Introduction

Plastic debris is ubiquitous and problematic in global marine ecosystems1,2,3. With the increasing public perception of plastic pollution, increasing research efforts have been devoted to revealing the impacts of plastic debris on fish4, invertebrates5, sea birds6, turtles7, and microorganisms8 in oceans. In recent years, these buoyant plastic debris pieces in marine environments have been reported to provide durable substrates for microorganisms, facilitating colonization and biofilm formation9,10. Microbial communities on plastic debris have been called “plastisphere” and represent a novel feature in the Anthropocene, even sometimes called the eighth continent, sparking global interest9,10,11,12,13,14. Recent studies have shown that plastisphere exhibited great differences in microbial community and assembly from surrounding bulk water12,14. Further, plastisphere microbial communities could contribute to the acquisition and spread of pathogens and antibiotic resistance in aquatic environments15. These works suggest that the plastisphere is indeed capable of inducing negative effects at the ecosystem levels. However, the crucial roles of the plastisphere in biogeochemical cycling in marine environments are largely unknown.

Estuaries are the transitional areas between inland waters and remote oceans16, acting as hot zones for biogeochemical cycling. An estimated 1.2–2.5 million tons of plastic debris enter estuarine and coastal regions from inland rivers and lakes annually17. Therefore, any influence of plastics and the associated plastisphere on estuarine ecology and biogeochemical cycling could be particularly important. In addition to plastic contamination, estuaries are also subjected to unprecedented reactive nitrogen loads from anthropogenic activity, which leads to water eutrophication and ecological perturbations in estuaries18,19. Under anoxic conditions, microbially mediated denitrification in estuarine waters, stepwise reductions of nitrate or nitrite to potent greenhouse gas N2O or further to N220, is the major process to eliminate the reactive nitrogen21. This pathway can alleviate the effects of eutrophication21,22, but might contribute to global warming and ozone depletion23. Therefore, attention on denitrification and associated N2O productions should be paid in eutrophic estuarine regions. In the presence of plastic debris, microbial biomass in the plastisphere is likely considerable10, and anoxic microenvironments would occur during biofilm formation24,25. Thus, the emerging plastisphere could harbor great but unrecognized potential in nitrogen removal and N2O production. Elucidating the role of plastisphere in denitrification, especially in eutrophic waters, is of great significance for water quality remediation and climate change evaluation.

Research on denitrification processes in terrestrial and aquatic environments mainly focuses on bacterial denitrification (BD)22,26,27. More evidence has shown that fungal denitrification (FD) and chemodenitrification (CD) could also contribute to nitrogen removal and formation of N2O28,29,30,31,32,33,34. Recent studies suggest that a high abundance of fungi, such as parasitic and saprophytic ones, could be attached to the plastic surfaces in aquatic environment35. Moreover, fungal denitrifiers generally lack N2O reductase, leading to N2O as the end product36,37. Thus, fungal denitrifiers may represent a potent contributor to N2O production relative to bacterial denitrifiers (reduction of nitrate to N2 as the end product). In addition, in organic matter- and iron-rich environments, iron (II) reacts with the reactive nitrogen to increase N2O production from abiotic denitrification38,39,40, which may also be a great contributor relative to BD. In fact, the importance of FD and CD in nitrogen transformation and N2O production has been recognized across terrestrial ecosystems31,38,41. However, their contributions in marine waters are largely unknown. Comparing to BD, whether FD and CD processes have similar importance for nitrogen removal and N2O yield in marine plastisphere requires further investigation.

Here, we choose four types of plastics and establish in situ and lab-scale incubations in an estuary of Xiamen, China to investigate BD, FD, and CD potential in the plastisphere and to differentiate the contributions of these processes to N2O production (experimental design, see Supplementary Fig. 1). We hypothesize that (i) the surface of plastic debris could be a site of plastisphere formation that provides the necessary conditions for denitrification, (ii) plastisphere has higher denitrifying activity than surrounding bulk water, and (iii) FD and CD processes may have a great contribution to N2O production relative to BD. To test these hypotheses, we first measure messenger bis(3′−5′)-cyclic dimeric guanosine monophosphate) (c-di-GMP), extracellular polymeric substances (EPS), and intracellular lipid/fatty acid levels to explore the plastisphere formation. Then the denitrifying activities in plastisphere and bulk water are detected by 15N isotope-pairing technique, and the relative contributions of BD, FD, and CD to total N2O production are estimated using N2O isotopocules analysis. Finally, the keystone bacterial and fungal denitrifiers of plastisphere and bulk water are identified. Our results fill a knowledge gap regarding biogeochemical cycling in the plastisphere and indicate that the new plastic niche may be a potential source of N2O emission under global environmental change.

Results

Biofilm structure and activity

After the 30-d incubation in the estuary, the four types of plastic debris and surrounding bulk water were harvested. Using scanning electron microscope (SEM), we found that microorganisms in water were attached to the plastics and formed dense microbial clusters (Supplementary Fig. 2), suggesting biofilm formation on the surface of plastics.

We further explored anoxic microbial activities in the plastisphere and bulk water (Fig. 1). When cultured with heavy water (D2O), assimilation of D by metabolically active cells into new biomass can be explicitly and quantitatively detected by Raman as a new C–D band (Fig. 1a). Thus the ratio of (C–D)/(C–D + C–H) is a quantitative indicator for microbial phenotypic activity15,42. We acquired the Raman spectrograms of 40–50 microbial cells in each plastisphere and bulk water after 30 days, and calculated their (C–D)/(C–D + C–H) ratios. The ratios in each plastisphere (mean values 0.18–0.29) showed no statistical variations (Tukey, P = 0.203–0.953, Fig. 1b), but were greater than those in bulk water (mean value 0.09) (P < 0.001). This indicated a higher phenotypic activity in the plastisphere than in bulk water.

Fig. 1: D2O-labeled single-cell Raman spectroscopy showing microbial metabolic activities under anoxic conditions in the plastisphere and bulk water.
figure 1

a Schematic diagram of the D2O-labeled single-cell Raman technique. 1 represents the incubation and preparation of microorganism liquids; 2 represents Raman spectroscopy. Raman characteristic bands assigned to intracellular lipid/fatty acid biomolecules are pinpointed. b The ratios of (C–D)/(C–D + C–H) calculated from 40 to 50 single cells in the plastisphere and bulk water. Each point is a single cell. Different letters (a and b) indicate the significant differences (P < 0.05) among each plastisphere group and bulk water group.

Full size image

Biofilm composition, c-di-GMP, and lipid/fatty acid signals

A schematic diagram of microbial colonization processes on plastic surfaces in aquatic environments is illustrated in Fig. 2a. We detected the EPS contents (i.e., extracellular polysaccharide (PS), protein (PN), and DNA (eDNA)) in both the plastisphere and bulk water. As expected, the concentrations of PS, PN, and eDNA were all significantly greater than those in bulk water (P < 0.001, Fig. 2b–d). Except for PN and eDNA in the PVC plastisphere, no significant difference (P = 0.125–0.669) was detected among each plastic debris (Fig. 2b–d).

Fig. 2: Concentrations of extracellular polymeric substances (EPS) in the plastisphere and bulk water.
figure 2

a Schematic overview of biofilm formation and dispersion on plastic surfaces in water, including cell initial attachment, irreversible attachment, maturity, and dispersion. This diagram is modified from previous studies25,48. b–d Concentrations of extracellular polysaccharide (PS), extracellular protein (PN), and extracellular DNA (eDNA). Error bars are the standard error. Different letters (a, b, and c) indicate the significant differences (P < 0.05) among each plastisphere group and bulk water group.

Full size image

The intracellular second messenger c-di-GMP is associated with biofilm formation43. We observed higher concentrations of c-di-GMP in the plastisphere (0.013–0.019 μg mg–1 protein) than in bulk water (below detection limit) (P < 0.001, Supplementary Fig. 3). This showed that high levels of c-di-GMP commonly support microbial growth in a sessile mode, whereas the low level could result in the planktonic mode25. There were no obvious differences among the plastisphere formed by different plastic types (P = 0.135–0.987, Supplementary Fig. 3), suggesting that plastic types could not influence microbial c-di-GMP levels.

Microbial lipid and fatty acid levels may also be involved in biofilm formation and dispersion44. Using D2O-labeled single-cell Raman spectroscopy, a total of five characteristic bands assigned to lipid or fatty acids (i.e., 972, 1032, 1080, 1250, and 1735 cm–1) were identified (Fig. 3). Among the five bands, the plastisphere had no striking difference in peak intensities with bulk water at the bands of 972 and 1735 cm–1 (P = 0.268–0.883, Fig. 3a, e). At 1032, 1080, and 1250 cm–1, however, the peak intensities in the plastisphere were lower than those in bulk water (P < 0.001, Fig. 3b–d), highlighting the different lipid/fatty acid levels in microorganisms between the plastisphere and bulk water.

Fig. 3: Raman profiling of intracellular lipid/fatty acid levels in microorganisms of the plastisphere and bulk water.
figure 3

a 972 cm–1. b 1032 cm–1. c 1080 cm–1. d 1250 cm–1. e 1735 cm–1. Different letters (a and b) indicate the significant differences (P < 0.05) among each plastisphere group and bulk water group.

Full size image

Denitrification and N2O production pathway

During the in situ incubation, the plastic debris was suspended in the water depth of ~4 m, where the dissolved oxygen (DO) concentration was 1.19 mg L–1 (<2 mg L–1, generally considered as hypoxia45). Thus, we focused on the denitrification-derived N2O in the lab-scale experiments of this study. Plastisphere and bulk water showed pronounced differences in denitrification rate and N2O production (Fig. 4). After 24 h, N2O concentrations were 0.44–0.59 fmol cell–1 in the plastisphere, greater than 0.41fmol cell–1 in bulk water (P < 0.001–0.042, Fig. 4a). N2O concentrations in the PS plastisphere were lower than the other types of plastisphere (P < 0.001–0.012). For denitrification rate, the plastisphere registered higher rates ranging from 3.36 to 4.07 fmol cell–1 h–1, while the bulk water had a lower rate (mean value 2.97 fmol cell–1 h–1) (P = 0.001, Fig. 4b). These corroborated that the plastisphere could be a potential source of denitrification and N2O production. Comparing with Anammox process, denitrification contributed 73–84% of total N2 productions (Fig. 4c), meaning the dominant role of denitrification in nitrogen removal in the estuary.

Fig. 4: Denitrifying activity and N2O production in the plastisphere and bulk water.
figure 4

a Concentrations of N2O after the 24 h denitrification experiment. b Denitrification rates quantified by 15N isotope-pairing technique. Both N2O production and denitrification rate are normalized with microbial cell numbers by using the absolute abundances of 16S rRNA in the plastisphere and bulk water (see “Methods”). c Contributions of denitrification and Anammox processes to N2 production. Error bars are the standard error. Different letters (a, b, and c) indicate the significant differences (P < 0.05) among each plastisphere group and bulk water group.

Full size image

We further measured N2O isotope values in the plastisphere and bulk water (Supplementary Fig. 4). Coupling N2O-δ18O with N2O-SP (site preference, see “Methods”) values, we could investigate the N2O dynamics during denitrification. In the plastisphere, N2O-δ18O were 64.14–77.82‰ and N2O-SP were 15.56–26.21‰, while they were 55.62–59.93‰ and 7.53–8.68‰ in bulk water, respectively (P < 0.001, Supplementary Fig. 4), signifying a shift in N2O production pathways. After calibration for oxygen exchange, a multiple-isotope plot (N2O/H2O-δ18O vs N2O-SP) was generated (Fig. 5a). Considering N2O mixing and reduction, we established a N2O isotope-based mass balance model to differentiate the relative fractions of BD, FD, and CD processes to N2O productions (see “Methods”). We found that BD was the major N2O source, contributing 51.08–67.06% and 77.10–79.59% of total N2O production in the plastisphere and bulk water, respectively (P = 0.008, Fig. 5b). FD unexpectedly contributed 8.55–46.40% of total N2O production, and the contribution in the plastisphere was higher than that in bulk water (P < 0.001, Fig. 5b). These demonstrated that the plastisphere exhibited a different N2O production pattern from bulk water. Comparing with BD and FD, the proportion of the CD process (0–10.28%) can be ignored, for both plastisphere and bulk water.

Fig. 5: Multiple-isotope signatures of N2O illustrating the relative contributions of bacterial denitrification (BD), fungal denitrification (FD), and chemodenitrification (CD) in the plastisphere and bulk water.
figure 5

a Relationships between N2O-SP and N2O-δ18O (N2O/H2O) values in each group. N2O-δ18O values were corrected for H2O-δ18O assuming almost complete O exchange. Ranges of putative isotope signatures of N2O (δ18O (N2O/H2O)) from BD, FD, and CD are indicated as the three color boxes. The corresponding references are listed in the figure. The area enveloped by gray dash lines represents the N2O reduction through BD process based on fractionation effects (slope=SPεN2O/δ18OεN2O = 0.25). This mapping approach is modified from the method in ref. 83. b Relative proportions of N2O production via BD, FD, and CD in the plastisphere and bulk water. Errors were estimated by Monte Carlo simulation and reflected the accuracy of N2O isotope mass balance model. Although the contributions of each denitrification process would shift with the prescribed extent of oxygen change and N2O-SP and N2O-δ18O values, all ranges suggest that biotic denitrification instead of chemodenitrification was the main contributor to N2O production in the plastisphere and bulk water.

Full size image

Denitrifier communities

We first quantified the abundances of 16S rRNA and ITS genes in the plastisphere and bulk water (Supplementary Fig. 5), and found that the abundances of bacteria and fungi were higher in the plastisphere than those in bulk water (P < 0.001–0.027). Denitrifier communities were then investigated by high-throughput sequencing based on three denitrifying marker genes (i.e., bacterial nirS (bnirS), bacterial nirK (bnirK), fungal nirK (fnirK)). As indicated by α-diversity, the plastisphere had a higher diversity and richness than bulk water (P < 0.05, Supplementary Fig. 6). The diversity of bnirS-type denitrifiers reached the highest, followed by fnirK-type and by bnirK-type denitrifiers. In light of the Bray-Curtis distances (Fig. 6a), the three denitrifier communities between plastisphere and bulk water clearly formed two separated clusters (P < 0.01, R = 0.873–0.927). No obvious or separated clusters among the four plastic debris were observed. As shown in Supplementary Fig. 7, the major phylum was Proteobacteria in bnirS-type and bnirK-type denitrifier communities, and it was Ascomycota and Basidiomycota in fnirK-type denitrifier communities. However, their relative abundances varied greatly between plastisphere and bulk water, implying a distinct denitrifier structure between these compartments. We also found that plastic types did not obviously change the denitrifier compositions and relative abundances (Supplementary Fig. 7), albeit with some variations in bnirK-type denitrifier community at the genus level for the polystyrene (PS) type.

Fig. 6: Denitrifier community in the plastisphere and bulk water.
figure 6

a Principal coordinates analysis with ANOSIM test showing the differences in community structures of bnirS-type, bnirK-type, and fnirK-type denitrifiers between plastisphere and bulk water. b Co-occurrence network analysis illustrating the keystone denitrifiers (nodes, R2 > 0.9, P < 0.05) in the plastisphere and bulk water. The parameters of the networks are listed in Supplementary Table 1. bnirK-type denitrifiers dominated in the keystone taxa in the plastisphere, while bnirS-type ones were the main keystone in bulk water. c, d Niche breadth and overlap based on the three types of denitrifier communities. Error bars are the standard error. Different letters (a and b) indicate the significant differences (P < 0.05) between plastisphere and bulk water.

Full size image

Co-occurrence network analysis showed that the plastisphere harbored more complex interactions among the three denitrifier communities in comparison with bulk water (Fig. 6b and Supplementary Table 1). We found that the keystone genera of denitrifiers were shifted between plastisphere and bulk water. In the plastisphere, the bnirK-type denitrifiers (Pseudomonas, Bosea, Ochrobactrum, and Starkeya) accounted for the main keystone genera, while bnirS-type denitrifiers (α,β-Proteobacteria and Azoarcus) were dominant in bulk water (Fig. 6b). fnirK-type keystone denitrifiers were also changed from Talaromyces, Chaetomium, and Penicillium in the plastisphere to Fusarium, Alternaria and Trichosporon in bulk water. Such shifts probably explained the different denitrifying activities between plastisphere and bulk water. In addition, we also quantified the relative abundances of narG, nirS, nirK, and nosZ (Supplementary Fig. 8). Plastisphere generally contained higher abundances than the bulk water (P = 0.001–0.038), partially explaining the higher denitrification rates in the plastisphere.

We further explored the denitrifier-based niche breadth and overlap at community level. Niche breadths of all three denitrifier communities in the plastisphere were greater than those of bulk water (P < 0.001, Fig. 6c). This indicates that plastic surfaces could offer more opportunities for denitrifiers attachment and biofilm development than the surrounding water46. In line with the results of niche breadth, the plastisphere also presented a higher niche overlap level (P < 0.001, Fig. 6d), which reflects a sharp competition in denitrifier communities for energy and nutrients in comparison with bulk water47.

Discussion

The potential of plastisphere in global nitrogen cycling and associated greenhouse gas flux is rarely examined. Our study provides evidence of the plastisphere exhibiting a higher denitrifying activity compared to bulk water and functioning as a neglected source of greenhouse gas N2O production. Below, we discuss the likely reasons for these findings.

In our study, higher levels of PS, PN, and eDNA were observed in the plastisphere comparing with bulk water (Fig. 2b–d). This indicates that the estuarine microbes secrete EPS and then grow within the self-produced extracellular matrix to form biofilms during their colonization on plastic debris24,48. We also found that the EPS concentrations were coordinated with the c-di-GMP levels that were higher in the plastisphere and lower in bulk water (Supplementary Fig. 3). This is likely because microbial secretion of EPS and biofilm formation are mainly regulated by the intracellular c-di-GMP level, especially in the Proteobacteria phylum25,43. Recent studies have demonstrated that higher c-di-GMP levels could enhance EPS yields through repressing the expression of transcription factor FleQ (P. aeruginosa)49 or increasing the surface adherence and immobility organelles such as T4P (V. cholerae and C. difficile)50. In addition, c-di-GMP is also capable of downregulating flagellar expression (P. aeruginosa), reducing flagellar assemblies (C. crescentus)43, or perturbing flagellar motor function (E. coli)51, which suppresses the motility and gradually forms biofilms. Therefore, the elevated c-di-GMP levels in the plastisphere are important for EPS production and biofilm formation on the plastic debris.

Another important finding is that microbial lipid and fatty acid levels were lower in the plastisphere than in bulk water (Fig. 3). Intracellular lipid/fatty acid signals are reported to be involved in biofilm formation and dispersion44. Accumulations of lipid/fatty acids in microorganisms could activate the activity of phosphodiesterases (catalyzing c-di-GMP consumption pathway), leading to lower c-di-GMP levels and biofilm dispersion25. Moreover, lipid/fatty acid signaling molecules, such as cis-2-decenoic acid, cis-11-methyl-2-dodecenoic acid, and cis-2-dodecenoic acid, are capable of dismantling small microaggregates and inhibiting biofilm formation52,53. For example, previous studies have shown that biofilm roughness, thickness, and biomass were significantly reduced in the presence of cis-2-decenoic acid52. This is likely because these signaling molecules could maintain bacterial and fungal cells metabolically active, benefit for dispersive growth. Thus, the lower intracellular lipid/fatty acid levels observed here might be another important reason for the biofilm formation on the plastics.

During biofilm formation, surfaces of plastics are more likely to recruit anoxic denitrifiers, which is substantiated by the higher abundances of denitrifiers and denitrifying genes in the plastisphere than in bulk water (Supplementary Figs. 7 and 8). Low concentrations of dissolved oxygen in biofilms may account for the result. In biofilms, a steep gradient of dissolved oxygen could occur from outer to inner spaces, and thus produce anoxic conditions54,55. Since denitrifiers can utilize nitrate/nitrite as electron acceptors instead of O2 to support their metabolisms20, the oxygen-limited conditions are favorable for their growth and activity. The higher anoxic activities in the plastisphere (Fig. 1) further supported this explanation. In addition, we also found that the bacterial denitrifiers in the plastisphere mainly belonged to Proteobacteria phylum (Supplementary Fig. 7), consistently with previous studies indicating that Proteobacteria was the globally most abundant on surfaces of marine plastics56. Many members of Proteobacteria (such as Pseudomonas) display strong capacities of EPS secretion and have evolved a “holdfast”57. Thus, these denitrifiers may be favored compared to other microorganisms for surface attachment onto plastic debris. Taken together, our study revealed that, on contact with the plastic debris, estuarine microorganisms produced EPS to form biofilm by the regulation of c-di-GMP and lipid/fatty acid signals. The subsequently generated anoxic microsites could provide opportunities for denitrifiers, and thereby resulted in the higher denitrifying activity in the plastisphere, supporting our first hypothesis.

Denitrification and N2O production were recorded in the plastisphere and bulk water (Fig. 4). The observed denitrification rates are of comparable magnitude to previous studies conducted in remote oceans58, coastal, and estuarine sediments21, suggesting the importance of the denitrification process in marine ecosystems. Comparing with bulk water, plastisphere had greater denitrification rates and N2O productions (Fig. 4), which supports our second hypothesis. It can be soundly attributed to the differences of denitrifier communities and biomasses between plastisphere and bulk water. Our study showed that the niche breadths of denitrifier communities in the plastisphere were higher than that of bulk water (Fig. 6c), implying that more denitrifiers are likely to be attached to the plastics than to freely live in water14,46. Network analysis further indicated that bnirK-type denitrifiers dominated in the keystone taxa of plastisphere, whereas bnirS- type ones were the core residents in bulk water (Fig. 6b). In comparison with bnirS-type denitrifiers, bnirK-type ones could display strong adaptability to fluctuating environments (i.e., dissolved oxygen)59, and thus are more easily attached onto the plastics. Furthermore, bnirK-type denitrifiers are more effective for nitrite reduction and N2O production than bnirS-type ones60. This is further evidenced by the greater ratios of (nirS + nirK)/nosZ (1.5–2.3) in the plastisphere than those (0.33–0.41) in bulk water (Supplementary Fig. 8). All of these accounted for the stronger denitrifying capability in the plastisphere.

Using the mass balance model, we found that BD dominated in N2O production in both the plastisphere and bulk water (Fig. 5b). Generally, the oxygen-limited conditions are more favorable for BD process. The bacterial denitrifiers (bnirS- or bnirK- type), identified as the main keystone taxa in the plastisphere and bulk water (Fig. 6b), further supported the bacterial dominance in N2O productions. The bacterial proportions in this study (51–79%) were higher than those of previous studies conducted in coastal (14–60%)28 and estuarine (15–21%)61 sediments, suggesting a spatial heterogeneity of BD in marine ecosystems. In addition, we noticed that the bacterial contributions to N2O production had no striking difference among the four types of plastics (Fig. 5b), which signifies the intrinsic importance of BD, rather than an opportunistic response in the estuarine plastisphere. The enhanced contribution of BD is favorable for reactive nitrogen removal by the transformation to N2O and N2. This implies that BD is an important source for greenhouse gas N2O.

Fungal contributions to N2O production in estuarine water have been less well studied. It generally derives from two pathways including FD and fungi-mediated codenitrification31. During codenitrification with the addition of 15N-nitrate/nitrite, fungi would form more hybrid 45N2O and 29N2 than 46N2O and 30N262,63. In our 15N-labeling experiment, however, we detected a large portion of 30N2 rather than 29N2, indicating that the codenitrification contribution to N2O production was less in this study. Fungal N2O production could thus be mainly from FD process. Our study further showed that the contributions of FD to total N2O productions were 8–46% in the plastisphere and bulk water (Fig. 5b). This highlighted the non-negligible role of fungal denitrifiers in N2O production, despite the proportions being lower than those in marine sediments (40–70%)28, which partially supports our third hypothesis. Moreover, we found higher proportions of N2O production from FD in the plastisphere (26–46%) than in bulk water (8–22%, Fig. 5b). Fungal niche selection may account for the different N2O production patterns between plastisphere and bulk water, because the aquatic microorganisms are prone to nesting in more fragmented habitats like plastic debris rather than in water14. The biofilms formed by attached microbes promote cellular interactions, thus likely leading to the distinct N2O dynamics between plastisphere and bulk water.

The small contribution of CD to N2O production (0–10.28%) indicated that CD is not an important N2O contributor in both the plastisphere and bulk water (Fig. 5b). By contrast, Wankel et al.28 and Wang et al.38 documented that the contributions of CD to N2O production could reach 31–75% in paddy soils and coastal sediments. The different results were probably caused by the different iron contents that play an essential role in the CD process64. In soils and sediments, the chemical reactions of particle-bound reduced iron with various enzymatically-derived electron acceptors (i.e., hydroxylamine, nitric oxide or nitrite) can contribute to pronounced yields of N2O64. In our study, however, the iron concentrations were low (2.89–7.95 mg L–1), thus likely limiting the importance of the CD process in estuarine water.

It is worth noting that the nitrification process in the plastisphere and bulk water, including hydroxylamine oxidation and nitrifier denitrification41,65, may also contribute to N2O production. We further conducted a lab-scale experiment under in situ dissolved oxygen concentration (~1.2 mg L–1, hypoxic conditions) to explore nitrifier potential in N2O production in the plastisphere and bulk water (Supplementary Fig. 9). There was no significant difference observed between groups with and without adding allylthiourea (a nitrification inhibitor66), indicating that N2O production via hydroxylamine oxidation was minimal under in situ conditions. By contrast, nitrifier denitrification was likely a potential contributor in the plastisphere and bulk water (Supplementary Fig. 9b), but the concentration was lower than that of denitrification-based N2O. More importantly, the nitrifier denitrification-based N2O production in the plastisphere was higher than in bulk water (Supplementary Fig. 9). This further supports that the plastisphere is an overlooked source of N2O production. Although in-depth discussions on the nitrifier-related sources of N2O are beyond the scope of this study, we believe that it is urgent to investigate nitrification and associated N2O dynamics in future research, which will deepen the understanding of the role of aquatic plastisphere in biogeochemical cycling.

Plastisphere derived from the widespread dissemination of plastic debris in estuarine environments8, is changing the stability and function of the ecosystem12,14. Our results reveal that the plastisphere produces approximately twofold greater emissions of the potent greenhouse gas N2O from denitrification pathways than the bulk water. By further assessing Global Warming Potential (GWP) of N2O production67, we also find that plastisphere exhibits the higher warming potential (34.86–41.12 mg CO2 eq L–1 d–1) than the bulk water (13.15–17.26 mg CO2 eq L–1 d–1). Both of these indicate that the emerging plastisphere could be a great but neglected source of anthropogenic N2O. Although our data could not make precise estimations of N2O flux and its warming potential from global plastisphere, the importance of estuarine plastisphere in N2O production via anthropogenic sources should be recognized, especially under global climate change. It is estimated that annual emissions of plastic debris to marine ecosystems are 5–13 million tons8, and the biomass on these plastics could be 0.2–6% of total microbial biomass in marine surface waters10. Given the substantial plastic waste and microbial biomass, future research at larger scales is needed to make reliable predictions of N2O fluxes from plastisphere and to evaluate the warming potential of N2O production from pervasive plastisphere across global marine ecosystems.

Methods

Study site, plastic preparation, and experimental design

In this study, the in situ incubations were conducted in an estuary of Xiamen, China (118°07’E, 24°61’N–118°09E’, 24°59’N) during April and May 2021. This region possesses a subtropical maritime monsoon climate, and has an average temperature of 21 °C and receives 1100 mm of rain68. Owing to human activities, pollutants in rainwater runoff and a portion of wastewater have been recently transported to the estuary, leading to reduced water quality and slight eutrophication. We sampled the estuarine water at 4 m depth and determined its characteristics. The water (pH 6.5, 20 °C) contained 1.19 mg L–1 dissolved oxygen, 21.2–25.7 mg L−1 total organic carbon, 0.72~0.99 mg L−1 nitrate, 0.19–0.43 mg L−1 ammonium, 0.09~0.17 mg L−1 nitrite, and 2.89~7.95 mg L−1 iron.

Four types of common commercial plastics, i.e., polyethylene (PE), polystyrene (PS), polypropylene (PP), and polyvinylchloride (PVC), were selected. They have a low bioreactivity with densities of 0.88–0.97 g cm−31. The PE-type food bags were from Cleanwrap Co., China. PS was obtained from transparent plastic cups (Chuanguan Co., China). PP was from sealable plastic bags from Xingmin Co., China. The PVC-type cling films were donated by Jusu Co., China. The four types of plastics were cut into a size of ~10 × 10 cm for subsequent experiments.

To investigate the denitrifying capability of the plastisphere, we performed (1) an in situ incubation for 30 days (30d), and (2) a subsequent denitrification experiment for 24 h (24 h). The detailed experimental design is illustrated in Supplementary Fig. 1. At the initial stage of in situ incubation, the obtained 10 × 10-cm plastic debris was sterilized using 70% ethanol and connected with cotton cords, and we used stone debris placed under the plastics to maintain buoyancy. This led to a ~4 m water depth for the plastics suspended in estuarine water. After 30 days, the plastic debris and surrounding bulk water (10 L) were harvested with a polymethyl-methacrylate sampler (Haifuda, China). A portion of plastics and bulk water was used for water quality analysis, SEM, and D2O-labeled single-cell Raman spectroscopy. The remaining material was used for the preparation of plastisphere suspensions and the subsequent 24 h experiment. Methods for water quality detections and SEM are listed in Supplementary Methods 1 and 2.

Prior to the denitrification experiment, a portion of bulk water was first filtered through a 0.22-μm polycarbonate membrane to remove microorganisms and then transferred into 250-mL sterile beakers. Thereafter, the collected plastic debris was placed in these beakers, and biofilms were repeatedly shaken, agitated, and washed with the sterile bulk water until a large proportion of the biofilms were detached69. Plastisphere suspensions were then collected for the determinations of denitrification rate, N2O reduction ratio (fR), EPS and c-di-GMP levels, and further experiments (Supplementary Fig. 1). During the 24 h denitrification experiment, we established plastisphere groups (i.e., plastisphere suspensions of PE, PS, PP, and PVC) and a bulk water group (estuarine water). Each plastisphere group was presented in triplicate, and the bulk water group (in quintuplicate) was set as control. Forty milliliters of plastisphere suspensions and bulk water were transferred into 120-mL serum bottles, respectively. Next, these bottles were sealed, purged with high-purity He (>99.99%) for 5 min, and incubated at 20 °C for 24 h. No additional carbon or nitrogen source was added, aiming to explore the basal capacities of denitrification in the plastisphere and bulk water. Each lab-scale experiment was conducted in parallel, and variations in the concentrations of organic matter during the lab-scale experiment are shown in Supplementary Fig. 10. After 24 h, N2O concentration and isotopocules, denitrifier community, and gene abundance in the plastisphere and bulk water were measured.

In this study, we prepared the plastisphere suspensions, aiming to explore the biofilms on the plastics under more controlled conditions during the lab-scale experiments. To examine if the growth pattern (attached or detached biofilms) has significant impacts on microorganisms during the 24 h experiment, we further investigated N2O production, anoxic microbial activities, and community structure of the attached and detached biofilms (Supplementary Fig. 11). These results showed no significant differences between attached and detached biofilms, suggesting that denitrifier activities during 24 h were not likely to be significantly impacted by growth patterns. In addition, we selected the water volume of 40 mL to conduct the 24 h experiment, because it is the minimum volume for a large proportion of biofilm detachment from plastic debris. We also conducted the same experiments with different volumes (80 and 120 mL) to avoid bias induced by incubation water volume (Supplementary Fig. 12). The results showed that N2O production in the plastisphere was higher than that in bulk water, regardless of water volumes.

In addition, it should be noted that marine particle debris is mainly in the form of plastics, occupying up to 60–80%70. Moreover, other particle fragments such as glass, metal, or stone, could be deposited to marine sediments, and the direct effects of these particles on marine waters are minor. Therefore, we only consider the plastic debris in this study.

EPS and c-di-GMP detections

After the 30-d incubation, the EPS and c-di-GMP concentrations in the plastisphere and bulk water were measured. Extractions of EPS were following the formaldehyde-NaOH method71. Plastisphere suspension (3 mL) or bulk water (3 mL) together with 2 mL sterile phosphate buffer were transferred into a 15-mL centrifuge tube. After centrifugation (5000×g, 4 °C, 5 min) and removal of the supernatant, 5 mL of sterile phosphate buffer and 50 μL of 37% formaldehyde were added to re-suspend the pellet. The tube was shaken for 5 min and stored at 4 °C for 60 min, and then 2 mL of 1 M NaOH was added. After being stored at 4 °C for 3 h, the supernatant was harvested by centrifugation (5000×g, 4 °C, 10 min) and used for PS, PN, and eDNA detections after filtering through a 0.22-μm polycarbonate membrane. PS concentration was measured at λ = 625 nm following the Anthrone method using glucose as standard. PN concentration was determined at λ = 562 nm by a Protein Assay Kit (Beyotime Biotechnology, China) using BSA as standard. eDNA concentration was quantified at λ = 260 nm by a DNA Detection Kit (QuantiFluor, dsDNA system, USA).

Microbial c-di-GMP in the plastisphere and bulk water were extracted and measured based on previous studies72,73. The extraction solvent consisted of acetonitrile/methanol (50/50, v/v). Cells in the plastisphere suspension and bulk water were disrupted using ultra-sonication for 5 min (120 W, 4 °C), and then transferred into a 15 mL centrifuge tube. After adding 8 mL extraction solvent and vortexing for 60 s, the suspensions were maintained at −80 °C for 2 h, and the supernatants were collected by centrifugation (14,000×g, 15 min). Thereafter, a pressured gas blowing concentrator (HGC-12A. Hengao, China) was used to dry the supernatants under continuous flows of N2. The residues were resuspended in 500 µL of ultrapure water for subsequent detection of c-di-GMP. The concentration of c-di-GMP was detected with an LC-MS/MS equipped with an electrospray ionization source (Applied Biosystems 6500, USA) at a reversed-phase. The m/z + 689/150 and + 689/343 quantifiers were c-di-GMP transitions. Declustering potential (DP) was −185 and collision energy (CE) was −45 and −48. The obtained peak areas were compared with a calibration curve generated with a c-di-GMP standard (Sigma-Aldrich, Germany). Concentrations of intracellular proteins were measured using a Protein Assay Kit (Beyotime Biotechnology, China).

Denitrification rate and N2O isotope measurement

After the 30-d incubation, denitrification rates of the plastisphere and bulk water were measured using the 15N isotope-pairing technique21,74. Five milliliters of plastisphere suspension or bulk water were prepared and transferred to a 12 mL gas-tight vial (Labco Exetainer, UK). The vials were purged with high-purity He (99.99%) for 5 min to replace the headspace air. The samples were first pre-incubated at 20 °C for 24 h to remove residual nitrate. After the removal, the vials were spiked with K15NO3 solution (99.8% atom, Macklin Co. China) to reach a concentration of 200 μM 15NO3– on the basis of in situ concentrations, and then re-purged with high-purity He. The samples were incubated in dark for 8 h at 20 °C. After the incubation, 2 mL of headspace gas was taken by a syringe and transferred to a new 12 mL vial pre-filled with He gas. The concentrations of 28N2, 29N2, and 30N2 in the headspace were quantified with a GasBench-II/isotope ratio mass spectrometry (Thermal Delta V Advantage, Germany). The detection limit is 0.1 μM. The denitrification rate was calculated as follows:

$${{{{{\rm{Denitrification}}}}}}\,{{{{{\rm{rate}}}}}}\,({{{{{{\rm{mg}}}}}}}^{15}\,N\,{L}^{-1}{h}^{-1})=\frac{{T}_{30}\times 2\times (1-{F}_{n})}{{F}_{n}}+2\times {T}_{30}$$
(1)

where T30 (mg 30N2 L−1 h−1) represents the production rates of 30N2 during the 8 h. Fn (99%) is the 15N fraction in added 15NO3− to residual 15NO3− after pre-incubation.

After the 24-h denitrification experiment, the headspace gas from the 120-mL serum bottle was taken to measure N2O concentrations and isotopic compositions. Three milliliters of headspace gas was used to analyze N2O concentration with a gas chromatograph (7890A, Agilent Technologies, USA), and 1 mL of headspace gas was used to detect N2O isotope compositions, including δ15Nbulk, δ15Nα, δ15Nβ, and δ18O, with a Precon+Gasbench/isotope ratio mass spectrometer (Delta V plus, Thermal, USA)75. Of these,15Nα (central,14N–15N–16O) and 15Nβ (terminal, 15N–14N–16O) are the isotopocules76. N2O gas in samples was first enriched in liquid N2 trap, purified, and then separated by a gas chromatography column (30-m long, 0.32-mm i.d.). Next, high-purity He flowing at 2 mL min−1 was applied to transport the N2O gas to the mass spectrometer. N2O isotopocules were detected by simultaneously capturing NO+ (m/z 30, 31) and N2O+ (m/z 44, 45, 46). The scrambling factor was determined as 0.085. In this study, the isotopocule values of the internal reference gas (high-purity N2O (>99.99%)) were pre-analyzed in Thünen Institute of Climate-Smart Agriculture, Germany. Two standard gases provided by Dr. Anette Goeske and Dr. Reinhard Well were applied to perform two-point calibrations for values of N2O site preference (N2O-SP, AK1 N2O-SP = −2‰, Mix 1 N2O-SP = 15‰, 5ppm). Values of δ15Nbulk, δ15Nα, δ15Nβ, δ18O, and SP of N2O were calculated as follow:77,78

$${{{{{{\rm{N}}}}}}}_{2}{{{{{\rm{O}}}}}}-{{{\updelta }}}{\!\,}^{15}{{{{{{\rm{N}}}}}}}{\!\,}^{{{{{{\rm{i}}}}}}}(\permil )=({\!\,}^{15}{{{{{{\rm{N}}}}}}}_{{{{{{\rm{sample}}}}}}}{\!\,}^{{{{{{\rm{i}}}}}}}{-}{\!\,}^{15}{{{{{{\rm{N}}}}}}}_{{{{{{\rm{standard}}}}}}})/{\!\,}^{15}{{{{{{\rm{N}}}}}}}_{{{{{{\rm{standard}}}}}}}({{{{{\rm{i}}}}}}=\alpha \,{{{{{\rm{or}}}}}}\,\beta )$$
(2)
$${{{{{{\rm{N}}}}}}}_{2}{{{{{\rm{O}}}}}}-{{{\updelta }}}{\!\,}^{18}{{{{{\rm{O}}}}}}(\permil )=({\!\,}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{sample}}}}}}}{{-}}{\!\,}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{standard}}}}}}}){/}{\!\,}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{standard}}}}}}}$$
(3)
$${{{{{{\rm{N}}}}}}}_{2}{{{{{\rm{O}}}}}}-{{{\updelta }}}{\!\,}^{15}{{{{{{\rm{N}}}}}}}{\!\,}^{{{{{{\rm{bulk}}}}}}}(\permil )=({{{\updelta }}}{\!\,}^{15}{{{{{{\rm{N}}}}}}}{\!\,}^{{{\alpha }}}+{{{\updelta }}}{\!\,}^{15}{{{{{{\rm{N}}}}}}}^{{{\upbeta }}})/2$$
(4)
$${{{{{{\rm{N}}}}}}}_{2}{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}({{\permil}})={\updelta }^{15}{{{{{{\rm{N}}}}}}}^{{{\alpha }}}-{{{\updelta }}}^{15}{{{{{{\rm{N}}}}}}}^{{{\upbeta }}}$$
(5)

where δ18O and δ15Ni are the ratios of 18O/16O and 15N/14N, respectively. δ15Nα and δ15Nβ denote the ratios of 14N–15N–16O and 15N–14N–16O, respectively. All values were presented as ‰ relative to atmospheric N2 (15N) or Vienna Standard Mean Ocean Water (18O). The typical detection precisions are 0.3‰ (N2O-δ15Nbulk), 0.9‰ (N2O-δ15Nα), 0.9‰ (N2O-δ15Nβ) and 0.6‰ (N2O-δ18O).

Correction of N2O-o18O and N2O isotope mass balance model

During denitrification, the incorporation of the oxygen atom from H2O into N2O or nitrate could affect the measured δ18O values of N2O (N2O-δ18O)75,79,80. Previous studies conducted in soils suggest that an almost oxygen exchange occurred during denitrification79. Therefore, assuming a complete O exchange δ18O values of the ambient water (H2O-δ18O) could be used for interpretation. As the variation of H2O-δ18O values in global surface seawater is minor81, we used a theoretical H2O-δ18O value (0.91‰, between −2.98 and 1‰)81,82 reported in the literature for seawater to correct the measured N2O-δ18O values (subtracting H2O-δ18O value).

After correction, the multiple-isotope relationships between N2O-SP and N2O-δ18O were plotted based on the mapping approach83, which was first applied by Lewicka-Szczebak et al. Then, we further established a N2O isotope mass balance28 to explore the relative fractions of N2O-producing sources in the plastisphere and bulk water. As the denitrification experiment was anaerobically executed, N2O productions via BD, FD, and CD processes were taken into account. Fractions of BD (fBD), FD (fFD), and CD (fCD) pathways to total N2O flux were expressed as follow:

$${{{{{\rm{Total}}}}}}\,{{{{{{\rm{N}}}}}}}_{2}{{{{{\rm{O}}}}}}\,{{{{{\rm{production}}}}}}={{{{{{\rm{f}}}}}}}_{{{{{{\rm{BD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{FD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{CD}}}}}}}=1$$
(6)

For N2O-SP values, N2O isotope mass contributions of BD, FD, and CD processes to the total N2O-SPNR values production were expressed as:

$${{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{NR}}}}}}}={{{{{{\rm{f}}}}}}}_{{{{{{\rm{BD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{BD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{FD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{FD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{CD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{CD}}}}}}}$$
(7)

where N2O-SPNR refers to the N2O-SP value before reduction. N2O-SPBD, N2O-SPFD, and N2O-SPCD are the N2O-SP values of BD (−7.5–3.7‰)65,76,84, FD (27.2–39.9‰)75,76,85,86,87, and CD (10–22‰)29,32,76,88, respectively. Similarly, the contributions of the three endmembers to N2O-δ18O values of the three processes were expressed as:

$${{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{NR}}}}}}}={{{{{{\rm{f}}}}}}}_{{{{{{\rm{BD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{BD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{FD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{FD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{CD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{CD}}}}}}}$$
(8)

where N2O-δ18ONR refers to the N2O-δ18O value before reduction. N2O-δ18OBD, N2O-δ18OFD, and N2O-δ18OCD are the N2O-δ18O values of BD (16.4–23.3‰)65,76,84, FD (42.0–55.1‰)75,76,85,86,87, and CD (24.9–37.6‰)29,32,76,88, respectively.

Nitrogen and oxygen isotope fractionations during N2O reduction such as BD can increase the measured N2O-SP and N2O-δ18O values of remaining N2O. Fractionation-derived shifts in N2O-SP and N2O-δ18O values during N2O reduction were thus estimated according to a previous study and presented as:

$${{{{{{\rm{N}}}}}}}_{2}{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}={{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{NR}}}}}}}{-{{{{{\rm{f}}}}}}}_{{{{{{\rm{R}}}}}}}\times {{{{{{\rm{SP}}}}}}}_{{{{{{\rm{\varepsilon }}}}}}{{{{{\rm{N2O}}}}}}}$$
(9)
$${{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{\rm{O}}}}}}={{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{NR}}}}}}}{-{{{{{\rm{f}}}}}}}_{{{{{{\rm{R}}}}}}}\times {{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{\varepsilon }}}}}}{{{{{\rm{N2O}}}}}}}$$
(10)

where SP and δ18O are the N2O-SP and N2O-δ18O values after N2O reduction, respectively, which are the measured SP and δ18O values here; fR denotes the N2O reduction ratio (0.75 in this study). This was estimated using the C2H2 approach41, i.e., the proportion of the residual (without C2H2; residual N2O after N2O reduction) to total (with C2H2; residual N2O after N2O reduction) N2O concentrations. SPεN2O (−6‰) and δ18OεN2O (−25‰) were assumed to be the fractionation effect during N2O reduction76,89. Given the N2O reduction effect during experiments, Eqs. (7) and (8) were modified as:

$${{{{{{\rm{N}}}}}}}_{2}{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}={{{{{{\rm{f}}}}}}}_{{{{{{\rm{BD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{BD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{FD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{FD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{CD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{SP}}}}}}}_{{{{{{\rm{CD}}}}}}}{-{{{{{\rm{f}}}}}}}_{{{{{{\rm{R}}}}}}}\times {{{{{{\rm{SP}}}}}}}_{{{{{{\rm{\varepsilon }}}}}}{{{{{\rm{N2O}}}}}}}$$
(11)
$${{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{\rm{O}}}}}}={{{{{{\rm{f}}}}}}}_{{{{{{\rm{BD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{BD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{FD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{FD}}}}}}}+{{{{{{\rm{f}}}}}}}_{{{{{{\rm{CD}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{CD}}}}}}}{-{{{{{\rm{f}}}}}}}_{{{{{{\rm{R}}}}}}}\times {{{{{{\rm{N}}}}}}}_{2}{{{{{{\rm{O}}}}}}-{{{{{\rm{\delta }}}}}}}^{18}{{{{{{\rm{O}}}}}}}_{{{{{{\rm{\varepsilon }}}}}}{{{{{\rm{N2O}}}}}}}$$
(12)

We further applied the Gaussian estimation and Monte Carlo sampling method to obtain the mean values and standard errors of N2O-SP and N2O-δ18O for each endmember, and assess the errors of the mass balance model. The detailed MATLAB codes and Monte Carlo stimulations (10,000 sampling) are provided in Supplementary Table 2 and Supplementary Fig. 13. The values of SP and δ18O-(N2O/H2O) used in the model for BD were −1.5‰ and 19‰, respectively; for FD were 37‰ and 47‰; for CD were 16‰ and 30‰. Combining Eqs. (6), (11), and (12), the relative contributions of BD, FD, and CD to N2O productions are acquired in the plastisphere and bulk water. Proportions estimated by the equations are considered as zero if yielding negative values, and the other two proportions sum up to 100% accordingly.

D2O-labeled single-cell Raman spectroscopy

To compare microbial phenotypic activity and to explore intracellular lipid/fatty acid levels between plastisphere and bulk water, a D2O-labeled single-cell Raman spectroscopy experiment was performed15,90. After the 30-d incubation, the four types of plastics were cut to produce ~1.5 × 1.5 cm plastic fragments. For the plastisphere, 1 mL of sterilized bulk water and the plastic fragments were transferred to a 12-well plate and mixed with 1 mL D2O, finally reaching a concentration of 50% D2O. For bulk water, 1 mL of in situ bulk water and 1 mL D2O were added to another 12-well plate. Each group was conducted in triplicate. The 12-well plates were anaerobically incubated at 300 rpm and 20 °C for 24 h. Next, plastic fragments were transferred to 5 mL phosphate buffer with 0.5% Tween-20, and attached microorganisms were separated by vortexing for 10 min. After washing with sterilized water twice (4000×g, 5 min), 2 μL of microorganisms from bulk water and plastisphere were spotted on a tin foil-coated microslide and dried at 25 °C. Raman spectroscopy was acquired with a LabRAM Aramis (HORIBA Jobin-Yvon) confocal micro-Raman system, including a 300-g/mm diffraction grating, a 532-nm Nd:YAG laser, and a ×100 objective (Olympus, Japan). We used the Raman characteristic band of a silicon wafer (520.6 cm−1) to calibrate the Raman spectrometer. The generated spectra were subsequently analyzed in LabSpec-5 software (HORIBA Jobin-Yvon) with baseline correction and normalization. Peak intensities of Raman bands assigned to C–D (2040–2300 cm−1) and C–H (2800–3100 cm−1) were obtained to estimate (C–D)/(C–D + C–H) ratios42.

Denitrifier community

After the 24 h denitrification experiment, 2 mL of plastisphere suspension was taken for DNA extraction with a FastDNA Kit (MP, SA, USA). In situ bulk water samples were first filtered through sterilized 0.22μm polycarbonate filters that were then used for DNA extraction15. Bacterial nirS gene (bnirS-type denitrifier), bacterial nirK gene (bnirK-type denitrifier), and Internally Transcribed Spacer (ITS) gene (fungal community) were used for amplicon sequencing with the primer sets (Supplementary Table 3). The obtained raw data were quality-filtered, chimera checked and then grouped at 97% similarity. Taxonomy was compared with GeneBank and UNITE databases using QIIME. Information of fungal nirK denitrifiers (fnirK-type denitrifiers) was screened from fungal communities based on previous studies and NR database36,91. The coverage of screened operational taxonomic unit (OTU) to total OTUs of ITS data was >85%, which was used for subsequent analysis. The α-diversities including abundance-based coverage estimation (ACE), Chao1, Shannon, and Simpson indexes, were estimated in R. Principal coordinates analysis coupled with analysis of similarities (ANOSIM) test was applied to explore β-diversities of the three types of denitrifier communities. Co-occurrence network analysis was performed in R92 (nodes R2 > 0.90 and P < 0.05) and visualized by Gephi (Version 0.9.2). Sequences will be submitted under accession number SUB10278422.

We investigated denitrifiers-based niche breadth and overlap according to Levins’ niche breadth method and asymmetric α-model in R14,46,47. These two indicators reflect the potential of denitrifiers adapting to environments and competing for resources in the plastisphere and bulk water. The formulas were as follow:

$${{{{{{\rm{B}}}}}}}_{{{{{{\rm{i}}}}}}}=\frac{1}{\mathop{\sum}\nolimits_{{{{{{\rm{i}}}}}}=1}^{{N}}\times {{{{{{\rm{P}}}}}}}_{{{{{{\rm{ij}}}}}}}^{2}}$$
(13)

and

$${{{{{{\rm{\alpha }}}}}}}_{{{{{{\rm{ij}}}}}}}=\frac{\mathop{\sum}\nolimits_{{{{{{\rm{i}}}}}}=1}^{N}\times {{{{{{\rm{P}}}}}}}_{{{{{{\rm{i}}}}}}{{{{{\rm{\alpha }}}}}}}\times {{{{{{\rm{P}}}}}}}_{{{{{{\rm{j}}}}}}{{{{{\rm{\alpha }}}}}}}}{\mathop{\sum}\nolimits_{{{{{{\rm{i}}}}}}=1}^{{N}}\times {{{{{{\rm{P}}}}}}}_{{{{{{\rm{i}}}}}}{{{{{\rm{\alpha }}}}}}}^{2}}$$
(14)

where Bi and αij are the niche breadth and overlap of denitrifier i in the plastisphere or bulk water, respectively; N denotes the total number of denitrifiers in each type of denitrifier community; Pij denotes the proportion of denitrifier i in communities.

Gene abundance

The abundances of bacteria (16S rRNA) and fungi (ITS) in the plastisphere and bulk water were quantified with qPCR using the universal primers 515F/907R and ITS1F/ITS2R, respectively68,92. The standard curves were obtained by a serial dilution of the plasmids with 16S rRNA and ITS genes. The mixing system (20 μL) included 2 μL of DNA sample, 10 μL of Mix-enzyme (LightCycler SYBR Green I), 0.8 μL of 515 F/907 R or ITS1F/ITS2R primers, and 7.2 μL of H2O. The thermal cycling conditions of 16 S rRNA: 95 °C (3 min), 40 cycles at 95 °C (30 s), 55 °C (30 s) and 72 °C (45 s), and 72 °C (10 min). The thermal cycling conditions of ITS: 94 °C (10 min), 40 cycles at 94 °C (30 s), 56 °C (30 s) and 72 °C (45 s), and 72 °C (7 min). Samples with sterile pure water were regarded as negative control. Amplification efficiencies were 106.8–115.5% with R2 > 0.995. Based on the Ribosomal RNA Operon Copy Number Database, microbial cell numbers could be assessed by the absolute abundances of 16S rRNA/4.193,94.

High-throughput qPCR-based SmartChip technique95 was applied to detect denitrifying genes (narG, nirS, nirK, nosZ) abundances in the plastisphere and bulk water. The mixing system included 3.1 μL of DNA sample, 24.8 μL of Mix-enzyme (LightCycler SYBR Green I), and 3.1 μL of denitrifying gene primers. They were dispensed into a SmartChip via a MultiSample NanoDispenser, and then quantified with a Real-Time PCR system (WaferGen, Biosystems, USA). The bacterial 16 S rRNA (515 F/907 R) gene was set as a reference. The thermal cycle of qPCR conditions included heating at 95 °C (10 min), followed by 40 cycles at 95 °C (30 s), annealing at 58 °C (30 s), and extension at 72 °C (30 s). The threshold cycle of 31 was regarded as the detection limit. Information of primer pairs and other procedures are listed in our previous study95 and Supplementary Table 4. Relative copy numbers of denitrifying gene and16S rRNA gene were estimated as follows:

$${{{{{{\rm{Relative}}}}}}\;{{{{{\rm{gene}}}}}}\;{{{{{\rm{copy}}}}}}\;{{{{{\rm{numbers}}}}}}=10}^{(31-{{{{{\rm{threshold}}}}}}\;{{{{{\rm{cycle}}}}}})}/(3.33)$$
(15)

Normalized relative denitrifying gene abundance was the proportion of the relative functional gene copy number to the 16S rRNA gene copy number.

Statistical analysis

In this study, the significance analyses on raw data were performed in IBM SPSS (Version 22.0). The significant differences in microbial activity, EPS concentration, c-di-GMP, and lipid/fatty acid level, N2O concentration, denitrification rate, the proportion of denitrification pathways to N2O production, functional gene abundance, denitrifiers-based niche breadth and overlap between plastisphere and bulk water were tested with one-way ANOVA method followed by a post-hoc Tukey test. P value < 0.05 was considered as significant.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.