Plastic Litter Emits the Foraging Infochemical Dimethyl Sulfide after Submersion in Freshwater Rivers

Plastic pollution is widespread throughout aquatic environments globally, with many organisms known to interact with and ingest plastic. In marine environments, microbial biofilms that form on plastic surfaces can produce the odorous compound dimethyl sulfide (DMS), which is a known foraging cue. This has been shown to increase the ingestion of plastic by some invertebrates and therefore act as a biological factor which influences the risks of plastic to marine ecosystems. In freshwater however, the production of DMS has been largely overlooked, despite the known sensitivity of some freshwater species to this compound. To address this gap, the present study analyzed the production of DMS by biofilms which formed on low‐density polyethylene and polylactic acid films after 3 and 6 weeks of submersion in either a rural or an urban United Kingdom river. Using gas chromatography–mass spectrometry, the production of DMS by these biofilms was consistently identified. The amount of DMS produced varied significantly across river locations and materials, with surfaces in the urban river generally producing a stronger signal and plastics producing up to seven times more DMS than glass control surfaces. Analysis of biofilm weight and photosynthetic pigment content indicated differences in biofilm composition across conditions and suggested that DMS production was largely driven by nonphotosynthetic taxa. For the first time this work has documented the production of DMS by plastic litter after submersion in freshwater rivers. Further work is now needed to determine if, as seen in marine systems, this production of DMS can encourage the interaction of freshwater organisms with plastic litter and therefore operate as a biological risk factor in the impacts of plastic on freshwater environments. Environ Toxicol Chem 2024;43:1485–1496. © 2024 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.


INTRODUCTION
Plastic pollution is now abundant throughout aquatic environments globally, and there is rising concern for the consequences of plastic exposure on the health of natural ecosystems (MacLeod et al., 2021).Shortly after submersion in natural waters, bacteria, algae, fungi, and small invertebrates accumulate on plastic surfaces, forming a distinct biofilm layer of living organisms within a matrix of extracellular polymeric substances and other particulate matter (Carpenter & Smith, 1972;Zettler et al., 2013).Microbial biofilms release chemical metabolites, such as volatile organic compounds (VOCs) like sulfurous compounds, polyunsaturated aldehydes, and cyclic hydrocarbons, which function as foraging cues for larger organisms in the environment (Fink, 2007).For example, the freshwater gastropod Radix ovata is known to show significant attraction to VOCs released from certain diatom and green algae species and is even able to differentiate between high-and low-quality food using only these odors (Fink et al., 2006a(Fink et al., , 2006b;;Moelzner & Fink, 2014).Odorous compounds are thought to mediate the interactions between plastic litter and larger organisms, with a growing number of studies demonstrating that the presence of a microbial biofilm on plastic significantly increases its ingestion by both marine and freshwater organisms, compared to the same plastic in its virgin state (DeMott, 1986;Polhill et al., 2022;Sucharitakul et al., 2021;Vroom et al., 2017).It has therefore been proposed that biofilm attachment to plastic and VOC production by microbial community members may mask the inedible nature of the plastic.This is thought to create a Trojan horse effect, increasing the probability that organisms will interact with and ingest plastic particles (Botterell et al., 2020;Fabra et al., 2021;Procter et al., 2019) and therefore also exacerbating the wider impacts of plastics within the ecosystem.
The VOC dimethyl sulfide (DMS) is produced by plasticassociated biofilms formed in the marine environment (Savoca et al., 2016).This compound is naturally present in the atmosphere and aquatic systems globally, and plays an important role in climate regulation and the global sulfur cycle (Campen et al., 2022;Lomans et al., 2002;Shaw, 1983).Dimethyl sulfide is also widely recognized as a key infochemical and foraging cue in marine systems (Owen et al., 2021;Savoca & Nevitt, 2014), with responses to aqueous and airborne DMS documented in procellariiform seabirds, seals, penguins, whale sharks, loggerhead turtles, reef fish larvae, phytoplankton, and zooplankton (Amo et al., 2013;Botterell et al., 2020;Dove, 2015;Endres & Lohmann, 2012;Foretich et al., 2017;Kowalewsky et al., 2006;Nevitt et al., 1995;Procter et al., 2019;Savoca et al., 2016;Shemi et al., 2021;Wright et al., 2011).Furthermore, a significantly higher ingestion rate of microplastics infused with DMS, compared to virgin microplastics, has been seen in marine copepods and lobster larvae (Botterell et al., 2020;Procter et al., 2019).While the ecological importance of DMS within freshwater remains largely unexplored, one previous study has documented the sensitivity of three freshwater fish species to DMS and identified its ability to induce feeding behavioral responses in these organisms (Nakajima et al., 1989), indicating the potential role of DMS as a foraging cue in freshwater.
In the marine environment, most DMS is produced through the bacteria-mediated degradation of dimethylsulfoniopropionate (DMSP), which is widely synthesized by marine algae, bacteria, and some higher plants and is thought to be an important osmoprotectant and cryoprotectant (Bentley & Chasteen, 2004;Lomans et al., 2002).By contrast, while the DMSP-to-DMS pathway does occur in freshwater (Ginzburg et al., 1998), it is generally thought to be less dominant.Instead, most DMS in freshwater is thought to be produced from a mixture of other microbially mediated mechanisms such as the degradation of methoxylated aromatic compounds and sulfur-containing amino acids, as well as the methylation of methanethiol (Carrión et al., 2015;Lomans et al., 2002).Although not directly measured, the genetic potential of various common bacteria to produce DMS from the methylation of methanethiol in freshwater and terrestrial environments has been identified (Carrión et al., 2015), and the production of DMS by several freshwater photosynthetic species, including cyanobacteria and green algae, has been directly observed (Bechard & Rayburn, 1979;Steinke et al., 2018).
Previous studies have demonstrated the unique composition and metabolic functionality of biofilms which form on plastic compared to other surrounding natural surfaces and indicate that it could lead to impacts on ecosystem-scale processes such as biogeochemical cycling (Chen et al., 2020;Hu et al., 2021;Miao et al., 2021;Mincer et al., 2016;Su et al., 2022;Xue et al., 2020).However, very little is known about how the composition of plastic biofilms may drive metabolic activities which can influence interactions between plastic litter and larger organisms within freshwater environments.Despite its importance in marine systems, there is far less research surrounding DMS in freshwater, and only one study to date has examined DMS production by plastic in a freshwater system (Zink & Pyle, 2019).Although no evidence of production was found in that study, the detection limit (32 nmol/L) was relatively high, and the storage of samples before analysis is likely to have led to rapid DMS depletion (Li et al., 2020).Studies which use more specialized and sensitive methods are therefore required.
The aim of the present study was to use sensitive methodology, highly adapted to sulfuric-VOC detection (limit of detection [LOD] 0.015 nmol/L), to establish the ability of plastic materials within United Kingdom rivers to produce DMS.Comparisons with surrounding nonplastic surfaces were made to ascertain the potential relative risk of DMS production by plastic within the river.Furthermore, given that surroundinglocation conditions are known to drive microbial biofilm composition (see Nguyen et al., 2022;Yang et al., 2021), materials were also analyzed to determine the mass and photosynthetic pigment content of biofilms to identify broad differences in biofilm characteristics driven by location.

Materials
Low-density polyethylene (LDPE) and polylactic acid (PLA) films were used in the present study, derived from commercially available plastic bags.Both films were clear, with a thickness of 50.0 and 40.6 μm, respectively.Their identity was confirmed using Fourier-transform infrared spectroscopy (Supporting Information S1, Figure S1).Glass used in the study was derived from 1-mm-thick noncoated Academy-branded microscope slides.Plastic was laser-cut into 7 × 82-mm coupons, and glass slides were cut into 26 × 38-mm coupons before materials were secured inside 240 × 240 × 130-mm custom-built stainless steel woven mesh (3.5 mm aperture) cages for deployment in the field.

Sample incubation
Cages containing plastic and glass samples were deployed at two river locations, one representing rural riverine conditions and one representing urban riverine conditions.Cages were placed just under the surface of the water with coupons orientated vertically within the water column, allowing both sides of the coupon to be colonized.Materials were then sampled after either 3 or 6 weeks of submersion, with the deployment and sampling regime staggered by 1 week between locations.Rural river samples were deployed on August 10, 2021, in the River Ouse upstream of York city center, United Kingdom (54°00′30.7″N1°11′28.7″W).One week later, urban river samples were deployed into the River Foss within York city center, United Kingdom (53°58′35.4″N1°04′26.6″W).River locations were categorized using the 2011 England Rural-Urban Classification system (Government Statistical Service, 2011; Office for National Statistics, 2020), with the River Ouse location classified as a "rural hamlet and isolated dwellings" area and the River Foss location classified as an "urban city and town" area.These locations are shown in Figure 1.Deployment in the urban river was logistically restricted to a more shaded area; light flux above cages was 95 µmol/s/m 2 in the rural location and 22 µmol/s/m 2 for urban samples.The water depth of cage deployment was approximately 3 m at the rural location and <1 m at the urban location.
After 3 or 6 weeks, materials were removed, transported back to the laboratory, and rinsed thoroughly with Milli-Q water to remove loose material before the next stage of analysis.At each sampling point the pH, dissolved oxygen, conductivity, and temperature of the river water were measured.The nitrate and phosphate levels of rivers were also determined by analyzing surface water samples (filtered to 0.2 µm with Sartorius surfactant-free cellulose acetate filters), with API colorimetric testing kits.At the 6-week sampling point for each location, surface water samples were also collected in gas-tight glass vials for measurement of background DMS levels in the river; because of logistical constraints, these measurements could only be made for this sampling point.

Biofilm composition and characteristics
Separate coupons from each sampling point were used to quantify the amount of biofilm present on each material along with the biofilm chlorophyll-a and pheophytin content, with six replicates per treatment for each endpoint.The weight of biofilm was measured by drying coupons and attached biofilm for 3 h at 30 °C before weighing with an analytical balance.The biofilm was then completely removed using a soft natural bristle brush and Milli-Q water before drying and reweighing the clean coupons to determine mass loss.Biofilm removal and drying times were previously validated with preliminary work (Supporting Information S1, Figure S2).
To measure the chlorophyll and pheophytin content of biofilms, coupons were placed in 90% high-performance liquid chromatography (HPLC)-grade acetone (Fisher Scientific).Low-density polyethylene and PLA were extracted using 23.5 mL of solvent, while glass was extracted in 35 mL, and pigment extraction was carried out by incubating samples at 4 °C for 24 h.The chlorophyll and pheophytin contents of acetone were then measured with a Turner Designs Trilogy fluorometer using the acidification method procedure outlined in the user manual (Turner Designs, 2019) to obtain raw fluorescence unit (RFU) values.The RFU was related to pigment concentration by creating an external calibration curve.To do this, approximately 1 mg of chlorophyll-a powder was dissolved in 90% acetone.The resulting solution concentration was determined by measuring the absorbance at 664 nm using a Shimadzu UV-1800 spectrophotometer and the equation given by Jeffrey and Humphrey (1975).A series of solutions were then made with 90% acetone and measured in the fluorometer to create a calibration curve which was used along with equations provided by Turner Designs (2019) to calculate the chlorophyll-a and pheophytin contents of biofilm samples (Supporting Information S1, Figure S3).

Measurement of DMS production
Plastic and glass coupons were placed in 35-mL clear glass vials filled with filter-sterilized (0.22 µm) artificial river water (ARW; Naylor et al., 1989).Vials were capped with US Environmental Protection Agency gas-tight lids and incubated for 44 h at 15 °C under a light intensity of 120 µmol/s/m 2 and a 14: 10-h light: dark cycle.For each sampling point there were four replicates per material, with each LDPE and PLA replicate vial containing five 64 × 7-mm coupons (4480 mm 2 colonized material) and each glass vial replicate containing three 35 × 23-mm coupons (4830 mm 2 ) colonized material.Concentrations of DMS were subsequently normalized to 1 cm 2 of material surface area.For each sampling point, blank control samples consisting of only ARW were also analyzed.Three blanks were analyzed prior to the 44-h incubation, and three were analyzed after-alongside the treatment samples.To verify that any DMS detected was being produced by the biofilm, rather than the material it was attached to, a separate virgin plastic control analysis was also performed.For this, material from the same source and batch of plastic as used for the treatment samples was analyzed using the same methodology; however, these samples were clean and had never been previously exposed to a riverine environment (n = 3 for each material).
After 44 h of incubation, the DMS concentration of ARW samples was measured using a custom-built purge-and-trap system coupled with an Agilent 7890B gas chromatograph (GC) and an Agilent 5977A mass spectrometer (MS).Artificial river water samples were inverted several times before 20 mL of liquid was drawn into a gas-tight glass syringe and spiked with 20 µL of internal standard (described below).The sample was then injected through a sterile 0.45-µm Sartorius surfactant-free cellulose acetate filter into a glass purging tube, and nitrogen gas was bubbled through the sample at a rate of 20 mL/min, for 10 min.The gas stream was dried using two in-line Nafion driers filled with 4A molecular sieves, and gases were collected in a polytetrafluoroethylene (PTFE) loop trap cooled to -150 °C with liquid nitrogen and an in-house-built liquid nitrogen boiler.After 10 min, the trap was rapidly heated to 100 °C by submerging the loop in boiling water for 1 min to inject gases into the GC-MS system.All tubing for the purge-and-trap system was PTFE, and sample-contact fittings were polyether ether ketone.Gases were separated on an Agilent DB-VRX column (length 60 m, inner diameter 320 µm, film thickness 1.8 µm).Injection of samples was splitless with injector temperature set at 250 °C and 11.8 psi.Helium was used as a carrier gas with a constant flow of 2 mL/min.The GC oven temperature regime was initial temperature of 32 °C held for 5 min, temperature ramp to 110 °C at 20 °C/min, temperature ramp to 160 °C at 40 °C/min held for 1 min.The transfer line from GC to MS was held at 280 °C, and the MS source temperature was 230 °C with an electron ionization of 70 eV.A solvent delay of 2 min was in place.Between samples the glass purging tube was thoroughly rinsed with Milli-Q water five times, and frequent checks confirmed that no cross-contamination of samples occurred.
Instrument response was calibrated to analyte concentration using a DMS standard from Sigma-Aldrich, which was diluted to a tertiary working standard solution using HPLC-grade methanol.Deuterated DMS (DMS-d6) from Sigma-Aldrich diluted in methanol was used as an internal standard and spiked into all samples and calibration standards at a concentration of 655 pmol/L.Single ion monitoring mode was used to look for four ions belonging to DMS and DMS-d6, which were determined from the National Institute of Standards and Technology's MS spectra database.For DMS, ions with a mass-tocharge ratio (m/z) of 62 and 47 were measured; and for DMS-d6, ions with an m/z of 68 and 50 were measured.Analyte response was normalized to internal standard response to account for small-scale variations in analyte recovery.Calibration curves were created (Supporting Information S1, Figure S4), and the equation of the regression line was used to relate sample response to DMS concentration.The LOD for these methods was 14.86 pmol/L, the limit of quantification (LOQ) was 45.03 pmol/L, and percentage recovery was 101.4%These values were calculated from the analysis of 10 replicate ARW samples spiked with DMS standard to a concentration of 65.7 pmol/L and internal standard.The LOD was defined as 3.3 × (σ + b)/S, and the LOQ was defined as 10 × (σ + b)/S, where σ is the standard deviation of the calculated concentration of the 10 samples, S is the slope of the calibration curve (Shrivastava & Gupta, 2011), and b is the average peak area of blank samples.

Statistical analysis
The successional changes of microbial biofilms are well documented (Eich et al., 2015;Pinto et al., 2019); therefore, direct statistical comparisons of biofilm changes over time were not of interest in the present study.Instead, differences in biofilms between material types and locations were the focus, with endpoints (biofilm weight, pigment content, and DMS production) analyzed using a two-way analysis of variance (ANOVA) or a nonparametric aligned ranks ANOVA, with material type and river location as factors.Statistical analysis and figure construction were carried out in RStudio (Ver.1.2.1335) using packages ggplot2, car, and ARTool.All data were tested for normality and equal variance using the Shapiro-Wilk and Levene tests.Data which did not meet these assumptions were transformed, or an aligned ranks ANOVA was carried out when assumptions were still not met.Details of each test and transformation performed are given in Supporting Information S1, Table S1.The significance level for the present study was set at 0.05.

Surrounding water conditions
Conductivity, temperature, and pH levels were similar between the two river locations, while nitrate and phosphate concentrations were considerably higher at the urban location and dissolved oxygen levels were lower (Table 1).This indicates that the shallower urban river was a more eutrophic environment than the deeper rural river.The average DMS concentration of river water from the rural location at the 6-week sampling point was 577 ± 5.79 pmol/L, which was significantly lower than that of the river water from the urban location (t (4) = −34.64,p < 0.001), where it was 2680 ± 105.02 pmol/L (mean ± SD), n = 3 for rural and urban locations.

Biofilm composition and characteristics
All samples at both river locations had a detectable biofilm on their surface after both 3 and 6 weeks of submersion.The amount of biofilm attached to materials varied considerably between locations, ranging between 0.01 to 1.20 and 0.17 to 1.30 mg/cm 2 after 3 and 6 weeks of submersion, respectively (Figure 2A).Examples of these samples and the attached biofilm can be seen in Supporting Information S1, Figure S5.After 3 weeks, a significant interaction effect between material and location was present (F (2,29) = 8.71, p = 0.001), with glass possessing the lowest biofilm weight in rural samples (0.70 mg/cm 2 ) but the highest biofilm weight among urban samples (0.01 mg/cm 2 ).A significant and dramatic difference in biofilm weight was present between materials from different locations, with biofilm weight on rural materials approximately two orders of magnitude higher than that on urban materials after 3 weeks of submersion.Interestingly, after 6 weeks, no significant effect of material or interaction was present (F (2,30) < 1.8, p > 0.05); but, similarly to Week 3, there was a strong and significant effect of location (F (1,30) = 58.35,p < 0.001), with biofilm weight on rural material approximately five times higher than that on urban material.
To better understand biofilm composition, the total photosynthetic pigment content of biofilms (chlorophyll-a + pheophytin) was normalized to the average weight of attached biofilms for each treatment condition (micrograms per milligram of biofilm).Similar to biofilm weight, pigment content varied across both material type and location (Figure 2B).A significant interaction effect between material and location was seen after 3 (F (2,30) = 13.58,p < 0.001) and 6 (F (2,10) = 4.28, p = 0.023) weeks.After 3 weeks, the pigment content of biofilm on glass was significantly lower than that on LDPE and PLA at both locations (p < 0.01 for all); however, the scale of difference varied, with plastic-associated biofilm possessing approximately four and two times more pigment than glass at rural and urban locations, respectively.The pigment content of all urban materials after 3 weeks was approximately an order of magnitude higher than for rural materials, with significant differences across locations present for all materials (p < 0.001).After 6 weeks, while the pigment content between rural materials (1.9-2.2 µg/mg) did not differ significantly (p > 0.05 for all), for urban materials, glass-associated biofilm had significantly less pigment than for LDPE (p = 0.002) but not PLA (p = 0.09).In contrast to Week 3, the pigment content of all urban materials (0.18-0.72 µg/mg) was significantly lower than that of rural materials (p < 0.05 for all).
Pheophytin is a breakdown product of chlorophyll and can therefore indicate the health of a photosynthetic community.Therefore, the proportion of total pigment that was pheophytin was also investigated (Figure 2B; Supporting Information S1, Figure S6).A significant interaction effect between material and location was found after 3 (F (2,30) = 118.40,p < 0.001) and 6 (F (2,30) = 4.36, p = 0.02) weeks of incubation.After 3 weeks, biofilm on glass at the rural location had a significantly higher proportion of pheophytin in the pigment compared to rural LDPE (p < 0.001), and urban materials all differed significantly from each other (p < 0.05 for all), with PLA-associated biofilms possessing a considerably lower pheophytin proportion in the pigments.Between locations, LDPE and PLA differed significantly (p < 0.001 for both), whereas glass did not.After 6 weeks, biofilms from rural material had an average pigment pheophytin content of 19% to 26%, with no significant variation between materials (p > 0.05 for all).Among urban materials LDPE had a significantly lower pigment pheophytin content than PLA and glass (p < 0.05 for both); however, the standard deviation for LDPE and PLA was remarkably high.Between locations pheophytin proportion did not differ significantly for any materials after 6 weeks.

DMS production
All materials at both locations consistently produced a detectable DMS signature after 3 and 6 weeks of submersion, with an average DMS production of 0.14 to 1.46 pmol/cm 2 of surface area (Figure 3).No DMS production from virgin materials or blank ARW samples was detected, confirming that DMS was produced from the biofilms associated with samples.After 3 weeks of incubation, a significant interaction effect  between material type and location and DMS production was found (F (2,15) = 5.47, p = 0.017), with LDPE producing significantly more DMS than glass at the rural location (p = 0.005) but no significant differences between materials at the urban location (p > 0.05 for all).Production of DMS by glass at the rural site (0.39 pmol/cm 2 ) did not significantly differ from that at the urban site after 3 weeks (0.46 pmol/cm 2 ; p = 0.059), whereas significant differences were present for LDPE and PLA  (p < 0.05 for both), with DMS production being higher for rural materials.After 6 weeks of submersion, similar patterns in DMS production were seen at both locations, with a significant effect of material (F (2,17) = 39.93,p = 3.77 E−7 ) and location (F (1,17) = 14.046, p = 0.002) but no interaction effect.Low-density polyethylene and PLA produced significantly more DMS than glass at both locations (p < 0.01 for all), and rural materials produced between 1.2 and 2.9 times more DMS than the same material from the urban location.
To further investigate differences in biofilm composition, DMS production was normalized to the average biofilm weight for each treatment and expressed in picomoles per milligram of biofilm (Figure 4).This analysis also identified considerable differences across material type and locations.After 3 weeks, there was an overall significant interaction between material and location (F (2,15) = 10.67,p = 0.001).While rural materials produced an average of 0.83 to 1.58 pmol/mg and did not differ significantly from each other (p > 0.05 for all), among urban materials, the normalized DMS production of PLA (35.38 pmol/mg) was significantly higher than that of LDPE (17.44 pmol/mg) and glass (12.94 pmol/mg; p < 0.001 for all).Notably, DMS production of all urban materials was approximately an order of magnitude higher than in samples from the rural river, with significant differences across locations for all materials (p < 0.001 for all).A significant interaction between material and location was also present after 6 weeks (F (2,17) = 9.68, p = 0.002).Normalized DMS production was significantly lower for glass compared to LDPE among rural samples (p = 0.008) and significantly lower than LDPE and PLA among urban samples (p < 0.001 for both).Furthermore, differences between plastics and glass were considerably larger for urban material, where plastics produced approximately seven times more DMS per milligram than glass.Similar to Week 3, the normalized DMS production of LDPE and PLA from the urban location was significantly higher compared to materials at the rural location (p < 0.001 for both).However, no significant differences were present between locations for glass (p = 0.70).

DISCUSSION
Dimethyl sulfide is an important foraging cue in the marine environment (Savoca & Nevitt, 2014); it has been shown to be produced by microbially colonized plastic from marine systems (Savoca et al., 2016) and is known to increase the ingestion of microplastics in some species (Botterell et al., 2020;Procter et al., 2019).However, despite previous studies also demonstrating the sensitivity of some freshwater fish species to the compound (Nakajima et al., 1989), little is known about the DMS production arising from plastic pollution in freshwaters.The only previous study to explore this (Zink & Pyle, 2019) did not detect the presence of DMS on plastics from freshwaterpossibly because of the low sensitivity of the analytical method used.In contrast, using specialized and sensitive methodology, the present study demonstrates for the first time that plastic pollution can consistently acquire a distinct DMS signature after submersion in two different types of United Kingdom rivers.

Material and location differences in DMS production and biofilm composition
In the present study, all plastic and glass materials acquired an evident biofilm layer over their surface after 3 and 6 weeks of submersion in rivers.Community members of this biofilm were consistently found to produce the odorous compound DMS, leading to acquisition of a clear DMS signature by materials.For all but one treatment (urban, 3 weeks) LDPE and PLA films produced significantly more DMS (per square centimeter of material) than glass under the same conditions.The DMS signature also differed between the same material at different river locations, indicating that the development of a DMS signature by materials in freshwater is influenced by multiple factors.After 3 weeks in the rural river, variation in the DMS signature between materials was accompanied by similar differences in biofilm weight, while normalized DMS production (picomoles per milligram of biofilm) did not differ.This suggests that under these conditions most variation in the DMS signature was driven simply by the amount of biofilm attached to samples, which is in line with previous studies which have noted that differences in substrate properties can mediate the amount of biofilm that accumulates (Li et al., 2019;Miao et al., 2020).For all other treatments, normalized DMS production showed significant variation between materials in the same treatment.Of particular note were the differences seen between materials submerged in the urban river for 6 weeks, for which the weight of biofilms differed very little between materials, while the DMS signature and normalized DMS production were significantly and substantially lower for glass than for LDPE and PLA.Furthermore, excluding glass after 6 weeks, there were significant differences between the normalized DMS production of materials at rural and urban locations.
These findings suggest that the taxonomic composition and metabolic functionality of biofilms differed between material type and location, and there is also evidence that river location influenced the relative differences between materials under each treatment.These results are in keeping with previous work which has found plastic biofilm composition to be shaped by surrounding water conditions (Nguyen et al., 2022;Yang et al., 2021), as well as studies which have found distinct differences in community functioning between plastic and natural surfaces within freshwater systems (Hu et al., 2021).The ability of the surrounding water conditions to influence the ability of plastic-specific biofilms to form has also been previously noted (Oberbeckmann et al., 2018).Interestingly, despite normalized DMS production being higher for urban materials compared to rural ones, their DMS production per square centimeter of material was lower.Given that rural materials also had a higher biofilm weight, the degree of DMS signature acquired by materials in the present study was likely to have been driven by both the amount and the composition of the associated biofilms.
The significant variability observed in the photosynthetic pigment content of biofilms further supports the notion of differences in community composition between materials and locations.With the exception of materials after 6 weeks in the rural river, the total pigment content of biofilms on LDPE and PLA was consistently higher than on glass and may indicate different levels of heterotrophy (Hoellein et al., 2014).The pigment content of biofilms also differed between the same materials at different locations for all treatments, and, while significant differences were present between materials in the urban river after 6 weeks, they were not for materials at the rural location.While the existence of a true plastic-specific microbiome remains controversial, as seen in previous studies, interactions between substrate properties and surrounding environmental conditions, such as nutrient concentration and temperature, are likely to determine biofilm composition and substrate specificity (Oberbeckmann et al., 2018).In the present study, conditions of the urban river were more eutrophic than in the rural location, with higher nutrients levels and lower dissolved oxygen; and these, therefore, are likely to have contributed to the differences in biofilm composition and metabolic functioning observed.Interestingly though, normalized DMS production and pigment content (indicative of biofilm composition) differed most between materials in the urban river, which is in contrast to findings from Oberbeckmann et al. (2018), who report the largest differences between materials under conditions with lower nutrient levels.However, many other factors are likely to shape these outcomes, and this further highlights the complexity of plasticassociated biofilms in freshwater systems.
As well as influencing biofilm composition, the trophic status of the river may have driven the amount of biofilm which attached to materials.The lower biofilm weight seen on urban samples is in keeping with previous studies which found a significantly higher biofilm mass on microplastics in oligotrophic freshwater lakes compared to eutrophic ones (Arias-Andres et al., 2018).Furthermore, this appears to be a known phenomenon, whereby lower nutrient concentrations trigger substrate attachment and biofilm formation in microorganisms, and higher ambient nutrient levels reduce this drive, with more cells remaining planktonic (Du et al., 2022;Stanley & Lazazzera, 2004).It should, however, be noted that while the trophic conditions of the rivers are likely to have driven many biofilm differences, some variation could have been due to the light levels under which the samples were incubated, which were higher at the rural location.Nevertheless, given that the photosynthetic pigment content of urban biofilms was much higher than that of rural ones after 6 weeks, the role of light may have been negligible.
As well as surrounding water conditions, the biofilm differences observed may have been driven by river depth and the proximity of samples to the river sediment.Although all samples were submerged just below the water surface, samples in the rural river were sitting above approximately 3 m of water, whereas samples in the urban river were in <1 m of water.High levels of DMS production are known to take place in freshwater sediments, with the DMS concentration of overlying water decreasing with increasing distance from the sediment (Lomans et al., 1997).A recent study found that the core bacterial microbiome of microplastics was most strongly derived from the core microbiome of nearby sediment (Zhang et al., 2023).Therefore, there may have been some connectivity between DMS-producing species in the sediment and those which colonized experimental materials, contributing to the differences in biofilm composition and DMS production between locations.This may also explain the large difference in the DMS concentration observed in water samples from the two locations.
Overall, the normalized DMS production and pigment content of biofilms observed in the present study demonstrates the importance of surrounding-water and location conditions in shaping the plastic biofilm composition and functionality, as well as the occurrence of distinct plastic-specific communities.

Drivers of DMS production within the biofilm
Although several common freshwater photosynthetic species are known to produce DMS (Bechard & Rayburn, 1979;Steinke et al., 2018), the normalized DMS production of biofilms in the present study was not consistently and convincingly explained by either their total pigment content or their pheophytin proportion.For example, after 3 weeks, normalized DMS production did not differ significantly between rural materials, whereas total pigment content and pheophytin percentage did.Conversely, after 6 weeks in the rural river, normalized DMS differed significantly between materials, but total pigment content and pheophytin percentage did not.Similarly, there were also no discernible links between normalized DMS production and pigment for materials submerged in the urban river for 3 weeks.On the other hand, after 6 weeks in the urban river, differences in normalized DMS values between materials were mirrored by the total pigment content; and, although highly variable, LDPE samples possessed both a significantly higher normalized DMS and a significantly lower pheophytin percentage.Comparisons between locations also show a similarly unclear pattern: After 3 weeks, urban materials had a higher normalized DMS production and total pigment content than rural ones; however, after 6 weeks, urban materials unexpectedly showed a higher normalized DMS production but a lower total pigment content.Therefore, despite previous studies finding a correlation between DMS production and chlorophyll-a concentration in freshwater (Steinke et al., 2018), the present data suggest that a mixture of DMSproduction pathways was likely to have occurred in the present study and that, while photosynthetic organisms may have played a role, they were unlikely to have been the only driver.Instead, bacterial communities, such as Novosphingobium, Pseudanabaena, and Pirellula, which have the genetic potential to produce DMS and have been observed within the biofilm community on plastic surfaces in other studies (Carrión et al., 2015;González-Pleiter et al., 2021;Huang et al., 2022) may have produced the majority of DMS observed.

Wider implications of DMS production by plastic pollution in freshwater
Many freshwater species are known to interact with plastic pollution, yet the reasons for these interactions often remain unknown.For example, the caddisfly larvae Agrypnia sp. has recently been shown to use microbially colonized plastic film to build their protective cases, despite plentiful natural material being available; during this process they also actively fragmented the plastic into hundreds of submillimeter-sized fragments (Valentine et al., 2022).Similarly, a specific preference for anthropogenic litter compared to natural substates by freshwater invertebrates has been demonstrated by Wilson et al. (2021), who found that Limnophora spp.and Bathyomphalus contortus were exclusively associated with flexible fabric and plastic in United Kingdom urban rivers.However, the reasons for these preferences could not be discerned in the study.Other recent work has demonstrated the increased ingestion of microbially colonized plastic by Daphnia magna compared to virgin plastic, with the specific factors which drove this response remaining unknown (Polhill et al., 2022).Understanding the processes which drive the attraction to, and ingestion of, plastic is an important step in making accurate risk assessments of plastic pollution in the environment and is likely to increase the environmental realism of laboratory-based ecotoxicological studies.It may also help to identify certain types of plastic or geographical locations that could be particularly high-risk-based on the composition of biofilms which form.
Dimethyl sulfide is a known key foraging cue in marine systems and can increase the ingestion of microplastics by certain marine species (Botterell et al., 2020;Procter et al., 2019;Savoca & Nevitt, 2014).Furthermore, although the sensitivity of freshwater species to DMS is largely unexplored, DMS has been shown to increase feeding behavior responses in three species of freshwater fish and to stimulate the olfactory nerve response of one species (Nakajima et al., 1989).However, a DMS concentration of 1 × 10 3 pmol/L was used to test this olfactory response, and a concentration of 1 × 10 9 pmol/L was used to test feeding responses; therefore, the sensitivity of these fish species to DMS concentrations produced by plastic in the present study remains unclear.Nevertheless, given its importance in marine systems and the evidence of detection by some freshwater species, the production of DMS by plastic-associated biofilms has the potential to influence interactions between plastic litter and larger organisms within United Kingdom rivers.Furthermore, after only 6 weeks of submersion, LDPE and PLA films produced significantly more DMS per square centimeter of material compared to glass, and plastic litter may therefore host biofilms which particularly enhance interactions between organisms and plastics, compared to biofilms on other surrounding surfaces.
Although the DMS concentration produced by plastics in the present study (0.24-1.16 nmol/g plastic) was considerably lower than that previously found for marine plastics (9.7 and 450 nmol/g plastic; Savoca et al., 2016), the methodology used in the present study quantified DMS production from living biofilms and is therefore likely to closely reflect the flux of DMS occurring from plastics within the environment.In comparison, the study of marine plastic involved freezing samples before analysis, which can increase cell lysis and may artificially enhance DMS flux.Furthermore, the lower levels of DMS found within freshwater generally may mean that freshwater organisms possess a higher sensitivity to the compound and are able to detect it at lower concentrations.If freshwater taxa do detect and respond to DMS production from plastic litter, the overall effects of this are also likely to be dependent on the DMS concentration of the surrounding water.For example, in areas with higher DMS levels, such as the urban river in the present study, the DMS signal from plastic may not contrast strongly against the background signal, whereas in water with lower DMS levels, plastic may act as a hotspot for DMS production, enhancing the attraction of organisms.There is also likely to be a temporal element to this, which should further be considered.In addition, DMS is only one of a multitude of different VOCs that are produced by freshwater microorganisms.For example, geosmin and 2-methylisoborneol are produced widely by freshwater algae and bacteria, sometimes in concentrations high enough to interfere with the odor and taste of drinking water (Lee et al., 2017).Work to understand the ability of freshwater organisms to detect and respond to environmentally relevant levels of DMS, and its relative importance compared to other VOCs, is therefore now needed to elucidate the role of DMS in mediating interactions between plastic and freshwater species.

CONCLUSIONS
The findings of the present study demonstrate, for the first time, that the formation of microbial biofilms on plastic surfaces can result in the acquisition of a distinct DMS signature by plastic pollution in United Kingdom rivers.Given the role of DMS as a foraging infochemical in marine environments and the sensitivity of some freshwater species to DMS, this could therefore have implications for the interactions of freshwater organisms with plastic pollution and has the potential to operate as a biological risk factor which influences the impacts of plastic on freshwater environments.
Materials in the present study consistently acquired a DMS signature in an urban and a rural river environment after 3 and 6 weeks of submersion, indicating that this could be occurring across freshwater environments widely.Furthermore, material type and river conditions appeared to interact to influence the composition and DMS production capacity of these plasticassociated biofilms.The DMS signal acquired by plastic was found to be significantly stronger than that of other surrounding nonplastic surfaces under most conditions and may therefore result in an increased attraction of aquatic organisms to plastic, further enhancing its impacts within the environment.The trophic status and water depth of river locations were also thought to mediate the strength of the DMS signature acquired by materials by influencing the composition and amount of biofilm which formed.Materials in the deeper and less eutrophic rural river showed a lower DMS production per unit weight of biofilm but a larger biofilm weight and stronger overall DMS signature per square centimeter of material.Location conditions also influenced the relative differences observed in DMS production between materials in each river.
While the production of DMS by biofilms in the present study may have been partially driven by phototrophic organisms, such as cyanobacteria and green algae, the data suggest that heterotrophic organisms, such as bacteria, played a more major role.Further studies which combine the measurement of DMS production with taxonomic characterization and identification of DMS-producing genes would be useful to determine the main producers of DMS in plastic-associated biofilms.Overall, it is clear that DMS signature acquisition by plastic in freshwater is highly dynamic and that the strength of the attained signal is context-dependent.More plastic types, water conditions, and temporal factors should therefore be investigated to identify further influences on biofilm DMS production and to determine which combinations of material and water conditions result in the strongest signal.The sensitivity of freshwater organisms to DMS also remains largely unknown, and further work to determine this for environmentally relevant DMS concentrations is therefore required to understand the potential ecological impacts of this DMS production by microbially colonized plastic.Finally, to enhance our knowledge of the wider relative risks of plastic pollution within freshwater ecosystems, the overall role of the biofilm in the attraction of freshwater taxa to plastic and the relative importance of DMS compared to other VOCs also needs to be determined.
Supporting Information-The Supporting Information is available on the Wiley Online Library at https://doi.org/10.1002/etc.5880.

FIGURE 1 :
FIGURE 1: Locations of the rural and urban river sites where plastic and glass samples were submerged.This map was produced with resources from ©OpenStreetMaps under the Open Data Commons Open Database License.

FIGURE 2 :
FIGURE 2: (A) Weight of biofilm attached to low-density polyethylene, polylactic acid, or glass.(B) Photosynthetic pigment content (chlorophyll-a and pheophytin) normalized to biofilm weight.For all plots, note the scale of axes on Week 3 urban plots (highlighted with red axes).Letters above bars indicate significant differences between materials within each individual subplot only; asterisk above bars in rural plots indicates that the amount of dimethyl sulfide produced by these materials differed significantly from the amount produced by the same material in the urban location after the same incubation time.Letters and asterisks in (B) indicate statistical analysis results for the total pigment content (chlorophyll-a + pheophytin).LDPE = low-density polyethylene; PLA = polylactic acid.

FIGURE 3 :
FIGURE 3: Dimethyl sulfide (DMS) produced per square centimeter of low-density polyethylene, polylactic acid, and glass material, after they were incubated in a rural or urban river for either 3 or 6 weeks.Week 3 and Week 6 data were analyzed separately.Letters above bars indicate significant differences between materials within each individual subplot only; asterisk above bars in rural plots indicates that the amount of DMS produced by these materials differed significantly from the amount produced by the same material in the urban location after the same incubation time.LDPE = low-density polyethylene; PLA = polylactic acid.

FIGURE 4 :
FIGURE 4: Dimethyl sulfide (DMS) production normalized to average biofilm weight for each treatment condition.Note the scale of axes on Week 3 urban plots (highlighted with red axes).Letters above bars indicate significant differences between materials within each individual subplot only; asterisk above bars in rural plots indicates that the amount of DMS produced by these materials differed significantly from the amount produced by the same material in the urban location after the same incubation time.LDPE = low-density polyethylene; PLA = polylactic acid.

TABLE 1 :
Environmental parameters of river water measured at each of the sampling time points and locations