2.1 Study area

The BS and the YS are marginal seas of the western Pacific Ocean. The BS is a shallow and almost enclosed sea with a surface area of 77×10³ km² and a volume of 1.39×10³ km³. Its maximum depth is 83 m, with an average depth of only 18 m. The Yellow River is a major source of freshwater to the BS whereas the YS is the source of salt for the BS through water exchange via the Bohai Strait. The YS has a surface area of 380×10³ km² and total volume of 16.7×10³ km³. Maximum depth is 140 m with an average depth of 44 m, and like the BS, it is a continental shelf sea. The Yangtze River at the southwest corner of the YS is the major source of freshwater for the southern Yellow Sea (SYS) and the East China Sea. The hydrographic properties of this region are mainly influenced by the Yellow Sea Cold Water Mass (YSCWM) on the shelf (Lü et al., 2010; Li et al., 2016) and the Yellow Sea Coastal Currents on the western side of the basin (Su, 1998; Lee et al., 2002; Zhang et al., 2004). The BS and the YS are regions surrounded by areas of high population growth and economic development in China and Korea (Zhan et al., 2010; Jiang et al., 2014).

2.2 Sampling

Samples were collected from 13 to 28 June 2011 on board the R/V Dong Fang Hong 2 in the BS and the YS. Sampling for NO from both the surface (at 1 m) and the bottom layer (=1 m above the ocean bottom) were performed at 75 sampling stations (including one 24 h anchor station: B65) shown in Fig. 1 and listed in Table 1 (Tian, 2019). Water samples were collected using 8 L Niskin bottles equipped with silicon O-rings and Teflon-coated springs and mounted on a Sea-Bird CTD (conductivity, temperature, depth) instrument (Sea-Bird Electronics, Inc., USA). A 500 mL Wheaton glass serum bottle was rinsed with in situ seawater three times and was then filled with seawater quickly through a siphon. When the overflowed sample reached the half volume of the bottle, the siphon was withdrawn rapidly, 0.5 mL saturated HgCl₂(aq) solution was added, and the bottle was sealed quickly. The surface water samples were immediately analyzed after collection, and samples from other depths were temporarily placed in the dark in a water bath with a surface seawater circulation system and analyzed within 1 h (Liu et al., 2017). Samples for the analysis of ${\mathrm{NH}}_{\mathrm{4}}^{+}$-N, ${\mathrm{NO}}_{\mathrm{2}}^{-}$-N, and ${\mathrm{NO}}_{\mathrm{3}}^{-}$-N were immediately filtered through 0.45 µm Millipore membrane (pre-ignited at 450 ^∘C for 6 h). The filtrates were stored in polyethylene bottles (pre-marinated with 1:10 HCl for 24 h) at −20 ^∘C. Photoproduction rates were determined in surface water samples from 22 stations in the YS and 4 stations in the BS (shown in Fig. 1), respectively.

Table 1Description of sampling stations and seawater temperature (T), salinity (S), Chl a concentrations, DO, illuminance (I), and NO concentrations in the BS and the YS.

NA: not available. < LOD: below the limit of detection.

Download XLSX

Figure 1Locations of the sampling stations in the BS and the YS during summer. Solid dots (•) represent the stations for incubation experiments.

2.3 Analytical procedures

Analysis for dissolved NO was conducted with the improved method of Liu et al. (2017) by a combined purge-and-trap and fluorometric detection method. The precision of the analytical method was better than ±7 %, and the limit of detection (LOD) was 32 pmol L⁻¹ in the field experiment (Liu et al., 2017).

The photolysis experiments were conducted under natural light conditions on deck. Surface water samples were filtered with 0.45 µm Millipore membrane and transferred into 10 mL cleaned quartz vials wrapped with aluminum foil with no headspace. Then 200 µL NaN₃ solutions were added to remove the microbial influence. After adding 10 µL of $\mathrm{1}\times {\mathrm{10}}^{-\mathrm{3}}$ mol L⁻¹ 2, 3-diaminonaphthalene (DAN, a NO probe) solution and gentle mixing, the fluorescence of the mixed solution was measured before irradiation (Liu et al., 2017). Then the duplicate sample made without the addition of DAN was capped with a Teflon-lined silicone septum and without aluminum foil; following this, the vial was placed in a shallow circulating seawater bath. After 0.5 h irradiation time, 10 µL of $\mathrm{1}\times {\mathrm{10}}^{-\mathrm{3}}$ mol L⁻¹ DAN was added into the duplicate sample without addition of DAN before and the fluorescence of the solution was analyzed. The NO concentrations were measured with the method described in Liu et al. (2017). The NO apparent photolysis production rates were computed based on the time-dependent difference between the NO concentrations before and after irradiation. For dark controls, vials were wrapped in aluminum foil. The quartz vials and syringe used in the experiment were soaked in a 10 % (v∕v) HCl bath for 24 h, rinsed with Milli-Q water, and baked at 500 ^∘C for 4 h.

Chlorophyll a (Chl a) was fluorometrically measured by an F-4500 fluorescence spectrophotometer after filtration of 200 mL seawater through a Whatman glass fiber filter and extraction in 90 % acetone according to Strickland and Parsons (1968). The wind speeds were measured at a height of 10 m above the sea surface using a Model 27600-4X shipborne weather instrument (Young, USA). Illuminance was measured by a digital illuminometer (TES-1330A, Shenzhen, China). Dissolved oxygen (DO) was measured in discrete water samples by the Winkler method (Grasshoff et al., 2009). The concentrations of dissolved inorganic nitrogen (nitrate, nitrite, and ammonium) were analyzed using a nutrient automatic analyzer (Auto Analyzer 3, SEAL Analytical, USA) in the laboratory. The limits of the detection were 0.14 µmol L⁻¹ for nitrate, nitrite, and ammonium, with the precision of the method found to be better than 3 % (Liu et al., 2005).

2.4 Calculation of NO flux and NO saturation

Fluxes of NO across the sea-to-air interface were estimated following the approach of McGillis et al. (2000) for a sparingly soluble gas that is also moderately reactive in the atmosphere:

where F is the flux density (mass area⁻¹ time⁻¹) across the air–sea interface, k_sea is the gas transfer velocity (length per time), and c_sea is the measured concentration of NO in the surface seawater (mol per volume). The partial pressure of the atmospheric NO (pNO_air) was calculated as follows:

where x^′NO_air is the mole fraction of atmosphere NO (dimensionless). We used the value of 2.13 ppb for x^′NO_air, which is the average atmospheric NO mole fraction over the YS (Hu Min, Peking University, personal communication, 2018). p_ss is the barometric pressure at sea surface, which was set to 101.325 kPa as the average pressure, and p_w is the water vapor pressure at sea surface, which was calculated following Weiss and Price (1980):

H^cp is the Henry's law constant, which is calculated as follows:

where $-\mathrm{\Delta }\mathrm{sol}\frac{H}{R}=\frac{\mathrm{d}\ln H}{\mathrm{d}\ln \left(\frac{\mathrm{1}}{T}\right)}$. H^Θ and −ΔsolH∕R are tabulated following Sander (2015).

k_sea was calculated as follows:

where k_w is the water side air–sea gas transfer coefficient for sparingly soluble gases (length per time) calculated according to Wanninkhof (2014), γ_a is the fraction of the entire gas concentration gradient across the air side boundary layer, as a fraction of the entire gradient from the bulk water to the bulk air (dimensionless) (McGillis et al., 2000), k_a is the air side air–sea gas transfer coefficient (length per time) according to McGillis et al. (2000), H^cc is the Henry coefficient (dimensionless) (Sander, 2015), M_NO and $M_{{\mathrm{H}}_{\mathrm{2}}\mathrm{O}}$ are the relative molecular mass of NO and H₂O (dimensionless), and u is the wind speed at 10 m height under neutral boundary conditions (length per time).

The Schmidt number (Sc) is the kinematic viscosity of water divided by the molecular diffusion coefficient of the gas in (sea)water (Jähne et al., 1987; Wanninkhof, 2014). Seawater dynamic viscosity (μ_sw) is a function of temperature (T) and salinity (S) and was estimated using the following equations (Sharqawy et al., 2010):

Seawater density was estimated using Millero's empirical equation (Millero et al., 1976), and NO diffusion coefficient D_L ($\times {\mathrm{10}}^{-\mathrm{5}}$ cm² s⁻¹) in water was calculated according to Wise and Houghton (1968):

The saturation factor (α) is defined as $\mathit{\alpha }=c_{\mathrm{sea}}/(p{\mathrm{NO}}_{\mathrm{air}}\times H^{\mathrm{cp}})$, α>1 represents that NO was supersaturated and that the flux was from sea to air.

3.1 NO in the surface and bottom layers

The NO concentrations from the surface and bottom layers of the BS and the YS, as well as the local sampling time, bottom depth (D), temperature (T), salinity (S), Chl a, wind speed (u), and DO are listed in Table 1. In the study area, temperature varied from 8.8 to 21.7 ^∘C and salinity varied from 29.57 to 32.73 ‰ in the surface water. DO fluctuated from 178 to 271 µmol L⁻¹. The average concentrations of ${\mathrm{NH}}_{\mathrm{4}}^{+}$-N, ${\mathrm{NO}}_{\mathrm{2}}^{-}$-N, and ${\mathrm{NO}}_{\mathrm{3}}^{-}$-N were 2.11, 0.20, and 2.59 µmol L⁻¹, respectively. While in the bottom layer, temperature varied from 3.8 to 21.6 ^∘C, salinity varied from 30.00 to 33.71 ‰, and DO varied from 147 to 266 µmol L⁻¹.

The surface concentrations of NO ranged from below the LOD to 616 pmol L⁻¹, with an overall average of 186±108 pmol L⁻¹, and exhibited a considerable spatial variability (Fig. 2a). The mean NO surface concentrations in the BS (203±107 pmol L⁻¹) and northern YS (NYS) (212±130 pmol L⁻¹) were higher than the mean NO concentration in the SYS (159±84 pmol L⁻¹). The NO concentrations in the bottom layer ranged from below the LOD to 482 pmol L⁻¹, with an overall average of 174±123 pmol L⁻¹. The mean concentrations of NO in the bottom layers of the BS and NYS were 228±116 and 210±138, respectively, and were higher than the mean (127±98 pmol L⁻¹) for the SYS. The maximum NO surface and bottom concentrations were measured at stations B21 and B28 in the NYS, respectively (Fig. 2b), whereas the lowest NO surface and bottom concentrations were measured in the center of the SYS (Table 1). Overall, there were no statistically significant differences (p>0.05) between the mean NO concentrations in the surface and bottom layers.

Figure 2Horizontal distributions of NO (pmol L⁻¹) in the surface water and bottom water.

NO surface concentrations did not show any statistically significant relationship (p>0.05) with sea surface temperature, salinity, DO, Chl a, and illuminance. Trends of NO concentrations with salinity and DIN were only found for the stations affected by the outflow of the Yellow River in the southern BS (Fig. 3), where we found an inverse relationship between salinity and NO surface concentrations. Here high NO concentrations were associated with high DIN concentrations (data not shown; see Liu et al., 2015; Yang et al., 2015) indicating that DIN, especially ${\mathrm{NO}}_{\mathrm{2}}^{-}$, was a prerequisite for enhanced NO concentrations.

Figure 3Variations in salinity, NO surface concentrations, and ${\mathrm{NO}}_{\mathrm{2}}^{-}$ concentrations from station B65 to station B70.

Download

The water columns of the BS and YS were well-oxygenated during our study, and thus no suboxic or anoxic conditions were detected. Therefore, we did not find any enhancement of NO concentrations in the bottom layers with low DO or ${\mathrm{NO}}_{\mathrm{2}}^{-}$ concentrations as observed in the OMZ of the eastern tropical South Pacific Ocean off Peru (Lutterbeck et al., 2018). However, NO concentrations in the bottom layer showed negative correlations with salinity (p<0.05, $r=-\mathrm{0.272}$, n=60) and bottom depth (p<0.05, $r=-\mathrm{0.298}$, n=61) indicating a decrease in NO concentrations from the coast toward offshore waters. It is known that both the BS and the YS are affected by submarine ground water discharge (Kim et al., 2005; Taniguchi et al., 2008; Liu et al., 2017a, b). Moreover, NO has been detected in groundwater and aquifers (Smith et al., 2004; Smith and Yoshinari, 2008). Therefore, we suggest that input of NO from submarine groundwater discharge contributed to the distribution of NO bottom concentrations as well.

An overview of published NO surface concentrations is given Table 3. The average NO surface and bottom concentrations from this study are comparable to the concentrations measured in the Jiaozhou Bay, in the waters off Qingdao, and in the Seto Inland Sea. However, our mean concentrations are considerably higher than the NO concentrations reported from the central equatorial Pacific and the eastern tropical North Pacific Ocean. The maximum concentration reported here is at the lower end of the NO concentrations recently reported from the anoxic OMZ off Peru. Overall, NO surface concentrations seem to be generally higher in coastal waters compared to those found in offshore waters.

3.2 Diurnal variability

The diurnal variability of surface NO concentrations, illuminance, DO, and Chl a were investigated at anchor station B65 (Fig. 4). NO concentrations varied from 64 to 424 pmol L⁻¹, exhibiting a significant diurnal variation with the maximum concentration 8-fold higher than the minimum concentration. The NO concentrations reached the maximum concentration in the early afternoon (about 13:00 LT) and then decreased to the minimum concentration at 22:00 LT. A less pronounced second maximum (209 pmol L⁻¹) was reached at 04:00 LT. DO showed a similar diurnal cycle but shifted by 3 h with maxima at 16:00 LT and 07:00 LT. Chl a concentrations peaked at 19:00 LT and 07:00 LT. The illuminance had its maximum at 13:00 LT, coinciding with the NO maximum indicating that the first NO maximum was indeed resulting from a photochemical production during daytime (Zafiriou and McFarland, 1981). However, the second maximum of NO at 04:00 LT, when it was still dark, must have resulted from an alternative chemical and/or biological production.

Figure 4Diurnal variations in NO concentrations, illuminance (I), DO, and Chl a concentrations in the surface water at anchor station B65.

Download

3.3 Photoproduction rates

The results of the NO photoproduction experiments are listed in Table 2. The photoproduction rates of NO in the BS, the NYS, and the SYS varied from not detectable to $\mathrm{5.07}\times {\mathrm{10}}^{-\mathrm{11}}$, 0.09 to $\mathrm{0.69}\times {\mathrm{10}}^{-\mathrm{11}}$, and 0.32 to $\mathrm{1.54}\times {\mathrm{10}}^{-\mathrm{11}}$ mol L⁻¹ s⁻¹, respectively. The average photoproduction rate of the whole study area was $\mathrm{1.14}\pm \mathrm{1.37}\times {\mathrm{10}}^{-\mathrm{11}}$ mol L⁻¹ s⁻¹. The photoproduction rates from the BS and YS are in good agreement with the rates reported from the Seto Inland Sea (Olasehinde et al., 2009, 2010). However, the mean NO photoproduction rates are higher than those from the central equatorial Pacific Ocean (Zafiriou and McFarland, 1981) and lower than those from Kurose River, Japan (Olasehinde et al., 2009) (Table 2), which mirrors the available ${\mathrm{NO}}_{\mathrm{2}}^{-}$ concentrations that are low in the open ocean and but high in a river.

Table 2Description of sampling stations and their seawater temperature (T), illuminance (I), mean surface downward UV radiation flux (UV), and photoproduction rates in the BS and the YS.

Download Print Version | Download XLSX

Table 3NO concentrations and flux densities from different regions.

Download Print Version | Download XLSX

The average photoproduction rate in the NYS was obviously lower than those in the BS and the SYS, consistent with illuminances of 22 450, 20 433, 27 852 lux and mean surface downward UV radiation flux 34.7, 32.1, 40.6 W m⁻² (European Centre for Medium-Range Weather Forecasts (ECMWF), reanalysis data) for the BS, the NYS, and the SYS, respectively. Enhanced photoproduction rates ($\ge \mathrm{2.00}\times {\mathrm{10}}^{-\mathrm{11}}$ mol L⁻¹ s⁻¹) occurred in the SYS, especially in the central part of the SYS and stations influenced by the Yellow Sea Cold Current (see Fig. 1). This is apparently in contrast to the distribution of NO, which showed the lowest concentrations in the central SYS (see above). The high illumination observed in the SYS (Table 2) not only leads to enhanced NO photoproduction (see Table 2) but also generates reactive oxygen species like ${\mathrm{O}}_{\mathrm{2}}^{\mathrm{•}-}$, ROO^•, and other OH-derived radicals, which in turn would efficiently scavenge NO (Olasehinde et al., 2010). Overall, the NO photoproduction rates showed a positive relationship between illuminance (p<0.01, r=0.884, n=26) and the mean surface downward UV radiation flux (p<0.01, r=0.865, n=26), indicating that the NO concentrations in the surface layer during daylight were dominated by photochemical production. However, we did not find a significant difference between the mean NO concentrations sampled during the day (179±80 pmol L⁻¹) and at night (195±140 pmol L⁻¹). This suggests that there was also a non-photochemical NO dark production in the surface layer, such as the nitrification process, or other chemical process, such as the process of ammonium (${\mathrm{NH}}_{\mathrm{4}}^{+}/{\mathrm{NH}}_{\mathrm{3}}$) oxidation into ${\mathrm{NO}}_{\mathrm{2}}^{-}$ and ${\mathrm{NO}}_{\mathrm{3}}^{-}$ (Joussotdubien and Kadiri, 1970). Caranto and Lancaster (2017) found that NO is an obligate bacterial nitrification intermediate produced by hydroxylamine oxidoreductase; Ward and Zafiriou (1988) also found that NO might be an intermediate for soluble byproducts of nitrification such as hydroxylamine.

Besides, Olasehinde et al. (2010) found that filtered and unfiltered seawater samples collected from the Seto Inland Sea showed no significant difference in NO photoformation rates, which suggested a negligible contribution of NO produced by photobiological processes from particle matter in seawater. However, Liu et al. (2017) reported that the rate difference between filtered (0.45 µm, $\mathrm{1.46}\times {\mathrm{10}}^{-\mathrm{12}}$ mol L⁻¹ s⁻¹) and unfiltered ($\mathrm{1.52}\times {\mathrm{10}}^{-\mathrm{12}}$ mol L⁻¹ s⁻¹) seawater samples from coastal waters of the YS indicated that particles in seawater could increase the NO production rate. The difference might be due to the composition of sample, filter membrane, etc. Thus, further research is needed.

3.4 Sea-to-air fluxes of NO

In the present study, the NO supersaturation was ubiquitous at all investigated sites. The supersaturation factors varied from 8 to 154, with an average of 47. Together with the fact that NO was supersaturated during daytime and nighttime at the 24 h station B65, which indicates that the BS and YS were a source of NO to the atmosphere. The sea-to-air flux densities ranged from $\mathrm{5.8}\times {\mathrm{10}}^{-\mathrm{19}}$ to $\mathrm{3.6}\times {\mathrm{10}}^{-\mathrm{15}}$ mol cm⁻² s⁻¹, with an average value of $\mathrm{4.5}\times {\mathrm{10}}^{-\mathrm{16}}$ mol cm⁻² s⁻¹ (Fig. 5). The comparison of wind speeds and flux densities reveal that the flux densities are mainly driven by the wind speed (Fig. 5) and that they showed significant positive relationship (p<0.01, r=0.938, n=20). Anifowose and Sakugawa (2017) also found that wind speed was an important factor governing NO flux at the air–sea interface. Our flux densities from the BS and YS were similar to those computed for the central equatorial Pacific Ocean and the Seto Inland Sea, while they are slightly lower than those computed for of Jiaozhou Bay waters (Table 2). Based on the YS area of 380×10³ km² and the BS area of 77×10³ km², the emission of NO to the atmosphere was estimated to be 9.0×10⁸ g N yr⁻¹ or 7.6×10⁷ g N month⁻¹. Ding et al. (2018) reported a satellite-derived average NO_x emission estimate for June over the BS and the YS in the period from 2007 to 2016 of about 1.3×10¹⁰ g N month⁻¹. The obvious very large discrepancy between the satellite-derived emission estimate and the one presented here results from the fact that Ding et al.'s (2018) estimate is dominated by the NO_x emissions from ships' diesel engines. This indicates that oceanic NO emissions to the atmosphere only account for a negligible fraction (∼0.6 %) of the NO_x emissions observed over the BS and YS.

Figure 5Wind speeds and flux densities of NO in the Bohai Sea and the Yellow Sea.

Download

3.5 NO mixed-layer budget

In order to estimate the contribution of different sources and sinks of NO in the surface layer of the BS and YS, we applied a simple box model. We assume that the surface layer is represented by the mixed layer, with a mean water depth of 15 m (Qiao et al., 2004). At steady state the loss of NO by air–sea exchange (F_ase) must be equal to the sum of the photoproduction rate (F_pp) in the mixed layer, the input from below into the mixed layer by diapycnal diffusion (F_dia) and other production or consumption pathways (F_poc), and advection into or out of the BS and YS mixed layer (F_adv). To this end, F_ase is given by

The mean F_ase was $\mathrm{4.5}\times {\mathrm{10}}^{-\mathrm{16}}$ mol cm⁻² s⁻¹ (see above). We assume that the NO photoproduction decreases linearly from the average rate of $\mathrm{1.14}\pm \mathrm{1.37}\times {\mathrm{10}}^{-\mathrm{11}}$ mol L⁻¹ s⁻¹ at the surface (see above) to 0 at 15 m. The mean F_pp was calculated to be $\mathrm{8.6}\times {\mathrm{10}}^{-\mathrm{12}}$ mol cm⁻² s⁻¹ in mixed layer. Since the mean surface NO concentration in the surface layer was statistically not different from the mean NO concentration in the bottom layer (see above), it is reasonable to assume that F_dia=0. Moreover, it seems reasonable to assume that advection of NO into or out of the BS and YS surface layer is zero. Since F_ase < F_pp, we conclude that F_poc should be negative, indicating that the bulk of the produced NO was rapidly consumed in the surface layer before its release to the atmosphere. Please note that we have some indications that there is also a dark production of NO (see above), meaning that the “true” NO surface production might be even higher. Chemical reactions with DO, OH, or ROO^• etc. are potential sinks for NO in the surface layer of the BS and YS (Ford et al., 1993; Olasehinde et al., 2010; Carpenter and Nightingale, 2015).