 2 gas (SI Science Co. Japan, for this gas, contaminations of nitrate, nitrite, and ammonium were determined to be low (< nM level), indicating that the overestimation of nitrogen fixation rates due to the uptake of 15 N-labeled contaminants (Dabundo et al., 2014) was minimal (Shiozaki et al., unpublished data)) were injected directly into the incubation bottles through a septum using a gastight syringe.The tracer-added samples were covered with neutral-density screens to adjust the light level and incubated for 24 h in an on-deck incubator filled with flowing surface seawater.After the incubation, the samples were filtered onto precombusted GF/F filters.The isotopic analyses were performed as described previously (Shiozaki et al., 2009).The rate of nitrogen fixation was calculated using the equations of Montoya et al. (1996).
To examine the possibility of underestimation of nitrogen fixation as determined by the bubble method (Mohr et al., 2010;Großkopf et al., 2012), we compared the nitrogen fixation rates determined using the 15 N 2 gas dissolution method (hereafter, the dissolution method; Mohr et al., 2010) with those determined using the bubble method (see above) during the KK-13-6_Sep and KS-14-2_Mar cruises.For the dissolution method, 15 N 2 -enriched seawater was prepared according to Mohr et al. (2010) and Großkopf et al. (2012).Briefly, filtered seawater was degassed using a Sterapore membrane unit (20M1500A: Mitsubishi Rayon Co., Ltd., Tokyo, Japan) at a flow rate of ∼ 500 mL min −1 (recirculation period, 10 min).Degassed seawater was stored in 1 L Tedlar bags without headspaces and 15 N 2 gas was added at a ratio of 10 mL 15 N 2 per 1 L seawater.After complete dissolution, the 15 N 2 -enriched seawater was added to seawater samples contained in 2 L PC bottles, which were incubated and used for isotopic analyses as described above.The 15 N 2 -enriched seawater was prepared at each station, and was added to the incubation bottles within 1 h after preparation.The nitrogen fixation rate was calculated according to Mohr et al. (2010).For this comparison, triplicate samples were used for both the dissolution and bubble methods.

Statistical analysis
Pearson's correlation coefficient was used to examine the relationships between nitrogen fixation activities and environmental variables including temperature, nitrate, ammonium, phosphate, and the ratio of nitrate + nitrite + ammonium to phosphate (N / P ratio) in the entire water column (the data used for the calculation are shown in Table S1).When the nutrient concentration was below the detection limit, the value of the detection limit was used for the analysis.When nitrogen fixation was undetectable, the value was assumed to be zero.

DNA extraction, sequencing, and phylogenetic analysis
Samples (0.38-1 L) for DNA analysis were filtered through 0.2 µm pore-sized Nuclepore filters and stored in a deep freezer (−80 • C) until onshore analysis.Total DNA was extracted using a ChargeSwitch Forensic DNA Purification Kit (Invitrogen, Carlsbad, CA, USA) with slight modification of the manufacturer's protocol (Shiozaki et al., 2014a).Partial nifH fragments were amplified using a nested PCR strategy (Zehr and Turner, 2001) from samples collected from surface water at Stns.OT4, ON1, ON5, and ON7 during the KT-12-20_Aug and KT-12-27_Oct cruises, at Stns.OT4, ON1, and ON5 during the KT-13-2_Jan and KS-14-2_Mar cruises, at Stns.OT4, ON1, ON5, and ON8 during the KK-13-1_Jun cruise, and at Stns OT4, ON5, ON7 during the KK-13-6_Sep cruise (Table 1).PCR reagents were applied as described by Shiozaki et al. (2014a).The first and second PCRs were run using the same cycling conditions: 95 • C for 30 s followed by 30 cycles of 98 • C for 10 s, 52 • C for 30 s, and 72 • C for 30 s, followed by a final extension at 72 • C for 7 min.Sterile distilled water was used as the negative control.After PCR analysis, we confirmed that the negative control showed no bands in the gel.The PCR products were cloned and sequenced according to Shiozaki et al. (2014a).The present study obtained 197 nifH sequences in total.The nifH sequences were translated into amino acid sequences and searched against the protein database of the National Center for Biotechnology Information using the Basic Local Alignment Search Tool (BLASTp) algorithm.
Clones with 100 % amino acid sequence similarity were defined as the same operational taxonomic unit (OTU) using the CD-HIT suite (Huang et al., 2010).The amino acid sequences were aligned using multiple sequence comparisons by the log-expectation (MUSCLE) module in the MEGA5 package (Tamura et al., 2011).A phylogenetic tree was constructed using the maximum likelihood method employing the Dayhoff matrix-based mode, and 1000 bootstrap replicates were run.cluster within the phylogenetic tree (Zehr et al., 2003a).The sequences from this study were deposited in the DNA Data Bank of Japan (DDBJ) as accession numbers LC013480 to LC013676.

Quantitative PCR (qPCR) analysis
The clone library analysis showed that UCYN-A, Trichodesmium, and γ -proteobacterial phylotype γ -24774A11 (hereafter γ -24774A11) were likely important diazotrophs from early summer to fall when nitrogen fixation occurred (see below).Therefore, the present study quantified these nifH phylotypes by qPCR analysis to examine their relative importance during these seasons.In addition, UCYN-B which is considered to be a major diazotroph in the tropical and subtropical oligotrophic ocean (Moisander et al., 2010), was quantified.TaqMan primer and probe sets previously designed for these four nifH phylotypes were used for quantification (Shiozaki et al., 2014a, c;Moisander et al., 2014).The 20 µL qPCR reactions contained 10 µL 2× Premix Ex Taq (Probe qPCR; Takara), 5.6 µL of nuclease-free water, 1 µL each of the forward and reverse primers, 0.4 µL of TaqMan probe, and 2 µL of template DNA.The qPCR assays were performed using LightCycler 480 System (Roche Applied Science, Germany).The qPCR assays were run in triplicate reactions.Linear regression r 2 values for the standard curves were > 0.99 for all reactions.The efficiency of the qPCR assays ranged from 90.9 to 98.4 %, with an average of 95.1 %.
As the negative control, sterile distilled water was used, from which no amplification signals were detected.The detection limit was 75 copies L −1 .

Comparison of the bubble method and the dissolution method
Nitrogen fixation rates determined by the bubble and dissolution methods were compared during the KK-13-6_Sep and KS-14-2_Mar cruises (Fig. 2).Both methods failed to detect nitrogen fixation in samples collected during the KS-14-2 cruise.During the KK-13-6_Sep cruise, the nitrogen fixation rates determined by the dissolution method were significantly higher (1.5-2.2 fold) than those determined by the bubble method at Stns.OT6 and ON5 (p < 0.05).At Stns.OT4 and ON7, the nitrogen fixation rates determined by the two methods did not differ significantly.Thus, the bubble method may have significantly underestimated the nitrogen fixation rates in some, if not all, of the samples that we analyzed.Although the nitrogen fixation rates reported in the rest of this paper are those obtained using the bubble method, which was used as the standard protocol during all cruises, the possibility that some of these rates could be underestimated must be kept in mind.cruise.An asterisk indicates a significant difference between the two methods (p < 0.05).

Seasonal variations in nitrogen fixation rates
According to the temperature-salinity (TS) diagram proposed by Hanawa and Mitsudera (1987), both the offshore and bay waters collected during this investigation mostly belonged mostly to either the surface layer water system (SW) or the Tsugaru Warm Current water system (TW) (Fig. 3).Exceptions included the waters collected from the 1 % light depth (119 m) at Stn. ON5 during the KT-13-2_Jan cruise (classified as the Oyashio water system (OW)) and those collected at the surface of OT5 during the KS-14-2_Mar cruise (classified as the Coastal Oyashio water system (CO)).These water classifications based on the TS diagram were generally consistent with the geostrophic current field of the investigated region (Fig. S1).Based on these results, it was assumed that surface waters collected during the same cruise in a particular season generally belonged to the same water system that was prevalent in the investigated region at the time of our sampling.
During the four cruises conducted in early summer (KK-13-1_Jun), mid-summer (KT-12-20_Aug), late summer (KK-13-6_Sep), and fall (KT-12-27_Oct), nitrogen fixation was measurable in most of the samples collected from surface waters: the nitrogen fixation rates varied in the range of 0.33-13.6nmol N L −1 d −1 (Figs.4c and S2).Relatively high nitrogen fixation rates were determined for samples collected during the KT-12-20_Aug cruise, although the highest value was obtained at Stn. ON7 during the KK-13-6_Sep cruise.Nitrogen fixation was below the detection limit in seawater samples collected during the winter and spring cruises.For those samples, nitrogen fixation was undetectable even after the addition of mannitol (KS-14-2_Mar).Also, nitrogen fixation was undetectable in DIN-replete water collected at Stn. OT1 in late summer (KK-13-6_Sep).

Relationship between nitrogen fixation rates and environmental variables
Nitrogen fixation rates tended to increase with temperature (p < 0.01) (Fig. 6a and Table 2).Nitrogen fixation was detected only when seawater temperatures exceeded 11.7 • C, with higher rates (> 6 nmol N L −1 d −1 ) observed in waters warmer than 19.5 • C.However, there were exceptions to this general relationship between the nitrogen fixation rate and temperature.For example, from the data collected during the KK-13-1_Jun cruise we observed that the nitrogen fixation  rate was highest at 15.4 • C, while it was low (below the detection limit) at higher temperatures.Nitrogen fixation rates were negatively correlated with nitrate and phosphate concentrations (p < 0.01) (Table 2), whereas they were not significantly correlated with ammonium concentrations (p > 0.05) (Table 2).We also found no significant correlation between nitrogen fixation rates and the ratio of total inorganic nitrogen (nitrate + nitrite + ammonium) to phosphate (Table 2).Nitrogen fixation was generally detectable only when nitrate was depleted KT-12-20_Aug KT-12-27_Oct KT-13-2_Jan KK-13-1_Jun KK-13-6_Sep KS-14-2_Mar

Diazotroph community
PCR reagents have been suggested to be a potential source of nifH genes during analysis of the diazotroph community (Zehr et al., 2003b).Although we confirmed the absence of any bands from the negative control in agarose gel electrophoresis, some sequences recovered from the samples obtained during the KK-13-6_Sep and KS-14-2_Mar cruises (10 clones in total) were judged to be the contaminants in PCR reagents (> 97 % similarity at the amino acid level was used as a criterion).We did not include these sequences in our data analysis.
The recovered cyanobacterial sequences belonged to Trichodesmium, UCYN-A, and Leptolyngbya.The nifH sequences of UCYN-B, UCYN-C, and Richelia intracellularis were not recovered.The nifH sequence of Trichodesmium was recovered only during the KT-12-27_Oct cruise (Table 1).UCYN-A was generally recovered from early summer to fall, while nifH of Leptolyngbya was recovered during winter.The present study detected the sequences of γ -24774A11 during the KT-12-27_Oct and KK-13-6_Sep cruises.This heterotrophic bacterial phylotype is considered to significantly contribute to nitrogen fixation in a wide range of oceanic environments (Moisander et al., 2014).During the KS-14-2_Mar cruise, all of the sequences that we recovered were derived from heterotrophic bacteria, and were dominated by Cluster III diazotrophs at Stns.OT4 and ON5.The Cluster III diazotroph nifH sequences were recovered during all cruises except for the KK-13-1_Jun cruise.Note that 58 out of 187 sequences displayed > 97 % similarity, at the amino acid level, to terrestrial diazotroph sequences derived from soil, mudflats, and lakes (Figs.S3, S4, and  S5).These sequences were mainly affiliated with α-and δproteobacterial diazotrophs, with 29 of 39 α-proteobacterial sequences and 22 of 24 δ-proteobacterial sequences being similar to terrestrial diazotroph sequences.

Diazotrophs abundances
The nifH sequence of Trichodesmium was detected by qPCR assay during the .During these two cruises, the abundance of Trichodesmium ranged from below the detection limit to 8.7 × 10 4 copies L −1 at all depths.Trichodesmium abundance at the surface was higher than those of UCYN-A, UCYN-B, and γ -24774A11 at most stations during the KT- [Log 10 ((copy+1) L −1 )] at the surface during each cruise.When the target nifH gene was not detected, the copy number was assumed to be zero.

Seasonal variations in nitrogen fixation rates in the temperate coastal ocean
Nitrogen fixation rates were measurable mainly from early summer to fall when nitrate was generally depleted in sample seawaters, although there were some exceptions.Our estimates of the nitrogen fixation rates (0.33-13.6 nmol N L −1 d −1 ) were significantly (p < 0.05) higher than the corresponding values previously reported in the temperate region of the eastern North Pacific (0.15-0.31 nmol N L −1 d −1 ; Needoba et al., 2007) and the olig-otrophic region of the western and central North Pacific (0.17-3.62 nmol N L −1 d −1 ; Shiozaki et al., 2010), whereas they were comparable to those determined in the Kuroshio (0.54-28 nmol N L −1 d −1 ; Shiozaki et al., 2010) and the western Atlantic coastal regions (1.3-49.8nmol N L −1 d −1 ; Mulholland et al., 2012).Higher nitrogen fixation rates have been determined in other temperate oceans, including the western English Channel (18.9 ± 0.01 and 20.0 nmol N L −1 d −1 ; Rees et al., 2009) and the Baltic Sea estuaries (47-83 nmol N L −1 d −1 ; Bentzon-Tilia et al., 2015).
In our study, spatiotemporal variability in nitrogen fixation rates appeared to be partly related to the Tsugaru Warm Current path.This current, which flows from the north (after passage through the Tsugaru Strait) to the study region (Fig. S1), may carry active diazotrophs and therefore enhance nitrogen fixation in our study region.This is supported by the fact that nitrogen fixation rates during individual cruises tended to be higher at Stn. OT4 than at Stn. ON5.These stations were located up-and down-stream of the Tsugaru Warm Current, respectively.In addition, variations in nitrogen fixation rates among stations and seasons might also be related to the extent of vertical mixing in the Tsugaru Warm Current.It has been suggested that vertical mixing may introduce iron-rich subsurface water to the surface of the Tsugaru Strait (Saitoh et al., 2008).Such input of iron may enhance nitrogen fixation rates.Consistent with this notion, our results showed that the nitrogen fixation rate was relatively high at Stn. OT4, where the nitracline was relatively deep.Blais et al. (2012) proposed that nitrogen fixation can occur even in nutrient-replete waters, if large amounts of iron and organic materials are available for consumption by bacterial diazotrophs.In the present study, this possibility was examined by conducting mannitol addition experiments using surface seawaters collected during spring.These waters, which belong to the Oyashio Current system (Nishioka et al., 2007(Nishioka et al., , 2011;;Shiozaki et al., 2014b), were considered to be rich in iron during spring, as indicated by a previous study (iron conc., 0.79-8.46 nM;Nishioka et al. 2007).Despite potentially high iron concentrations, our results showed that nitrogen fixation was undetectable even after the mannitol addition, suggesting that, contrary to the Blais et al. proposition, diazotrophs remained inactive under our experimental settings.
Our data showed that nitrogen fixation rates were below the detection limit during winter, spring, and late summer (KK-13-6_Sep), when nitrate concentrations were high.These results were consistent with the results of previous studies in the Pacific Ocean, which indicated that nitrogen fixation rates were low or undetectable in DIN-replete waters (Shiozaki et al., 2010).In contrast, Mulholland et al. (2012) reported that, in temperate regions of the Atlantic Ocean, nitrogen fixation rates were high even in DIN-replete (> 1 µM) and cold (< 10 • C) surface seawaters.Their study was conducted downstream of the Gulf Stream, where diazotrophs  0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 KT-12-20_Aug KT-12-27_Oct KT-13-2_Jan KK-13-1_Jun KK-13-6_Sep KS-14-2_Mar mid-summer fall winter early summer late summer spring could be delivered from subtropical oceans where DIN is depleted.Previous studies have suggested that cyanobacterial diazotrophs can travel over long distances (> 1000 km) in currents, without losing their capacity for N 2 fixation (Shiozaki et al., 2013), and that activity is not lost immediately even after mixing with DIN-replete seawaters (Holl and Montoya, 2005;Dekaezemacker and Bonnet, 2011).In our region, because the Tsugaru Warm Current flows from north to south, diazotrophs entrained by the current have little chance of meeting DIN-rich water at the surface.DIN-replete water during mid-summer was observed at the inside bay station OT1 (Fig. S2).Concomitantly, low-salinity surface waters spread offshore along the OT transect line (Fig. S7), suggesting that anomalously high DIN concentrations were likely attributable to terrestrial surface discharge enhanced by Typhoon Man-yi, which passed over the region immediately before the cruise.Subramaniam et al. (2008) reported that nitrogen fixation rates near the Amazon River estuary, with low salinity and high nitrate levels, were fairly low.
Their results are consistent with ours.Ammonium inhibits nitrogen fixation, especially when ammonium concentrations exceed 1 µM, as demonstrated for Trichodesmium (Mulholland et al. 2001).In our study, ammonium concentrations were generally low (≤ ∼ 1 µM) throughout the investigation, and no negative relationship between nitrogen fixation and ammonium concentration was found.Our data showing that nitrogen fixation rates were negatively correlated with nitrate concentrations (Table 2) are consistent with the general notion that nitrogen fixation rates are generally low in nitrate replete waters (Falkowski, 1983).Our data also showed nitrogen fixation rates tended to increase with increasing temperature and with decreasing phosphate concentrations (Table 2).
Because temperature and phosphate concentrations were correlated with nitrate concentrations, these factors would not necessarily influence nitrogen fixation directly.Rather, one or more factors that varied with nitrate could synergistically influence nitrogen fixation.

Seasonal variation in the diazotroph community in the temperate coastal ocean
The qPCR analysis demonstrated that the target groups were quantifiable even at stations at which their sequences were not recovered by the clone library analysis, suggesting that the number of clones was not sufficient to capture the diazotroph community structure on each cruise.Despite this limitation, the sequences more frequently recovered in the clone library generally corresponded to the most abundant group revealed by the qPCR analysis.For example, UCYN-A was frequently recovered in the library during the KT-12-20_Aug, KK-13-1_Jun, and KK-13-6_Sep cruises; for these samples, the qPCR results showed that UCYN-A was the most abundant group among the four examined.Similarly, qPCR data indicated that Trichodesmium was the most abun-dant group during fall, when this group was frequently recovered in the library (during the KT-12-27_Oct cruise).Therefore, the diazotrophs targeted by the qPCR analysis were likely important for nitrogen fixation in this study region.In the discussion below, we mainly discuss possible factors responsible for seasonal variation in the diazotrophs targeted by the qPCR analysis.UCYN-A was detected in all seasons except spring (KS-14-2_Mar), suggesting that this group of diazotrophs could be important agents of nitrogen fixation in this region.Especially from early to late summer, the abundance of UCYN-A was generally higher than that of Trichodesmium, UCYN-B, and γ -24774A11.UCYN-A has been widely detected in temperate regions, and is considered to be one of the major diazotrophs of these locations (Needoba et al., 2007;Rees et al., 2009;Mulholland et al., 2012;Bentzon-Tilia et al., 2015).UCYN-A is known to be most abundant in relatively warm waters around ∼ 20 • C (Needoba et al., 2007;Moisander et al., 2010).In our study, UCYN-A was detected during winter at some stations.It appears that UCYN-A abundance decreased with decreasing temperature from fall to winter, and then became undetectable in spring.
Trichodesmium was detected from late summer to fall, when water temperatures ranged from 19.1 to 23.4 • C at the surface.Given that the optimal growth temperature for Trichodesmium has been reported to be high (24-30 • C) (Breitbarth et al., 2007), Trichodesmium detected in the investigated region likely existed under suboptimum conditions.The relatively high abundance of Trichodesmium observed during fall, despite the suboptimal temperature conditions, might indicate that Trichodesmium was transported from the adjacent subtropical region where seawater temperatures were high (> 24 • C).In the western North Pacific subtropical region, Trichodesmium is abundant from July to September (Marumo and Nagasawa, 1976;Chen et al., 2008).Trichodesmium that flourished in the subtropical region during summer could be transported by the Tsugaru Warm Current, displaying peak abundance during fall in the investigated region.This could support the above discussion that waters containing active nitrogen fixation were delivered to this region by the Tsugaru Warm Current.
We detected γ -24774A11 during all cruises except for the KS-14-2_Mar cruise.γ -24774A11 is considered to be one of the most important heterotrophic diazotrophs in the tropical and subtropical oligotrophic ocean (Moisander et al., 2014).However, the γ -24774A11 sequence has not been detected previously in other temperate oceans (Needoba et al., 2007;Rees et al., 2009;Mulholland et al., 2012).The γ -24774A11 sequence was similar (94 % similarity at the amino acid level) to the nifH sequence of Pseudomonas stutzeri, which has been reported to be present in temperate estuaries (Bentzon-Tilia et al., 2015).Bentzon-Tilia et al. (2015) reported that P. stutzeri-like nifH genes (99 % similarity at the nucleotide level) were the most abundant sequences among their samples collected from the Baltic Sea estuary.In the present study, we recovered P. stutzeri-like nifH genes (> 97% similarity at the amino acid level) only at Stn. OT4 during the KT-13-2_Jan cruise by the clone library analysis, and γ -24774A11 was not detected on that occasion by qPCR analysis probably due to the difference in the sequence between γ -24774A11 and P. stutzeri.The ecology of γ -24774A11 is still fairly unknown.It remains to be seen whether this phylotype contributes to the nitrogen fixation in this region, a topic for future studies.
UCYN-B was not detected except at one station.This result is consistent with previous knowledge.UCYN-B becomes abundant with increasing temperature, similar to Trichodesmium (Moisander et al., 2010), and is rarely observed in the temperate region (Needoba et al., 2007;Rees et al., 2009;Mulholland et al., 2012;Bentzon-Tilia et al., 2015).Furthermore, UCYN-B abundance is low in shallow nitracline regions (Shiozaki et al., 2014a, c).The nitracline depth in this region (≤ 60 m) was shallower than that of > 100 m depths of regions where UCYN-B is abundant (Shiozaki et al., 2014a).Therefore, although UCYN-B might also have been delivered from subtropical region, it could not have survived in the shallower nitracline region.
In nitrate-rich water during winter and spring, Cluster III diazotrophs were detected at most of the stations.Furthermore, from early summer to fall, nifH sequences of Cluster III diazotrophs were recovered by the clone library analysis in samples from all cruises (except KK-13-1_Jan).Therefore, Cluster III diazotrophs appeared to be present throughout the investigation period.Cluster III diazotrophs are putative anaerobes (Hamersley et al., 2011;Farnelid et al., 2013;Bentzon-Tilia et al., 2014), and hence, they are usually dominant in the diazotrophic community of oxygen-depleted waters (Hamersley et al., 2011;Farnelid et al., 2013) or marine sediments (Bertics et al., 2013).In this study, dissolved oxygen was not depleted (> 3.16 mL L −1 ) in the upper winter maximum mixed layer depth in this region (∼ 200 m;Shiozaki et al., 2014b) (Fig. S8).Therefore, the Cluster III activity was likely strongly suppressed in the water column because of the high oxygen concentration.
Many nifH sequences recovered by the clone library analysis were similar to terrestrially derived sequences.These results agree with previous data collected in coastal regions, where terrestrially derived nifH sequences were also found (Rees et al., 2009;Mulholland et al., 2012;Blais et al., 2012).We obtained a Leptolyngbya-like nifH gene during the KT-13-2_Jan cruise.The organism has been found on beaches and in coastal land areas (Brito et al. 2012), but not in the open ocean.Because nitrogen fixation was not detected during the KT-13-2_Jan cruise, the organism was considered not to perform nitrogen fixation.
This study demonstrated that nitrogen fixation can and does proceed at high rates, depending on the season, in the temperate coastal region of the northwestern North Pacific, although we failed to detect nitrogen fixation in DIN-replete cold waters.nifH sequences were omnipresent and recovered throughout the year, displaying a marked seasonality in their composition.UCYN-A was a major diazotroph during summer, whereas Trichodesmium was abundant during fall, despite low temperatures.It has been suggested that Trichodesmium was laterally transported from the adjacent subtropical region, which displays high temperatures.Although the Cluster III diazotrophs were recovered almost throughout the year, they were considered to be inactive in oxic water columns.
The Supplement related to this article is available online at doi:10.5194/bg-12-4751-2015-supplement.

Figure 3 .
Figure 3. Temperature-salinity diagram at each sampling point.The water classification was defined by Hanawa and Mitsud-era (1986).SW, KW, TW, OW, and CO denote the surface layer water system, Kuroshio water system, Tsugaru Warm Current water system, Oyashio water system, and Coastal Oyashio water system, respectively.

Figure 5 .
Figure 5. Time-series variations in the vertical profiles of temperature [ • C] (black), ammonium (purple) and nitrate (green) concentration [µM], and nitrogen fixation (red) [nmol N L −1 d −1 ] at Stns (a) OT4 and (b) ON5.Open symbols indicate that nitrogen fixation was not detected.The horizontal dashed line indicates the nitracline depth.The strait lines of temperature and nitrate were ascribable to strong mixing.
Numbers in parentheses indicate the number of sequences with > 97% similarity at the amino acid level to terrestrial diazotroph sequences.

Table 2 .
Pearson's correlation matrix of N 2 fixation rates and water properties in the entire water column (n = 73).