Polystyrene and Azithromycin Co-Contamination: Comparative Analysis of Microbial Shifts in the Rhizosphere vs. Bulk Soil under Wheat Growth

Abstract

The high rate of entry of microplastics (MPs) and pharmaceuticals into the agroecosystem leads to progressive accumulation within the soil year after year, raising concerns about both soil ecological health and crop productivity. The combined contamination of macrolide antibiotics and nanoplastics in the rhizosphere remains poorly understood. In this work, microbial shifts in the rhizosphere and bulk soil under wheat growth as affected by the polystyrene (PS) and azithromycin antibiotics (AZM) co-contamination were investigated. In this study, the toxicity effects of different sizes and concentrations of PS (i.e., 100 nm; 300 nm; 50 and 500 mg/kg, respectively) and AZM (i.e., 5 and 20 mg/kg) were combined under wheat growth in the greenhouse for 110 days. A total of 15 treatments with three replicates each were prepared and arranged in a completely random design. Findings revealed that combined contamination (i.e., PS+AZM) negatively affected microbial richness and diversity compared to individual contamination, and it was more prominent in the rhizosphere than in bulk soil. This, in turn, affected the composition of the bacterial community, abundances, functions, and interactions. It also influenced functions involved in biogeochemical cycles within the system. The co-occurrence network showed that PS+AZM inhibited wheat growth in general and reduced bacterial interactions (nodes: 770, edges: 6184, positive interactions: 4656, negative interactions: 1528) compared to individual PS (nodes: 1301, edges: 18,190, positive interactions: 11,597, negative interactions: 6593) and AZM application (nodes: 866, edges: 7464, positive interactions: 5192, negative interactions: 2272). The same trend occurred in the rhizosphere vs bulk soil. PS+AZM drove distinct microbial community shifts, where the rhizosphere was enriched with mostly bacteria involved in nitrogen biogeochemical cycling (Microvirga; Pseudorhodoplanes; Polaromonas; Hydrogenophaga; and Bauldia) and bulk soil was enriched with bacteria adapted to low nutrients and stress resistance (i.e., Opitutus; Adhaeribacter; Blastococcus; Ilumatobacter; IS-44; Bryobacter; Subgroup_10; Gemmatimonas). These results shed light on the co-occurrence of MPs and antibiotics in agroecosystems. They may help in developing mitigation measures against the effects of MPs and antibiotics in agroecosystems.

1. Introduction

As a source of food for humans, agroecosystems are increasingly subjected to contamination from microplastics (MPs) , and antibiotics due to intense anthropogenic activities. As a model MP, polystyrene (PS) is a versatile polymer due to its chemical and physical characteristics. It is commonly used in the production of cutlery, egg trays, spectacle frames, building insulation, and food packaging, leading to its waste being widespread in the environment vis-à-vis soil ecosystems. On the other hand, azithromycin (AZM), along with clarithromycin (CLA), and erythromycin, are major macrolide antibiotics and have been found in relatively large amounts (reaching tens of μg/L) in wastewater and enter soil ecosystems via wastewater irrigation and manure usage. − Keep in mind that macrolides are among the antibiotic classes that have been detected in farmland, which raises concern about their contamination. Due to their small size and large surface area, MPs in soil can sorb organic contaminants like AZM, potentially altering their bioavailability, transport, and joint toxicity, which might inflict considerable damage to microbial community structure, plant health, and soil fertility. ,

The rhizosphere, a plant-derived microbial hotspot, is fundamentally distinct from bulk soil due to continuous root exudation, which modifies nutrient availability and pH and shapes microbial composition and interactions that promote plant growth. , Recent studies in the rhizosphere involving MPs confirm that MPs act as substrates for microbial colonization, induce oxidative stress, and alter nutrient availability in major global staple crops such as wheat (Triticum aestivum) and maize. − For instance, applying PS at 10.0% released toxic chemical compounds in the soil, significantly reducing maize root dry weight (32%), shoot dry weight (13%), chlorophyll content (29%), leaf area (13%), and plant height (29%) relative to control (0.0% PS).

Exposing soil to MPs or antibiotics may result in rhizosphere microbial communities variability. − For instance, plastic-degrading (e.g., the genus Streptomyces) and disease-causing microbes (e.g., the family Nocardiaceae) are vulnerable to MPs exposure. , Similarly, the co-exposure of PS and phenanthrene in soybean rhizosphere decreased the relative abundance of Proteobacteria while enriching that of Cyanobacteria and Chloroflexi. Contrarily, MPs might increase bacterial communities. For instance, Zhang et al. reported that MPs could create ecological niches through their physicochemical properties, which in turn facilitate the growth of bacteria or even facilitate the adsorption of antibiotics. ,, Again, exposing PS (5 mg/kg) to sulfamethazine (SMZ) (1 mg/kg) alleviated the negative effects of SMZ on bacterial diversity, composition, and structure and inhibited sul1 transmission by decreasing the intl1 abundance due to the adsorption of SMZ to PS through nonspecific van der Waals forces and π–π interactions.

The spread of macrolide antibiotics in farmland soil continues to increase at an alarming rate despite extensive research on MPs and antibiotics in agroecosystems, , leading to accumulation in edible plants and resulting in potential health problems. , Both PS and AZM can persist in soil, yet their coexistence and interactive effect on the rhizosphere remain poorly characterized. This raises concerns about their environmental contamination, soil microbial alterations, and crop production. − ,,

To the best of our knowledge, no study on macrolide antibiotic plastics has been conducted that systematically examines how PS and AZM co-exposure differentially shapes microbial network interactions and functional associations with a sole focus on rhizosphere and bulk soil. By combining PS and AZM, root exudates may either mitigate antibiotic toxicity via enhanced biodegradation or exacerbate MP-induced stress by altering surface biofilm formation. To address this gap, we conducted the current study. We hypothesized that co-exposure to PS and AZM would synergistically alter rhizosphere microbial network complexity and functional gene abundance compared to individual exposures in the rhizosphere more than in the bulk soil.

An understanding of this co-contamination is paramount because shifts in microbial community composition and functions could potentially affect wheat growth rates, soil fertility, and the spread of antibiotic resistance genes (ARGs). This study aims to (1) characterize the individual and combined effects of PS and AZM contamination in the rhizosphere and bulk soil microbiomes; (2) identify key microbial taxa responsive to co-contamination; and (3) assess potential functional implications for nutrient cycling and plant health. By integrating high-throughput sequencing, we provide novel insights into how emerging contaminants interact to reshape the soil microbial ecosystems. Our findings will not only contribute to the risk assessments of co-contamination but also inform possible strategies for mitigating ecological impacts of MPs and antibiotics in agroecosystems.

2. Materials and Methods

2.1. Preparation of the Tested Soils and Characteristics

The soil was collected (2 cm layer) from a farmland in Shanxi Province (Shanxi, China). The soil samples were pulverized, air-dried naturally, and sieved (2 mm mesh screens). Basic physical and chemical characteristics of the soil prior to use were pH: 6.04, total nitrogen: 0.43 g/kg, total phosphorus: 0.34 g/kg, total potassium: 25.3 g/kg, and soil organic matter: 7.34 g/kg.

2.2. Polystyrene and Azithromycin: Contaminants and Soil Mixing Procedures

Different sizes of PS (100 and 300 nm) (≥ 99% purity) were obtained from Wuxi Ruige Biotechnology Co., Ltd (Wuxi, China), AZM (≥99% purity) was obtained from Xilong Scientific, and methanol (99.5%) was obtained from Shanghai Macklin Biochemical Technology Co., Ltd. (Shanghai, China). Given that most agricultural studies focused on micrometer-sized polystyrene (from tens to hundreds), herein we chose (a) 100 and 300 nm anticipating that they would behave differently in soil, as 100 nm would be taken up by plant roots and transported to aerial parts as previously reported, , and (b) 5 and 20 mg/kg of AZM because these concentrations fall in the range of antibiotics concentrations (i.e., 0.5–50.0 mg/kg) commonly reported in contaminated soils. ,

Contaminants and soil mixing procedures were modified following a previously used method by Gao et al. First, AZM was dissolved in methanol (1000 mg/L) and sprayed onto 1.9 kg of soil to achieve 5 or 20 mg/kg AZM, then thoroughly mixed. Second, after methanol volatilization, PS powder was gradually incorporated, and each rhizobox was filled to a total mass of 1.9 kg of the soil-PS mixture. The mixture was air-dried overnight, then added to each rhizobox after 48 h. Third, to facilitate germination, seeds were soaked for 12 h, enclosed in moistened paper towels, sealed in polyethylene bags, and then left in the greenhouse (i.e., 25 °C for 48 h).

2.3. Rhizobox Construction, Operation, and Sampling Procedures

A 13 × 13 × 13 cm³ cuboid rhizobox was divided into a central rhizosphere (1 cm wide × 13 cm long × 13 cm high) and two lateral bulk compartments (each 6 cm wide × 13 cm long × 13 cm high). A 30-μm stainless-steel mesh (27 cm × 13 cm) was folded and placed vertically in the middle, with its lower edge 1 cm above the container bottom. It is important to note that the stainless-steel mesh restricts root growth to the bulk soil but allows the movement of water and nutrients (Figure S1). The experiment consisted of 15 treatments as follows: CK (the control), T2 (5 mg/kg of AZM and 50 mg/kg of PS100), T3 (20 mg/kg of AZM and 50 mg/kg of PS100), T4 (5 mg/kg of AZM and 500 mg/kg of PS100), T5 (20 mg/kg of AZM and 500 mg/kg of PS100), T6 (5 mg/kg of AZM and 50 mg/kg of PS300), T7 (20 mg/kg of AZM and 50 mg/kg of PS300), T8 (5 mg/kg of AZM and 500 mg/kg of PS300), T9 (20 mg/kg of AZM and 500 mg/kg of PS300), T10 (50 mg/kg of PS100), T11 (500 mg/kg of PS100), T12 (50 mg/kg of PS300), T13 (500 mg/kg of PS300), T14 (5 mg/kg of AZM), and T15 (20 mg/kg of AZM). They were divided into four categories: Control (CK), PS+AZM double effect (T2-T9), PS single effect (T10-T13), and AZM single effect (T14-T15). When compared to other treatments, CK served as the baseline. Each treatment was replicated three times, as shown below (Table ).

1. Treatments: Design and Composition .

Major groups	Treatments	Contaminants	Concentration (mg/kg)
No effect (control)	CK	NA	PS	AZM
Double-effects group (PS+AZM)	T2	Polystyrene (PS) size: 100 nm	50	5
	T3		50	20
	T4		500	5
	T5		500	20
	T6	Polystyrene (PS) size: 300 nm	50	5
	T7		50	20
	T8		500	5
	T9		500	20
Single-effect group (PS)	T10	Polystyrene (PS) size: 100 nm	50	NA
	T11		500
	T12	Polystyrene (PS) size: 300 nm	50
	T13		500
Single-effect group (AZM)	T14	Azithromycin (AZM)	NA	5
	T15			20

^aNote: PS, polystyrene; AZM, azithromycin; AZM; CK, control.

Soil was added to the rhizobox and mixed with deionized water to reach 60% of maximum water-holding capacity. Rhizoboxes were arranged in a completely randomized design and then placed in a greenhouse at 25 °C and 43% relative humidity for 2 days before use. Overall, a total of 20 wheat seeds (1 cm apart) were grown in the rhizosphere compartment of the rhizobox. After germination, 10 healthiest seedlings were kept per rhizobox. They were watered, monitored, and exposed to sunlight throughout the experimental growth stages. Monitoring of the experiment was done daily, and water was replenished by irrigation with deionized water.

At the tillering stage (∼25 days), rhizosphere and bulk soil were sampled. Soil samples were collected at three points spaced 2 cm apart. Wheat plants were gently removed, and shoots and roots were separated. Rhizosphere soil was obtained by gently shaking the roots to dislodge small clumps of soil adhering to the roots, following the previously used procedure by He et al. A small portion was taken and freeze-dried at −60 °C and sieved through a 0.25 mm screen for testing physicochemical parameters, while samples for microbiological analysis were cryopreserved at −80 °C. The processes outlined above were repeated throughout the harvesting stage (110 days).

2.4. Physical and Chemical Analysis

Basic chemical and physical characteristics were determined using samples of soil taken during the tillering and harvesting periods. The pH level and electrical conductivity (EC) of the soil were determined using a pH meter and a conductivity meter at a soil-to-water ratio of 1:5 (w:v).

2.5. Biological Analysis

2.5.1. Soil Microbial DNA Extraction, PCR Amplification, and Library Construction

Microbial DNA from the soil extracts was extracted using a Fast DNA SPINKit for Soil (MP Biomedicals, CA, USA). The quantity of the extracted DNA and the purity were assessed by agarose gel electrophoresis. The DNA was then diluted to 1 ng μL^–1 using sterile water. Polymerase chain reaction (PCR) was performed with universal primers 515F (GTGCCAGCMGCCGCGGTAA) and 907R (CCGTCAATTCCTTTGAGTTT) targeting the hypervariable V4 region of the 16S rRNA gene to amplify nearly all bacterial taxa from the soil samples as per in silico prediction. The primers employed for sequencing comprised an Illumina-compatible adapter, a padlinker, and a gene-specific primer. In addition, a unique 6-nucleotide barcode (Table S1) per sample was added to the reverse primer. All PCRs were performed in 30-μL reactions containing 15 μL of Phusion High-Fidelity PCR Master Mix (New England Biolabs), 0.2 μM forward and reverse primers, and 10 ng of template DNA. Thermal cycling comprised an initial denaturation at 98 °C for 1 min, 30 cycles of denaturation for 10 s at 98 °C, annealing for 30 s at 50 °C, elongation for 30 s at 72 °C, and finally 72 °C for 5 min. PCR amplicons were combined with 1× loading buffer with SYBR green and subjected to 2% agarose gel runs. The samples with a distinct and bright band at 300 base pairs were selected for processing. PCR products were pooled in equidensity proportions and then purified with a gel extraction kit (Qiagen; Hilden, Germany). Index codes were added following sequencing library preparation with a TruSight DNA PCR-Free Sample Preparation Kit (Illumina, San Diego, CA, USA) according to the manufacturer’s protocol. The library quality was examined using an Agilent Bioanalyzer 2100 and a Thermo Scientific Qubit 2.0 fluorometer. Finally, the library was sequenced on an Illumina HiSeq 2500 platform (Illumina) by Novogene Bioinformatics Technology Co., Ltd. (Tianjin, China) to generate 250-bp paired-end reads.

According to specific barcodes, paired-end reads were assigned to samples and merged with FLASH (V1.2.7) to gain raw tags. These raw tags were spliced (flash version: 1.2.11, parameters: --min-overlap) and filtered to obtain clean tags (Table S2). Then, operational taxonomic units (OTUs) were aligned, and species classifications were assigned to the taxonomic classification at 97% similarity threshold in terms of the Silva Database. The OTU (e.g., ternary phase diagram) and Venn diagram analyses were conducted based on the species richness and information about common and particular OTUs. α diversity values (e.g., redundancy analysis or MetaStat) based on the weighted UniFrac distance or the Spearman correlation were calculated by the QIIME software.

2.5.2. Microbial Data Analysis

The analysis process was as follows: The raw data obtained by sequencing contain a certain proportion of interference data (dirty data). To make the results of information analysis more accurate and reliable, the raw data were first stitched and filtered to obtain effective data (Clean Data). Then, based on valid data, OTUs (operational taxonomic units) clustering and species classification analysis were performed using Uparse software (Uparse v7.0. 1001 http://drive5.com/uparse/). Sequences with ≥97% similarity were assigned to the same OTUs. A representative sequence for each OTU was screened for further annotation. For species annotation: 16S: For each representative sequence, the Silva Database (http://www.arb-silva.de/) was used based on the Mothur algorithm to annotate taxonomic information. OTUs were analyzed for abundance, α diversity calculations, Venn charts, and petal charts to obtain information on species richness and evenness within samples, as well as information on shared and unique OTUs between different samples or groups. To further explore community structure differences among group samples, MetagenomeSeq statistical analysis was selected to test the significance of differences in species composition and community structure of group samples. The amplicon annotation results were linked with corresponding functional databases. FAPROTAX was used for functional prediction and analysis of microbial communities in ecological samples. FAPROTAX through python plays a great role in elucidating the possible biochemical processes and elements in play. Details on software used were shown in the supplementary materials (Table S3).

2.6. Statistical Data Analysis

The means ± standard deviations were used to express the examined data. The data were subjected to a one-way analysis of variance (ANOVA) test using SPSS software (version 18.0, U.S.A.). Normality (Kolmogorov–Smirnov and Shapiro–Wilk) and homogeneity of variance (Levene’s test) were first tested (Table S5) for nonparametric data, and Kruskal–Wallis tests followed by Duncan’s post hoc test with Bonferroni correction were used. The significant differences among various treatments were analyzed using Duncan’s multirange test (P < 0.05). Origin software (version 8.0, U.S.A.) was used to create graphs, and Inkscape was used for editing.

3. Results and Discussion

3.1. Effects on Physical and Chemical Properties at Tillering and Harvesting Stages

The soil pH (Figure S2A,B) of each treatment indicated high alkalinity, which was more prominent in the bulk soil than in the rhizosphere region at both the tillering and harvesting stages. Interestingly, the values of pH increased with increasing size and concentration of PS and the concentration of AZM. CK exhibited lower pH values compared to other treatments but remained elevated compared to the original soil (OS). The pattern for soil pH variations was as follows: CK < PS (single effect) < AZM (single effect) < PS+AZM (double effect), suggesting that AZM inhibited the negative effects of PS on soil pH. Rassaei (2023) confirmed that the effect of PS on soil pH depended on various factors, including such as environmental conditions, coexisting contaminants, and PS size and concentrations.

The rhizosphere showed lower pH values compared to bulk soil. This pH drop can likely be attributed to the acidifying effects of root and microbial activity in the rhizosphere. Plant roots release organic acids (such as citric and malic acids) and protons (H⁺) during nutrient intake, causing the rhizosphere to become acidic. Another reason is that PS might release some low-molecular-weight compounds during degradation or additives that might lower pH either by decreasing hydroxyl (OH^–) ions or increasing hydronium (H₃O⁺) ions in the rhizosphere compared to the bulk soil. The co-occurrence of PS and AZM showed higher pH values compared to the single effect of PS or AZM alone. Herein, AZM can suppress some microbial degraders of PS, reducing acid-producing metabolic pathways. The remaining microbial community may shift toward less acid-producing taxa, leading to a net pH increase.

In contrast, the soil EC exhibited an opposite pattern to the soil pH, which was as follows: CK < PS+AZM (double effect) < AZM < PS. This pattern was maintained for all treatments at both tillering and harvesting stages (Figure S2C,D). The combination of PS+AZM reduces soil EC primarily because AZM inhibits microbial activity, limiting PS degradation and ion release. Nutrient cycling (e.g., nitrification) is disrupted, reducing soluble ion accumulation. PS may bind ions, while AZM’s adsorption onto PS reduces its ionic contribution.

3.2. Biological Analysis

In this study, the high-throughput sequencing of 93 samples revealed 3,930,073 effective tags, with rhizosphere (45 samples), bulk soil (45 samples), and original soil (3 samples) accounting for 1,905,995, 1,908,939, and 115,139, respectively. The average length of the 93 samples was 373 bp. Additionally, the goods coverage estimators were from 97.99 to 98.86%, and the observed species ranged from 1432 to 3066. Therefore, it was demonstrated that the size of the gene libraries had a high degree of coverage for the bacterial community.

3.2.1. Effects on Community Richness and Diversity Indices

Regarding community richness and diversity using Chao1 and ACE (richness: Figure A,C), Shannon and Simpson (diversity: Figure B,D) were used. The higher the Shannon and Simpson indices, the greater the bacterial diversity. The general pattern was as follows: PS > CK > AZM > PS+AZM; and bulk soil > rhizosphere, except in CK and AZM groups alone. This suggests that there is greater biodiversity, stability, and complexity in single-effect groups than in double-effect groups. The co-occurrence of PS+AZM had inhibitory effects on bacterial diversity in the rhizosphere more than in bulk soil. The same trend was observed in ACE and Simpson indices (Figure C,D). It is obvious that lower bacterial richness and diversity are strongly related to acidic conditions, whereas alkaline conditions are associated with high values. These results are consistent with those of a previously conducted study by Gao et al. in wheat rhizosphere. First, this proved that soil acidity plays a crucial role in determining the fate and behavior of pollutants in soil ecosystems. Second, because of their accommodating properties, PS may serve as substrates for microbial colonization. They grow as a biofilm on the surfaces and protect the microbes from AZM actions, as previously shown. Briefly, richness and diversity were suppressed by the combined effects of PS+AZM, compared to individual applications, and the rhizosphere was at risk compared to bulk soil.

1.

α diversity indices. (A) Chao1; (B) Shannon index; (C) Ace; (D) Simpson. The values are the means of three replicates. Different letters indicate significant differences among treatments by ANOVA (Duncan test at p < 0.05).

3.2.2. Effects on the Variability of Bacterial Communities at Phylum and Genus Levels

The top six bacterial phyla accounted for >80% of the overall bacterial population across all treatments, with T2 and T5 exceeding 85%. Proteobacteria was the most abundant phylum. At the phylum level (Figure A), slight differences were observed between rhizosphere and bulk soil. Rhizosphere enriched Proteobacteria (+10.3%), Bacteroidota (+1%), and Chloroflexi (+5%), whereas bulk soil favored Acidobacteriota (+4.3%), Gemmatimonadota (+21.5%), and Actinobacteriota (+20.3%). To further characterize differences, each of the phyla was subsequently analyzed at the genus level (Figure A–H). Previous studies have linked Proteobacteria, Bacteroidetes, and Firmicutes with antibiotic resistance genes (ARGs), supporting the role of PS in AZM enrichment and ARG-carrying bacterial movement. Notably, several Proteobacteria genera strongly correlated with Chloroflexi and Bacteroidota (Figure B). For example, Ramlibacter was significantly correlated with Flavisolibacter and OLB13. This suggests a synergistic relationship in the rhizosphere (Figure S3A). In both rhizosphere and bulk soil, Proteobacteria abundance followed the same trend: single PS (decrease of 6% and 24%, respectively) < CK < single AZM (increase of 26.7% and 20%) < double treatments (increase of 30% and 30.5%). Higher abundances in the rhizosphere than in bulk soil are consistent with Proteobacteria’s preference for nutrient-rich conditions: root exudates supply labile carbon and energy sources, while the bulk soil has lower organic carbon and slower decomposition. Similar results were recently observed in soybean rhizosphere under PS and phenanthrene co-contamination, where the relative abundances of Proteobacteria were inhibited.

2.

Relative abundances of microorganisms in the rhizosphere and bulk soil at different levels. (A) Phylum level; (B) genus level; and the top 6 phyla accounting for >85% and their dominant genera distributions within (C) Proteobacteria; (D) Acidobacteriota; (E) Bacteroidota; (F) Gemmatimonadota; (G) Actinobacteriota; (H) Chloroflexi. Note: The values are means of three replicates, and error bars are added.

In Acidobacteriota (Figure D), it is obvious that the growth of subgroup_14 and RB41 was suppressed by single-effect AZM and by the co-occurrence of PS+AZM; however, they were less affected by the single application of PS. As a member of Proteobacteria, RB41 is known for its involvement in nitrogen cycling and may be sensitive to antibiotics that disrupt bacterial protein synthesis or DNA replication. On the other hand, subgroup_14 is often associated with oligotrophic conditions, making it vulnerable to antibiotic-induced stress, which can alter soil pH and nutrient availability. We speculated that the low values are a result of AZM suppressing sensitive taxa, which is in line with Guo et al. Concerning Bacteroidota (Figure E), the rhizosphere affected its abundance. AZM alone played a vital role in its growth, leading to an increase of 167%. In contrast, PS alone showed a slight change, whereas their co-occurrence caused an increase of 106% compared to CK. Similarly, in the bulk soil, the trend was as follows: single-effect PS (decrease: 18%) < CK < double-effect group (increase: 108%) < single-effect AZM (increase: 191%).

Subsequent analysis at the genus level showed that the top 6 genera belonged to Proteobacteria (i.e., Ramlibacter; Delftia; Massillia; Polaromonas), Bacteroidota (i.e., WCB1-32), and Gemmatimonadota (i.e., Gemmatimonas), and they accounted for more than 85% of the total bacterial community (Figure B). Ramlibacter and Gemmatimonas were strongly associated (p < 0.001), as indicated by the Pearson correlation heatmap (Figure S3A,B) in the rhizosphere and bulk soil, highlighting their role in plant growth. In Proteobacteria (Figure C), a slight change was observed throughout the whole treatment compared to CK. Compared to CK, the co-occurrence of PS+AZM increased the genera of Ramlibacter, Delftia, Massillia, and Polaromonas, contrary to that of Delftia. Interestingly, the genera Arenimonas preferred bulk soil, whereas Delftia preferred the rhizosphere. As the dominant genus in Proteobacteria, Ramlibacter’s growth and activities were suppressed by PS application. The reason behind this could be the effects of PS on sensitive bacteria. Proteobacteria is known for its metabolic versatility and is very sensitive to variations in soil carbon inputs, consequently inhibiting the growth rate of bacterial communities.

Moreover, in Acidobacteriota (Figure D), the co-occurrence of PS and AZM and that of single-effect AZM contributed to the disappearance of the genus RB41. Furthermore, a remarkable reduction (86%) of subgroup_10 was observed compared to CK, and the decrease was pronounced in the PS and AZM groups.

In Chloroflexi (Figure H), high genus dominance was observed in the rhizosphere compared to the bulk soil. PS+AZM negatively affected the growth of the genera Anaerolinea and UTCFX1 (70% and 96% decrease), and yet that of OLB13 increased by 612% compared to that of CK. Interestingly, Anaerolinea growth was more pronounced in the rhizosphere than in bulk soil. Briefly, genera in Actinobacteria were shown to thrive in environments containing PS more than in those of AZM alone or those of PS+AZM. This result is consistent with that of Zhou et al., explaining that PS is not only a physical entity but also contributes to shaping ecological processes due to its properties.

3.2.3. Impact of Environmental Factors on Bacterial Communities: Redundancy Analysis

RDA showed that both rhizosphere and bulk soil show a strong gradient (RDA1) explaining most of the variation (∼74–76%) (Figure A,B), suggesting that the soil pH and soil EC at tillering and harvesting stages were dominant in shaping microbial communities in general. On the other hand, RDA2 explains additional variation, with bulk soil having a slightly higher contribution (18.33%) compared to rhizosphere (14.06%). Proteobacteria and Bacteriota were strongly correlated with soil pH at both the tillering and harvesting stages (Figure C,D). Again, the PS+AZM group was strongly correlated with Proteobacteria and Bacteriota in the rhizosphere, and they were positively correlated with pH. The higher cumulative variance in bulk soil (93.20%) compared to rhizosphere (90.46%) suggests that the environmental drivers in bulk soil may be more clearly defined or that microbial communities in bulk soil are less complex. Another aspect is that the rhizosphere, being more dynamic due to root exudates and plant-microbe interactions, might have additional unmeasured factors contributing to the above variation.

3.

Redundancy analysis (RDA) of the top 10 dominant bacteria versus environmental parameters in red arrows (i.e., soil pH and soil EC at tillering and harvesting stage) in rhizosphere (A) in bulk soil (B); genera in Proteobacteria (in blue arrows) versus environmental parameters in rhizosphere (C), and bulk soil (D). Various treatments are represented by red circles, respectively.

3.2.4. Prediction of Soil Microbial Functional and Key Biogeochemical Cycles: FAPROTAX

A predictive analysis based on FAPROTAX conducted on the top 35 most abundant genera in each treatment revealed that the major biogeochemical processes linked to nitrogen, carbon, sulfur, and iron cycles were abundantly present. The carbon cycle was dominant, yet the nitrogen cycle encompasses the majority of the functions. However, there were discrepancies in both bulk soil and rhizosphere (Figure A,B). It is evident that in bulk soil, the major carbon biogeochemical processes vis-à-vis Chemoheterotrophy; Aerobic-Chemoheterotrophy were abundant in the PS+AZM group and single-effect AZM application compared to the PS group and CK. On the other hand, alterations in the rhizosphere indicate that PS, AZM, and PS+AZM affected the microbial diversity and distribution, and those alterations could play, as predicted by FAPROTAX, major roles in deciding the biogeochemical cycles (i.e., nitrogen, carbon, and sulfur). ,,,

4.

Microbial function prediction analysis heatmap by FAPROTAX based on the top 35 most abundant genera in each treatment (n = 3 replicates): (A) rhizosphere and (B) bulk soil.

3.2.5. Co-Occurrence Networks: All Networks and a Summary Table

A variety of biogeochemical and ecological processes rely on the associations and interactions between microorganisms. The potential interactions in the rhizosphere and bulk soil (Figure A,B) and potential groups as influenced by the different sizes and concentrations of PS and AZM (Figure S4) were examined using the co-occurrence networks. Sparse InversE Covariance estimation for ecological association inference (SPIEC-EASI) was used to construct networks in order to examine and contrast co-occurrence patterns of OTUs and their interactions. Initially, we used occurrence frequency to simplify each OTU dataset (i.e., OTUs occurrence in at least 80% of each group). A total of 14,381 OTUs were present in each dataset across different groups; however, following simplification, each dataset ranged between 700 and 800 OTUs. The processed OTU dataset served as the basis for network analysis after the application of R software. Model selection was performed by means of the stability approach to regularization selection (StARS) with the application of a threshold for variability at 0.05%. Network visualization and editing were attained through Gephi (ver. 0.9.2). Microbial network research essentially reveals how certain species coexist in niches and interact with the environment. Additionally, it can determine which “keystone” taxa are most influential in a community. Bacteria with higher degrees were designated as keystone taxa (Table S4) for preserving the composition, structure, and function of bacterial communities.

5.

Regional influence on the co-occurrence network analysis of OTUs. Notes: (A) rhizosphere and (B) bulk soil. Node size is an indication of the degree rank at the phylum level. Lines connecting nodes indicate interactions (i.e., pink lines represent positive interactions, while black lines represent negative interactions). The abundances of the phyla in the network are shown by their percentages.

Keystone taxa in the rhizosphere: Hub taxa (i.e., degree range: 18–20) were Microvirga, Pseudorhodoplanes, Polaromonas, Hydrogenophaga, and Bauldia, and belonged to the phylum Proteobacteria. These results were in line with those presented in Figure C, where the rhizosphere favored the growth of Proteobacteria compared to the bulk soil. It is key to note that Pseudorhodoplanes and Microvirga are often found in the rhizosphere due to their involvement in nitrogen biogeochemical cycling. These findings suggest a high adaptability to root exudate-rich environments. It is key to note that adaptive microbes may become closely associated with each other in the rhizosphere in response to stress conditions and fulfill specific ecological functions.

Keystone taxa in the bulk soil: Hub taxa (i.e., degree range: 18-20) for bulk soil were Opitutus; Adhaeribacter; Blastococcus; Ilumatobacter; IS-44; Bryobacter; Subgroup_10; Gemmatimonas, and they belonged to Verrucomicrobiota; Bacteroidota; Actinobacteriota; Proteobacteria; Gemmatimonadota; and Acidobacteriota. It is evident that bulk soil is home to taxa adapted to low nutrients (i.e., oligotrophy), stress resistance (e.g., drought, low C), and decomposer lifestyles (i.e., those involved in the breakdown of complex organics rather than root exudates). Briefly, these findings show that the rhizosphere network is dominated by a narrow phylogenetic group (i.e., Proteobacteria) with involvement in specific metabolic functions (i.e., nitrogen cycling), while the bulk soil features a phylogenetically diverse set of taxa adapted to low nutrients and stress, thus highlighting the role of root exudates.

Analyzing single effects, in CK, hub taxa (degrees 11–12) belonged to Gemmatimonadota, Proteobacteria, Planctomycetota, and Acidobacteriota, whereas PS alone markedly increased the degree range (40–42) and enriched known PS-degrading genera (e.g., Bacillus, Streptomyces). Zhang et al. showed that, sometimes MPs in soil alter bacterial genera involved in their own biodegradation. On the other hand, AZM alone showed intermediate degree ranges (28–31) that further shifted with concentration (5 mg: 23–24; 20 mg: 23–29) toward different Proteobacteria, Bacteroidota, and Chloroflexi taxa. Overall, PS induced stronger keystone connectivity and recruited putative degraders, while AZM altered keystone composition in a dose-dependent manner (Table S4).

In the PS+AZM co-occurrence networks, hub taxa differed by size and concentration of PS, where smaller PS size (100 nm) yielded a lower degree range (20–23) with keystone taxa like Shinella, Lautropia, and Pirellula, whereas larger PS size (300 nm) increased connectivity (degree 23–28) and shifted keystone taxa toward Abditibacterium, Gemmatimonas, and Gaiella. Regarding PS concentration, 50 mg/kg PS (degree 23–24) favored Opitutus and Gemmata, while 500 mg/kg PS (degree 20–26) broadened the range and introduced Planifilum, Bradyrhizobium, and Altererythrobacter, indicating size- and dose-dependent effects as well on network structure (Table S4).

The co-occurrence network sheds light on linkages, organizations, niche structures, resource partitioning, and keystone species (influential taxa), providing a novel perspective on bacterial community associations and interactions. Bacteria were more adapted in the bulk soil compared to the rhizosphere, as shown by the high number of nodes and edges, with the number of connections increasing by 1.56% (Table ). Despite this adaptation, the rhizosphere favors positive interactions, as shown by the percentage of positive links compared to that of bulk soil. Many ecological and biochemical factors can explain these variations. For instance, despite being a tiny zone (1 cm width), the rhizosphere is a resource-rich environment enriched with root exudates (i.e., sugars, amino acids, organic acids), providing abundant carbon and energy sources. This promotes cross-feeding, where different bacteria cooperate to metabolize complex compounds, leading to more positive interactions. Another aspect is that high nutrient availability supports denser microbial populations, increasing the likelihood of facultative mutualism when conditions are favorable for growth; again, the rhizosphere is also characterized by reduced competition due to niche specialization. The rhizosphere offers microniches (e.g., root tips, mucilage layers), allowing bacteria to specialize rather than compete directly. Herein, plants are known to select bacteria that cooperate to enhance nutrient uptake in the rhizosphere. On the other hand, the bulk soil had an increase of 14.6% of negative connections; this is due to the fact that bulk soil is a resource-limited, leading to stronger competitive exclusion, which may result in more negative connections.

2. Characteristics of Co-Occurrence Networks Constructed for Bacteria in Figures A,B and S4 .

no	categories	low/high	TC	nodes	edges	PE	PE%	NE	NE%	Avg.dg	Av.W.dg	MD	avg.C.C	avg. P.L
Double effect	Size of PS (nm)	100 nm	T2-T3-T4-T5	732	5512	4348	78.88	1164	21.12	15.06	0.699	0.637	0.078	2.791
		300 nm	T6-T7-T8-T9	770	6184	4656	75.29	1528	24.71	16.062	0.677	0.662	0.077	2.749
	Content of PS (mg/kg)	50	T2-T3-T6-T7	767	6120	4729	77.27	1391	22.73	15.958	0.705	0.647	0.078	2.765
		500	T4-T5-T8-T9	754	6011	4637	77.14	1374	22.86	15.944	0.676	0.641	0.074	2.744
	Concentration of AZM (mg/kg)	5	T2-T4-T6-T8	727	5646	4320	76.51	1326	23.49	15.532	0.688	0.655	0.079	2.767
		20	T3-T5-T7-T9	781	6401	4980	77.8	1421	22.22	16.392	0.706	0.621	0.075	2.739
Single effect	PS	NA	T10-T11-T12-T13	1301	18190	11597	63.75	6593	36.25	27.963	0.573	0.762	0.064	2.551
	AZM	NA	T14-T15	866	7464	5192	69.56	2272	30.44	17.238	0.666	0.845	0.107	2.792
NA	By location	NA	Bulk soil	525	3523	2968	84.25	555	15.75	13.421	0.679	0.624	0.102	2.881
		NA	Rhizosphere	524	3469	2985	86.05	484	13.95	13.24	0.733	0.612	0.104	2.891
No effect	NA	NA	CK	1029	3904	2341	59.96	1563	40.04	7.588	0.007	2.365	0.176	4.759

^aNote: Symbols and abbreviations are as follows: PS: polystyrene nanoplastics; AZM: azithromycin antibiotic; []: concentration; NA: non-applicable; TC: treatment combinations; T1–T15: treatment groups 1 up to 15; CK: control group with no addition; PE: positive edges; PE%: positive edge percentage; NE: negative edges; NE%: negative edge percentage; Avg. dg: average degree; Avg. W. dg: average weighted degree; MD: modularity class; Avg. C. C: average clustering coefficient; and avg. P. L: average path length.

On the other hand, CK revealed higher adaptability of bacteria compared to single (i.e., PS or AZM) and double effects (i.e., PS and AZM) as follows: Double effect < Single effect < No effects (Table ). The co-occurrence of PS and AZM resulted in the amplification of their impact, which can be seen in prolonged toxicity to microbes. This is associated with the fact that when MPs and antibiotics co-occur, MPs act as vectors of antibiotics by adsorbing them onto their surface via various interactions (i.e., hydrophobic interactions), concentrating them in hotspots and fostering antibiotic resistance. −

It is apparent that the application of PS and AZM alone resulted in densely populated groups and connected into closely related species with greater interactions, whereas PS+AZM was characterized by low interactions and populated with distant species (Figure S4, Table ). Even though CK was densely populated, it displayed low positive interactions compared to the single and double applications (co-occurrence), and this suggests that the competitive or inhibitory relationships between microbial taxa are weak or infrequent in this group. Bacterial interaction networks depend on nutrient abundance (i.e., food, oxygen, and space), can be antagonism (i.e., one microbe inhibits another), or exclusion (i.e., one taxon’s presence strongly correlates with another’s absence). However, in terms of a greater average degree, the microbial network in the bulk soil was more compact than that in the rhizosphere (Table ). This suggests that the complexity of the bulk soil may affect the ability of microbes to cope with complex environments. Overall, the above-mentioned results provided more useful insights into the co-contamination of PS and AZM in the rhizosphere and bulk soil of the soil, emphasizing their synergistic and/or antagonistic relationship, which is helpful for future mitigation measures.

4. Conclusion

In this study, we investigated the effects of emerging contaminants (i.e., PS nanoplastics and AZM macrolide antibiotic) on microbial shifts in wheat rhizosphere using a pot experiment; findings demonstrate that microbial adaptation was greater in the bulk soil than in the rhizosphere. This is contrary to the existing knowledge, which suggests that bulk soils have a low nutrient content due to the absence of root exudates. The effects brought by the application of PS, AZM, and PS+AZM affected the microbial communities and played major roles in determining their abundances and interactions. There were low interactions in PS+AZM treatments than in individual PS or AZM. Shifts in microbial communities were related to PS characteristics (i.e., size and concentration), AZM (i.e., concentration), and plant species. In addition, the interaction between PS and AZM was strongly correlated with the soil pH and soil EC. The rhizosphere was selected for fast-growing, copiotrophic bacteria that exploit root exudates and interact with plants, while bulk soil favored slow-growing, stress-tolerant microbes adapted to nutrient limitation and complex organic matter degradation. Moreover, findings revealed that the keystone taxa in the rhizosphere were Microvirga; Pseudorhodoplanes; Polaromonas; Hydrogenophaga; and Bauldia, while in the bulk soil, they were Opitutus; Adhaeribacter; Blastococcus; Ilumatobacter; IS-44; Bryobacter; Subgroup_10; and Gemmatimonas. The key contributors to these fluctuations are the impacts produced by single (PS or AZM) and combined effects (PS+AZM), which may influence root exudates and, hence, lead to alterations in the rhizosphere and bulk soil microbiota. For future prospects, further studies will be required to elucidate how this combination affects the spread of ARGs as well as the above-ground matters. Again, the incorporation of biochar merits investigation in combination, as biochar has been proven to alleviate negative effects of PS , in soil due to its properties. In conclusion, these results shed light on microbial interactions between the rhizosphere and bulk soil, between single (PS or AZM) and double (PS+AZM) application, and how the environment is prone to the spread of ARGs.

5. Environmental Implications

As emerging contaminants, MPs and antibiotics have raised concerns because of their environmental implications for agricultural soil health and crop production. Their co-contamination often leads to new types of soil pollution. This study focused on how combined (PS+AZM) and individual PS or AZM affect microbial shifts (i.e., abundances, composition, structure, interactions, functional predictions) in the rhizosphere under wheat growth (i.e., from germination to harvest stage). The findings revealed the co-occurrence of antibiotics and MPs within the rhizosphere and bulk soil in bacterial networks and predicted the possible functions among bacteria. Results may help in developing mitigation measures against the effects of MPs and antibiotics in agroecosystems.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c06521.

• Rhizobox construction diagram (Figure S1); soil pH and EC at various growth stages (tillering and harvesting) (Figure S2); Pearson correlation analysis of the top 35 genera in the rhizosphere and bulk soil (Figure S3); co-occurrence network analysis of OTUs under different treatments (Figure S4); barcoding information and primer sequences (Table S1); data processing statistical results and quality control (Table S2); list of analyses and software used (Table S3); keystone taxa based on degree centrality under different groups (Table S4); and normality test results using SPSS (Table S5) (PDF)

I.E.R.: Conceptualization, writingoriginal draft. M.G.: Funding acquisition, resources, formal analysis, writingreview and editing. J.Z.: Formal analysis, review and editing; Y.D.: Formal analysis, review & editing. Z.S.: Project administration, writingreview and editing.

The authors declare no competing financial interest.