1 Introduction

Nitrogen (N) is an essential element for biological organisms, playing a key role in promoting primary productivity and maintaining lake ecosystem health (Saggar et al. 2013; Heffernan and Cohen 2010; Zhu et al. 2020). With the rapid development of agriculture and industry in lake watersheds, a substantial amount of reactive N is transported into lake ecosystems (Finlay et al. 2013; Zhu et al. 2020; Yang et al. 2022). The excessive input of reactive N has caused various environmental issues, particularly aquatic eutrophication and nitrous oxide (N2O) production (Jiang et al. 2020; Yang et al. 2022; Zhang et al. 2022). In lake ecosystems, sediments play an important role in nitrate removal and the mitigation of N2O emissions (Mulholland et al. 2008; Nizzoli et al. 2010; Finlay et al. 2013). However, continuous N loading and nitrate retention have converted lake sediments into significant sources of nitrate pollution and N2O emissions (Bogard et al. 2020; Kortelainen et al. 2020; Yang et al. 2022). Consequently, the problems of N pollution and N2O production in lake ecosystems have attracted global attention (Schmadel et al. 2018; Rex et al. 2021; Zhang et al. 2022). Sediment nitrate dissimilatory reduction processes, including anaerobic ammonium oxidation (anammox), denitrification (DEN), and dissimilatory nitrate reduction to ammonium (DNRA), are the primary pathways of nitrate removal or retention in aquatic environments (Stelzer et al. 2014; Reisinger et al. 2016; Xiong et al. 2017; Li et al. 2018). Understanding the mechanisms and environmental implications of these processes in lake ecosystems is necessary for controlling N pollution and reducing N2O emissions.

In lake ecosystems, sediment DEN and anammox processes lead to the reduction of NO3- to N2 gas, thus removing it from the lake ecosystem (Jiang et al. 2020; Tan et al. 2022). In the DEN process, N2O is an intermediate product that can be released into the atmosphere if the process is incomplete (Wu et al. 2021; Li et al. 2023; Xiang et al. 2023). N2O is a powerful greenhouse gas with a warming potential approximately 298 times that of carbon dioxide (Lauerwald et al. 2019; Ashiq et al. 2022; Kirkby et al. 2023). Sediment DNRA processes convert nitrate into ammonium, leading to nitrate retention in aquatic environments (Nogaro and Burgin 2014; Kim et al. 2016; Lin et al. 2017; Shelley et al. 2017). The competition among DEN, anammox, and DNRA processes in lake sediments is controlled by the high variability of sediment properties, such as C and N contents (Plummer et al. 2015; Shan et al. 2016; Gao et al. 2019; Jiang et al. 2021a). Previous studies have extensively investigated spatial and temporal distribution and environmental factors affecting sediment dissimilatory nitrate reduction processes in lake sediments. However, these processes in river-lake ecotone systems remain unclear.

The river-lake ecotone system is an important transition zone between a lake and its inflow rivers, playing a crucial role in absorbing and dissipating surface runoff pollution and being influenced by both river and lake ecosystems (Yan et al. 2022). Sediments in the river-lake ecotone reflect the environmental impacts of the river watershed on the lake ecosystem, such reactive N loading and land use changes affecting sediment N transformation. Compared with lake ecosystems, reactive N loading from the upstream river watershed is often concentrated in the river-lake ecotone, significantly regulating sediment environmental properties (such as reactive N, pH, and TOC contents) and nirS, nirK, nosZ, nrfA, and anammox 16S rRNA bacteria abundances and community structures (Wei et al. 2020; Yan et al. 2022). As a result, sediments in the river-lake ecotone become potential hotspots for N transformation and N2O emissions. Therefore, understanding the spatial distribution and mechanisms of sediment dissimilatory nitrate reduction processes in the river-lake ecotone is crucial for controlling N pollution and enhancing N removal in lake ecosystems.

With industrial and agricultural development in the Poyang Lake watershed, N inputs into the lake ecosystem have greatly increased, posing a severe eutrophication risk (Yao et al. 2016). Understanding the mechanisms and environmental implications of sediment nitrate reduction is important for maintaining the health of Poyang Lake and reducing eutrophication. Although DEN and DNRA processes in Poyang Lake have been studied (Yao et al. 2016; Zhang et al. 2016; Wang et al. 2017; Jiang et al. 2020), the nitrate dissimilatory reduction processes, including DEN, anammox, DNRA, and N2O production, as well as the differences and driven mechanisms between the river-lake ecotone system and the lake ecosystem, have not been clarified. The objectives of this study were to: (1) reveal variations in sediment dissimilatory nitrate reduction processes and N2O production rates in different river-lake ecotones; (2) identify the mechanisms of sediment nitrate reduction rates and N2O production in the river-lake ecotone; and (3) analyze nitrate fates in Poyang Lake and reveal the environmental implications of sediment nitrate reduction. These finding are of great significance for improving the understanding of nitrate removal in lake ecosystems.

2 Material and methods

2.1 Study area and sampling

Poyang Lake is the largest freshwater lake in China, located in the northern part of Jiangxi province. The lake area covers approximately 4000 km2 during the wet season, with annual water level fluctuations of around 10 m (Wang et al. 2016; Liao et al. 2018). The Poyang Lake watershed spans 162,200 km2, of which 96.62% is within Jiangxi province (Yao et al. 2016). Five large rivers flow into the lake: Ganjiang River (GJ), Raohe River (RH), Fuhe River (FH), Xiushui River (XS), and Xinjiang River (XJ) (Fig. 1).

Fig. 1
figure 1

Location of the study area and sampling sites

Full size image

A total of 26 surface sediment samples (0–5 cm) were collected from five river-lake ecotones and lake ecosystem (DC) from February 21st to 23rd (dry season) and May 16th to 18th (wet season) in 2021 using a Van Veen grab sampler (Fig. 1). In total of 26 surface sediment samples, six samples were collected in GJ, five samples were collected in RH, three samples were collected in DC, four samples were collected in FH, XS, and XJ. For each sample, we collected five sub-samples within a 50 m transect and mix then together. Adjacent sampling sites within the same river-lake ecotone was spaced 5-10 km apart. To minimize the impact of human activities on the results, sampling points were located at least 1 km away from urban and industrial area. Samples were collected in airtight, acid-cleaned plastic bags and transferred to the laboratory within 4 hours of collection, maintained at 4 ℃. In each season, the samples are brought back to the laboratory every night to ensure the reliability of experimental analysis data. The longitude and latitude of each sample were recorded using GPS. In the laboratory, the well-mixed sediment samples were divided into three fractions: (1) frozen at -20 °C for measuring sediment physicochemical characteristics; (2) stored at 4 °C for analyzing sediment dissimilatory nitrate reduction processes rates and N2O production; and (3) frozen at −80 °C for measuring genes abundance related to nitrate reduction.

2.2 Soil physicochemical analysis

Sediment total organic carbon (TOC) content was measured using the dichromate oxidation method (Jiang et al. 2023). Before measuring TOC content, the sediment was freeze-dried using a vacuum-freeze dryer and passed through a 0.15 mm mesh sieve. Amorphous Fe oxides in the sediment were extracted using 1 M HCl, 1g of fresh sediment was placed into a 50 mL centrifuge tube, and 30 mL of 1 M HCl was added. After shaking for 16 hours, Fe2+ and Fe3+ contents in the supernatant were measured using the photometric analysis of the Ferrozine assay (Gao et al. 2019). An Orion-Sure-flow combination silver-sulfide electrode (Thermo Scientific Orion) was used to measure sediment sulfide concentration (Hou et al. 2012). Sediment NH4+, NO3−, and NO2− were extracted using a 2 M KCl solution, and their contents were determined using a continuous-flow analyzer (Yin et al. 2017). Sediment pH was measured using a pH meter at a sediment-to-water ratio of 1:2.5 (Zheng et al. 2014). Properties of Fe2+, Fe3+, sulfide, pH, NH4+, NO3−, and NO2− were measured using both fresh sediments frozen at -20 °C and fresh sediment that was converted into dry sediment based on the sediment’s water content. All sediment physicochemical properties were determined in triplicate.

2.3 Measurements of sediment DEN, anammox, and DNRA rates

Sediment potential DEN and anammox process rates were measured using the 15N isotope tracing method (Cheng et al. 2016). Fresh sediment samples stored at 4 °C were stirred into a slurry state by mixing with deionized water at a mass ratio of 1:7. Dissolved N2 and O2 in the slurry samples were removed by aeration with helium gas to ensure the dissolved N2 was not saturated. Subsequently, the slurry samples were transferred into 12.5 mL gastight borosilicate vials and pre-incubated for 48 hours at temperatures similar to the sampling site (30℃ in May and 12℃ in February) to ensure an anaerobic environment. After pre-incubation, 100 μL of 15NO3− (15N at 99%) was added to the vials containing the slurry samples. To prepare the initial samples, 200 μL of ZnCl2 solution was immediately added to half of the vials to stop microbial activity. For the final samples, microbial activity in the remaining vials were stopped after 8 hours of incubation. Dissolved 29N and 30N in the final and initial samples were measured using membrane inlet mass spectrometry (MIMS). This 15N isotope tracing method requires immediate testing of samples in an anaerobic environment with unsaturated dissolved N2. Rates of DEN and anammox in sediment were calculated by the difference in 29N2 and 30N2 between the final and initial samples. The respective contributions of DEN and anammox to total 29N2 production were quantified using the equation (Deng et al. 2015):

$${P}_{29}={A}_{29}+{D}_{29}$$
(1)

where P29 (nmol N g−1 h−1) denotes the total 29N2 production rate during the slurry experiments, and D29 and A29 (nmol N g−1 h−1) are the production rates of 29N2 from DEN and anammox, respectively. Given the random isotope pairing of 14N and 15N generated from 14NO3− or 15NO3−, D29 can be estimated by the equation (Gao et al. 2017):

$${D}_{29}={P}_{30}\times 2\times (1-{F}_{N})\times {F}_{N}-1$$
(2)

where P30 (nmol N g−1 h−1) denotes the total 30N2 production rate and FN (%) is the fraction of 15N in NO3−, calculated from the concentrations of added 15NO3− and residual NO3− in the incubation slurries. Consequently, the potential rates of DEN and anammox were estimated using the equations (Gao et al. 2019):

$${D}_{t}={D}_{29}+2\times {D}_{30}$$
(3)
$${A}_{29}={P}_{29}-{D}_{29}$$
(4)

where Dt and A29 (nmol N g−1 h−1) represent the potential rates of DEN and anammox, respectively.

The OX/MIMS method (15NH4+ oxidation technique combined with MIMS analysis) was used to measure the sediment DNRA rate (Yin et al. 2014). The same incubation and 15N isotope addition procedures as for DEN and anammox rates were applied. Dissolved N2 initial samples and final samples were removed using helium gas. In all vials, 200 μL of hypobromite iodine solution was injected to oxidize 15NH4+ (produced by DNRA) into 15N2, which was then measured using MIMS. Potential DNRA rates were estimated by the changes in 15NH4+ concentration during incubation, calculated using the equation (Jiang et al. 2021a):

$${R}_{DNRA}=\frac{\left({\left[{15}_{N{H}_{4}^{+}}\right]}_{F}-{\left[{15}_{N{H}_{4}^{+}}\right]}_{I}\right)}{W\times T}$$
(5)

where RDNRA (nmol N g−1 h−1) denotes the total DNRA rate, and [15NH4+]F and [15NH4+]I are the concentrations of 15NH4+ in the final and initial samples of the soil slurries, respectively. V(L) denotes the volume of the vial, W(g) denotes the dry weight of the soil, and T(h) is the incubation time.

2.4 Potential N 2 O production rate

The headspace equilibrium technique was used to measure potential N2O production (Hou et al. 2015). The incubation method was consistent with that used for the DEN rate. After incubation, 2 mL helium gas was injected into the vials to replace the water phase and create a headspace. To equilibrate the liquid and gas phases, the vials were shaken vigorously for 1 hour, and the N2O concentration of the headspace gas was measured by gas chromatography immediately. The potential N2O production rate was calculated using the following equation (Hou et al. 2015):

$$\text{P}=\frac{\left({O}_{i}-{O}_{f}\times V\right)}{T}$$
(6)

where P (nmol g-1 h-1) represents the potential N2O production; Oi (nmol L-1) represents the initial dissolved N2O concentrations; Of (nmol L-1) represents the final dissolved N2O concentrations; and V (h) represents the incubation times.

Sediment samples for DEN, anammox, DNRA, and N2O production rates need to be determined immediately, as prolonged storage at 4 °C can increase experimental bias. Rates of sediment DEN, anammox, DNRA, and N2O production were measured using wet sediment directly and then converted to dry sediment based on sediment water content.

2.5 DNA extraction and quantitative PCR (q-PCR) analysis

DNA from the sediments was extracted using a Powersoil™ DNA Isolation Kit according to the manufacturer’s specifications. The abundance of nirS, nirK, nosZ, anammox bacterial 16S rRNA gene, and nrfA genes were quantified through real-time q-PCR using the SYBR Green method on the ABI 7500 Detection System. Plasmid DNA containing the target gene fragment was constructed by cloning and quantified using an ultramicro spectrophotometer. A 10-fold gradient dilution series (102–109 copies) was used as a reaction template for quantitative PCR reaction with environmental sample DNA under the same conditions (Jiang et al. 2021a, 2023). The amplification efficiencies under q-PCR conditions were 92.4, 94.68%, 95.2%, 92.7%, and 93.5% for nirS, nirK, nosZ, anammox bacterial 16S rRNA, and nrfA genes, respectively. Details of the gene primers and q-PCR conditions are provided in Table S1. The abundance of these genes was calculated according to the standard curve and then converted into copies per gram of dry sediment based on sediment water content.

2.6 Statistical analysis

Spatio-temporal differences in the nitrate dissimilatory reduction processes of sediments were evaluated through one-way analysis of variance (ANOVA) and Dunn's test for multiple comparisons. Relationships among sediment nitrate dissimilatory reduction processes, environmental factors, and gene abundance related to nitrate reduction were analyzed using Pearson’s correlation analysis, redundancy analysis (RDA), and stepwise multiple regression analysis, and stepwise multiple regression analysis. A significance level of p < 0.05 was considered statistically significant. One-way analysis of variance, stepwise multiple regression, and Pearson’s correlation analyses were performed using SPSS 19.0. Stepwise multiple regression analysis identified the main factors influencing sediment nitrate dissimilatory reduction processes. Multivariate relationships among sediment nitrate dissimilatory reduction processes, environmental factors, and gene abundance related to nitrate reduction were examined using RDA. Pearson’s correlation analysis was used to reveal relationships between DEN, anammox, DNRA, N2O production rates, and sediment properties.

3 Results

3.1 Sediment physicochemical properties

The physicochemical properties of sediments in different river-lake ecotones and the lake ecosystem are shown in Fig. 2. Sediment Fe2+ and TOC contents were significantly higher in the wet season than in the dry season (T-test, p < 0.05). The average Fe2+ contents were 0.50 mg g-1 in the wet season and 0.42 mg g-1 in the dry season. Fe2+ contents in RH and DC were significantly lower than those in FH and XJ (one-way ANOVA, p < 0.05). NH4+ contents in RH and DC were significantly lower than those in GJ and XJ (one-way ANOVA, p<0.05), with the highest NO3- content observed in GJ. TOC contents in GJ, RH, DC, and XS were significantly lower than those in FH and XJ (one-way ANOVA, p < 0.05). NH4+ and TOC contents were 10.44 and 8.64 μg g-1, and 14.13 and 11.28 mg g-1 in the wet and dry season, respectively. Sulfide contents were 24.82 and 22.22 μg g-1 in the wet and dry season, respectively, with FH and XJ significantly higher than RH (one-way ANOVA, p < 0.05). Sediment pH in RH was significantly lower than in DC during the wet season (one-way ANOVA, p < 0.05).

Fig. 2
figure 2

Physiochemical characteristics of sediment in Poyang Lake (mean ± SE). Different letters (a, b, c, etc.) indicate significant differences (p < 0.05). GJ, RH, DC, FH, XS and XJ were represent samples in Ganjiang River, Raohe River, lake ecosystem, Fuhe River, Xiushui River and Xinjiang River, respectively

Full size image

3.2 Abundance of functional genes

The abundance of nirS, nirK, nosZ, anammox bacterial 16S rRNA, and nrfA genes showed no significant different between the wet and dry seasons, but significant differences were observed among the river-lake ecotones (one-way ANOVA, p < 0.05) (Fig. 3). The abundance of the nirS gene ranged from 7.19 to 7.85 log10 copies g−1 in the wet season and from 7.22 to 7.84 log10 copies g−1 in the dry season. The abundance of the nirS gene in RH was significantly lower than in FH and XJ (one-way ANOVA, p < 0.05). The nirK gene abundance in FH was significantly higher than in GJ, RH, DC, XS, and GJ (one-way ANOVA, p < 0.05) with FH showing the highest abundance of nirS and nirK genes in both seasons. The nosZ gene abundance in GJ, RH, and DC was significantly lower than in FH, XS, and XJ (one-way ANOVA, p < 0.05), with abundance ranging from 7.33 to 8.60 log10 copies g−1 in the wet season and from 7.06 to 8.60 log10 copies g−1 in the dry season. The highest abundance was observed in XS (the wet season) and FH (the dry season). The abundance of anammox bacterial 16S rRNA genes in DC was significantly lower than in GJ and XJ, with abundances ranging from 6.79 to 7.98 log10 copies g−1 in the wet season and from 6.66 to 7.85 log10 copies g−1 in the dry season. The abundance of the nrfA gene in DC and RH was significantly lower than in GJ, FH, and XJ (one-way ANOVA, p < 0.05), with abundances ranging from 6.66 to 8.10 log10 copies g−1 in the wet season and from 6.66 to 8.18 log10 copies g−1 in the dry season.

Fig. 3
figure 3

Sediment abundances of functional genes in Poyang Lake. Different letters (a, b, c, etc.) indicate significant differences (p < 0.05). GJ, RH, DC, FH, XS and XJ were represent samples in Ganjiang River, Raohe River, lake ecosystem, Fuhe River, Xiushui River and Xinjiang River, respectively

Full size image

3.3 Rates of dissimilatory nitrate reduction processes

The rates of sediment dissimilatory nitrate reduction processes at different river-lake ecotones and the lake ecosystem are shown in Fig. 4. Nitrate reduction rates in Poyang Lake were significantly higher in the wet season than in the dry season (T-test, p < 0.01). The rate of DEN was significantly higher than that of anammox and DNRA in both seasons (one-way ANOVA, p < 0.01). The average rates of potential DEN in the wet and dry seasons were 6.65–30.81 and 2.28–10.65 nmol g-1 h-1, respectively. Potential DEN rates in FH and XJ were significantly higher than those in GJ, RH, DC, and XS (one-way ANOVA, p < 0.05), with the highest DEN rate observed in FH. Potential anammox rates were 0.07–1.86 nmol g-1 h-1 and 0.03–0.72 nmol g-1 h-1 in the wet and dry seasons, respectively. Anammox rates in RH and DC were significantly lower than those in GJ and FH (one-way ANOVA, p < 0.05). The potential DNRA rate in Poyang Lake was significantly higher than the anammox rate (T-test, p < 0.01), with DNRA rates of 1.52–4.52 and 0.54–1.80 nmol g-1 h-1 in the wet and dry seasons, respectively. Average DNRA rates in FH, GJ, XS, and XJ were significantly higher than those in DC and RH (one-way ANOVA, p < 0.05).

Fig. 4
figure 4

Potential dissimilatory nitrate reduction rates of sediment in Poyang Lake. Different letters (a, b, c, etc.) indicate significant differences (p < 0.05). GJ, RH, DC, FH, XS and XJ were represent samples in Ganjiang River, Raohe River, lake ecosystem, Fuhe River, Xiushui River and Xinjiang River, respectively. DEN and DNRA represent denitrification rate and dissimilatory nitrate reduction to ammonium, respectively

Full size image

3.4 Potential N 2 O production rates

Potential N2O production at different river-lake ecotones and the lake ecosystem are shown in Fig. 5. Sediment N2O production was significantly higher in the wet season (0.14–11.55 nmol g-1 h-1) than in the dry season (0.09–6.47 nmol g-1 h-1) (T-test, p<0.01). Potential N2O production in GJ and RH was significantly lower than in DC, FH, XS, and XJ (one-way ANOVA, p < 0.05), with the highest N2O production observed in FH in both seasons. Sediment N2O/(N2O+N2) in FH was significantly higher than in other river-lake ecotones (one-way ANOVA, p<0.05). N2O/(N2O+N2) in the dry season (0.03–0.42) was significantly higher than in the wet season (0.02–0.32) (T-test, p < 0.05).

Fig. 5
figure 5

Sediment N2O production and N2O/(N2 + N2O) in Poyang Lake. Different letters (a, b, c, etc.) indicate significant differences (p < 0.05). GJ, RH, DC, FH, XS and XJ were represent samples in Ganjiang River, Raohe River, lake ecosystem, Fuhe River, Xiushui River and Xinjiang River, respectively

Full size image

The contributions of anammox, DEN, and DNRA to nitrate reduction are shown in Fig. 6. Potential DEN was the dominant process contributing to nitrate reduction, accounting for 83.99% and 83.67% in the wet and dry seasons, respectively. DNRA contributed 12.81% and 12.61% to nitrate reduction in the wet and dry seasons, respectively, with the largest contribution observed in RH. Sediment anammox contributed 3.20% and 3.52% to nitrate reduction in the wet and dry seasons, respectively.

Fig. 6
figure 6

Contributions of potential anammox (ANA), denitrification (DEN), dissimilatory nitrate reduction to ammonium (DNRA) and N2O production rates to nitrate reduction. a is represent Wet season and b is represent Dry season. GJ, RH, DC, FH, XS and XJ were represent samples in Ganjiang River, Raohe River, lake ecosystem, Fuhe River, Xiushui River and Xinjiang River, respectively

Full size image

3.5 Influencing factors of sediment dissimilatory nitrate reduction processes and N 2 O production

The results of Pearson’s correlation analysis and RDA analysis are shown in Tables S2, S3, and Fig. 7. The anammox rate was significantly positively correlated with anammox bacterial 16S rRNA abundance and NH4+ content. Sediment DEN, DNRA, and N2O production rates were significantly positively correlated with Fe2+, TOC, and sulfide contents. The DEN rate and N2O production were significantly positively correlated with the abundance of the nirS, nirK, nosZ genes, while the DNRA rate was significantly positively correlated with the abundance of the nrfA gene. The abundance of the nirS, nirK, nosZ, and nrfA genes was significantly positively correlated with Fe2+, TOC, and sulfide contents. N2O production was positively correlated with DNF, DNRA, nirS, nirK, nosZ, Fe2+, TOC, and sulfide.

Fig. 7
figure 7

Redundancy analysis (RDA) of dissimilatory nitrate reduction processes rates, associated functional gene abundances and physicochemical characteristics. ANA represent anammox rate; AMX represent anammox bacterial 16S rRNA gene, DEN represent denitrification rate and DNRA represent dissimilatory nitrate reduction to ammonium. GJ, RH, DC, FH, XS and XJ were represent samples in Ganjiang River, Raohe River, lake ecosystem, Fuhe River, Xiushui River and Xinjiang River, respectively

Full size image

Stepwise multiple regression analysis further identified the dominant factors affecting dissimilatory nitrate reduction processes. Based on the results (Table S4), the anammox rate was positively correlated with anammox bacterial 16S rRNA abundance. The DEN rate was positively correlated with nirK gene abundance and sulfide content in both the wet and dry seasons. N2O production and DNRA rates were positively correlated with the DEN rate and TOC content, respectively, in both seasons.

4 Discussion

4.1 Effects of river-lake ecotone on dissimilatory nitrate reduction processes in Poyang Lake

Compared with other aquatic ecosystems, sediment DEN and DNRA rates in Poyang Lake were similar to those in estuary, river, and lake ecosystems but significantly lower than in eutrophic and saline lakes (Table S5). The anammox rate in Poyang Lake sediments was significantly lower than in other aquatic ecosystems but similar to that in urban closed lakes. Seasonality variations in sediment potential anammox, DEN, DNRA, and N2O production rates were observed in Poyang Lake, indicating a close relationship between temperature and sediment dissimilatory nitrate reduction processes, consistent with previous studies (Korol et al. 2019; Jiang et al. 2021b; Tan et al. 2022). Temperature influences microbial metabolism, directly increasing nitrate reduction activities (Tan et al. 2022).

Nitrate reduction processes in lake sediments are regulated by differences in environmental factors (Plummer et al. 2015; Gao et al. 2019; Jiang et al. 2021a). This study observed that nitrate reduction processes in Poyang Lake sediments are significantly affected by TOC, sulfide, Fe2+, and the abundance of genes related to nitrate reduction processes. Sediment DEN, DNRA, and N2O production rates in river-lake ecotone systems were significantly higher than in the lake ecosystem. Furthermore, significantly higher TOC, sulfide, and Fe2+ contents were observed in river-lake ecotone systems, suggesting that the surrounding river watershed alters sediment environmental factors, ultimately enhancing nitrate dissimilatory reduction processes. Land use, agriculture, and urbanization in the watershed are likely to increase TOC, sulfide, Fe2+, and nitrate loading into the lake ecosystem, thereby changing sediment environmental factors and promoting nitrate dissimilatory reduction processes (Liu et al. 2016).

The sediment DEN rate was the dominant contributor to nitrate reduction, with an average contribution of 73.45%, similar to other aquatic environments (Gao et al. 2017; Yang et al. 2022). Average DEN rates in the river-lake ecotones and the lake ecosystem were 12.21 and 8.57 nmol g-1 h-1, respectively, indicating that river-lake ecotones increased DEN rates. RDA analysis showed that the sediment DEN rate in Poyang Lake is significantly affected by sediment TOC, sulfide, and Fe2+ contents, as well as the abundance of the nirS, nirK, and nosZ genes. Previous studies confirm that higher TOC content provides sufficient substrates and electron donors for denitrifying microorganisms, promoting the growth of denitrifying bacteria and increasing heterotrophic denitrification (Reisinger et al. 2016; Wang et al. 2018; Chang et al. 2021). In Poyang Lake, sediment TOC content was significantly positively correlated with the abundance of the nirS, nirK, and nosZ genes and the DEN rate, indicating that TOC content can significantly increase the abundance of these genes and ultimately promote the DEN rate. Sulfides can promote the DEN rate by serving as electron donors for denitrifier (Hou et al. 2012; Deng et al. 2015). Stepwise multiple regression analysis indicated that sulfide content and nirK gene abundance were the dominant factors affecting the DEN rate. Higher sulfide content was significantly positively correlated with the abundance of the nirK genes and the DEN rate, indicating that higher sulfide content in Poyang Lake promotes nirK gene abundance and increases the DEN rate. Changes in sediment Fe2+ contents can strongly regulate DEN rates (Shan et al. 2016; Li et al. 2022), as low-redox conditions and high substrate availability of Fe2+ favor the DEN process by promoting NO2- reduction (Picardal. 2012; Ding et al. 2014; Cheng et al. 2016). Therefore, higher Fe2+ content promotes the increase of nirS, nirK, and nosZ genes as well as the DEN rate.

N2O production contributed 12.43% to nitrate reduction, and stepwise multiple regression analysis results demonstrated that the DEN rate was the dominant factor regulating N2O production. Therefore, incomplete denitrification may be widespread in the sediments of Poyang Lake, and higher DEN rates significantly increase N2O production. Sediment N2O production showed a strong positive correlation with Fe2+, TOC, and sulfide contents and nirS and nirK gene abundance, indicating that these factors may promote incomplete denitrification in Poyang Lake. Higher contents of Fe2+, TOC, and sulfide have been reported to increase nirS and nirK gene abundance, ultimately promoting DEN rates and stimulating N2O production (McCrackin and Elser 2011; Gao et al. 2019; Jiang et al. 2021a). The nirS and nirK gene abundance also showed significant positive correlations with Fe2+, TOC, and sulfide contents, similar to previous studies. In the sediments of Poyang Lake, nosZ gene abundance, which enhances the conversion of N2O to N2, was influenced by sediment physicochemical properties. However, FH and XJ still showed higher N2O production, possibly due to the lower nosZ/(nirS+nirK) ratios in these locations. The ratio of nosZ/(nirS+nirK) is associated with higher N2O/(N2O+N2) ratios, and a higher ratio of nosZ/(nirS+nirK) may increase N2O reduction to N2 and decrease N2O production. (Duan et al. 2019; Zhang et al. 2021). Therefore, sediment N2O consumption is lower than N2O production under the influence of sediment physicochemical properties in Poyang Lake, indicating that higher nitrate removal may increase N2O production in Poyang Lake.

The DNRA process contributed 11.16% to nitrate reduction in Poyang Lake, representing nitrate retention in sediments. The higher DNRA rate showed significant positive correlations with TOC, sulfide, and Fe2+ contents and nrfA gene abundance (Gao et al. 2019; Jiang et al. 2020) (Fig. 7). The DNRA process is the preferred energy pathway, and higher sediment organic carbon content promotes NO3− consumption through the DNRA process (Hardison et al. 2015). Furthermore, TOC serves as the electron donor for the DNRA process, providing a favorable environment for DNRA microorganisms (nrfA gene) (Burgin and Hamilton 2007; Deng et al. 2015; Yin et al. 2017). Stepwise multiple regression analysis showed that TOC content was the dominant factor regulating the DNRA rate, indicating that TOC is crucial factor for increasing nrfA gene abundance and the DNRA rate in Poyang Lake. Sulfide content is significantly positively correlated with the DNRA rate because sulfides provide extra free energy for the nrfA gene, stimulating the DNRA rate (Simon et al. 2011; Yin et al. 2017). Under anoxic conditions, Fe2+ can be used by chemolithoautotrophic microbes, increasing the DNRA rate (Yin et al. 2017). Therefore, higher Fe2+ contents in sediments provide a favorable environment for the DNRA process (Giblin et al. 2013).

The sediment anammox rate contributed an average of 2.96% to nitrate reduction and was significantly correlated with NH4+ content and anammox bacterial 16S rRNA gene abundance, indicating that NH4+ content and bacterial activity can influence the anammox rate. NH4+ is an important substrate in the anammox process, and higher NH4+ content can increase anammox bacterial 16S rRNA gene abundance (Plummer et al. 2015; Shan et al. 2016). Therefore, higher NH4+ content promotes anammox bacterial 16S rRNA abundance and enhances the anammox rate. Sediment TOC, sulfide, Fe2+ and NH4+ contents were significantly higher in river-lake ecotones, indicating that these river-lake ecotones can alter sediment properties and ultimately increase nitrate reduction.

4.2 Environmental implications of dissimilatory nitrate reduction processes in Poyang Lake

Understanding the fate of nitrate in lake sediments is crucial for addressing lake N removal, N2O production, and N pollution problem issues. According to calculations, the average N removal from the anammox process was 8.73 g m-3 yr-1, while the DEN rate contributed 254 g m-3 yr-1 (217 g m-3 yr-1 for N2 and 37 g m-3 yr-1 for N2O) to nitrate removal. The rate of NO3- transformation to NH4+ through DNRA was 33 g m-3 yr-1, indicating that nitrate transformed to NH4+ and was retained in the sediments, potentially aggravating N pollution in Poyang Lake. Nitrate removal from DEN and anammox in river-lake ecotones and the lake ecosystem were approximately 234 g m-3 yr-1 and 157 g m-3 yr-1, respectively. Nitrate removal in river-lake ecotones significantly exceeded that in the lake ecosystem. However, sediment nitrate reduction in river-lake ecotones and the lake ecosystem were approximately 34 g m-3 yr-1 and 22 g m-3 yr-1, respectively, with N2O production rates of approximately 38 g m-3 yr-1 and 25 g m-3 yr-1, respectively. Therefore, while river-lake ecotones increased nitrate removal, they also increased nitrate reduction and N2O production, which may elevate the risk of lake eutrophication and enhance the greenhouse effect.

The fate of N in the sediments of Poyang Lake is illustrated in Fig. 8. The rate of N removal was 262 g m-3 yr-1. Over the area of 4000 km2 area of Poyang Lake, the total N removal is 1.05×106 t yr-1 (1.47×105 t for N2O and 9.02×105 t for N2). According to a previous study, the annual net anthropogenic N input into the lake was approximately 7443 kg km-2 (Chen et al. 2016). Given that the area of the Poyang Lake watershed is 163,242 km2, the total N input into the lake is 1.22×106 t yr-1, which is similar to the N removal rate. Therefore, net anthropogenic N inputs into Poyang Lake have not yet become a serious environmental problem. Approximately 11.16% of anthropogenic nitrate from the Poyang Lake watershed can be converted to NH4+ (1.32×105 t yr-1 in Poyang Lake), indicating that the DNRA process is an important pathway for N retention in Poyang Lake. Additionally, approximately 12.43% of anthropogenic nitrate from the Poyang Lake watershed can be converted to N2O, suggesting that the N2O production problem in Poyang Lake may be aggravated in the future. Nitrate fate in Poyang Lake indicates higher N2O production and nitrate retention in the river-lake ecotone, suggesting that river-lake ecotones are hot-spot regions for nitrate retention and N2O production.

Fig. 8
figure 8

Sediment nitrate removal, input and retention in Poyang Lake. DEN and DNRA represent denitrification rate and dissimilatory nitrate reduction to ammonium, respectively

Full size image

Lake watershed land use changes can regulate river hydrological and biogeochemical functions, controlling seasonal N input into lake ecosystems (Basu et al. 2022; Liu et al. 2022). The discharge of anthropogenic wastewater from agriculture land or urban land into lake ecosystems is recognized as a significant factor affecting the competition among DEN, anammox, and DNRA processes (Liu et al. 2016; Yang et al. 2022). Furthermore, with increased N fertilizer input and changes in irrigation and drainage management practices, agricultural land in the watershed can alter properties such as TOC, pH, and nitrate in lake sediments, increasing N loading and regulating lake dissimilatory nitrate reduction processes (Liu et al. 2016). Therefore, lake watershed land use is a critical factor causing lake nitrate pollution and N2O production. Strengthening land use management has become crucial for controlling pollutant transfer in the watershed (Du et al. 2023). Understanding the relationships between river watershed land use and sediment N transformation may become a future research direction.

5 Conclusions

This study investigated the dissimilatory nitrate reduction processes in lake sediments across different river-lake ecotones and the lake ecosystem in Poyang Lake, China. The rates of sediment DEN, N2O production, DNRA, and anammox in the river-lake ecotone ecosystem were significantly higher than in the lake ecosystem, with these rates being higher in the wet season than in the dry season. DEN was found to be the dominant process for N removal, followed by DNRA and N2O production rates. Sediment TOC, sulfide, and Fe2+ contents are key factors controlling the abundance of denitrifying and nrfA genes, ultimately increasing the rates of DEN, DNRA processes, and N2O production. The sediment anammox rate was significantly influenced by the abundance of the anammox bacterial 16S rRNA gene and NH4+ content, but it had a minimal contribution to N removal. The fate of N in sediments indicated that the annual input of reactive N to the lake from the watershed is almost equivalent to the annual N removal. However, approximately 23.59% of the annual N input to the lake may be transformed into N2O or NH4+ through the DEN and DNRA processes, suggesting that N input to the lake increases warming potential or retention in sediments, thereby threatening water quality and ecological functions. Although nitrate removal rates were higher in river-lake ecotones, the increased N2O production and nitrate retention in these areas still pose a threat to lake N pollution and N2O emissions. Strengthening land use management has become crucial for controlling the transfer of pollutants from the river into the watershed. Understanding the relationships between river watershed land use and sediment N transformation may become a key focus of future research.