Nitrogen dynamics in a highly urbanized coastal area of western Japan: impact of sewage-derived loads

Study area

The study area is located in the northeastern part of Osaka Bay in the Seto Inland Sea in western Japan (Fig. 1). The Seto Inland Sea is the largest semi-enclosed coastal sea in Japan. The area is characterized by a temperate climate, with a mean annual precipitation of 1270 mm and a mean temperature of 15.1 °C. Osaka Bay is one of the most enclosed bays in the inland sea and has a water surface area of 1447 km² and mean depth of 30.4 m (International EMECS Center 2008). The watershed population of the bay is approximately 16 million, including residents of the major cities of Osaka, Kobe, Kyoto, and Sakai (Nishida et al. 2012). Rapid economic growth progressed urbanization significantly in the Osaka metropolitan area from the 1950s to the 1970s. During this period, increased nutrient loads from domestic and industrial activities induced serious eutrophication and an oxygen deficit in the seawater of the bay (ANSIOB 1996; Nakatsuji 1998). Since the 1970s, the regulation of chemical oxygen demand (COD), phosphorus, and nitrogen loads and the development of sewage treatment systems have decreased nutrient loads and the presence of organic matter in the Bay (International EMECS Center 2008; GESAP 2017).

The target area for this study is the inner bay, near the cities of Osaka and Sakai (Fig. 1). Almost 100% of these cities are covered by sewage systems, with 12 public sewage treatment plants (STPs) in Osaka and 3 in Sakai (OPG 2016). However, the removal of nitrogen from this wastewater remains insufficient for preventing nitrogen loading, as most STPs have not yet installed advanced treatment facilities for dissolved nitrogen, such as nitrification–denitrification processes. In Osaka, the average concentration of total nitrogen (TN) and TN load of the influent to these plants are 26 mg L⁻¹ and 44.0 t day⁻¹, respectively, which are reduced to 11 mg L⁻¹ and 18.6 t day⁻¹ in the treated sewage effluent (Osaka City 2016). In Sakai, the average concentration of TN and TN load of the influent to the Sambo STP are 30 mg L⁻¹ and 2.3 t day⁻¹, respectively, which are reduced to 3.4 mg L⁻¹ and 0.3 t day⁻¹ in the treated sewage effluent (SCWSB 2015). In addition, DIN, such as nitrate (NO₃⁻) and ammonium (NH₄⁺), is a major component of the TN in the treated sewage effluent (TSE) (Osaka City 2017). This indicates that a relatively large proportion of the DIN is discharged in the effluent of these STPs to rivers and coastal areas. Most of the sewage systems in Osaka are “combined systems,” which result in increased discharge after high rainfall events because of an inflow of storm water (Broadhead et al. 2013). Therefore, an increase in sewage-derived nitrogen loads from heavy rain has been recognized as a new problem in recent years. We focused on three STPs: Konohana (KH) and Suminoe (SU) in Osaka and Sambo (SA) in Sakai, which are all located in areas of reclaimed land near the coastal bay (STP in Fig. 1). The designed treatment populations for the STPs are 64,302 for KH, 386,431 for SU, and 175,500 for SA (OPG 2017). The treated effluents from KH and SU are discharged directly to the bay, whereas the effluent from SA is discharged to the mouth of the Yamato River.

Field campaigns and sample collection

Vertical profiles of water temperature and electrical conductivity (EC) were measured about every 300 m along the observation lines using a conductivity-temperature-depth (CTD) sensor (CTD Diver, Schlumberger) and that of chlorophyll-a (Chl-a) concentration via a fluorescent chlorophyll and turbidity sensor (INFINITY-CLW, JFE Advantech Co., Ltd.).

Seawater samples were collected at the surface at approximately 100 m intervals and from less than 1 m above the bottom of the coastal seabed at 300 to 1000 m intervals along the observation lines. At 600 to 1000 m intervals along the observation lines, sediment samples from the surface of the seabed were collected using an Ekman grab sampler. TSE, river water, sediment pore water, and offshore seawater were collected as endmembers of coastal seawater in the study area. TSE samples were collected from land at the effluent discharge point from SA in August 2013. River water samples were collected 2 km upstream from the mouth of the Yamato River in August 2013 and September 2014. Sediment pore water was collected using piezometers installed 50 cm deep using a PVC pipe at the same location for the river water sampling in the Yamato River in August 2013. Offshore seawater samples were collected approximately 7 km from the STP on the SA line in September and October 2014. Water samples were filtered in the field using 0.2 μm cellulose ester filters for analysis of nutrients and stable isotopes and 0.7 μm glass microfiber filters for analysis of particulate organic matter (POM). The samples of water and filtrates were stored in a freezer until analysis.

Chemical and isotopic analyses

DIN as a combination of NO₃⁻, nitrite (NO₂⁻), and NH₄⁺; DTN; and dissolved inorganic phosphorus (DIP) measured as phosphate (PO₄³⁻) in water samples were determined using a continuous-flow automated analyzer (SWAAT; BL TEC K. K.). Concentrations of dissolved organic nitrogen (DON) were calculated from the difference between DTN and DIN. The organic nitrogen content and δ¹⁵N of POM and surface sediment samples were determined using an isotope ratio mass spectrometer (Delta V Advantage and Flash EA 1112; Thermo Fisher Scientific, Inc.). The δ¹⁵N in NO₃⁻ was determined using the bacterial denitrifier method (Sigman et al. 2001) with an isotope ratio mass spectrometer (Delta V Plus; Thermo Fisher Scientific, Inc.).

Impact zone of sewage-derived nitrogen

Figure 2 shows the spatial variations in 2014 of EC, DIN, DON, the molar ratio between DIN and DON (DIN/DON), and DIP in the surface seawater along the three observation lines (KH, SU, and SA), starting from the discharge point of the treated sewage and ending approximately 5 km offshore. The KH and SA lines were observed in October, whereas the SU line was observed in August. The tidal and weather conditions, excluding the air temperature, were similar between these two periods (Table 1). The EC was lowest (< 21 mS cm⁻¹) and the concentration of nutrients and DON was highest (DIN > 3.0 mg L⁻¹, DIP > 0.17 mg L⁻¹, and DON > 0.7 mg L⁻¹) at the STPs. DIN, mostly NO₃⁻ and NH₄⁺, was a few times higher than DON, because DIN is a major form of nitrogen derived from TSE, as reported previously (Osaka City 2017). The DIN concentration for the SU line was higher than that for the KH line within 1–2 km of the STPs (Fig. 2b). As already mentioned, the DTN concentration in the TSE was lower for SU (8 mg L⁻¹) than for KH (14 mg L⁻¹) (Table 1); however, the designed treatment population for SU is six times larger than that for KH (OPG 2017). This indicates that the impact of TSE-derived nitrogen is more significant for the SU line. In addition, the higher temperature conditions in August may have accelerated the decomposition of organic matter and the nitrification process, resulting in elevated DIN concentrations. EC increased and nutrients and DON decreased significantly within 1–2 km of the STPs through mixing with seawater of the KH and SU lines. These results indicate that the impact zone of TSE-derived nitrogen is about 1–2 km offshore from the STPs. The significantly low EC (< 10 mS cm⁻¹) and relatively high DON and DIP in this zone of the SA line may have been influenced by the Yamato River (Fig. 2a–e). The DTN and dissolved total phosphorus (DTP) concentrations in the TSE from SA were approximately 4 and 0.1 mg L⁻¹, respectively (SCWSB 2015). This suggests that the DIP increased (> 0.2 mg L⁻¹) as a result of phosphorus desorption via mixing of seawater (Fox et al. 1986). However, the DTN (DIN+DON) concentration in this zone was approximately 4 mg L⁻¹, which was a similar level to that of the TSE. In addition, DIN/DON was similar among the lines. These results suggest that TSE-derived nitrogen as well as the Yamato River may have influenced the SA line.

Seasonal and temporal changes in the sewage impact zone

The spatial variations of water temperature, EC, Chl-a, NO₃⁻-N, and NH₄⁺-N in the surface seawater of the KH line between October 2014 and February 2015 are shown in Fig. 3. Water temperature was lower in February than in October because of the difference in air temperature (Table 1). The EC was lower and the NO₃⁻-N and NH₄⁺-N concentrations were significantly higher in February than in October from the STP to 2 km offshore. The relatively high Chl-a concentration suggests that phytoplankton were still actively taking up nutrients in February (Fig. 3c). These results suggest that the sewage impact zone expanded in February. As noted previously, the flow of the Yodo River may affect the results for the KH line through the small channel approximately 2.5 km from the STP (Fig. 1). This indicates the effects of the TSE caused the expansion of the impact zone.

A slightly higher DTN concentration in the TSE in February (18 mg L⁻¹) than in October (14 mg L^− 1) may have increased the NO₃⁻-N and NH₄⁺-N levels near the STP (Table 1). The tidal range was greater in February than in October, and the February observation was conducted during ebb tide (Table 1). This suggests that the plume of TSE-derived nitrogen was transported offshore with less mixing and dilution by seawater. The water temperature was nearly equal to the air temperature in October; however, it was higher than that of air temperature (> 10 °C) from the STP to approximately 2 km offshore in February (Fig. 3a, Table 1). This suggests that another possible cause for the expansion of the impact zone is the difference in water temperature between the TSE and the surface seawater. TSE is usually discharged at more than 20 °C throughout the year because of the operating parameters of the biological activated sludge treatment process (Hashimoto and Sudo 1986). This would intensify the buoyancy of the effluent plume and allow the plume to extend further offshore during periods of low seawater temperature. The effect of thermal effluent from STPs on water bodies has previously been reported for urban river systems (Nakamuro et al. 2006; Kinouchi et al. 2007).

Next, we discuss the nitrogen dynamics at approximately 2.5 km from the STP, to examine the effect of TSE-derived nitrogen. As noted above, the influence of the Yodo River inflow should be considered for more than 2.5 km offshore from the STP on the KH line. Vertical profiles of EC and Chl-a from the STP to 2.5 km offshore on the KH line during the ebb and rising tide periods in September 2016 are shown in Fig. 4. The surface EC at 2.5 km from the STP was relatively high compared to that of the other sites (Fig. 4a). This suggests that the influence of the Yodo River inflow was small within 2.5 km of the STP. The EC at depths of less than 5 m was relatively high during the ebb tide because the tidal levels in the observation period were higher during the ebb tide than during the rising tide. Both the EC and the Chl-a concentrations varied significantly between the surface and bottom for both periods. The EC was relatively low and the Chl-a concentration was high at the surface compared to those at the bottom. These results suggest that the surface seawater is significantly affected by TSE-derived nitrogen. The vertical profiles in Fig. 4 suggest that the spatial and temporal variations in DIN should be compared between these two depths.

Figure 5 shows the spatial variations of EC, NO₃⁻-N, and NH₄⁺-N at the surface and less than 1 m above the bottom of the seabed for seawater on the KH line in October 2014 and September 2016. The EC was lower at the surface than at the bottom in both periods (Fig. 5a, b). The NO₃⁻-N concentration was relatively higher at the surface than that at the bottom; however, it was significantly lower in September than in October at the surface (Fig. 5c, d). These EC and NO₃⁻-N results suggest that the TSE-derived nitrogen was diluted by rainwater in September because of the antecedent rainfall (Table 1). The NH₄⁺-N concentration was at a similar level (< 0.6 mg L⁻¹) between the two depths in October; however, it was significantly higher at the bottom than at the surface in September (Fig. 5e, f).

The results during the ebb and rising tides were compared in September 2016. The EC was lower during the rising tide than during the ebb tide from the STP to 1.5 km offshore at the surface (Fig. 5a) but only near the STP at the bottom (Fig. 5b). Moreover, the NO₃⁻-N and NH₄⁺-N concentrations were relatively high during the rising tide in the lower EC zone at the surface (Fig. 5c, e). These results suggest that the TSE-derived nitrogen was transported about 1.5 km offshore from the STP. In contrast, both the NO₃⁻-N and NH₄⁺-N concentrations were relatively high during the ebb tide at the bottom. The NH₄⁺-N concentration, in particular, was detected at a level more than 10 times higher than the concentration at the surface (Fig. 5d, f). These results suggest that the sources of nitrogen were different between the surface and the bottom, especially in September.

Previous research confirmed that estuarine circulation flow is dominant in the inner part of Osaka Bay (Nakajima and Fujiwara 2007; Kobayashi et al. 2017). They imply that nitrogen at the bottom layer of the KH line originated from the bottom seawater of Osaka Bay. However, the reported NH₄⁺-N concentrations of the bottom seawater in the bay were less than 0.3 mg L⁻¹ (RIEAFOP 2006-2015), which is much lower than the NH₄⁺-N concentrations detected in the bottom layer of the KH line in September (Fig. 5f). On the other hand, Onodera et al. (2013) confirmed exchange between surface water and sediment pore water caused by tidal variation in the study area. These results suggest that nitrogen supply via exchange between seawater and sediment pore water occurs in the study area.

Nitrogen dynamics in the sewage impact zone

TSE-derived nitrogen was confirmed to decrease significantly within 1–2 km of the STP (Figs. 2, 3, and 5). Figure 6 shows the relationship between the EC and DIN concentrations and between the Chl-a and particulate organic nitrogen (PON) concentrations in the surface seawater on the KH line in October 2014 and September 2016. Most of the observed DIN concentrations were lower than the concentration derived via the mixing of TSE and offshore seawater (Fig. 6a). This suggests that the dilution effect via mixing with offshore seawater is a major factor in the reduction of DIN concentrations; however, the effect of nitrogen uptake by phytoplankton should also be considered. Figure 6b indicates that the Chl-a and PON concentrations correspond relatively closely to each other and that most of the PON originates from phytoplankton. This result also suggests that the biomass of phytoplankton was about 10 times higher in September 2016 than in October 2014. The Chl-a in September 2016 was also significantly higher than in February 2015 (Fig. 3c). As noted above, the bottom materials in the observation lines are mainly composed of sludge with a high content of organic matter. These results suggest that the sediment in these observation lines originated from POM as phytoplankton.

Figure 7 shows a comparison of spatial variation in PON concentrations and in δ¹⁵N values for NO₃⁻, PON, and surface sediment of the KH line. The PON results for October 2014 and the rising tide period in September 2016 are shown (Fig. 7a, b). The δ¹⁵N values forNO₃⁻ and sediment for October 2014 are shown (Fig. 7b). As also shown in Fig. 6, the PON concentration was at a high level in September (Fig. 7a). In Fig. 7b, the average values of δ¹⁵N(NO₃⁻) for offshore seawater, river water, pore water, and TSE were also shown as endmembers of the coastal seawater in the study area. The δ¹⁵N(NO₃⁻) value for the KH line was + 8.8 to + 10.8‰; however, the offshore seawater, river water, and TSE showed higher values (+ 12 to + 14‰), which suggests the δ¹⁵N(NO₃⁻) for the KH line may have been influenced by the isotope fractionation via nitrification (Kendall et al. 2007). On the other hand, the lower δ¹⁵N(NO₃⁻) value for pore water (+ 5‰) also suggests a contribution of sediment-derived nitrogen to the NO₃⁻ in the KH line.

Patterns of spatial variation of δ¹⁵N(PON) were similar between the two periods (Fig. 7b). It was high at the STP owing to the influence of the TSE, and then decreased before increasing again in the offshore direction, influenced by the mixing with offshore seawater and river water. δ¹⁵N was higher in NO₃⁻ than in PON, excluding at the STP in October. The result that NO₃⁻-N was a major form of DIN at the surface (Fig. 5c) suggests that δ¹⁵N(PON) reflected δ¹⁵N(NO₃⁻) and that δ¹⁵N(NO₃⁻) was enriched by the nitrogen uptake of phytoplankton in October (Waser et al. 1998; Sigman et al. 1999). Waser et al. (1998) estimated through a batch culture method in a laboratory that δ¹⁵N fractionation is about 5.2‰ during uptake by a coastal diatom (Thalassiosira pseudonana). Thalassiosira spp. are major diatom species observed in the study area (Oshima et al. 2009). Sigman et al. (1999) estimated that δ¹⁵N fractionation by phytoplankton uptake is 4–6 ‰, based on field observations in the Southern Ocean. For the results of the current study, the average δ¹⁵N(NO₃⁻) value was 9.6‰ and δ¹⁵N(PON) was 6.9‰ in October. The enrichment ratio was smaller than the isotope fractionation reported previously.

The δ¹⁵N(PON) in September was lower within 1 km of the STP and higher than in October in the offshore direction (Fig. 7b). The fact that significantly higher NH₄⁺-N concentrations were detected in September than in October (Fig. 5e, f) suggests two possibilities: δ¹⁵N(PON) may have been influenced more by NH₄⁺ than by NO₃⁻ and the δ¹⁵N(PON) fractionation by phytoplankton uptake of lower δ¹⁵N(NO₃⁻) was derived through nitrification. On the other hand, the tidal range was more than two times greater in September than in October (Table 1). This suggests that a greater tidal range would have increased the exchange between seawater and pore water through the coastal sediment (intrusion and discharge) and increased the contribution of pore water with a low δ¹⁵N(NO₃⁻) in September. δ¹⁵N was lower than PON for the surface sediment samples in the same period and had a similar value with the δ¹⁵N(NO₃⁻) for pore water (Fig. 7b). The discharge volume of TSE would have been larger in September because of the antecedent rainfall (Table 1). This condition may have intensified the estuarine circulation flow, and exchange between seawater and pore water may have increased in September. It also would have been supported by the high concentration of NH₄⁺-N observed in both the bottom and surface seawater samples in September (Fig. 5e, f). The increase in nitrogen supply through sediment during this period would have caused an increase in phytoplankton and PON. However, we require data on δ¹⁵N(NH₄⁻) to understand these processes in greater detail. Also, seasonal variation in nitrogen concentration and δ¹⁵N(NO₃⁻) for the endmembers, such as the Yodo River water, will need to be confirmed because previous research pointed out that δ¹⁵N(NO₃⁻) for river water may change according to river discharge (Takagi et al. 2016).

Based on these results, a schematic diagram of nitrogen dynamics in the coastal area, as influenced by sewage-derived loads, is shown in Fig. 8. The sewage-derived nitrogen impact zone (SNIZ) via direct discharge of TSE was estimated to be 1–2 km from the STPs on the KH and SU lines. This result also suggests that the spatial variation in SNIZ is controlled by tidal variation and the water temperature difference between the TSE and seawater. On the other hand, the SA line was confirmed to be significantly influenced by the Yamato River as well as the TSE. However, similar values of δ¹⁵N(NO₃⁻) for the TSE and Yamato River water suggest the Yamato River is influenced by the STPs located upstream. Conclusively, this study indicates that coastal sediment is another potential source of nitrogen and is transported via exchange between seawater and pore water. More specifically, the sediment may have originated from POM as phytoplankton, which is a secondary product of TSE-derived nitrogen. This suggests that the TSE-derived nitrogen in the coastal environment is not only from direct discharge but also from resupply through the sediment.

However, as the study area is influenced by tidal variation and mixing of several endmembers, the nitrogen dynamics should be quite complex compared with closed systems and the open ocean. Our results indicate that temporal changes in mixing ratios of endmembers and nitrogen uptake by phytoplankton influence both the nitrogen concentration and δ¹⁵N in this study area, although further research is needed for quantitative evaluation of nitrogen transport processes.