Sediment samples were collected in July 2023 from the coastal intertidal zone of Haizhou Bay, Western Yellow Sea, China, following low tide, and simultaneously, ambient seawater was also collected at the same site. The sediments were sieved on-site to remove larger impurities and macroorganisms, and subsequently dried in air and sieved through a 2mm stainless steel mesh to achieve complete homogenization and removal of small impurities and interstitial water (Seeley et al., 2020). Concurrently, ambient seawater was filtered on-site using a 0.45 µm Millipore membrane prior to use.

The microcosm experimental design includes four commonly used MP treatments (PP, PVC, PA, and PE, reviewed in Wang et al., 2024) and a no MP control treatment, with replicates per treatment ( Supplementary Figure S1 ). These MPs are commercially available (Shenzhen Guangyuan Plasticisation Co., Ltd., China; high purity, greater than 99%) and are made in specific sizes as per the requirements of our experiments. MPs of different polymer types were further treated at different sizes (0.02 mm, 0.1 mm, and 1 mm; n = 26; Supplementary Figure S1 ). It needs to be clarified that the specific size of the MPs selected may not be fully representative of the size characteristics of MPs in diverse sediments, but rather reflect the range of sizes in which MPs may be present in environmental sediments (reviewed in Wang et al., 2024). The choice of these MPs was based on the main types of MPs commonly observed in the water column at the location where the sediment samples were collected (Song et al., 2023), in order to ensure a realistic representation of actual conditions. Furthermore, these are also the types of MPs that are frequently selected in other studies (reviewed in Wang et al., 2024), thus enabling a meaningful comparison. A relatively practical concentration of MPs (0.5% w/w) was set for the experiment based on available studies (Baysal et al., 2020; Fang et al., 2024; Seeley et al., 2020; Wang et al., 2024). We weighed 300 grams of the above fully homogenized sediment into a 1L beaker (sediment sample depth of 3 cm) and added the different polymer types and sizes of MPs mentioned above to achieve a consistent concentration (weight ratio to wet sediment), followed by seawater addition without disturbing the sediment as much as possible (Seeley et al., 2020). Prior to the start of the incubation experiment, the microcosm was gently aerated to maintain oxygen in the overlying water and to allow an oxic/anoxic gradient to develop in the sediment over a 24-hour period (Seeley et al., 2020). The control treatment followed the same steps as the MP treatment except that no MPs were added. Subsequently, the experiments were operated at room temperature (20°C), and each beaker was covered with a piece of tin foil for the duration of the experiments to prevent water evaporation or other disturbances caused by other factors.

The overlying water samples were collected at 30 mL on days 1, 3, 7, 18, and 25 of the experimental system. For NO₂ ⁻ concentration analysis, the samples were measured immediately or frozen for later measurement. For NO₂ ⁻ dual isotope analysis, 6 M sodium hydroxide was added to preserve the samples (prevent O isotope exchange between NO₂ ⁻ and H₂O) for isotopic measurement (Casciotti et al., 2007; Chen et al., 2021, 2022; Chen and Chen, 2022).

NO₂ ⁻ concentration was determined on a QuAAtro flow analyzer (Seal, Germany) by the flow injection technique using a Greiss-Ilosvay colorimetric method (Strickland and Parsons, 1972), with steps including reagent preparation (sulfanilamide and N-(1-naphthyl) ethylenediamine), and correction of blanks and refractive index.

The δ¹⁵N_NO2 and δ¹⁸O_NO2 were determined using the well-established azide reduction method (McIlvin and Altabet, 2005). Depending on the NO₂ ⁻ concentration, an appropriate sample volume (calculated based on an optimal injection size of 20 nmol for N₂O formed by the sample) was pipetted into a pre-acid washed and scorched 20 mL headspace vial. The vial was then sealed with an aluminum cap and silica pad, and purged with helium to remove any N₂O present (Chen et al., 2021; Chen et al., 2022a; Chen and Chen, 2022). The reaction reagents in the azide reduction method include sodium azide and acetic acid solutions, which were prepared and used on the same day. These solutions were added to a 50 mL headspace vial in a 1:1 ratio (the volume added is determined by the number of samples) and purged with helium for half an hour (Chen et al., 2021; Chen et al., 2022a; Chen and Chen, 2022). Since the sample had a high pH with the addition of sodium hydroxide, the concentration of the acetic acid solution needed to be increased to 7.84 M in order to achieve the conditions for optimal reaction rate (pH = 4.5, Chen et al., 2021; Chen et al., 2022a; Chen and Chen, 2022; Hu et al., 2016). After purging, 0.9 mL of sodium azide-acetic acid buffer solution was transferred to the sample vial with a syringe for reaction, and the reaction was ended by adding 0.5 mL of 10 M sodium hydroxide at 1 hour (Chen et al., 2021; Chen et al., 2022a; Chen and Chen, 2022). The N₂O produced by the reaction was automatically extracted, purified, concentrated, and analyzed online using a Thermo Finnigan Gasbench II purge-trap system interfaced with a DELTAplus XP isotope ratio mass spectrometer. To calibrate the isotope values of the samples, the NO₂ ⁻ isotopic standards (RSIL-N23, RSIL-N7373, and RSIL-N10219, Casciotti et al., 2007) were preserved and processed in the same procedure as the samples. A set of reference standards was inserted every 6–10 samples for analysis. The volume of the reaction solution was kept the same for each sample and standard to minimize the effect of pH in the azide reduction reaction (Granger et al., 2020; Kobayashi et al., 2021). The N content of the standards was matched to the samples to minimize the effect of sample size on the isotopic analysis (Granger et al., 2020; Kobayashi et al., 2021). The mean standard deviations for all replicate measurements of samples and standards in this study were 0.6‰ for δ¹⁵N and 0.4‰ for δ¹⁸O, respectively.

This study first reports the impact of MPs on the isotopic signatures of NO₂ ⁻ ( Figure 3 ) and analyzes the NO₂ ⁻ transformation pathways accordingly ( Figures 3 , 4 ). The impact of MPs on the source of NO₂ ⁻ was mainly analyzed on day 18 in our experiments, as NO₂ ⁻ was continuously produced, although there may also have been a simultaneous partial consumption of NO₂ ⁻. As for the subsequent NO₂ ⁻ consumption stage, the regulation of the NO₂ ⁻ pool by the NO₂ ⁻ consumption process was mainly considered, and therefore the analysis of the source process in this stage was not made here. Before an analysis of the specific source process of NO₂ ⁻, one basic framework that needs to be stated is that compared with δ¹⁵N_NO2, δ¹⁸O_NO2 has the advantage of distinguishing signals from two potential sources (NH₃ oxidation and dissimilatory NO₃ ⁻ reduction) of NO₂ ⁻, since the δ¹⁸O_NO2 values produced by these two processes are on opposite sides of the expected equilibrium value (δ¹⁸O_{NO2, eq}, when O isotope exchange equilibrium is reached between NO₂ ⁻ and ambient H₂O) (Buchwald and Casciotti, 2013; Casciotti, 2016a, 2016b; Chen et al., 2021; Chen et al., 2022a; Chen and Chen, 2022; Figure 4 ). Specifically, NO₂ ⁻ from dissimilatory NO₃ ⁻ reduction has a higher δ¹⁸O_NO2 than that from NH₃ oxidation, with the latter being lower than the equilibrium value (Buchwald and Casciotti, 2013; Casciotti, 2016a, 2016b; Chen et al., 2021; Chen et al., 2022a; Chen and Chen, 2022; Figure 4 ). Thus, although we did not measure the isotopes of NO₃ ⁻ and NH₄ ⁺, the major source of NO₂ ⁻ could be inferred from the measured δ¹⁸O_NO2 compared to the δ¹⁸O_{NO2, eq}. Using the equations from Buchwald and Casciotti, 2013, along with the measured temperature in the overlying water for each treatment and the assumed δ¹⁸O value of the H₂O (0‰, Buchwald and Casciotti, 2013), we then estimated the δ¹⁸O_{NO2, eq} in our experimental system to be 13.2 ± 0.1‰. During the NO₂ ⁻ accumulation stage, both the MPs and the control treatments were relatively close to each other in terms of δ¹⁸O_NO2 and all were slightly lower than the δ¹⁸O_{NO2, eq} ( Figure 3A and Figure 4 ), implying that the NO₂ ⁻ accumulation in all treatments was mainly contributed by NH₃ oxidation (Buchwald and Casciotti, 2013; Casciotti, 2016a, 2016b; Chen et al., 2021, 2022; Chen and Chen, 2022; Figure 4 ).

Thus, based on isotopic evidence, the higher or lower accumulation of NO₂ ⁻ in the MP treatment compared to the control may indicate that the polymer type and size of the MPs either promoted or impeded NH₃ oxidation, a specific NO₂ ⁻ source process. For instance, in the case of PE-0.02 mm, the accumulated NO₂ ⁻ was lower than that of the control ( Figure 1C ). Furthermore, given that its δ¹⁸O_NO2 was lower than the equilibrium value ( Figures 3A , 4 ), this may imply that this treatment may have impeded NH₃ oxidation, leading to a lower accumulated concentration of NO₂ ⁻. However, the assessment regarding this impediment impact should be viewed with caution. This is because, although we focus on the role of NO₂ ⁻ source processes in regulating the NO₂ ⁻ pool at this stage, NO₂ ⁻ consumption processes may also be operating simultaneously. These consumption processes may alter δ¹⁸O_NO2, which, in turn, may affect the assessment of the impact of MPs on NO₂ ⁻ production. Therefore, in conjunction with the analysis of δ¹⁵N_NO2, we provide a comprehensive analysis of this aspect below.

The fate of NO₂ ⁻ determines whether N is lost or retained in the environment, influencing greenhouse gas N₂O production and the environmental N budget. However, the impact of MP pollution on NO₂ ⁻ consumption pathways remains unclear (Fang et al., 2024; Green et al., 2016; Hope et al., 2020; Seeley et al., 2020). To investigate this, we used NO₂ ⁻ dual isotope data ( Figures 3A, B ), similar to analyzing the effect of MPs on NO₂ ⁻ source processes. Before proceeding, we must also clarify the effects of NO₂ ⁻ oxidation, denitrification (dissimilatory NO₂ ⁻ reduction), and anammox on NO₂ ⁻ dual isotopes to better assess the specific processes impacted by MPs. NO₂ ⁻ oxidation, either driven by NO₂ ⁻-oxidizing bacteria (Buchwald and Casciotti, 2010; Casciotti, 2009) or anammox bacteria (Brunner et al., 2013; Kobayashi et al., 2019), is expected to decrease both δ¹⁵N_NO2 ( Figure 3C ) and δ¹⁸O_NO2 ( Figure 3D ) (Brunner et al., 2013; Buchwald and Casciotti, 2010; Casciotti, 2009; Kobayashi et al., 2019). Denitrification is discussed here as dissimilatory NO₂ ⁻ reduction, since dissimilatory NO₃ ⁻ reduction, although also part of the denitrification, is a process that acts as the source of NO₂ ⁻. Dissimilatory NO₂ ⁻ reduction will result in an increase in δ¹⁸O_NO2 ( Figure 3C ) and δ¹⁵N_NO2 ( Figure 3D ) (Martin and Casciotti, 2016). Similarly, the operation of anammox will also result in an increase in δ¹⁵N_NO2 during NO₂ ⁻ reduction to N₂ (Brunner et al., 2013; Kobayashi et al., 2019), but its effect on δ¹⁸O_NO2 remains unclear (Casciotti, 2016b; Kobayashi et al., 2019).

In light of the effects of these consumption processes on NO₂ ⁻ concentration and isotopes, we herein first reevaluate the reliability of the above discussion of the impacts of MPs on NO₂ ⁻ production, focusing on treatments with lower NO₂ ⁻ accumulation compared to the control. Some treatments, such as PA-0.02 mm, PE-0.02 mm, and PE-0.1 mm, had lower NO₂ ⁻ concentrations compared to the control ( Figure 1C ), implying that NO₂ ⁻ production may have been impeded. However, the relationship between NO₂ ⁻ accumulation and δ¹⁵N_NO2 revealed a deviation in the PA-0.02 mm and PE-0.02 mm treatments ( Figure 5 ), indicating enhanced NO₂ ⁻ oxidation, which can simultaneously decrease both NO₂ ⁻ concentration and δ¹⁵N_NO2 (Brunner et al., 2013; Buchwald and Casciotti, 2010; Casciotti, 2009; Kobayashi et al., 2019). A similar effect was observed in the PE-0.1 mm treatment ( Supplementary Figure S3 ). However, as the MP size increased, the impact of NO₂ ⁻ oxidation on the NO₂ ⁻ pool became limited ( Supplementary Figure S4 ), likely due to the maximal NO₂ ⁻yield ( Figure 2 ), making the NO₂ ⁻ oxidation insufficient to significantly reduce NO₂ ⁻accumulation. Consequently, the lower NO₂ ⁻ concentrations in small-sized MP treatments may not be solely due to reduced NO₂ ⁻ production, but rather the promotion of NO₂ ⁻oxidation.

It is important to note that stronger NO₂ ⁻ oxidation in these treatments may confound δ¹⁸O -based evaluations of NO₂ ⁻ source processes, as this process would lower δ¹⁸O_NO2 (Brunner et al., 2013; Buchwald and Casciotti, 2010; Casciotti, 2009; Kobayashi et al., 2019), potentially masking the contribution of dissimilatory NO₃ ⁻ reduction to NO₂ ⁻ production ( Figure 4 ). But here, we focus mainly on whether the lower NO₂ ⁻ concentrations in these MP treatments relative to the control were due to source processes being impeded, regardless of the specific source process. Obviously, the above analysis is a good example to show the ability of natural abundance stable isotopes in addressing the impact of MPs on the N cycle.

Furthermore, during the NO₂ ⁻ consumption stage, the δ¹⁵N_NO2 was higher in the MP and the control treatments than in the previous stage ( Figure 3D ). This shows that the NO₂ ⁻ in some MP treatments, including the control, was consumed primarily through dissimilatory NO₂ ⁻ reduction or anammox ( Figure 4 ). However, some MP treatments showed outliers in δ¹⁵N_NO2 relative to the control ( Figure 3D ), meaning that there may have been variations in the pathway or strength of NO₂ ⁻ consumption. The presence of these outliers further supports size as an important factor in considering the impact of MPs on the N cycle. By analyzing these outliers, we can capture valuable information about the impact of MPs on the biogeochemical cycling of NO₂ ⁻. Accordingly, we performed a discussion of the relationship between NO₂ ⁻ concentration (residual), degree of NO₂ ⁻ consumption, δ¹⁵N_NO2 of residual NO₂ ⁻, and δ¹⁵N_NO2 variation in the MPs and control treatments during the NO₂ ⁻ consumption stage ( Figure 6 , Supplementary Figures S5 - S7 ). Among these, the degree of NO₂ ⁻ consumption and δ¹⁵N_NO2 variation represent the quantity of NO₂ ⁻ consumed and the impact of the NO₂ ⁻-consuming process on δ¹⁵N_NO2 over the period of 18–25 days, respectively.

The first case is enhanced NO₂ ⁻ oxidation (arrow 1 in Figure 6 ). The δ¹⁵N_NO2 decreased with increasing NO₂ ⁻ consumption in the PP treatment of different sizes ( Figure 6 , Supplementary Figures S5 - S7 ). In particular, for the PP treatment at 1 mm, the NO₂ ⁻ consumption was higher than that of the control, yet δ¹⁵N_NO2 was lower ( Figure 6 ). The low δ¹⁵N_NO2 signal during NO₂ ⁻ consumption could either be due to a low δ¹⁵N_NO2 during the previous NO₂ ⁻ accumulation stage compared to the control or a smaller increase in δ¹⁵N_NO2 during NO₂ ⁻ consumption. The first possibility can be excluded, as δ¹⁵N_NO2 values were consistent across treatments and the control during NO₂ ⁻ accumulation ( Figure 3B ). Therefore, the second possibility must be considered. As shown by the relationship between NO₂ ⁻ consumption and δ¹⁵N_NO2variation, the δ¹⁵N_NO2 increase in the PP treatment was smaller than in the control as NO₂ ⁻ consumption increased ( Supplementary Figures S5 - S7 ). This suggests a contribution of dissimilatory NO₂ ⁻ reduction or anammox (Brunner et al., 2013; Martin and Casciotti, 2016; Kobayashi et al., 2019), but the limited increase implies these processes were weaker than in the control. If these processes were weaker, NO₂ ⁻ consumption would be lower, which contradicts the higher NO₂ ⁻ consumption in the PP treatments ( Figure 6 , Supplementary Figures S5 - S7 ). Thus, another process must be consuming NO₂ ⁻ and limiting δ¹⁵N_NO2 increase. We propose that NO₂ ⁻ oxidation played a more significant role in NO₂ ⁻ consumption in the PP-0.1 mm and PP-1 mm treatments ( Figure 6 , Supplementary Figure S6 , and Supplementary Figure S7 ), as it can both consume NO₂ ⁻and reduce δ¹⁵N_NO2 ( Figure 4 , Brunner et al., 2013; Buchwald and Casciotti, 2010; Casciotti, 2009; Kobayashi et al., 2019). This suggests that in environments polluted by these MPs, NO₂ ⁻ may be retained as bioavailable N, potentially reducing N₂O greenhouse gas production.

The second scenario is enhanced dissimilatory NO₂ ⁻ reduction or anammox (arrow 2 in Figure 6 ). The higher δ¹⁵N_NO2 values and NO₂ ⁻ consumption in some MP treatments, especially PA-1mm, imply a stronger operation of dissimilatory NO₂ ⁻ reduction or anammox ( Figure 4 , Brunner et al., 2013; Martin and Casciotti, 2016; Kobayashi et al., 2019). This is further supported by the observation that in the PA-1mm treatment, the increase in δ¹⁵N_NO2 during NO₂ ⁻ consumption was greater than in the control ( Supplementary Figure S6 and Supplementary Figure S7 ). Unlike the previous scenario, this indicates that more NO₂ ⁻ is being lost from the environment as gas (N₂O or N₂) in this scenario, suggesting that the production of the greenhouse gas N₂O may be enhanced.

The final scenario may involve an enhanced dissimilatory NO₃ ⁻ reduction and dissimilatory NO₂ ⁻ reduction (or anammox) (arrow 3 in Figure 6 ). This scenario is distinct because, although NO₂ ⁻ consumption was limited, there was an increase in δ¹⁵N_NO2 compared to the control (PE-0.02 mm, PE-0.1 mm, PA-0.02 mm, Figure 6 , Supplementary Figure S5 , and Supplementary Figure S6 ). The increase in δ¹⁵N_NO2 suggests that dissimilatory NO₂ ⁻ reduction or anammox are the primary processes responsible for NO₂ ⁻ consumption ( Figure 4 ; Brunner et al., 2013; Martin and Casciotti, 2016; Kobayashi et al., 2019). However, since NO₂ ⁻ consumption was not significant, we propose that dissimilatory NO₃ ⁻ reduction, the first step of denitrification, may also be regulating the NO₂ ⁻ pool. This could offset NO₂ ⁻ consumption, and the δ¹⁸O_NO2 analysis supports this hypothesis. The δ¹⁸O_NO2 values of these MP treatments were all higher than the control and equilibrium values ( Figure 3C ), indicating a contribution from dissimilatory NO₃ ⁻ reduction ( Figure 4 ). Consequently, the imbalance between the source and consumption of NO₂ ⁻ in this scenario likely led to insufficient NO₂ ⁻ consumption coupled with increased δ¹⁵N_NO2. When considered in terms of NO₂ ⁻ consumption pathways, this also reveals a promotion of dissimilatory NO₂ ⁻ reduction or anammox in these MP treatments compared to the control. Without the isotope data, such low consumption of NO₂ ⁻ in these MP treatments might have been considered to be an inhibition of the dissimilatory NO₂ ⁻ reduction or anammox by these MPs relative to the control. Thus, this is once again a prime example of the value of natural abundance stable isotopes in identifying the real impact of MPs on the N cycle.

Overall, the isotopic evidence clearly indicates that the consumption pathways of NO₂ ⁻ vary depending on the polymer type and size of the MPs, which plays a crucial role in determining whether the production of the greenhouse gas N₂O in the environment increases or decreases. This suggest that MPs may contribute to local variations in greenhouse gas emissions, potentially altering greenhouse gas inventories in affected regions (Li et al., 2022; Ren et al., 2020; Su et al., 2022; Wang et al., 2024). These findings underscore the importance of considering MPs as an emerging environmental factor in greenhouse gas management. Future studies should aim to explore the long-term effects of MPs on N transformations and their cumulative impact under various environmental stressors, thereby improving the accuracy of greenhouse gas emission inventories and informing more effective mitigation strategies.