1 INTRODUCTION

Dissolved oxygen (DO) is an important factor influencing the stability of riverine and marine ecosystems. Hypoxia, which is characterized as the DO decreasing to a certain level that is unable to meet the DO demand, may lead to mass mortality, biodiversity reduction, and alterations of the community structure and the whole ecosystem (Diaz and Rosenberg, 1995; Rabalais and Turner, 2001). Nixon (1995) defined the threshold of hypoxia as 2 mg/L, which has been widely used in many studies. Since the onset of the observation history of DO, hypoxia has been frequently detected in the world's major coastal and estuarine ecosystems, together with a negative secular trend of DO (Diaz, 2001), which is likely to be caused by increasing nutrient inputs from land due to human activities (Anderson et al., 2002; Zhang et al., 2007; Danielsson et al., 2008; Diaz and Rosenberg, 2008; Halpern et al., 2008; Rabalais, 2015). It has been suggested that global warming driven by greenhouse gases also reduces DO in seawater (Stramma et al., 2008; Solomon et al., 2009). On the other hand, a series of reactions such as denitrification or metal oxide reduction could occur in hypoxic water (Middelburg and Levin, 2009), which might generate greenhouse gases and promote global warming (Naqvi et al., 2010).

The Bohai Sea is a marine system vulnerable to bottom deoxygenation (Jiang et al., 2016). The Bohai Sea is a semi-enclosed shallow sea in northern China with an average depth of 18 m, and the Bohai Strait is the only channel through which it can exchange water with the northern Yellow Sea. The topography of the Bohai Sea is characterized by a shallow central ridge and a deep channel (Laotieshan Channel) in the northern Bohai Strait. Due to the influence of the East Asian monsoon, the prevailing surface wind direction is southeasterly wind in summer and northwesterly wind in winter, while the winter monsoon has a significantly larger magnitude. With increasing human activities, in recent years the Bohai Sea has become more vulnerable to marine environmental problems such as eutrophication and red tides (Song et al., 2016). According to Lin et al. (2008) and Zhang et al. (2016), red tides are likely to occur from May to September, with the highest frequency in June. Thus, our study is of great importance due to it further revealing the characteristics and intrinsic mechanisms of DO in the Bohai Sea.

Using discrete observation data, the mechanism of DO variations in the Bohai Sea has been discussed in previous studies (Jiang et al., 2016; Zhang et al., 2016; Zhao et al., 2017). However, these data rarely satisfy both temporal and spatial continuity. Only a small proportion of the data have both long-term time series and comprehensive spatial coverage. Nevertheless, previous studies have found that the bottom layer of the Bohai Sea is prone to deoxygenation in summer (Jiang et al., 2016; Zhao et al., 2017; Wei et al., 2019a), and there is some consensus on its mechanisms. Zhang et al. (2016) indicated that the blocking effect of stratification was the key mechanism for the generation of oxygen-poor zones, and the decomposition of organic matter was an important reason for the development of oxygen-poor zones and bottom acidification. Zhao et al. (2017) studied the DO distributions in the Bohai Sea in August 2014, suggesting that the DO distribution at the bottom was correlated with the water depth and the stratification and decomposition of organic matter. Wei et al. (2019a) proposed that in summer, the bottom layer of the Bohai Sea tended to form two oxygen-poor zones corresponding to cold pools, indicating that vertical stratification had a significant influence on the distributions of DO. In summary, the occurrence of summer oxygen-poor zones in the Bohai Sea is probably due to seasonal stratification, and the decomposition of bottom organic matter contributes to the development of bottom deoxygenation. In addition, some studies have suggested that the ocean circulation and the residence time of water column also influence the oxygen-poor zones of the Bohai Sea (Wang, 2009; Zhai et al., 2012). Ocean circulation transports water masses with different properties, influencing the spatial distributions of temperature, salinity, or DO (Zhou et al., 2017). Moreover, circulation may indirectly affect DO by modulating the deposition of marine systems by inducing the aggregation and sedimentation of organic matter (Hu et al., 2011, 2013). Water columns with long residence times more easily suffer from hypoxia (Zhang et al., 2019).

Overall, studies concerning DO in the Bohai Sea have obtained considerable results and reached a consensus to some extent. However, the seasonal variability in DO and the controlling mechanisms in different seasons are still unclear. Previous studies seldom involve long-term and large-scale DO variations. Due to the limited possible survey times, many studies have focused only on the summer of one year. Generally, the data of those studies were discrete, with poor tempo-spatial coverage. Studies concerning long-term observations have mainly focused on the Huanghe (Yellow) River estuary-Liaodong Bay transection, which has difficulty displaying spatial variations. In addition, DO variations in spring and autumn are less studied and comprehensive studies should be performed to further uncover the seasonality of DO. Our study illustrates the spatial distributions of DO in each season in recent years (2008-2017). The observation data used in our study, which is introduced in Section 2.1, have high tempo-spatial continuity, meeting the need for a study of DO seasonality. This study aimed to display the seasonal variations in DO and to explore the potential mechanisms that modulate the variations to provide appropriate guidance for the development of marine activities in the Bohai Sea. In addition, this work will help optimize relevant ecological models, achieve more reliable hypoxia forecasting, and reduce the extent and number of disasters induced by hypoxia.

2 MATERIAL AND METHOD 2.1 In-situ observation

In-situ observation data were collected by the North China Sea Environmental Monitoring Center, State Oceanic Administration. Large-scale voyage surveys were conducted in the spring (May), summer (August), and autumn (October) every year during 2008-2017. The survey stations were encompassed by our study area (Fig. 1), covering the whole Bohai Sea. Notably, the spatial coverage of the stations was similar in each survey, with a minimum number of 37 stations (Table 1). The distributions of the stations in each survey are shown in Supplementary Figs.S1 & S2. The observed variables included the temperature, salinity, DO, and dissolved inorganic nitrogen (DIN). These variables were recorded in two layers, with the surface layer recorded 1 m under the free surface and the bottom layer recorded 2 m above the seabed. Water samples were taken by a multi-bottle CTD-type water sampler at the surface and bottom. The temperature and salinity were recorded by corresponding sensors, while DO and DIN were determined onboard by iodometric titration and spectrophotometry, respectively, using the collected water samples. In addition, DIN was used in our study since it occupies a high proportion of the total concentration of nutrients in the Bohai Sea. The DIN concentration was considered the sum of the NO₃-N, NO₂-N, and NH₄-N concentrations, following Shi et al. (2009).

2.2 Chlorophyll-a concentration

Monthly mean sea surface chlorophyll-a concentration data used in this study were derived using the Moderate-Resolution Imaging Spectroradiometer (MODIS) from January 2008 to December 2017. The Level-3 Equidistant Cylindrical Mapped Image with a spatial resolution of 4.64 km×4.64 km was acquired from the U.S. National Aeronautics and Space Administration (NASA) website at https://oceandata. sci.gsfc.nasa.gov. The spatial range of the chlorophyll-a concentration data fits our study area, and the May, August, and October mean data were used when determining the climatological seasonal mean chlorophyll-a concentration field.

2.3 Method

Due to the varying number of stations with high spatial coverage (Table 1) in each survey, it is necessary and reasonable to interpolate the in-situ observations to our study area. By applying convergent weighted-averaging interpolation (Barnes, 1964), all the observation data are interpolated to a 56×41 regular grid. The interpolation method is based on the Fourier integral, which can be applied in the practical form of

where g(x, y) is the interpolated field determined by a reasonable r value (radius of influence) and all M number of observations circled by it, and f_j is the j^th observation within the radius of influence. The weight factor η can be expressed as η=exp(-r²/4k), which is influenced by the distance between a data point and the grid point. k is a parameter determining the shape of the weight factor, which is calculated using k=r²/16. We choose a wide range of r values and calculated several two-dimensional fields, e.g., temperature, as well as the corresponding mean absolute error (MAE) between the raw observations and the interpolated values. Then, the field with the minimum MAE (Supplementary Fig.S3) was selected as the best field for further analysis. Additionally, near-shore interpolations were removed to eliminate the marginal values with relatively high errors caused by the method.

According to García and Gordon (1992), the DO solubility equation can be represented as:

where O_2S is DO solubility, t is temperature and S is salinity. The rest of the parameters are constants (Supplementary Table S1) listed by García and Gordon (1992). Apparent oxygen utilization (AOU) is calculated using the observed temperature, salinity and DO, which can be expressed as:

where O_2S is the DO solubility at equilibrium with the atmosphere. O_2O represents the observed DO.

To quantify the stratification in the Bohai Sea in Section 4.2, the bottom-surface density gradient (∆ρ/z) is calculated as an indicator of the stratification intensity following Wei et al. (2019a), which is determined by

where ∆ρ denotes the difference between the bottom density (ρ_b) and surface density (ρ_s). z represents the depth. In fact, the density gradient here might be smaller than that in the stratified layers because the calculated ∆ρ/z only reflects the mean state of stratification intensity between the surface and the bottom.

3 RESULT 3.1 Seasonality

Obviously, both the surface and bottom DO displayed strong seasonal variability (Fig. 2). The concentrations of spring surface DO (Fig. 2a) were higher in the northwestern Bohai Strait and the southeastern Liaodong Bay, displaying a tongue structure spreading from the northern Bohai Strait to Bohai Bay. For the bottom layer (Fig. 2b) in spring, the bottom DO displayed nearly the same distribution as that of the surface, although it did not show an extended tongue structure, unlike the surface DO. The difference was clearly captured in Fig. 2c, with a high-difference region (> 0.5 mg/L) located in the northeastern Huanghe River estuary. Additionally, some of the domains were covered by negative differences, such as the southern Bohai Strait, southern Liaodong Bay, and east of the Luanhe River estuary, although their absolute values were much lower (< 0.3 mg/L).

Surface and bottom DO distributions showed significant differences in summertime (Fig. 2d-e). For the surface layer, there were three high-DO regions located in the northeastern Huanghe River estuary, the northwestern Bohai Strait, and the southeastern Liaodong Bay, which were similar to the distributions in the summer of 2014 presented by Zhao et al. (2017). Since these high-DO regions coincided well with the spring surface DO (Fig. 2a), we assumed that the surface DO retained its spring structure until summer, accompanied by overall decreased magnitudes and weak spatial variability (note the smaller summertime statistical parameters relative to spring in Table 2). On the other hand, the summer bottom DO has been widely studied (Jiang et al., 2016; Zhang et al., 2016; Zhao et al., 2017; Wei et al., 2019a) because hypoxia is most likely to occur at the bottom in this season. Our results showed that summer bottom DO distributions (Fig. 2e) presented a dualcore spatial pattern, with two distinct low-DO cores in the south and north. The low-DO regions spread from the southern trough to the northern trough, bypassing the central shoal. These distributions were also similar to those detected by Zhang et al. (2016), Zhao et al. (2017), and Wei et al. (2019a), suggesting that the summer bottom of the Bohai Sea was vulnerable to deoxygenation from 2008 to 2017.

Autumn DO (Fig. 2g-h) showed the minimum spatial variability compared to the other seasons, with the smallest standard deviations (SDs) of 0.19 mg/L and 0.21 mg/L at the surface and bottom, respectively (Table 2). The surface DO in autumn was lower in the southwestern Bohai Strait and northeastern Huanghe River estuary, while it was higher in Liaodong Bay. By comparing Fig. 2g-h, it could be found that the distributions of the surface and bottom DO in autumn were slightly different. Specifically, the DO values in the northwestern Bohai Strait were slightly lower at the bottom than at the surface, which is clearly depicted in Fig. 2i as well (with differences larger than 0.3 mg/L). Nevertheless, the surface-bottom DO differences were small (< 0.1 mg/L) in most areas of the domain. The seawater was highly vertically homogeneous, indicating the strong vertical mixing in this season.

By computing the spatial mean values of each climatological seasonal mean DO field, the statistical characteristics of DO are given in Table 2 to show its general variations. From Table 2, obviously all statistical values (maximum, minimum, and mean) except the SD were the largest in the spring and smallest in the summer. By combining Table 2 with Fig. 2, we found that the DO generally peaked in spring, and then decreased rapidly to its minimum in summer, and then, it finally recovered in autumn.

3.2 Oxygen-poor zones

The summer dual-core distributions in previous studies and our work both indicated that the summer bottom Bohai Sea deserves more attention because of its potential to experience hypoxia. We counted the years when the summer bottom DO was lower than 5 mg/L from 2008 to 2017. The resultant frequency distributions are shown in Fig. 3, as well as the summer bottom AOU distributions. Obviously, the regions with high frequencies were coincident with the summer bottom DO (black isolines, also refer to Fig. 2e). Simultaneously, these regions were covered by high AOU values, as well, suggesting drastic DO consumption in these regions. This phenomenon further demonstrated the high possibility of these regions suffering from bottom deoxygenation. As shown in Fig. 3, we considered the regions encompassed by the black isoline of 5.5 mg/L as the dual-core region in the subsequent analysis.

4 DISCUSSION

In general, previous studies of DO are based on discrete observations or ocean models, in which potential factors that affect DO variations are extensively discussed (Jiang et al., 2016; Zhang et al., 2016; Zhao et al., 2017). There are a series of factors affecting DO distributions in seawater, which can be approximately classified as physical and biochemical factors. The physical factors include seawater temperature, salinity, and stratification. Specifically, a rise in temperature and salinity would reduce the solubility of DO in seawater, resulting in a decrease in DO (Carstensen et al., 2014; Schmidtko et al., 2017). In contrast to temperature, salinity exerts a weak influence on DO solubility (García and Gordon, 1992). Additionally, the stratification and mixing process can also affect DO distributions by restraining/stimulating vertical water exchange (Obenour et al., 2012). On the other hand, biochemical factors mainly consist of photosynthesis, air-sea interactions, and the decomposition of organic matter. These factors are also considered the main sources or sinks of DO in seawater (Diaz, 2001; Zhao et al., 2017). In particular, the natural balance between the DO supply and consumption can be disequilibrated by eutrophication, which stimulates primary production and brings a large amount of organic matter to the subsurface, resulting in an increasing DO demand for the decomposition of organisms (Carstensen et al., 2014; Rabalais et al., 2014). The traditional view of the circulation pattern in the Bohai Sea is mainly based on discrete observations and does not have a consensus, causing difficulties in researching the circulation and its impact on DO. Based on the study of Guan (1994), two cyclonic circulations are located in the northern and southern trough, and a basin-scale cyclonic circulation constitutes the summer circulation pattern of the Bohai Sea. Furthermore, since more than 80% of sedimentary organic matter in the central Bohai Sea comes from algae (Liu et al., 2015), these cyclonic circulations could induce the aggregation and sedimentation of organic matter, which would cause intensified decomposition at the bottom (Hu et al., 2011, 2013). Nevertheless, more observations are needed to accurately determine the real circulation patterns in the Bohai Sea and the resulting variable transport.

Based on in-situ observations and ocean models, previous studies have revealed that DO is influenced by different factors, including temperature, salinity, stratification, circulation, wind, and biochemical processes (García and Gordon, 1992; Babin et al., 2004; Obenour et al., 2012; Zhao et al., 2017; Wei et al., 2019b). Furthermore, the dominant factors are variable among different marine systems (Levin, 2018). For the Bohai Sea, the modulation of DO is hard to explain by any single factor. In the following subsections, we focus on the main factors which include temperature, stratification, and biochemical factors, to explore the relationships between DO and these factors.

4.1 Sea water temperature

The temperature controls DO by physically affecting the DO solubility in seawater (Schmidtko et al., 2017). With increasing temperatures, the DO solubility drops, resulting in a reduced DO concentration. Based on a previous study, salinity has a much lower influence on DO than temperature (García and Gordon, 1992). We examined the relative contributions of temperature and salinity on DO solubility in the Bohai Sea. For the given mean temperature/salinity and the salinity/temperature change climatological seasonal field, we calculated the correlation coefficients between the raw DO solubility field (according to the equation of García and Gordon (1992)) and the solubility field computed with a fixed temperature or salinity. Moreover, their SD ratios were calculated as well, which are shown in Table 3. Obviously, both the correlations and SD ratios are significantly higher with a fixed salinity and varying temperature, further demonstrating that salinity has an almost negligible influence on DO solubility in the Bohai Sea. Thus, the salinity will not be subsequently discussed.

The correlation coefficients between the temperature and DO field (Fig. 2) were calculated and are shown in Table 4 to evaluate the similarity of their distributions. By comparing the surface and bottom temperatures in the spring, we see that the bottom cold pool has formed in this season, and the dual-core region displays high temperature differences (Fig. 4c), especially in the northern trough. In summer, previous studies have demonstrated that there are cold pools at the bottom of the Bohai Sea. Using observational data from large-scale surveys in different years, Lin et al. (2006), Zhang et al. (2016), Zhao et al. (2017), and Wei et al. (2019a) captured nearly the same summer temperature spatial pattern, which is well reproduced in our work (Fig. 4e). The summer surface-bottom temperature differences are strengthened compared to the spring temperature and display dual-core distributions. The similarity may imply the importance of the thermocline (or stratification), which will be discussed later. The autumn temperature distributions are similar at the surface and bottom, which reveal the smallest vertical differences, indicating that the seawater is vertically homogeneous in autumn.

By comparing Fig. 2 with Fig. 4, we find that the DO distributions coincide well with the temperature in spring and autumn (note the high R values in Table 4). Specifically, low (high)-DO regions correspond to high (low)-temperature regions, indicating that they are negatively correlated in spring and autumn. However, the summer temperature and DO distributions are not spatially correlated, with R=-0.32 and R=0.43 at the surface and bottom respectively, suggesting a decreased dominance of the temperature and, probably, other factors, such as stratification or biochemical processes, that modulate DO distributions. The strong connections in spring and autumn are also captured by the temperature-DO diagram (Fig. 5), with significantly higher negative slopes in spring and autumn relative to those in summer. The slopes are comparable in these two seasons, with a minimum value of -0.16 mg/(L·℃). Although the summer temperature and DO did not display a close relationship, their corresponding surface-bottom differences showed similar spatial patterns (comparing Fig. 4f with Fig. 2f), with R=0.82. Consequently, by physically regulating DO solubility in seawater, the temperature dominates the seasonal cycle of DO in spring and autumn. The summer DO is probably mainly controlled by other processes, which will be discussed later.

4.2 Stratification

It has been illustrated in Section 4.1 that the temperature has a poor correlation with DO in summer. However, the summer surface-bottom temperature differences display a close relationship with the summer bottom DO (or surface-bottom DO differences). Thus, we speculate that the thermocline influences summer bottom DO distributions by impeding vertical mixing of seawater. With the intensification of the thermocline, the vertical exchange of seawater will be further restrained, resulting in a reduced supply of DO from the surface to the bottom and thus affecting the bottom DO distributions. To confirm whether the summer stratification in the Bohai Sea is dominated by temperature or salinity, the temperature/salinity is fixed as its spatial mean value, while the salinity/temperature retains the original value. On this basis, the stratification intensity is computed and compared (Supplementary Figs.S4-S5). Using the algorithm (∆ρ/z) explained in Section 2.3, the distributions of the stratification intensity are depicted in Fig. 6. Upon comparing Supplementary Figs.S4-S5 with Fig. 6b, it is obvious that the contributions of the temperature are significantly higher than those of the salinity, especially in dual-core regions. Consequently, the stratification is actually dominated by the temperature in the Bohai Sea.

The dual-core regions are much deeper than the Liaodong Bay, Bohai Bay, and Laizhou Bay, as well as the central shoal. Vertical water mixing in these regions is weaker due to its greater depth. With this background, as depicted in Fig. 6, stratification begins to develop in dual-core regions because of intensifying solar radiation, although it is weak in this phase (< 0.07 kg/m⁴). Almost no stratification exists in shallow regions because of strong mixing. As summer arrives, the stratification has been highly strengthened in the dual-core region due to the strong solar radiation and weak summer monsoon, with a maximum of 0.11 kg/m⁴. Notably, the summer bottom DO (black isolines in Fig. 6b) has almost the same spatial pattern as that of the stratification (R=-0.86 in Table 5), demonstrating the importance of stratification in the formation and development of summer bottom deoxygenation. As the stratification corrupts in autumn, the dual-core pattern disappears due to strong vertical mixing. Additionally, the correlation coefficient between the summer surface-bottom DO differences and stratification reaches 0.85, indicating that the spatial distributions of the summer bottom DO are probably dominated by stratification, i.e., the summer bottom DO decreases with the strengthening of stratification.

Concerning the relationship between the stratification and bottom DO (or surface-bottom DO differences), their scatter diagram is shown in Fig. 6d. Obviously, high correlations are also revealed as shown in Fig. 6b and Table 5. Specifically, the linear regression shows that the bottom DO decreases with the strengthening of stratification, while the correlation between the stratification and surface-bottom DO differences is reversed, with relatively close absolute slopes of -21.57 and 23.22 (mg·m⁴)/(L·kg), respectively.

4.3 Biochemical processes

The biochemical processes that control DO variations mainly include photosynthesis and the decomposition of organic matter, which are considered important sources or sinks of DO in summer (Diaz, 2001; Zhao et al., 2017). Generally, in the Bohai Sea, phytoplankton blooms appear in spring and the concentration of chlorophyll a reaches its maximum in summer due to the strengthening of solar radiation and the rising seawater temperature. Moreover, spring blooms can provide a large amount of organic matter, resulting in intensified decomposition and consequent hypoxia in summer (e.g. Malone, 1987; Pomeroy et al., 2006). On the other hand, eutrophication also affects the decomposition of bottom organic matter in summer by intensifying photosynthesis in the euphotic zone, as Carstensen et al. (2014) and Rabalais et al. (2014) presented. Thus, in this section, we analyze the seasonal variations in the chlorophyll a and DIN concentrations and their impact on DO variations.

4.3.1 Chlorophyll a

The spatial distributions of chlorophyll a show distinct seasonal variability (Fig. 7a-c). The spatial mean values of the chlorophyll-a concentration are 4.71, 5.22, and 3.88 mg/m3 in spring, summer, and autumn respectively. Consequently, the surface chlorophyll-a concentration in the Bohai Sea is highest in summer, which is coincident with previous studies. Spring chlorophyll a is relatively higher on the eastern Qinhuangdao Coast. The newly produced organic matter from spring is continually deposited on the bottom. As summer arrives, the regions with higher chlorophyll-a concentrations expand to the whole dual-core region. By comparing the surface chlorophyll-a concentration (Fig. 7a-c) with the surface saturation of DO (Fig. 7d-f), it is determined that the regions east of Bohai Bay are slightly oversaturated in spring and summer, only covering the southern part of the dual-core regions. In addition, the surface DO is also higher in the southern regions (note the black isolines of 9.5 mg/L in Fig. 7a and 7.5 mg/L in Fig. 7b). This phenomenon implies that as a source of DO, photosynthesis mainly influences the southern dual-core regions. On the other hand, considering the bottom DO distributions in summer, the growth and flourishing of phytoplankton in the southern dual-core regions may provide more organic matter for the bottom, intensifying bottom decomposition. Under the combined effects of stratification and decomposition, deoxygenation appears at the bottom in summer. Notably, the chlorophyll-a concentration is generally high in coastal regions in each season, where the seawater is vertically homogeneous. The concentration is also not spatially variable, with standard deviations of 0.83, 1.04, and 0.48 mg/m³ in spring, summer, and autumn, respectively. Nevertheless, further studies should be performed with accurate observations or models to quantify the relationship between chlorophyll a and DO.

4.3.2 DIN

Eutrophication may stimulate the primary production of phytoplankton in the spring, causing superabundant organic matter to be transported to the bottom in summer and resulting in an increased need for DO for the decomposition of those organic matter types (Carstensen et al., 2014; Rabalais et al., 2014). With the impedance of strong stratification in summer, the DO consumed at the bottom can barely be replenished by surface water, inducing hypoxia (Wei et al., 2019a). In the Bohai Sea, the spring DIN (Fig. 8a-b) has similar distributions at the surface and bottom, with a higher DIN in Liaodong Bay, Bohai Bay, and Laizhou Bay. The summer bottom DIN distributions are noticeably different compared to those of the surface. Although the summer bottom DIN (Fig. 8e) does not seem higher in the dual-core regions, it does increase significantly in the regions relative to the surface DIN. The distributions of surface-bottom DIN differences (Fig. 8f) are coincident with the dual-core region (the correlation coefficient between the surface-bottom DIN differences and the summer bottom DO reaches 0.85), as well as the distributions of the summer bottom AOU (R=-0.89, note the black isolines in Fig. 8f). This phenomenon indicates that the bottom DIN may increase in summer due to the impedance of strong stratification and the strengthened decomposition of organic matter. In autumn, with the disappearance of stratification, the increased bottom DIN might rise to the surface through strong vertical mixing. Thus, the surface and bottom DIN (Fig. 8g-h) in autumn are approximately equal, with little surface-bottom differences (Fig. 8i).

To combine chlorophyll a, DIN, and DO, Fig. 9 shows the distributions of the climatological mean surface chlorophyll-a concentration, surface-bottom difference of DIN, and bottom AOU in summer, with the distributions of the summer bottom DO superimposed (note the black isolines). As a result, we found that the surface chlorophyll-a concentration (Fig. 9a) was slightly higher in the dual-core regions, while the surface-bottom differences in DIN and AOU were significantly high in these regions. This spatial pattern in Fig. 9b-c is also similar to the distributions of dissolved inorganic carbon (DIC) presented by Zhao et al. (2017). The significantly increased DIN and AOU in the dual-core regions suggest strong decomposition of organic matter and DO consumption. On the one hand, the organic matter newly produced by phytoplankton since spring is continually transported to the bottom layer, intensifying bottom decomposition. The process consumes more DO and generates more DIN. On the other hand, as summer arrives, the vertical exchange of water is blocked by strengthened stratification. As a result, the bottom DO is irreversibly consumed and high-DIN water is retained at the bottom.

In addition, to discover the different mechanisms of the two oxygen-poor zones in summer, we separated the dual-core regions into two parts. The dual-core regions to the north and south of 39°N are defined as the northern and southern cores, respectively. By computing the statistical mean values of the bottom DO, stratification intensity, bottom DO saturation, and bottom DIN in the two oxygen-poor zones in summer (Table 6), we find that DO in the southern core is slightly lower, with a difference of 0.063 mg/L. Meanwhile, the stratification seems stronger in the southern core, accompanied by a lower DO saturation and a higher DIN concentration. These results indicate that the southern oxygen-poor zone probably experiences more severe deoxygenation than the northern oxygen-poor zone under the combined effect of stronger stratification and bottom decomposition of organic matter. The stronger stratification further hinders vertical mixing, resulting in the greater consumption of DO and an increase in DIN concentration induced by bottom decomposition.

5 CONCLUSION

Using in-situ observation data collected in large-scale voyage surveys, the seasonal variation in DO and its controlling mechanisms were studied. The results show that the surface and bottom DO distributions are similar in spring and autumn, with overall small surface-bottom differences. Temperature dominates DO variations in the two seasons but has a minimal effect in summer. The situation is different at the bottom in summer, where the DO displays a dual-core spatial pattern, with two oxygen-poor zones located in the north and south of the Bohai Sea. This dual-core region has been demonstrated to be vulnerable to bottom deoxygenation. The analysis of the mechanisms showed that the summer bottom DO distributions are controlled by both stratification and bottom decomposition. On the one hand, stratification impedes vertical water exchange, restraining the replenishment of the bottom DO and the upward movement of high-DIN waters. On the other hand, the growth and flourishing of phytoplankton may provide more organic matter for the bottom, resulting in strengthened bottom decomposition, a reduced DO and elevated DIN concentrations. Furthermore, the differences between the two oxygen-poor zones in the dual-core regions in summer were studied. The stronger stratification in the southern part of the dualcore regions might be the key factor that induces more severe bottom deoxygenation in the south.

6 DATA AVAILABILITY STATEMENT

The in-situ observation data that support the findings of this study are available from the North China Sea Environmental Monitoring Center, but restrictions are applied to the availability of these data, which were used under license for the current study and so are not publicly available. However, the data are available from the authors upon reasonable request and with permission from the North China Sea Environmental Monitoring Center.

The chlorophyll-a concentration data that supports the findings of this study is available from MODIS. The hyperlink of the data is: https://oceandata.sci.gsfc.nasa.gov.

7 ACKNOWLEDGMENT

We thank the anonymous reviewers and the editor for their valuable comments that helped greatly improve the manuscript.

Electronic supplementary material

Supplementary material (Supplementary Table S1 and Figs.S1–S5) is availiable in the online version of this article at https://doi.org/10.1007/s00343-021-0235-6.