The Pearl River Delta (PRD, 112.9–114.4°E, 21.82–23.25°N) located on southern China covers an area of ~54,754 km² and is surrounded by several highly populated cities such as Hong Kong, Guangzhou, Macau, and Shenzhen with a total population of over 78.61 million people in 2021 (Li et al., 2000; Liu et al., 2018). It is plain river network area with the slope of 0.1–0.2‰ and the density of river network of ~0.8 km km^-2, which is characterized by subtropical humid climate with an annual average temperature of 21.8°C and an annual average rainfall of 1747.4 mm (Wang et al., 2012). A small irregular semidiurnal tide dominates this area with the average of 0.86 to 1.6 m and maximum of 2.29 to 3.36 m (Wang et al., 2012). The shallow aquaculture pools that were created by the removal of original marsh vegetation (mangrove). The PRD is experiencing massive industrialization and urbanization nowadays and makes great contributions to China’s economy. Take the year of 2021 as an example, this area provided ~12.3% of the national gross domestic product ($1406.18 billion) according to the China Statistical Yearbook 2022. Meanwhile, the coastal habitats of the PRD have been highly productive ecosystems, being an extremely important aquaculture area of China (Zhou et al., 2019). Coastal wetlands reclamation for aquaculture pond took the largest proportion of the total conversion from natural wetlands (estuarine water, mangrove forest and salt marsh) towards constructed wetlands in this region. For example, the breeding area of the PRD increased about threefold from 256.01 km² in 1980 to 826.72 km² in 2015 (Zhou et al., 2019). With rapid economic growth and urbanization of the PRD, local environment has suffered from serious environmental problems, especially the high N loading, which has great effects on water quality and N cycling processes (Dai et al., 2008; Li et al., 2013; Wu et al., 2020).

Our samples were selected from three separate coastal ponds aquaculturing different economic species on the western bank of the Pearl River Estuary, one for fish (Perca fluviatilis), one for shrimp (Penaeus vannamei), and one for crab (Scylla serrata) ( Figure 1 ). The sediments were collected using a core cylinder, a PVC pipe handle and a one-way valve from these sites in winter (January 19, 2019) and summer (July 12, 2019). At each site, triplicate surface sediments (0–5 cm) were taken from the cores, and sealed with air-tight, acid-cleaned plastic bags. Overlying water samples of each site were also collected in polyethylene bottles and filtered through 0.2 μm filters (Millipore, Bedford, United States). All the samples were transported to the laboratory on ice within 8 h. In the laboratory, filter samples were immediately stored at -20°C. Sediments of each sampling site were mixed thoroughly under a helium condition, and subsequently divided into three parts. One part was stored at 4°C for measuring microbial biomass carbon (MBC) and N-cycling rates, the second part was frozen at -20°C for determination of physicochemical parameters, and the third part was preserved at -80°C for molecular analysis.

Sediment water content was determined according to the weight loss from fresh sediment after freeze-drying until constant weight. Sediment pH was determined by measuring the pH of the mixture of fresh sediment and Milli-Q (MQ) water (1:2.5 in v/v) (Zhang et al., 2015). Sediment NH₄ ⁺, NO₂ ⁻, and NO₃ ⁻ were extracted by 2 M KCl purged with N₂ for 30 min and measured using 2 M KCl as standard curve substrate (Hou et al., 2013). Total extractable iron (Fe) and ferrous oxide (Fe²⁺) were extracted by 0.5 M HCl and 0.25 M hydroxylamine hydrochloride (both purged with N₂ for ~15 min) from fresh sediments, and the concentrations were determined using ferrozine-based colorimetric method using the spectrophotometer. The concentrations of ferric iron (Fe³⁺) were calculated by subtracting Fe²⁺ from Fe (Lovley and Phillips, 1987; Hu et al., 2022). Sediment texture was determined by laser particle size analyzer LS13 320 (Hou et al., 2013). After effective leaching of carbonates with 1 M HCl, sediment total organic carbon (TOC) and total nitrogen (TN) were measured on an elemental analyzer (Vario EL, Elementar, Germany) (Zhang et al., 2015; Wu et al., 2021). Sediment stable organic carbon isotopic composition (δ¹³C_org) were analyzed on a Thermo MAT 253Plu isotope ratio mass spectrometer. After oxidation with 333 mmol L^–1 KMnO₄, sediment easily oxidized organic carbon (EOC) was measured on a spectrophotometer colorimetry (Vieira et al., 2007). Sediment dissolved organic carbon (DOC) was extracted with MQ water and determined with a Shimadzu TOC-TN analyzer (Shimadzu Crop., Kyoto, Japan). Sediment MBC were extracted with K₂SO₄ and determined through chloroform fumigation-extraction method (Vance et al., 1987; Beck et al., 1997).

Gross N mineralization (GNM) and NH₄ ⁺ immobilization (GAI) rates were determined using ¹⁵N isotope dilution technique (Kirkham and Bartholomew, 1954). Briefly, twelve centrifuge tubes (50 mL) were prepared for each sample, and 5 g fresh sediment was weighed into each centrifuge tube. For determination of the GNM and GAI rates, six centrifuge tubes with a mixture of sediment and saltwater (1:5 in w/w) were vortexed thoroughly. Tubes were sealed and preincubated for 24 h under in situ temperature in dark on a shaker table (150 rpm). After the preincubation, ¹⁵NH₄ ⁺ (99 atom%, ¹⁵NH₄Cl) was added into the tubes, the final concentration of ¹⁵N was about 2 μg g^-1 and the final percentage of ¹⁵N was about 10–15%. Triplicate initial samples were terminated by adding 1 mL 50% Zinc chloride (ZnCl₂) solution and immediately frozen at -20°C, and triplicates incubation samples were sealed and were put back to the incubator. After 24-hour incubation, final samples were terminated and immediately frozen at -20°C.

M_i and M_f (μg N g^-1 dry weight) are total NH₄ ⁺ concentrations of initial (before incubation) and final (after incubation) samples; H_i and H_f (μg N g^-1 dry weight) are ¹⁵NH₄ ⁺ concentrations of initial and final samples; t (d) is the incubation time (24 h).

The relative NH₄ ⁺ immobilization (RAI) is calculated as the ratio of GAI to GNM rates, and RAI value of ≥ 1 indicates a N-limited sediment, whereas a value of approximately 0.5 indicates N saturation (Aber, 1992). The percentage of NH₄ ⁺ mineralized per day (PAM%) was referred to as the rates of GNM divided by sediment N contents, which can indicate the sediment available N by the internal N cycle.

Sediment DNA was extracted by the FastDNA spin kit for sediment (MP Biomedical, United States) based on the manufacturer’s instructions. The concentration and purity of extracted DNA were determined by NanoDrop spectrophotometer (NanoDrop 2000C, Thermo Scientific, United States), and the fragments size and quality were evaluated by 1.0% agarose gel electrophoresis. Quantitative PCR (qPCR) assays were used to measure microbial gene abundances on an ABI 7500 Fast real-time qPCR system (Applied Biosystems, United States). For bacterial 16S rRNA gene and fungal ITS gene, primer pairs 341F/519R (Bachar et al., 2010) and SSU081 and 1196R were used (Rousk et al., 2010), respectively. The standard curves for these two genes were created using a 10-fold dilution series (10²–10⁹ copies) of the standard plasmids DNA. Each sample was measured in triplicates.

Statistical analysis was performed using SPSS 19.0 (SPSS Inc., United States). All variables were tested for normality by Shapiro-Wilk test, and the non-normality variable were normalized by using Blom’s formula. Statistically significant difference of N transformation rates and sediment physicochemical properties between three ponds were determined by one-way ANOVA, Turkey test (p<0.05, equal variances assumed) or Dunnett’s test (p<0.05, equal variances not assumed). The correlations were analyzed by and linear regression analyses and Pearson test (two-tailed, p<0.05). The graphs were drawn by ArcGIS 10.2 (ArcMap 10.2, ESRI, United States) and Origin 2019 (OriginLab Corporation, United States).

Physicochemical properties of overlying water in the three types of relcaimed aquaculture ponds are shown in Table 1 and Table S1 . During winter sampling, crab pond was in the early period of pond-drying, and only a litter overlying water was remaing in the crab ponds, while the shrimp pond was in the late drying period, and the surface sediments were dried and cracked. Therefore, overlying water of the shrimp and crab ponds in winter was not sampled because of that those two ponds were in drying and dreging period. One-way ANOVA showed that overlying water salinity, pH, NH₄ ⁺, NO₃ ⁻, and NO₂ ⁻ of the fish pond showed significant seasonal differences (p<0.05), while no significant seasonal difference was found for the dissolved oxygen (DO) (p>0.05). The salinity and pH in winter was higher than those in summer. Influenced by aquaculture activities, NH₄ ⁺, NO₂ ⁻, and NO₃ ⁻ were all significantly higher in summer than those in winter (p<0.05). In summer, both DO and salinity showed significantly differences among three ponds (p<0.05 for all). The highest DO was observed in the shrimp ponds (10.30 ± 1.05 mg L^-1) followed by the crab ponds (7.90 ± 0.52 mg L^-1) and fish ponds (5.11 ± 0.14 mg L^-1), which may potentailly result from the most operating waterwheel-type aerators in the shrimp and crab ponds. Consistenly, the highest salinity was measured in the crap ponds (2.85 ± 0.00 ‰) followed by the crab ponds (2.25 ± 0.18 ‰) and fish ponds (0.9 ± 0.04 ‰). The overlying water quality of these ponds were relatively well controlled. Overlying water NH₄ ⁺ and NO₂ ⁻ of the fish ponds were both significantly higher that those in shrimp and crab ponds (p<0.05).

Physicochemical properties of surface sediments in the relcaimed aquaculture ponds are shown in Table 2 and Table S2 . Due to geographic proximity, most surface sediment physicochemical properties (excluding NO₂ ⁻, Fe²⁺, and TOC/TN) were found no significant spatial variation among those three ponds in summer. In winter, the shrimp and crab ponds were in the drying and dredging period, and the surface sediments of the shrimp ponds were dried and cracked. Thus we found that surface sediment moisture content, Fe²⁺, EOC, DOC, and MBC of shrimp pond in winter were significantly lower than those in summer (p<0.05 for all), and the sediment bulk density in winter were significantly higher that those in summer (p<0.05). Meanwhile, several sediment physicochemical properties in shrimp pond were significantly higher (including bulk density and δ¹³C_org) or lower (including moisture content, TOC/TN, DOC, MBC, and bacterial 16S rRNA abundances) that those in other ponds in winter (p<0.05 for all). In crab pond, sediment temperature, NH₄ ⁺, NO₃ ⁻, Fe²⁺, Fe³⁺, TOC, TN, EOC, and MBC in summer were significantly higher that those in winter (p<0.05 for all). In addition, all the surface sediments in those three ponds were mainly composed of fine silt and clay with low median grain size (7.69−57.77 μm).

The spatiotemporal distributions of sediment GNM, GAI, RAI, PAM, and microbial abundances in three reclaimed aquaculture ponds were shown in Figure 2 . GNM rates varied from 5.77 to 14.58 μg N g^-1 d^-1 with an average value of 9.64 ± 3.31 μg N g^-1 d^-1 in summer and from 1.75 to 6.45 μg N g^-1 d^-1 with an average value of 3.60 ± 1.19 μg N g^-1 d^-1 in winter. Significant seasonal variations in GNM rates were observed in those three ponds, with significantly higher values in summer than in winter (p<0.05 for all, Figure 2A ). No significant difference in GNM rates occurred among three ponds in summer (p>0.05 for all, Figure 2A ). In comparison, in summer, the GNM rates in crab pond were significantly lower than those in the other two ponds (p<0.05 for both, Figure 2A ).

The GAI rates ranged from 5.15 to 13.68 μg N g^-1 d^-1 with an average of 7.71 ± 2.49 μg N g^-1 d^-1 in summer and from 1.67 to 5.66 μg N g^-1 d^-1 with an average of 3.17 ± 1.06 μg N g^-1 d^-1 in winter. Similarly, GAI rates were higher in summer than in winter with a remarkable seasonal difference within these three ponds (p<0.05 for all, Figure 2B ). Spatially, no significant spatial difference in GAI rates occurred among three ponds in both summer and winter (p>0.05 for all, Figure 2B ). In addition, GNM rates were significantly positively correlated with GAI rates in the whole study area in both summer and winter (p<0.05 for both, Figure S1 ).

The relative NH₄ ⁺ immobilization (RAI) values ranged between 0.55 and 1.15 in summer and between 0.60 and 1.23 in winter ( Figure 2C ). There was no significant seasonal variation in RAI in those three ponds, and no significant spatial difference among three ponds (p > 0.05 for all; Figure 2C ). A previous study has confirmed that RAI ≤ 0.5 and RAI > 0.9 indicated N-saturated environment and N-limited environment, respectively (Aber, 1992). Here, the RAI values in most sampling sites were between 0.5 and 0.9, indicating these reclaimed aquaculture ecosystems were able to maintain N balance.

PAM values ranged from 0.36% to 1.25% in summer and from 0.15% to 0.59% in winter ( Figure 2D ). Spatially, PAM in fish pond were significantly higher than those in other ponds (p< 0.05 for all; Figure 2D ). The PAM in summer (0.92 ± 0.30%) was significantly higher than that in winter (0.45 ± 0.15%) in fish pond (p< 0.05; Figure 2D ). In shrimp and crab ponds, PAM in summer were also higher than those in winter, but with no significant seasonal difference (p > 0.05; Figure 2D ).

The bacterial 16S rRNA gene abundances were in a range of 3.93–13.94×10⁹ copies g^-1 with an average of 7.82 ± 3.22×10⁹ copies g^-1 in summer. In winter, the range and average value decreased to 1.91–11.22×10⁹ copies g^-1 and 5.12 ± 2.85×10⁹ copies g^-1. The bacterial 16S rRNA abundances in summer were significantly higher than those in winter in shrimp pond (p< 0.05; Figure 2E ).The values in sediment of fish pond were significantly higher than those in other two ponds in winter (p<0.05 for both, Figure 2E ). The fungal ITS gene abundances were in a range of 1.76–12.73×10⁷ copies g^-1 with an average of 5.46 ± 3.57×10⁷ copies g^-1 in summer and in a range of 2.00–8.88×10⁷ copies g^-1 with an average of 4.26 ± 2.15×10⁷ copies g^-1 in winter, showing no significant temporal differences in those three ponds and no spatial differences among three ponds (p< 0.05 for all; Figure 2F ).

Considering all sampling sites, GNM and GAI rates showed significant positive correlations with sediment temperature, organic matter (TOC, TN, EOC, DOC, and MBC), microbial abundances (bacterial 16S rRNA and fungal abundances), Fe²⁺, NH₄ ⁺, moisture content, as well as overlying water DO and NO₃ ⁻, while they were correlated negatively with sediment bulk density (p<0.05; Table 3 ). PAM were significantly positively correlated with sediment MBC, microbial abundances, Fe²⁺, Fe³⁺, as well as overlying water DIN (p<0.05; Table 3 ). RAI were positively correlated with sediment δ¹³C_org and correlated negatively with sediment microbial abundances, MBC, DOC, and TOC/TN (p<0.05; Table 3 ).

Temperature could have directly impacts on microbial physiological activities through the regulation of enzyme activities and is considered as a crucial factor of N mineralization and immobilization (Yang et al., 2010). In this study, GNM and GAI rates were both significantly higher in summer than in winter among those three ponds (p<0.05 for all). The seasonal variations are mainly due to temperature differences between summer (31.81 ± 0.86°C) and winter (17.54 ± 0.05°C). Also, the ratios of N-cycling rates in summer to rates in winter were used to characterize the temperature sensitivity of N mineralization and immobilization rates. A previous study found that total N immobilization rate is very sensitive to temperature while N mineralization rate shows relatively less sensitivity under low temperature condition (5–15°C) (Andersen and Jensen, 2001). However, we found that no significant difference between the temperature sensitivity of GNM and GAI ( Figure 3A ). The major reson is that the winter temperature in our study is over 15°C, and the temperature sensitivity of N mineralization has been found to decrease with increasing temperature (Kirschbaum, 1995). In addition, no significant spatial difference in temperature sensitivity of N mineralization and immobilization rates among those three ponds ( Figure 3A ). This is mainly due to similar physicochemical properties (including sediment grain size, organic matter, DIN, and TOC : TN) among those three ponds, and previous studies have found that temperature sensitivity of N mineralization and immobilization rates mainly depended on the above physicochemical properties in agricultural ecosystems (Liu et al., 2017; Miller and Geisseler, 2018).

Organic matter of aquaculture sediments originates from both autochthonous sources (microalgae, sediment and particulate organic matter) and bait inputs (Chen et al., 2016). According to the values of TOC/N ratio and δ¹³C_org, we found the surface sediment organic matter were mainly from marine algae, marine DOC, and marine POC ( Figure 3B ). This is caused mainly by the fact that the aquaculture ponds were located on the estuary, where a significant proportion of overlying water is derived from marine. More importantly, the main ingredients (~60%) of bait are fish meal and fish oil (Park et al., 2021), and they were made from marine fish (e.g. Engraulis japonicus).

As we all know, organic matter acting as energy source of microorganisms and substrates of N mineralization, which plays an important role in microbial-mediated N-cycling processes. To clarify how labile C impacts on N mineralization and immobilization in the reclaimed aquaculture sediments, several labile organic carbons including sediment EOC, DOC, and MBC, as well as potential N mineralization and immobilization rates, were measured. When considering all sites, the correlation strengths between labile organic carbons (EOC, DOC, and MBC) and N mineralization and immobilization rates were much higher than those between TOC and the rates. The order of above correlations was MBC>EOC>DOC>TOC ( Table 3 ). When considering seasonal effects, we found that this regularity and trend was only found in summer ( Figure 4 ), and the order of above correlations was changed slightly (MBC>DOC>EOC>TOC). This finding agreed well with several previous studies, which found that high availability of organic matter is favorable for soil/sediment N mineralization process in coastal wetland (Li et al., 2020; Yang et al., 2022), estuary (Huang et al., 2022) and forest (Jones and Kielland, 2012). Thus, our results suggest that organic quality rather than quantity was more crucial to the control of N mineralization and immobilization in aquaculture sediments.

No significant correlations existed between organic matter and N mineralization and immobilization rates in winter, this may be probably due to the profound effect of human activities (pond-drying) on these different aquaculture ponds. As mentioned in the preceding text, fish ponds are still being farmed, and shrimp and crab ponds are in a resting state in winter. Namely, shrimp and crab ponds were in the drying and dredging period, and surface sediment of the shrimp ponds was dried and cracked. Firstly, pond-drying can significantly reduce the sediment moisture content in shrimp ponds in winter, previous studies have found that N mineralization increased with the increasing soil/sediment moisture (Jia et al., 2019; Li et al., 2020; Huang et al., 2022). Sediment moisture is involved in the metabolism of microbes and is also the irreplaceable medium linking substrates supply and microbial activity (Greaver et al., 2016). Elevated sediment moisture has also been reported to increase the decomposition and leaching of organic matter and extracellular enzyme activities, promoting soil N transformations (Mooshammer et al., 2014; Greaver et al., 2016; Jia et al., 2017). Also, increased soil moisture could limit oxygen to penetrate soils and further lead to more strongly anoxic and reducing environments, thereby facilitating denitrification and N mineralization in coastal wetlands (Jia et al., 2019). Secondly, our measured sediment labile organic matter contents (including EOC, DOC, and MBC) in shrimp pond were significantly less in winter compared with those in summer ( Table 2 , p<0.05 for all). Thus, pond-drying can accelerate the decomposition of surface sediment organic matter, resulting in decreased the supply of energy source and substrate for microbial growth and mineralization. In addition, pond-drying can significantly reduce microbial biomass (MBC) and bacterial abundances in shrimp pond ( Table 2 , p<0.05 for both), thereby regulating the N mineralization and immobilization. Therefore, most physicochemical properties were significantly changed with aquaculture management activities (pond-drying), resulting in the profound reduction of sediment N mineralization and immobilization in aquaculture ecosystems.

No significant seasonal and spatial difference in microbial abundances was observed in our study excluding those in shrimp ponds in winter ( Table 2 ), this is largely due to small summer-winter temperature differences and low spatial heterogeneity in sediment physicochemical properties. When the surface sediment was dried and cracked after long-term pond-drying, this aquaculture management activity can significantly reduce the microbial biomass and bacterial abundances, but not for fungal abundances ( Table 2 ). This is mainly due to fungi can be remarkably drought tolerant, and they can remain active and even grow under extremely dry conditions (Treseder et al., 2010; Yuste et al., 2011). However, our study lack of data about microbial community structure and diversity. A previous study showed that pond-drying can change the sediment microbial community structure, and the abundance of Proteobacteria, Nitrospirae and Bacteroidetes were increased, while the pernicious microbe such as Cyanobacteria was decreased after pond-drying in aquaculture ecosystem (Wang et al., 2020). In addition, natural sunlight has effect on the abundance and microbial community structure in the sediments during pond-drying. Thus, we boldly speculated that pond-drying can alter microbial community composition from anaerobic, fast-growing, and copiotrophic to aerobic, slow-growing, and oligotrophic, thus promote the ecosystem reparation effectively.

Many previous studies have reported that N mineralization and immobilization rates were affected by the microbial biomass and activities, and thus they were generally closely related to the microbial respiration rates, enzymatic activities (including extracellular and intracellular enzymes), and ATP content (Bengtsson et al., 2003; Silva et al., 2005). N mineralization and immobilization rates were also controlled by the microbial biomass (MBC) and abundances (including bacterial 16S rRNA and fungal abundances) in this study ( Table 4 ), which is consistent with previous sudies in many estuarine and coastal zone (Lin et al., 2016a; Li et al., 2020; Huang et al., 2022; Yang et al., 2022). Thus, microbial abundances could be considered as important indicators of N mineralization and immobilization rates in aquaculture sediments.

In addition, GNM and GAI rates were also positively correlated with overlying water NO₃ ⁻ ( Table 3 , p<0.05). Previous studies have revealed significant influence of environmental effective electron acceptors on N mineralization. For example, NO₃ ⁻, as an electron acceptor of oxidized organic matter, could promote N mineralization (equation 3) and make high contribution (15−35%) to the mineralizaiont of organic matter especially under anoxic conditions (Khalil et al., 2018). Under anaerobic conditions, NO₃ ⁻, Fe³⁺, SO₄ ²⁻ as well as iron and manganese ions often work as active electron acceptors for coupling redox reactions with sediment organic matter (equation 4 and 5) (Pena et al., 2010). As sediment is flooded for long time and the vertical penetration ability of oxygen from overlying water into sediment is relatively weak, anaerobic conditions for the aforementioned reactions could be created in aquaculture ecosystem. Additional work regarding aerobic N mineralization can be done to compare the mechanisms between anoxic environments and aerobic environments.

The average N mineralization and immobilization rates of reclaimed aquaculture ponds in this study area were lower than the rates of other aquaculture, urban river network, and mangrove wentland sediments, and higher than the rates of estuarine and coastal sediments ( Table 4 ). The daily mineralization percentage of NH₄ ⁺ was 0.55−1.23%, which as comparable with the value in coastal wetlands ( Table 4 ). In addition, the value of RAI in this study (average: 0.86 ± 0.15; range: 0.58−1.23) implied that the ecosystem might be located between N limitation condition and N saturation condition (Aber, 1992). Significantly positive correlations between GNM and GAI identified in this study showed consistency with the observations in offshore seawater (Lin et al., 2016), wetlands (Jin et al., 2012), grassland (Corre et al., 2002) and forests (Bengtsson et al., 2003). It’s also noteworthy that a combination of sediment incubation experiment and ¹⁵N stable isotope dilution method was adopted in this study. As described in Lin et al. (2017), the added ¹⁵NH₄ ⁺ may stimulate microbial responses, along with the dilution ratio of 1:5 (fresh sediment: water) and aerobic incubation conditions, the measured GNM and GAI might not reflect in situ activities. Nevertheless, it can still imply potential activities of N mineralization and immobilization in aquaculture surface sediments.

Sediment N mineralization and immobilization are important N-cycling processes in coastal wetland reclamation and aquaculture ecosystems, which plays vital roles in the reactive N balance. However, very few systematic studies investigated environmental properties, N mineralization and immobilization, and their relationships, especially in reclaimed aquaculture ecosystems. Here, based on GNM and GNI measured in this study, annual surface sediment (0−5 cm) N mineralization and immoblization rates in reclaimed aquaculture ecosystems of the Great Bay Area were calculated (equation 6).

In which, F (t N yr^-1) is annual N mineralization or immoblization flux; mi and mj (μg N g^-1 d^-1) is GNM or GAI of surface sediment (0−5 cm) in summer and winter, respectively; di and dj (g cm^-3) is bulk density of surface sediment (0−5 cm) in summer and winter, respectively; a is a unit conversion factor; s (m²) is the area of reclaimed aquaculture ecosystems in the Great Bay Area (~395.57 km² according to remote sensing data in 2018 followed by the calclution using ArcGIS 10.2) ( Figure S2 ); h (cm) is sampling depth (5 cm); t (d) is time (365 days).

In total, annual sediment N mineralization and immoblization rates in reclaimed aquaculture ecosystems are approximately 4.55×10⁴ t N yr^-1 and 3.68×10⁴ t N yr^-1, respectively in the Guangdong-Hong Kong-Macao Greater Bay Area. GNM flux shows higher level than GAI flux, indicating that sediment is a nonnegligible DIN source in the aquaculture ecosystem, which plays a non-negligible role in the exacerbation of eutrophication in the estuarine and coastal ecosystems.