Microbial diversity within riverine plastispheres.
Pristine and weathered low-density PE films (LDPE and W-LDPE), together with wooden strips as a control surface, were incubated in-situ in the proximity of a wastewater treatment plant (WWTP) in the River Sowe for one week (Coventry, West Midlands, UK; Fig. S1 and S2) after which total DNA was extracted and sequenced (Table S1A). As expected, a distinct microbial community associated with the materials (i.e., wood, LDPE and W-LDPE) developed compared with the surrounding water (Figs. 1A and Table S1B). Principal Coordinate Analysis (PCoA) showed that all samples clearly clustered by sample type using Robust Aitchison’s distance (PERMANOVA R = 0.898, p = 0.001; ANOSIM R2 = 0.713, p = 0.001; Table S1C), with water being separated from all substrates on the first axis (representing 53.9% variation) and the three substrates being separated on the second axis (10.1% variation; Fig. 1A). Planktonic vs biofilm community differences are well documented (Wright et al., 2020), and come as a consequence of the different nature of surface-attached vs free-living communities and their capacity to become sessile (Berlanga and Guerrero, 2016). Furthermore, water samples represent only a snapshot of the community present at the time of sample collection. In contrast, the substrates represent a cumulative and changing microbial community in the river over the entire incubation period.
Differential microbial assemblages on surfaces have previously been reported between plastic and non-plastic materials (Delacuvellerie et al., 2019; Ogonowski et al., 2018; Vaksmaa et al., 2021; Zettler et al., 2013). However, the material preferences of microbial colonisers are unclear and may be dependent on residence time, location (Oberbeckmann et al., 2021, 2016) and nutrient availability (Oberbeckmann et al., 2018). The potential differences across microbial communities colonising plastics rely on the presence of specific early settlers and rare taxa (Wright et al., 2020), whereas the microbial community differences on wood seem to be shaped by the more degradable nature of this substrate per se (Oberbeckmann et al., 2021). Here it is likely that W-LDPE, as well as wood, also released readily available compounds for microbial biodegradation that caused an early selection of specific taxa, as discussed below. Consistent with previous studies (McCormick et al., 2016), our data showed wood to support the highest alpha diversity values across all samples (ANOVA p ≤ 0.05; Fig. 1B and Table S1D). Water had significantly lower (ANOVA p ≤ 0.05) richness than any other group, although it was similar in alpha diversity (Simpson’s Index of diversity) to W-LDPE (Fig. 1B). LDPE and W-LDPE were similar in richness to each other, with LDPE having significantly lower alpha diversity than any other sample group (ANOVA p ≤ 0.05). The differences in the microbial profiles between the pristine and weathered LDPE are probably related to physicochemical modifications of the weathered material, which reduce the polymer hydrophobicity (Arp et al., 2021) and promote the release of carbon leachates (Rummel et al., 2022; Zadjelovic et al., 2022), all being factors that influence microbial settling. Such a phenomenon was also evidenced at the community level during the early colonisation of PE in marine environments (Erni-Cassola et al., 2020).
Regarding the taxonomic affiliation, the vast majority of classified reads were related to Bacteria (98.83% on average within samples), with Eukaryota, Viruses and Archaea making up only 1.01%, 0.15% and 0.01%, respectively (Fig. S3 and Table S1E). Amongst Bacteria, the phylum Proteobacteria dominated across all microenvironments tested (i.e., LDPE, W-LDPE, wood, and water; averaging 90, 93, 85 and 77%, respectively), followed by Bacteroidota (averaging 6, 4, 7 and 8%, respectively; Fig. S3 and Table S1E), similar to a previous metagenomic analysis of water samples at a location nearby to our incubations (Borsetto et al., 2021). Amongst Proteobacteria, the most abundant class was Gammaproteobacteria, averaging > 70% of reads in all samples (Fig. S3), and this was mainly composed of the order Burkholderiales (average 82, 69, 68 and 58%, respectively; Fig. S3). The order Burkholderiales, previously classified as Betaproteobacteria (Depoorter et al., 2016) has been reclassified to Gammaproteobacteria in the GTDB taxonomy (Parks et al., 2018) based on the phylogenetic affiliation of their genomes. Interestingly, the dominance of the Betaproteobacteria in freshwater bodies identified in several previous publications (Jin et al., 2018; Llirós et al., 2014; Tiwari et al., 2021) was found to be driven by the dominance of the order Burkholderiales (Llirós et al., 2014). In this context, using amplicon sequencing, Lu et al. (2020) reported former Betaproteobacteria as the most dominant class (15.12–46.56%) and Limnohabitans (Burkholderiales) at the genus level in freshwater samples from the River Xiangxi. Additionally, critical roles in nutrient and carbon cycling have been related to versatile copiotrophs within the former Betaproteobacteria (Elifantz et al., 2005; Barberan and Casamayor, 2010), functions that can be attributed to the dominance of Burkholderiales in freshwater systems (Chiriac et al., 2022). In contrast to the high abundance of Burkholderiales in freshwater, this order is found in relatively low numbers in marine environments, most likely being outcompeted by other Gamma- and Alphaproteobacteria groups.
Dominant species belonging to the order Burkholderiales were similar across all solid substrates and distinct from the planktonic communities (Fig. 1C, Fig. S3 and Table S1B). Amongst them, the most abundant genus corresponded to Sphaerotilus (averaging 52, 35, 26 and 0.33% on LDPE, W-LDPE, wood and water, respectively; Fig. S3). Sphaerotilus is an aquatic filamentous iron bacterium –a taxon that can use iron as an energy source– also found in activated sludge in WWTP, and capable of forming sheaths that allow attachment to solid surfaces. This favours their growth in slow-running or nutrient-poor water and provides protection by shielding the bacteria from protozoa (Liu et al., 2002; van Veen et al., 1978; Mulder and Deinema, 1981). Such biofilm-forming bacteria clearly dominate all solid substrates assessed and could potentially serve as the main protective structure for other biofilm colonisers and even organisms that typically have a planktonic lifestyle. For example, the typically planktonic Burkholderiales genus Limnohabitans (Kasalický, 2019) was detected on plastic (LDPE and W-LDPE) and wood (averaging 11, 12 and 13%, respectively), although it was much more abundant in the surrounding water (44%; Fig. 1C). Another interesting Burkholderiales genus identified as part of the biofilms recovered from solid substrates was Methylotenera, present across all surfaces tested but more abundant on wood (4.4%) than plastic and water samples (Figs. 1C; 2.2, 1.4, and 0.6% on LDPE, W-LDPE and water, respectively). Methylotenera has been described as a putative cellulose degrader found in microbial communities associated with sunken wood logs in marine environments (Pop Ristova et al., 2017). Similarly, Duganella (Burkholderiales) and Microthrix (Acidimicrobiales) were found to be significantly more abundant in wood samples (Fig. 1C). Interestingly, Duganella was found to encode cellulases, xylan esterases and pectin lyases, all enzymes involved in the degradation of lignocellulosic carbon sources (Zhao et al., 2022). Microthrix species are abundant in active sludges and linked with the degradation of complex carbon compounds (Begmatov et al., 2022; Rossetti et al., 2005), however, no direct association with wood or wood derivate degradation has been previously reported. Finally, Hydrogenophaga (Burkholderiales) was more abundant on solid substrates, especially on W-LDPE (7%; Fig. 1C). Members of this genus have been detected within biofilm-forming bacteria on sand recovered from WWTP denitrification filters (Lemmer et al., 1997).
Even though aspects regarding the biodegradation of polyethylene are out of the scope of this investigation, it is important to point out that the genera Methylotenera and Pseudomonas (Gammaproteobacteria), identified in our samples, have both been associated with the degradation of Polycyclic Aromatic Hydrocarbons (PAHs) in sewage sludge (Guo et al., 2020) and that several species of Pseudomonas have had their degradative capacities widely explored (Baig et al., 2022; Ramos et al., 1995; Rojo, 2017). Curiously, most Pseudomonas were significantly more abundant on W-LDPE than in any other sample (Fig. 1C), with the total relative abundance of the family Pseudomonadaceae being 2.6, 18.6, 6.7 and 3.2% on LDPE, W-LDPE, wood and water, respectively (Fig. S3). W-LDPE releases large amounts of organic compounds that encourage the colonisation and growth of a distinct microbial community (Erni-Casola et al., 2020; Zadjelovic et al., 2022), however, these findings need to be further explored and the degradative capacity of this family elucidated.
The water samples showed a clear divergent microbial profile as compared with the solid substrates and contained the typical planktonic genera Limnohabitans (41%), Planktophila (4.4%), Polynucleobacter (4.1%) and Aquirufa (0.84%) (Fig. 1C). The most remarkable result from the water samples was the high abundance of potential human pathogens, such as the Enterobacterales Escherichia, Salmonella and Klebsiella (7.2, 3.7 and 1.5%, respectively), as well as Streptococcus (3.6%) –all described as frequent commensals in waterbodies in the proximity of cities, WWTP, and other industrial activities (Amarasiri et al., 2020; Kistemann et al., 2002; Liu et al., 2018; Rodrigues and Cunha, 2017; Rolbiecki et al., 2021; Sinton et al., 1993). These were also found as part of the microbial community on solid substrates but in much lower abundance (Fig. 1C). The lower abundance of the genus Escherichia on solid substrates correlates with previous results (Song et al., 2020) where Escherichia coli could not be isolated from plastics incubated across different points along the River Weser (Germany). Although most of these potential human pathogens were not abundant on the solid substrates, there were other examples of potential opportunistic human pathogens on these surfaces, such as Pseudomonas aeruginosa (Diggle and Whiteley, 2020) and Acinetobacter (Elhosseiny and Attia, 2018), which were more abundant in both wood and plastic samples than in water samples (Figs. 1C and S3).
While known biofilm-forming microbes such as P. aeruginosa abounded on material surfaces (especially on W-LDPE), Enterobacterales species did not seem to be good colonisers of plastics under in-situ environmental settings. Hence, further work is needed to determine whether these potential pathogens colonising plastics may survive, transfer and cause disease (Beloe et al., 2022) and whether they are able to compete with naturally biofilm-forming microbes in freshwater environments.
ARG distribution within the plastisphere and their surrounding freshwater compartments.
Our initial CARD analysis (The Comprehensive Antibiotic Resistance Database) for the identification of ARGs generated a comprehensive list of both known target point mutations for antibiotic resistance (e.g., gyrase and ribosomal mutations) and active antibiotic inactivating determinants (i.e., ARGs) (Table S1F and Fig. S4). However, we focus hereafter on the latter ARGs due to the elevated background noise that can occur when including point mutation ARGs from metagenomic data. As with microbial communities, ARG diversity and distribution also showed a noteworthy divergence between the solid substrates and the surrounding water (Fig. 2A and Table S1F). As stated above, taxonomic differences between planktonic communities and biofilms were expected; hence, it is not surprising that these differences in microbial assemblages also drive divergent ARG profiles (Fig. 2A). ARG richness was significantly lower in the planktonic community than in any of the biofilms (ANOVA p ≤ 0.05) while the alpha diversity of ARGs was similar between both LDPE substrates and water, with only wood having significantly higher alpha diversity than any other sample type (Fig. 2B). Thus, while biofilms are enriched in ARGs (Balcázar et al., 2015; Hall and Mah, 2017), the planktonic microbiome of our in-situ incubation site presented an interesting array of ARGs –as demonstrated by our genome-centric analysis below– which is possibly caused by the elevated number of pathogen-like microbes present in the water (e.g., Enterobacterales).
In total, we identified 242 ARG subtypes amongst all microbiomes (204, 203, 207 and 177 identified in the LDPE, W-LDPE, wood and water samples, respectively; Fig. 2C and Table S1F). While the number of ARGs detected kept constant across solid substrates (n = 198–214) the relative abundance of the ARGs was clearly higher in W-LDPE biofilms than any other microbiome (i.e., 321 reads per million in W-LDPE vs 154, 168 and 145 in wood, LDPE and water microbiomes, respectively; Fig. 2C). Multiple Drug Resistance ARGs (MDR) dominated the dataset in number (n = 87) and relative abundance, particularly in surface biofilms (79%, 87% and 83% of the reads per million in wood, W-LDPE and LDPE, respectively, vs 36% in water). MDR is known to dominate ARGs in soil microcosms (Cao et al., 2021) as well as in mining-impacted soil samples (Yi et al., 2022), or even in ready-to-eat food (Li et al., 2020). The reported levels of MDR in the literature are in line with our findings, where MDR accounts for a high proportion of the number of reads and the highest number of genes associated with AMR (Fig. 2C).
While MDR, beta-lactams and peptide antibiotics were abundant and similar between biofilm and planktonic microbiomes, other abundantly detected ARG subtypes showed large differences between both microbial communities (Fig. 2C and Table S1F). Hence, planktonic microbiomes were clearly enriched in ARG subtypes for aminoglycosides, tetracycline, aminocoumarin, fluoroquinolones, nitroimidazole, oxazolidinone and fosfomycin; whereas biofilms were enriched in ARGs that conferred resistance to Macrolides-Lincosamides-Streptogramins (MLS), rifamycin, sulfonamides, disinfecting agents and glycopeptides. Interestingly, microbiomes on W-LDPE were specifically enriched for triclosan, phenicol and diaminopyrimidine ARGs. These results suggest an intrinsic distinctness of ARG profiles within different environmental compartments, mostly driven by microbial community lifestyles, i.e., biofilm vs planktonic, but also influenced by the weathering of plastic surfaces.
As expected, similar profiles were observed at the individual ARG level (Fig. 3). Thus, water samples were dominated by predominant aminoglycoside resistance gene aph(3’)-Ia and tetracycline resistance gene tetC. On the other hand, the solid substrates showed high abundances of axyY, mex and mux genes, all belonging to the MDR ARG class (Figs. 3). Overall, our results confirm previous studies in which aquatic biofilms –independently of them growing on plastics or natural surfaces– showed high abundances of these MDR ARGs (i.e., (Wu et al., 2019; Sun et al., 2021; Yang et al., 2019)). Nevertheless, some of these mex and mux genes, as well as the triclosan ARG opmH, were substantially more abundant on W-LDPE in our study, emphasising for the first time the concerning enrichment of distinct ARGs on weathered plastics –an observation that requires further attention.
As shown here, several studies have reported that plastics support microbial communities harbouring resistance genes against a variety of antibiotics (Cheng et al., 2022; Wang et al., 2021; Yang et al., 2019). The fact that plastic microbiomes are enriched in ARGs has raised wide concern, although we demonstrate that this is only the case for a distinct set of ARG subtypes and greater attention should be given to all environmental compartments that are impacted by anthropogenic activities. Pathogens and encoded ARGs in planktonic communities may have lower survival and transport than those within the plastisphere, but are also much harder to filter out and may offer an increased risk of exposure to higher organisms. In this sense, fluoroquinolone resistance –enriched in planktonic microbiomes– confer protection against several second and third-generation drugs, such as ciprofloxacin, levofloxacin and ofloxacin (livertox.nih.gov). It is also worth highlighting that the main concern surrounding the presence of ARGs in the environment is for these to become reservoirs of resistance that can then be horizontally transferred to pathogenic bacteria. Horizontal gene transfer is more likely to occur on solid surfaces, and plastics have been shown to facilitate this process (Arias-Andres et al., 2018), but natural surfaces also need to be taken into account as demonstrated here.
To the best of our knowledge, there are only two other metagenomic datasets that analysed ARGs on plastics incubated in freshwater (Oberbeckmann et al., 2021; Wu et al., 2019). Oberbeckmann et al. reported a much higher association of ARGs with wood than with plastics (i.e., 20 putative ARGs conferring resistance to beta-lactams, fluoroquinolones and tetracycline were found in metagenomes from wood samples, while polystyrene (PS) and PE presented only one putative ARG related to beta-lactam resistance on PS and none on PE; (Oberbeckmann et al., 2021)). The second metagenomic analysis covered biofilms forming on polyvinyl chloride (PVC) pellets incubated in ex-situ 5 L bioreactor (Wu et al. 2019). Researchers determined that even though the biofilms associated with PVC pellets showed some degree of specificity, including a distinct profile of potential pathogens, major differences were only seen when comparing solid substrates vs surrounding water, as reported here in our analysis (Fig. 2). Nevertheless, we further show that plastic weathering prior to water submersion –a process that frequently occurs in nature– enhances the enrichment of particular ARGs.
Genome-centric insight of potential pathogens and associated ARGs within the plastisphere.
In an attempt to link ARGs to their host, we co-assembled the reads from all samples and generated 215 Metagenome Assembled Genomes (MAGs) with > 50% completion and < 10% redundancy (20 of these MAGs were > 90% complete and 73 MAGs > 75% complete; Table S3A; Fig. S5). Of the 215 MAGs, only one was predicted to be archaeal (MAG106, classified as the TA-21 genus from the Nitrosphaeraceae family; Thermoproteota phylum). The taxonomic classification of the other 214 bacterial MAGs revealed that the contribution of each class closely mirrored that of the read-based analyses: Gammaproteobacteria (113 MAGs), Bacteroidota (45), Myxococcota (11), Alphaproteobacteria (9), Verrucomicrobiota (9) and Actinobacteria (6) (Table S3A). Interestingly, MAGs were assembled for potential pathogens such as Escherichia flexneri (MAG1 with 100% completion and 0% redundancy; see Escherichia/Shigella reclassification in (Parks et al., 2021a)), Aeromonas spp. (MAG107) and Acinetobacter spp. (MAGs 98, 124, 92 and 214; Fig. 4). In accordance with our read-based analysis above, Escherichia was almost exclusively found in water samples, whereas Acinetobacter –typically found in soil and water samples (CDC.gov)– were mainly attached to plastic and wood materials (Fig. 4). While Acinetobacter members such as A. baumannii are related to pathogenesis in humans (Sarshar et al., 2021)(CDC.gov), it was not possible to assign a taxonomic affiliation to the species level for these Acinetobacter MAGs. Aeromonas spp. are also well-recognised disease-causing agents, not only for animals such as fish, but also for humans (Bhowmick and Bhattacharjee, 2018; Pessoa et al., 2022)
We used PathoFact (de Nies et al., 2021) to predict the ARGs, toxins and virulence factors present within the MAGs and found that of the 214 bacterial MAGs, 115 were predicted to have three or more ARGs in their genomes (Fig. 4). As above, we focussed only on the active antibiotic inactivating ARGs, and not the known target point mutations for antibiotic resistance. Expectedly, MDR genes were the most abundant ARG class (i.e., 254 MDR genes within all MAGs, averaging 1.19 MDR genes per MAG; Fig. S6). The maximum number of MDR genes predicted for a single genome was 22 in MAG1 (i.e., Escherichia flexneri). Other ARG classes that were both prevalent and abundant within the MAGs were aminoglycoside and beta-lactam resistance genes, with a total of 93 and 72 (mean 0.715 or 0.5 ARG copies per MAG), respectively (Fig. S6). Interestingly, MAG1 (i.e., E. flexneri) also had the most beta-lactam resistance gene copies (n = 6), as well as three aminoglycoside resistance gene copies. The highest number of aminoglycoside resistance gene copies (n = 4) were found in MAGs all belonging to the order Burkholderiales (i.e., MAGs 88, 172, 44, and 135; Fig. 4). Genes for bacitracin, MLS or Antimicrobial Peptide resistance were also prevalent, being present in 59, 40 or 49 MAGs, respectively (Fig. S6).
On top of the encoded ARGs, the presence of toxins and virulence factors within the MAGs provides further hints on their potential pathogenicity. Although water samples seem to be the main source of typical human pathogens (e.g., MAG1 E. flexneri, as well as raw read based detected Salmonella, Streptococcus and Klebsiella, Fig. 1 and Fig. S3), it is important to point out the potential of biofilms –established on either plastic or wood– to also harbour potential opportunistic pathogens (e.g., Acinetobacter spp. and Aeromonas spp., as well as P. aeruginosa, Fig. S3). The array of toxins and virulence factors across most MAGs suggest a wide diversity of pathogenic factors that may affect surrounding organisms; from co-occurring microbes to animal species or plants, as well as humans. For instance, some Flavobacterium spp. can cause disease in fish (Loch and Faisal, 2015). Regardless of the genomic indications of the potential pathogenicity of the plastisphere (i.e., ARGs and genes encoding virulence factors), it is not possible to draw conclusions on the eventual human risk of plastic pollution as a vector for pathogens without further experimentation. For this, additional assessments are needed to determine the actual pathogenicity of microbes within the plastisphere; these should take into account the potential transfer and ability to cause disease to the host organism –be it human, animal or plant (Beloe et al., 2022).
Altogether, this genome-centric analysis has allowed the assembly of MAG1, i.e., E. flexneri, one of many planktonically-found pathogen-like strains detected within our water metagenomes (e.g., Escherichia, Salmonella, Klebsiella and Streptococcus; see Fig. 1). Not surprisingly, this MAG showed the highest amount of encoded ARGs and an elevated potential to produce toxins and virulence factors. While these potential pathogens were not major components of the plastisphere, other taxonomical groups like Acinetobacter spp. and Aeromonas spp. did show a higher presence within the biofilms, in which case, their pathogenic capacity needs to be further elucidated.
Case study: sub-inhibitory antibiotics concentrations induce distinct ARG enrichments in different riverine compartments.
As our in-situ analysis showed a distinct enrichment of ARGs in different riverine compartments, we went on to test the selective pressure sub-inhibitory concentrations of antibiotics may have on the abundance of ARGs in the plastisphere. Antibiotic concentrations considerably below any ecotoxicological effect have been reported to be enough to select for resistances (Bengtsson-Palme and Larsson, 2016). For this, ex-situ microcosms containing river water and sediment were set up with PE, PP and wood fragments (as shown in Fig. S7) in the presence/absence of a cocktail of antibiotics: the macrolide azithromycin (AZM, 0.076 µg L− 1), the fluoroquinolone ciprofloxacin (CPFX, 0.136 µg L− 1) and the sulphonamide sulfamethoxazole (SMX, 4.8 µg L− 1). These are concentrations three orders of magnitude below susceptible breakpoints established by EUCAST (www.eucast.org) and in the range detected in WWTP effluent waters (Bhandari et al., 2008). We then used HT-qPCR to profile a range of ARGs (n = 48) in the microbiomes of PE, PP, wood, water and sediment (Resistomap results are shown in Table S4).
Microbiomes that developed on wood showed the highest detection of ARGs tested (21/48) regardless of the presence/absence of antibiotics (Fig. 5A). Surface biofilms had a significant impact on all antibiotic resistance classes (ANOVA p ≤ 0.05), whereas the presence of antibiotics significantly impacted the resistance to quinolones and tetracycline as well as MDR and ‘other’ genes (mainly resistance genes against quaternary ammonium compound (QACs); Fig. 5A); ARGs for tetracycline and MDR negatively correlated with the presence of antibiotics. Nevertheless, the most remarkable results are observed when analysing the effects at an individual antibiotic and corresponding ARG subtype level. After applying AZM, CPFX and SMX, we would expect an enrichment on ARGs related to MLS, quinolone and sulphonamide resistance, respectively. AZM did in fact cause a strong enrichment of MLS ARGs, particularly of the known resistance genes msrE and mphE (Chen et al., 2020), but this occurred mainly in the water samples (Fig. 5B). On the other hand, CPFX enhanced the presence of ARGs against quinolone antibiotics –i.e., gene qepA which encodes for a fluoroquinolone efflux pump (Yamane et al., 2007). The gene qepA was enriched in most compartments in the presence of antibiotics, but this was particularly evident in microbiomes from PE and wood surfaces (Fig. 5). Finally, SMX had an effect on ARGs against sulphonamides. While these genes were particularly high in all conditions, exposure to sub-inhibitory concentrations of SMX enriched for sulphonamide ARGs –i.e., mainly sul1– in microbiomes from plastics, PP and PE (Fig. 5). This is particularly interesting as, while previous experiments in our group observed almost negligible effects of such low SMX concentrations in riverine sediments and waters (Borsetto et al., 2021), this antibiotic has been shown to adsorb to PE (Xu et al., 2018) causing the potential enhanced (although not significant) effect of ARG enrichment observed here.
ARGs that confer resistance to antimicrobial compounds not included in the antibiotic cocktail (e.g., aminoglycoside, tetracycline, bacitracin or mechanisms for MDR) showed no observable increase in abundance (Fig. 5B and Table S4). On the other hand, the presence of sub-inhibitory concentrations of antibiotics did, curiously, produce an enrichment of ARGs against antiseptics such as QACs, mainly in the planktonic microbiome (i.e., qacE genes classed in the category ‘others’; Fig. 5). While sub-inhibitory concentrations of QACs are well documented to develop antibiotic resistance in the environment (Mulder et al., 2018; Zhang et al., 2015), the fact sub-inhibitory concentrations of antibiotics may enrich for QAC resistance –as reported here– has also been previously suggested (Murray et al., 2019). This is not surprising given that resistance genes to antibiotics and biocides co-occur in genetic clusters (Gaze et al., 2011; Pal et al., 2015). As performed in Borsetto et al. (2021), we also included sediment samples in an effort to better reflect the riverine environment (Borsetto et al., 2021); however, the presence of antibiotics produced no significant variations across all ARGs analysed.
Here we show that sub-inhibitory but environmentally relevant antibiotic concentrations can enhance ARGs in microbiomes from riverine systems. Specifically, our study shows a correlation between the presence of an antibiotic and the enrichment of its particular ARGs, and not a generic non-specific enrichment of ARGs, as well as the co-enrichment of QAC resistance genes. Antibiotic residues have been widely detected in riverine ecosystems (Wilkinson et al., 2022) and can adsorb to microplastics (Xu et al., 2018), but distinct ecological compartments seem to be affected by different antibiotics. The influence of an antibiotic on the plastisphere will most likely rely on its adsorption to the plastic’s hydrophobic surfaces or their biofilm penetrability. Hence, this will only occur on a case-by-case basis, opening a new area of investigation that will provide a more detailed view of the potential spread of particular ARGs across the environment using microplastics as vectors.