Microplastic characteristics differentially influence cyanobacterial harmful algal bloom microbial community membership, growth, and toxin production

Abstract

Terrestrial runoffs contribute to cyanobacterial harmful algal blooms (cHABs) by providing nutrients and other pollutants that may facilitate cyanobacterial growth. Microplastics (MPs) are being detected at increasing concentrations in various aquatic systems worldwide, including freshwater, yet the MP effects on cHAB formation, toxin production, and transport are largely unknown. We used the statistical design of experiments to elucidate microbe–plastic interactions with freshwater algal bloom communities obtained from a HAB event in the Great Lakes. These experiments measured the impact of differing sizes, concentrations, and UV aging times of polyethylene, polypropylene, and cellulose fibers on the chlorophyll-a content of Trichormus (previously Anabaena variabilis) and Microcystis aeruginosa and microcystin-LR content in M. aeruginosa. Additionally, we conducted metagenomic sequencing on the total community and 16S rRNA microbial community sequencing on members of the total community bound to plastics after 4 weeks of culturing. The results indicate that M. aeruginosa growth rate was inhibited in the presence of polymers, while production of microcystin-LR generally increased in the presence of MPs. Changes to growth of T. variabilis varied with polymer type, size, and UV aging time. These results suggest that specific MP characteristics, not just their presence, may influence the toxicity, growth, and dispersal of cHABs across aquatic systems.

harmful algal blooms, microplastics, microcystins, microbial communities

Sustainability Statement

As harmful algal blooms continue to increase in frequency and intensity, it is important to understand the role and impact of various anthropogenic pollutants in causing this. Studying how specific human-made products are interacting with microbial communities in the environment will enable us to better address how humans affect aquatic health and how the accumulation of waste will drive environmental change in the future. In determining how plastic interacts with algal bloom communities, this project is linked to Clean Water, Responsible Consumption and Production, and Life Below Water (SD6, SD12, and SD14).

Introduction

Increases in agricultural and urban runoff contribute to cyanobacterial harmful algal bloom (cHAB) formation in aquatic systems by providing essential nutrients and pollutants to stimulate growth of HAB-causing microbial species (Steffen et al. 2014). These cHAB events occur globally in freshwater and marine systems of all sizes and can significantly negatively impact water quality, fisheries, and even drinking water (US National Office for Harmful Algal Blooms). This is largely due to health impacts of cyanotoxins produced during these blooms, frequently from the class of hepatotoxins known as microcystins. Microcystins have hundreds of congeners, the most common being microcystin-LR, which is produced by a wide variety of cyanobacteria, including those belonging to the genus Microcystis. In recent years, cHABs appear to be occurring with increasing intensity and frequency, though a recent study has suggested these global trends may be due to an improved ability to detect and quantify HABs, rather than a global shift in their presence (Hallegraeff et al. 2021). Nevertheless, widespread eutrophication in tandem with global warming has provided toxin-producing cyanobacteria with improved conditions for rapid growth (Xiao et al. 2019).

Anthropogenic influences have led to an influx of nitrogen and phosphorus to aquatic systems from groundwater, sediments, or urban and rural runoff, commonly referred to as cultural eutrophication (Malone and Newton 2020). Overabundance of these nutrients has been found to increase the frequency and intensity of HAB formation. Eutrophication has also been found to preferentially increase the growth of toxin-producing cyanobacteria during algal blooms, as well as influence the toxicity of these blooms by increasing the rate at which toxins are released (Davis et al. 2010). Studies have widely focused on factors such as light availability, temperature, and nitrogen and phosphorus concentrations that influence the proliferation of toxin-producing cyanobacteria and the impacts of these factors on toxin production via regulation of the microcystin synthase gene cluster (Sevilla et al. 2010). While the impact of these factors on HAB formation and toxin production is well studied, it is suggested that a far wider variety of factors control the formation and ecology of these blooms (Yan et al. 2022). The impact of microplastics, which can also enter aquatic systems through similar sources as these key nutrients, is not well understood (Yokota et al. 2017).

Microplastics, characterized by a size <5 mm in diameter, have become ubiquitous in the environment, having been detected in water, soils, and even human blood (Sol et al. 2020). In large water bodies like lakes or oceans, microplastic inputs can be from fishing gear, litter or cargo spills, paint shedding from boats as well as wastewater treatment plant effluents, road and urban runoff, and agricultural runoff (Rossatto et al. 2023). Water containing microplastics from municipal sources often end up in drainage systems where they are processed in wastewater treatment plants (WWTPs). Microplastics account for roughly half of microparticles entering WWTPs, and while removal strategies have been found to recover a majority of microplastics that enter the system, roughly 10% may pass through into the final effluent (Bayo et al. 2020). Much of the 90% of microplastics that are removed during treatment end up in sewage sludge, with pathogenic materials and inorganic nutrients attached to their surface (McCormick et al. 2014). Ultimately, this sewage sludge ends up in landfills, and incinerators, or is applied to agricultural fields as fertilizer (Sol et al. 2020, Hou et al. 2021). During large storms or rain events, runoff from these agricultural lands can bring the nutrients essential to the proliferation of HABs alongside microplastics harboring additional nutrients (Rehm et al. 2021). While it is known that various environmental factors play a significant role in the formation and proliferation of HABs (Paerl et al. 2018), the impacts and influences of microplastics on HABs are not well understood. It is well documented that microplastics in various environments have distinct microbial communities on the plastic surface compared to the surrounding environment, commonly referred to as the “plastisphere” (Zettler et al. 2013). Microplastics have the potential to serve as a surface for the growth and transport of cyanobacteria and their associated toxins (Pestana et al. 2021), allowing for more rapid dispersal across the bodies of water (Desforges et al. 2014), but direct investigation of microplastic properties that may influence the formation of these plastisphere communities is largely unknown.

Of particular interest for HAB research is the Great Lakes, where HABs have caused severe shutdowns of drinking water in Lake Erie (Steffen et al. 2017) and have recently become a repeated occurrence in Lake Superior, which has historically had no observed HABs due to relatively low nutrient loading and colder temperatures. Occurrences of HABs in the Great Lakes are primarily driven by changes in nearshore agricultural management practices, wastewater discharge from sewage overflow, and extreme weather events, such as large quantities of rainfall in the spring followed by droughts in the summer, which facilitate pulses in nutrient loading to the nearshore (McLellan et al. 2018, Trainer et al. 2020, Fraker et al. 2023). Recent studies into influx and accumulation of such nutrients into the nearshore have focused on the role of tributary rivers, which have also been shown to carry microplastics into the Great Lakes system. As contaminants of emerging concern continue to accumulate across the Great Lakes, the accumulation of microplastics, small molecules (Al Harraq and Bharti 2022) bound to their surface, and the associated plastisphere creates the possibility for an amplified threat alongside the increase in HABs. While few studies have specifically investigated the plastisphere of HABs within the Great Lakes, studies in marine systems have found that HAB-associated assemblages are capable of being transported along microplastic surfaces (Masó et al. 2003).

Therefore, to determine the influences of microplastics on HABs in freshwater systems, we cultured two communities harvested from algal blooms in the Great Lakes, each with a single species of bloom-forming cyanobacteria and associated heterotrophic bacteria, in the presence of microplastic fibers of varying physiochemical properties including polymer type, size, concentration, and UV aging time guided by statistical design of experiments. We hypothesize that due to the different morphologies, the cyanobacteria would interact differently with microplastics. Additionally, we hypothesize that this impact would be greatest for petrochemical plastics, as these would be inaccessible to all members of the community for any form of biodegradation, while natural polymers, such as cellulose, may be accessible to cellulolytic bacteria in these freshwater communities. We further hypothesize that toxin concentration would increase over time mirroring growth as it varies across culture conditions. The findings from this study suggest plastic- and microbe-specific interactions and will allow for environmentally relevant understandings of how anthropogenic inputs affect aquatic systems. That understanding can be used to improve our ability to monitor and predict changes to environmental phenomena as a result of these inputs.

Methods

In order to test how varying polymer characteristics influenced the growth and toxin production of our cyanobacterial strains and their associated communities, we conducted two trials in which we either tested three different polymer types with varying physiochemical characteristics, referred to as the polymer trial, or two UV aging times of a single polymer type, referred to as the UV trial (Fig. 1). During the polymer trial, two petrochemical polymers (polyethylene and polypropylene) as well as a natural polymer (cellulose) were used in addition to varying polymer sizes (50 or 5000 µm) and concentrations (50 or 250 mg/l). During the UV trial, both virgin and artificially UV-aged polypropylene were used in addition to the varying sizes and concentrations. Each set of microplastics with distinct characteristics was added to cultures of our cyanobacterial communities, as outlined in Table 1. Samples were taken from each condition weekly in order to measure the impacts of polymer presence on the cyanobacteria and their associated communities.

Figure 1.

Design of experiments culture condition generation and experimental workflow.

Table 1.

Experimental conditions and abbreviations for design of experiments analysis.

Polymer type	Plastic size (µm)	Plastic concentration (mg/l)	Abbreviation
Polymer trial condition abbreviations
Polyethylene	50	50	PeSL
Polyethylene	50	250	PeSH
Polyethylene	5000	50	PeLL
Polyethylene	5000	250	PeLH
Polypropylene	50	50	PpSL
Polypropylene	50	250	PpSH
Polypropylene	5000	50	PpLL
Polypropylene	5000	250	PpLH
Cellulose	50	50	CeSL
Cellulose	50	250	CeSH
Cellulose	5000	50	CeLL
Cellulose	5000	250	CeLH
UV trial condition abbreviations
UV aging	Plastic size (µm)	Plastic concentration (mg/l)	Abbreviation
Non-aged	50	50	NSL
Non-aged	50	250	NSH
Non-aged	5000	50	NLL
Non-aged	5000	250	NLH
Aged	50	50	USL
Aged	50	250	USH
Aged	5000	50	ULL
Aged	5000	250	ULH

Microplastic preparation and verification

For both the polymer and UV trials, 200 m spools of polyethylene, polypropylene, and cellulosic fibers were ordered from Goodfellow (ET31-FB-000100, PP30-FB-000147, and AC32-FB-000117). In order to generate microplastic fibers, these spools were cut to 50 µm according to previously published methods (Cole 2016). Briefly, fiber spools were unraveled by coiling the strand around two screws fixed on a wooden block. The coil of plastic was then frozen into a block shape by coating in NEG-50 (Fisher Scientific 22-110-617) and freezing at −80°C for 15 min. This step was repeated until the fibers were completely encased. These frozen spools were then cut into four equal-sized blocks and frozen together in pairs using NEG-50 as a sealing agent at −80°C for 15 min. These paired and frozen blocks were mounted perpendicular to the blade on an ultracryotome (Cryostat Leica CM1950) and sliced in 50 µm increments to generate the smallest microplastics used in this study. Microplastics were also cut to 5 mm in length by freezing into paired blocks as described above, and then cut by hand using a ruler and scissors. Following the procedure in Cole et al. to generate microplastics, each of the polymer fibers was sputter coated with 5 nm of gold (Leica EM ACE600) before imaging via scanning electron microscopy (SEM; Zeiss GeminiSEM 450).

For the UV trial, microplastics were UV-aged by placing cut microplastic fibers on a Spectroline Ultraviolet Transilluminator (Model TVC-312R) for a period of 30 days, with occasional mixing to ensure even distribution of exposure to UV light at 312 nm wavelength. Following this 30-day period, the fibers were collected, rinsed with sterile water, dried, and analyzed via SEM for size verification and Fourier transform infrared spectroscopy (FTIR Nicolet iS50R with Pike IRIS diamond ATR) for confirmation of photooxidation.

Cyanobacterial microbial communities used and culture conditions

Cultures of two microbial communities were obtained by a request to the Boyer lab at SUNY ESF for use in this study. These cultures were originally obtained from algal blooms in Lake Erie by the Boyer Lab (Carmichael and Boyer 2016, Watson et al. 2016). Each community contains, and is dominated by, one of two species of cyanobacteria used in this study: Microcystis aeruginosa and Trichormus variabilis. These communities were selected due to M. aeruginosa having a single cellular morphology and toxin-producing capabilities, while the strain of T. variabilis used in this study is filamentous, floccular, and does not produce toxins, though some studies have found other strains of T. variabilis with toxin production capabilities (Yan et al. 2022).

Cultures were grown on the benchtop at room temperature in 12 h light–dark cycles under Life-GLO T-8 40 W bulbs at a distance of 30 cm in BG-11 media (Table S1). The pH of the media was adjusted to 7.5 by adding drops of 10 mM HCl as needed before autoclave sterilization. Media was then added to sterilized, 160 ml narrow-mouth glass milk dilution bottle (Corning 1367–160). Then, bottles were inoculated to 10% by volume with freshwater algal bloom communities containing either M. aeruginosa or T. variabilis. Cultures were agitated once daily by inversion and were randomly placed beneath the light after each agitation.

Design of experiments—batch culture trials and modeling

Design of experimental conditions

Experimental parameters were set, and statistical analysis using design of experiments was conducted using a custom design in JMP PRO 17 (version 17.0.0) in order to compare how growth and toxin production varied between experimental conditions. Prior to experimentation, the custom design of JMP was used to generate the list of experimental conditions that should be conducted in order to cover the sampling space and adequately assess the impact of each input variable on the output phenotypic responses (Fig. 1). In JMP, the responses were listed as total chlorophyll-a content and microcystin-LR content for both the polymer and UV trials. For input variables, it is necessary to provide ranges of values for each variable. We used the literature reports to find these ranges as follows.

For the polymer trial, polyethylene (Pe), polypropylene (Pp), or cellulose fibers (Ce) were selected as the polymer types, as these are among the most frequent polymer types found in the Great Lakes, and fibers were selected as the shape as they are among the most common forms of plastic found (Lenaker et al. 2019, Fuschi et al. 2022). For plastic size, 5000 µm was chosen for the largest size as it is the largest size of plastic considered to be a microplastic, and for the smallest plastic, we input 50 µm as particles smaller than 50 µm may be considered nanoplastics, and were denoted as small (S) or large (L), respectively (Hartmann et al. 2019). The concentration range of plastics was set from 5 to 250 mg/l, denoted as low (L) or high (H), corresponding to a range of microplastic concentrations found within the Great Lakes, though concentrations lower and greater than this have been noted (Yokota et al. 2017b). In the UV trial, only polypropylene was used, as out of the plastics we tested, it had the most visible damage after 30 days of UV aging. It was input as either its virgin, or non-aged, state (N) or exposed to UV light (U) for 30 days.

We then designated each variable as continuous or categorical. Size and concentration were each listed as continuous variables for both trials, while polymer type and UV aging were listed as categorical variables within their respective experiments. JMP PRO 17 then generated a list of test conditions for each of these experiments that are given in Table 1.

Experimental conditions for polymer and UV trials

Microbial communities containing either M. aeruginosa or T. variabilis were cultured in the presence of polymers for two distinct trials, aiming to determine the influence of either polymer types or UV aging time in addition to polymer size and concentration. All experimental conditions for both the polymer and UV trial, listed in Table 1, were run in triplicate. Cultures were grown in the same medium, bottles, volume, and light source as above but with the addition of plastics. Each week over 4 weeks, images were taken of the culture bottles, and 1 ml aliquots were taken to measure chlorophyll-a content via spectroscopy and toxin content via High Performance Liquid Chromatography (HPLC).

Experimental analysis and model evaluation

Following experimentation and collection of data, results were then analyzed within the software using standard least square analysis with an emphasis on effect screening. Three degrees of model effects were used in each study to look at interactions between size, concentration, and either polymer type or UV aging time in tandem. Each week was treated independently for analysis of variations between cultures. For each week, two-factor t-tests compared the mean chlorophyll-a or microcystin-LR content to controls in order to determine how the presence of each set of plastics influenced growth and toxin production. Additionally, JMP’s scaled estimate test determined the significance of each set of one, two, or three-factor variables on the recorded outcome via a t-test against the null hypothesis, as well as the direction and size (increase or decrease) of this influence compared to other experimental conditions. The accuracy of JMP’s DOE model estimates was predicted in actual by predicted plots prior to any additional analysis.

Microbial community membership

Metagenomic sequencing of inoculum cultures

In order to gain an understanding of membership, functional capabilities, and cyanotoxin biosynthesis within these communities, we conducted metagenomic sequencing of the initial cultures. Genomic DNA was extracted using Qiagen DNeasy Power Water Kit before being sent for processing at SeqCenter. The sample libraries were prepared using the Illumina DNA Prep kit and IDT 10 bp UDI indices, and sequenced on an Illumina NextSeq 2000, producing 2 × 151 bp reads. Demultiplexing, quality control, and adapter trimming were performed with bcl-convert (v3.9.3). After sequencing, raw sequencing data were processed in KBase following the protocol in Chivian et al. (2022). Reads were checked for quality using FastQC (v0.11.9), and Trimmomatic (v0.36) was used to filter low-quality reads. Reads were then aligned using Bowtie2 (v2.3.2) and assembled using metaSPAdes (v3.15.3). Contigs were then binned using FAS Tool (v1.1.2) before genome quality assessment and filtering using CheckM (v1.0.18). Microbes were then classified by bin using GTDB-TK (v1.7.0), and functional capabilities of each bin were determined using DRAM (v0.1.2).

16S rRNA amplicon sequencing of polymer-bound biofilms in UV trial

In order to identify which members of the microbial communities were forming biofilms on plastics following 4 weeks of culturing, we conducted 16S rRNA sequencing from the bound communities. For gDNA extraction, plastics were recovered from large aged and non-aged culture conditions by filtering onto a 0.22-µm PES membrane before extraction following the protocol in the Qiagen DNeasy PowerWater Kit (Qiagen 14900-100-NF). DNA amplification and sequencing followed the method of Chatman et al. (2024)​. Briefly, extracted gDNA was amplified using primers targeting the 16S rRNA gene, and the amplified genes were then verified using gel electrophoresis before being normalized to 20 µl using SequalPrep Normalization Kit (Life Technologies). Normalized products were then sequenced on an Illumina MiSeq to determine the composition of the microbial community (Dill-McFarland and Cox 2018)​. Sequencing results were then processed in Mothur (Schloss et al. 2009)​, and visual and statistical analysis was conducted in R.

Chlorophyll analysis

Chlorophyll-a concentration was determined using methanol extraction 2.1 in Zavřel et al. (2015). 1 ml of culture was collected each week during the 4-week experiment and immediately spun down for 5 min at 13 000 × g (Eppendorf Centrifuge 5424, Rotor FA-45–24-11) to recover cell pellets. Supernatant was decanted, and these pellets were resuspended in 1 ml of methanol (Fisher Chemical A412-4) that had been cooled to 4°C and vortexed to resuspend the pellet. The sample was then covered in aluminum foil and placed in a 4°C fridge for 20 min. Following cooling, the samples were centrifuged again for 5 min at 13 000 × g. The resulting supernatant contains the extracted pigment and was scanned in a ThermoScientific Genesys 30 Visible Spectrophotometer from 400 to 750 nm against a methanol blank. Chlorophyll-a concentration was then determined using Chl_a [µg/ml] = 12.9447 (A₆₆₅ − A₇₂₀) (Ritchie 2006). Chlorophyll-a content for each week was normalized to the amount of chlorophyll-a detected in the initial culture in week zero.

Toxin analysis

Each week over the 4 weeks of culturing, 1 ml of culture from each of the biological replicates was pooled for each condition and stored at –80°C before analysis. At the end of the 4 weeks, samples were thawed on ice and then ultrasonicated on a Branson Sonifier 450 for 2 min with duty cycle of 50% and output of 6 in order to lyse cells to obtain total microcystin content. Concentrations of microcystin-LR were then determined using high-performance liquid chromatography (Agilent 1260 with MWD detector). Samples were run in triplicate on a Porochell 120 EC C-18 Column (3 × 50 mm) at a temperature of 32°C and a flow rate of 1 ml/min. The mobile phase comprised of A: ultrapure water (Barnstead NanoPure Diamond) with 0.025% trifluoroacetic acid (TFA) (Sigma-Aldrich 302031-10X1ML) and B: acetonitrile (Fisher Scientific A998-4) with 0.025% TFA. A gradient elution was used to isolate the microcystin-LR peak using the following protocol: 0–1 min: 80%A, 20%B; 1–5 min: 80%A-50%A and 20%B-50%B; 5–7 min: 80%A, 20%B. Reference spectra were collected at 210 nm. A standard curve was generated using a microcystin-LR standards (Cayman Chemical 1007188) from 0.05 to 2.5 µg/ml in BG-11 culture media for comparison against culture extracts. Microcystin-LR content for each week was normalized to the amount of toxin detected in the initial culture in week zero.

Results

Microplastic size verification and characterization of changes from UV treatment

After preparing the microplastics from spools of commercial plastics, the 50 µm cut fibers of PE, PP, and cellulose were imaged with SEM to measure the size of the microplastics. Of the measured fibers from each of the three polymer types (n = 75), the longest fiber was 193.9 µm in length, while the shortest was measured at 28 µm. Cut fibers had an average length of 51.6 µm with a standard deviation of 14.09 µm, which matches the target length of 50 µm while still representing a distribution of sizes that would occur in an environmental setting.

To prepare polymer for the UV trial, following cutting polypropylene fibers to 50 µm on the cryotome, the fibers were exposed to 312 nm wavelength UV light, which mimics weathering common to environmental plastics (Binda et al. 2024). Following the exposure, SEM images were collected to compare the non-aged fibers (Fig. 2A) to UV-aged fibers (Fig. 2B). Visible cracks and breakages were noted on polypropylene fibers that were not present on the non-aged fibers. Additionally, FTIR-ATR analysis of polymer spectra before and after UV exposure indicated new peaks around 1731 and 1375 cm⁻¹, consistent with the formation of oxygenated functional groups (C = O and C-O), confirming photooxidation occurred (Fig. 2C) (Campanale et al. 2023).

Figure 2.

(A) SEM image of polypropylene fibers before exposure to UV light, (B) SEM image of polypropylene fibers after exposure to UV light, and (C) FTIR-ATR spectra of polypropylene fibers before and after exposure to UV light.

Metagenomics reveals that microbial community accompanying M. aeruginosa is highly similar to that of T. variabilis

While the primary cyanobacteria within these communities were previously classified as M. aeruginosa and Dolichospermum flosaquae, the other microbial community members had not been characterized. Following metagenomic sequencing, assembly, and binning of each of the two microbial communities, it was found that as expected each community was dominated by a specific cyanobacterium. The Lake Erie HAB-derived community with the single-cellular cyanobacterium was identified as M. aeruginosa, as was expected. However, the filamentous cyanobacterium that dominates in the second community was identified as T. variabilis, contrary to previous classifications of D. flosaquae and Anabaena.

The T. variabilis-associated community comprised eleven bins, and the M. aeruginosa-associated community comprised eight (Table 2). While it was found that the cyanobacterial presence was distinct in each of these communities, the rest of the community was nearly identical between the two cultures, with the seven non-cyanobacterial bins present in the M. aeruginosa community being present in the T. variabilis community. Three additional unique bins were identified within the T. variabilis community corresponding to members of the genera Hydrogenophaga and Lacibacter, as well as a member of the family SG8-39. Annotation of genomic capabilities within these communities confirmed that the cyanobacteria were the only chlorophyll-containing microorganisms present, and toxin-producing capabilities were found solely within our M. aeruginosa genome, though, unexpectedly, not within our T. variabilis genome. Additionally, functional predictions were conducted for members identified from metagenomic sequencing. From this functional prediction, microcystin degradation genes, mlrA and mlrC, were detected in members of the genera Blastomonas, Hydrogenophaga, Phreatobacter, and Microbacterium, though Hydrogenophaga was found only in association with T. variabilis communities, where toxins are not present. Bins associated with these communities were also found to contain genes responsible for both nitrogen fixation and denitrification. The presence of these functions within the heterotrophic members is important to nutrient cycling in the community and possibly in responding to inputs from runoff.

Table 2.

Taxanomic classification, identified using GTDB-Tk, of bins in each microbial community.

T. variabilis community bins	Classification
Bin.001	d__Bacteria;p__Proteobacteria;c__Alphaproteobacteria;o__Sphingomonadales;f__Sphingomonadaceae;g__Blastomonas;s__Blastomonas fulva
Bin.002	d__Bacteria;p__Proteobacteria;c__Alphaproteobacteria;o__Sphingomonadales;f__Sphingomonadaceae;g__Sphingopyxis;s__Sphingopyxis sp001468225
Bin.003	d__Bacteria;p__Proteobacteria;c__Gammaproteobacteria;o__Burkholderiales;f__Burkholderiaceae;g__Hydrogenophaga;s__
Bin.004	d__Bacteria;p__Proteobacteria;c__Alphaproteobacteria;o__Rhizobiales;f__Phreatobacteraceae;g__Phreatobacter;s__
Bin.005	d__Bacteria;p__Bacteroidota;c__Bacteroidia;o__NS11-12 g;f__UBA8524;g__;s__
Bin.006	d__Bacteria;p__Cyanobacteria;c__Cyanobacteriia;o__Cyanobacteriales;f__Nostocaceae;g__Trichormus;s__Trichormus variabilis
Bin.007	d__Bacteria;p__Proteobacteria;c__Gammaproteobacteria;o__Burkholderiales;f__SG8-39;g__;s__
Bin.008	d__Bacteria;p__Actinobacteriota;c__Actinomycetia;o__Actinomycetales;f__Microbacteriaceae;g__Microbacterium;s__
Bin.009	d__Bacteria;p__Bacteroidota;c__Bacteroidia;o__Chitinophagales;f__Chitinophagaceae;g__Lacibacter;s__
Bin.010	d__Bacteria;p__Proteobacteria;c__Gammaproteobacteria;o__Xanthomonadales;f__Xanthomonadaceae;g__Silanimonas;s__
Bin.011	d__Bacteria;p__Gemmatimonadota;c__Gemmatimonadetes;o__Gemmatimonadales;f__Gemmatimonadaceae;g__Gemmatimonas;s__
M. aeruginosa community bins	Classification
Bin.001	d__Bacteria;p__Proteobacteria;c__Alphaproteobacteria;o__Sphingomonadales;f__Sphingomonadaceae;g__Sphingopyxis;s__Sphingopyxis sp001468225
Bin.002	d__Bacteria;p__Proteobacteria;c__Alphaproteobacteria;o__Rhizobiales;f__Phreatobacteraceae;g__Phreatobacter;s__
Bin.003	d__Bacteria;p__Proteobacteria;c__Gammaproteobacteria;o__Xanthomonadales;f__Xanthomonadaceae;g__Silanimonas;s__
Bin.004	d__Bacteria;p__Actinobacteriota;c__Actinomycetia;o__Actinomycetales;f__Microbacteriaceae;g__Microbacterium;s__Microbacterium sp006715565
Bin.005	d__Bacteria;p__Cyanobacteria;c__Cyanobacteriia;o__Cyanobacteriales;f__Microcystaceae;g__Microcystis;s__Microcystis panniformis
Bin.006	d__Bacteria;p__Bacteroidota;c__Bacteroidia;o__NS11-12 g;f__UBA8524;g__;s__
Bin.007	d__Bacteria;p__Proteobacteria;c__Alphaproteobacteria;o__Sphingomonadales;f__Sphingomonadaceae;g__Blastomonas;s__Blastomonas fulva
Bin.008	d__Bacteria;p__Gemmatimonadota;c__Gemmatimonadetes;o__Gemmatimonadales;f__Gemmatimonadaceae;g__Gemmatimonas;s__

Certain polymer characteristics influence cyanobacterial growth in a species-specific manner

After 4 weeks of culturing, the average amount of chlorophyll-a measured in each polymer-containing condition for M. aeruginosa was compared to the control condition via a two-way t-test (Table S2). In each community, average chlorophyll-a content decreased compared to controls, regardless of polymer type, size, or concentration in the polymer (Fig. 3A, B). While the mean chlorophyll-a content for M. aeruginosa cultures after 4 weeks was 1.16 µg/ml, cultures containing microplastics, regardless of physiochemical characteristics, did not reach concentrations above 1.0 µg/ml (PE 0.53 µg/ml, PP 0.56 µg/ml, CE 0.75 µg/ml). Significant decreases were measured in 11 of 12 polymer conditions (P < .05) when compared against controls. Cultures with polyethylene at the small size and high concentration (MaPeSH) exhibited a decrease in average chlorophyll-a but was not significant (P = .085). While in the UV trial, the inverse was observed as the presence of UV-aged plastics, unexpectedly, did not vary chlorophyll-a content relative to controls without polymers present. 11 of 12 of the UV-aged conditions had no significant change in average chlorophyll content after 4 weeks. Only one condition, non-aged polypropylene at the large size and low concentration (MaNLL) resulted in a significant increase in chlorophyll-a content.

Figure 3.

Average chlorophyll-chlorophyll-a concentrations of M. aeruginosa polymer trial (A), UV trial (B), and T. variabilis polymer trial (C), and UV trial (D) during 4 weeks of culturing in the presence of plastics.

By contrast, chlorophyll-a concentrations in the T. variabilis culture varied depending on polymer type and properties when compared directly to controls (Table S3). Control cultures had an average chlorophyll concentration of 1.20 µg/ml after 4 weeks, which was higher than the averages for each of the three polymer conditions (PE 1.15 µg/ml; PP 1.02 µg/ml; CE 0.91 µg/ml). Average chlorophyll-a concentrations were significantly higher after 4 weeks for condition with small polyethylene fibers and small polypropylene fibers at the low concentrations (TvPeSL, TvPpSL) (P = .028, P = .025), and measurably higher for small polypropylene fibers at the higher concentration (TvPpSH), though not statistically significant (P = .062). However, average chlorophyll-a concentrations were significantly lower when in the presence of large fibers (P < .05), regardless of polymer type, with the exception of polyethylene at the large size and lower concentration (TvPeLL) where no significant difference was found (P = .135). No significant change was seen in chlorophyll-a concentrations of T. variabilis when cultured in the presence of small cellulosic fibers (TvCeSL, TvCeSH) (P = .346, P = .205). In the UV trial, no conditions had increased concentrations of chlorophyll-a when compared to controls, while conditions with large UV-aged fibers saw significant decreases in chlorophyll-a content (Fig. 3C, D).

In using JMP’s design of experiments feature to determine the influence of specific polymer characteristics, a plot comparing predicted to measured (actual) values was generated (Fig. 4). Outputs of actual by predicted plots allowed for assessment of JMP DoE model accuracy as well as the confidence of the software to determine variable influences on the growth of our cyanobacterial cultures. While the results of the root mean square error analysis of the experiments containing M. aeruginosa experiments did not show statistical significance (polymer trial: P = .6572, R² = 0.20; UV trial: P = .5321, R² = 0.30) indicated by the 95% confidence interval containing the null hypothesis (Fig. 4A, B), the results of the experiments containing the T. variabilis community resulted in a meaningful and statistically significant (polymer trial: P < .0001, R² = 0.91; UV trial: P = .0029, R² = 0.67) model to predict changes in chlorophyll-a content (Fig. 4C, D). Despite not being able to accurately predict outcomes based upon the generated model for M. aeruginosa, our analysis did find meaningful changes in chlorophyll-a content for both T. variabilis and M. aeruginosa, varying with specific polymer characteristics, some matching our predictions and some contrary, though JMP DoE model outputs for M. aeruginosa should be analyzed with caution going forward.

Figure 4.

Actual by predicted value plots generated by statistical design of experiment outcomes after 4 weeks using JMP for (A) M. aeruginosa polymer trial, (B) M. aeruginosa UV trial, (C) T. variabilis polymer trial, and (D) T. variabilis UV trial. Horizontal line represents the null hypothesis, while the red line indicates the alternative hypothesis. Shaded region depicts the 95% confidence interval.

After 4 weeks, chlorophyll-a content between experimental conditions of the M. aeruginosa cultures in the polymer trial was analyzed with standard least square analysis feature in JMP. Each set of polymer conditions was tested within the DoE software to determine significance and scale of the properties effect. It was determined that polymer type and size, both independently and as a two-factor variable, drove differences between the groups (Table S4). Over the course of the 4-week polymer trial, the presence of polyethylene was a significant driver of decreased chlorophyll-a content for the first 3 weeks (P < .001, P = .004, P < .001) when compared to the presence of both polypropylene or cellulose, which each drove increased chlorophyll-a content for two and three of the four weeks, respectively. Additionally, over the course of the first 3 weeks, the size of plastics significantly contributed to changes in chlorophyll-a content (P < .001, P = .001, P = .029), while concentration of plastics did not (P = .501, P = .204, P = .452). Two-factor variables containing polymer type and size together were found to influence chlorophyll-a content, with large polypropylene fibers driving increases in chlorophyll-a content compared to polyethylene and cellulose in weeks two and three, though no significance was present for any of these conditions in week four. Larger polyethylene fibers contributed to increased chlorophyll-a content in week one (P = .001), and lower content in week two (P = .044) but was no longer significant in weeks three and four (P > .10). Polypropylene exhibited an opposite trend to polyethylene, with a slight decrease in chlorophyll-a in week one (P = .020) and a significant increase in weeks two (P < .001) and three (P = .001). Cellulosic fiber size drove a decrease in average chlorophyll content compared to other polymers across all 4 weeks, with significant decreases noted in weeks two and three (P = .044, P = .013). No significance was found on the influence of multifactor variables containing polymer type, size, and concentration together. After 4 weeks in the UV trial, variations in mean chlorophyll concentrations between controls (0.52 µg/ml) and microplastic-containing cultures followed different trends, with the average chlorophyll-a concentrations in non-aged polymer trials reaching 0.69 µg/ml and cultures containing UV-aged fibers reaching 0.56 µg/ml. No single-factor variables were identified as driving significant changes in chlorophyll-a content between conditions (P > .05). Despite this, the two-factor variable of size and UV aging was significant in the first 3 weeks (P = .024, P = .008, P = .027), with smaller non-aged fibers having higher average chlorophyll-a content than small UV-aged polymers, and large UV-aged polymers having higher average chlorophyll-a content than large non-aged polymers.

For the T. variabilis cultures polymer trial, cultures were found to have different driving factors for chlorophyll-a content over the course of the 4 weeks (Table S5). The presence of polyethylene fibers drove significantly increased chlorophyll-a concentrations in weeks one, three, and four (P = .007, P = .040, P = .003), and the presence of cellulosic fibers drove an increase in week two (P = .002). Significantly decreased chlorophyll-a contents were driven by the presence of polypropylene in weeks two and three (P = .005, P = .010) and by cellulose in weeks one and four (P = .035, P = .005). Once again, concentrations of plastics alone were not a significant driving factor during any of the 4 weeks (P > 0.05), though polyethylene concentration drove a decrease in week one (P = .078), and cellulose concentration drove a significant increase in week two (P = .001). Polymer size only contributed to a significant decrease in growth in the third week of the trial (P = .012), while polypropylene size and cellulose size drove significant decreases during week two (P < .001, P < .001) and increases during week three (P = .003, P = .009) in chlorophyll-a concentrations. The only significant three-factor variable during the course of the polymer trial with T. variabilis cultures was polyethylene, size, and concentration during week two, which drove a significant decrease in growth (P = .010). In the UV trial, while the UV aging status alone only drove changes in average chlorophyll-a content in week four (controls 1.10 µg/ml; non-aged 0.99 µg/ml; UV-aged 0.69 µg/ml), the two-factor variable of size and UV aging of fibers was a significant driving factor from weeks two through four (P = .013, P = .008, P = .003), with large UV-aged fibers driving decreases in chlorophyll-a compared to the non-aged fibers. Additionally, the two-factor size and concentration variable were significant across all 4 weeks of the experiment (P < .05), suggesting that higher quantities of larger fibers drove decreases in chlorophyll content, though this was not seen in the polymer trials. Again, no significance was found on the influence of multifactor variables containing UV aging, size, and concentration together.

In using JMP’s DOE feature, predictions were generated on chlorophyll-a responses to changes in two-factor variables after 4 weeks of culturing. In analyzing predicted influence of two-factor variables within the T. variabilis cultures, a consistent response of decreasing chlorophyll content was seen for all polymer types with increasing fiber size (Fig. 5A). While no significant trend was noted in the relationship between either size and concentration or polymer type and concentration, increasing concentrations of polyethylene led to decreased average chlorophyll content, while increasing concentrations of cellulose led to slight increases. In the UV trial, the chlorophyll-a response to both size and concentration in relation to UV-aged and non-aged fibers showed opposite trends in relation to the other, with increasing size and concertation of non-aged fibers causing slight increases, while increasing size and concentration of UV-aged fibers led to decreases in chlorophyll-a content (Fig. 5B). No two-factor trends were analyzed for M. aeruginosa cultures due to the lack of fit for the predictive modeler generated in the DOE analysis.

Figure 5.

Predicted two-factor influences on T. variabilis chlorophyll-a content in the polymer trial (A) and UV trial (B).

Presence of virgin plastics drove changes in Microcystis toxin production levels, but UV-aged plastics had no effect

We hypothesized that toxin concentration would increase over time mirroring growth. Concentration of cyanotoxin was measured by HPLC and then, to account for differences in observed Microcystis biomass in each biological replicate, the cyanotoxin values were normalized to the chlorophyll-a content. Across all of the culture conditions in both the polymer and UV trials, no toxins were detected in week zero nor week one (Fig. 6). While most conditions in both the polymer and UV trial saw toxin production begin in week two, toxin was not detected in cultures containing small polyethylene or small polypropylene until week three.

Figure 6.

Average ratio of microcystin-LR to chlorophyll-chlorophyll-a in M. aeruginosa polymer trial (A) and UV trial (B).

In comparing toxin content between experimental conditions and controls with a two-factor t-test within the polymer trial after 4 weeks (Table S6), it was found that, despite the consistent drop in average chlorophyll-a content, toxin production was significantly increased in most conditions (P < .05) (Fig. 6A). The conditions that did not follow the trend of increased toxin production were those containing small polyethylene fibers at both concentrations (MaPeSL, P = .214; MaPeSH, P = .290) and large polypropylene fibers at the low concentration (MaPpLL, P = .222) where no significant differences were found. Unlike the polymer trial conditions and contrary to our hypothesis, no significant differences were observed in average toxin content for any condition in the UV polymer trials by week four, though conditions with high concentrations of small UV-aged fibers (MaUSH) saw a significant increase in week two (P = .006), and conditions with low concentrations of large UV-aged fibers (MaULL) saw a significant decrease in week two (P = .034) (Fig. 6B).

When comparing between experimental conditions using the JMP’s DOE, no statistical significance was found in the predictive model to determine changes in microcystin-LR content, and therefore analysis of results should be considered under scrutiny (Table S7). Despite this, few factors were identified as significant driving factors in differences between polymer conditions. In week two, polypropylene presence was found to be the only factor driving significantly decreased toxin content (P = .045), and cellulose and polymer size were found to be the only factor driving significant increase (P = .013, P < .001). No significant factors were found in week three, and polyethylene presence was the only significant factor driving increases to toxin content in week four (P = .039). Similarly, for the UV trial, no significant factors were identified to drive any changes in toxin content across weeks two and three, and polymer size was found to be the only driving factor in week four (P = .040).

MP-bound communities

Following the UV trial, DNA was extracted from the polymers of each of the experimental conditions containing large fibers at high concentrations in order to identify members of the communities that had adhered. No sequencing occurred on cultures with small plastics as insufficient DNA was recovered from plastic-bound biomass to conduct sequencing, as was also the case with the T. variabilis culture containing large UV-aged plastics (TvULH). Sequencing resulted in a coverage >98.5% for each of the three communities. Microcystis aeruginosa accounted for approximately 15.5% of the relative abundance of the particle-bound community for the non-aged fibers, while T. variabilis accounted for 36.4% (Fig. 7). The phylum Bacteroidota also had a much higher relative abundance in the T. variabilis non-aged polypropylene attached community than in either M. aeruginosa community. In UV-aged polymer communities, the relative abundance of M. aeruginosa increased to 19.1%, a 1.61-fold change compared to the non-aged polymer community. For both M. aeruginosa- and T. variabilis-associated communities, many members identified by metagenomics of the inoculum were also observed in the attached community by 16S rRNA gene amplicon sequencing. In comparing relative abundances for communities bound to virgin and aged polypropylene, members of the genus Phreatobacter (Metagenomics Bin 4) showed an approximately 0.5-fold change increased abundance in the UV-aged communities, while Lacibacter (Metagenomics Bin 9) and Silanimonas genera both showed higher abundances in the non-aged communities. Microbial alpha diversity within these attached communities was highest for the M. aeruginosa community containing the UV-aged plastics, followed by T. variabilis with no-aged plastics and then Microcystis with non-aged plastics, as indicated by calculated Shannon diversity indices of 2.21, 1.91, and 1.73, respectively.

Figure 7.

Relative abundance of plastic-bound communities following the UV trial. Ma NLH—M. aeruginosa with non-aged, 5000 µm fibers and 250 mg/l; Ma ULH—M. aeruginosa with UV-aged, 5000 µm fibers and 250 mg/l; Tv NLH—T. variabilis with non-aged, 5000 µm fibers and 250 mg/l. Insufficient DNA for sequencing of the Tv ULH condition—T. variabilis with UV-aged, 5000 µm fibers and 250 mg/l.

Discussion

In analyzing metagenomic results of the communities of both the M. aeruginosa- and T. variabilis-associated communities, it was found that all heterotrophic bacteria detected in the M. aeruginosa-containing culture were present in the T. variabilis-containing culture, though 16S amplicon sequencing data showed a far more complex and diverse community within each. The shared members of these communities belong to the genera of Sphingopyxis, Phreatobacter, Silanimonas, Microbacterium, Blastomonas, and Gemmatimonas, as well as one member best classified by the family UBA8524, all of which have been previously identified in association with algal blooms, and many within Lake Erie, where these cultures originated (Zhang et al. 2016, 2017, Le et al. 2024, Li et al. 2024). Members unique to the T. variabilis culture were best classified by the genera of Hydrogenophaga, Lacibacter, and by the family of SG8-39. Despite Hydrogenophaga not being detected in metagenomic sequencing, 16S amplicon sequencing data showed its presence in both Microcystis-associated communities sequenced during the UV-aged trial at a relative abundance of 5% in the non-aged polymer community and 23% in the UV-aged polymer community. Lacibacter was detected in the M. aeruginosa-associated community containing UV-aged polymers, though in a relative abundance of 0.2%. Both Hydrogenophaga and Lacibacter were again identified as members of the T. variabilis-associated community following 16S sequencing. Hydrogenophaga has previously been found in association with cyanobacteria, including both T. variabilis and M. aeruginosa, as both can produce hydrogen gas as a fermentation byproduct and a substrate usable by Hydrogenophaga (Kim et al. 2019). Previous studies investigating drivers of shifts in HAB-associated communities have found changes in communities over the course of a bloom, largely driven by nutrient requirements and cycling capabilities, including iron reduction at the onset, and denitrification, sulfur oxidation, and toxin degradation in the later stages (Zhou et al. 2020). Though community composition was not determined over the course of the study, the communities associated with our cyanobacteria showed capabilities for denitrification and sulfur oxidation, and future studies should investigate the role of these cyanobacterial-associated communities in driving changes in both growth and toxin production in the presence of plastics identified in this study.

The presence of UV-aged versus non-aged polymers appears to have shifted community composition at the phyla level primarily in the relative abundances of Gemmatimonadota, Cyanobacteria, and Actinobacteriota. While cyanobacterial relative abundance increased in the UV-aged polymer conditions compared to its non-aged counterpart, average chlorophyll concentrations were higher in the non-aged polymer communities. This indicates that, despite the shift in community composition, cyanobacterial growth was still overall greater in the presence of non-aged polymers compared to UV-aged polymers. Future studies could aim to determine the role of the microbial communities and metabolisms in facilitating the growth of cyanobacteria and in characterizing drivers of shifting community compositions.

While laboratory batch culture experiments are not directly representative of environmental systems, microplastic concentrations across global aquatic systems continue to increase, and their impacts on growth and toxicity of HABs continue to be examined. In this study, the presence of non-aged polymers, regardless of polymer type, size, or concentration, was found to decrease the average chlorophyll-a content in all M. aeruginosa-associated communities compared to control cultures, while T. variabilis-associated communities generally saw increases in chlorophyll-a content in the presence of small plastics, with the exception of small cellulosic fibers, and decreases in the presence of large plastics. This is consistent with previous literature on impacts of plastics presence on growth of M. aeruginosa, though many such studies did not vary polymer types or sizes (Li et al. 2022, Wang et al. 2023). In analyzing how chlorophyll-a concentrations differed between conditions with varying polymer characteristics, it was found that both polymer type and size significantly contributed to differences in chlorophyll-a content in both cyanobacteria-associated communities, though the two strains of cyanobacteria reacted differently to polymers of the same type. While M. aeruginosa saw the lowest chlorophyll-a content in the presence of polyethylene in the first 3 weeks, T. variabilis saw increases in this same time period. As the associated communities within each of these cultures were similar, variations in responses to polymer presence between the two cyanobacteria are likely driven by changes in the cyanobacteria themselves, rather than shifts in the overall community. While many studies have investigated how cyanobacteria and microplastics interact, the usage of different bloom-causing cyanobacteria will be important in identifying how microplastic runoff may drive which bloom-formers dominate cHABs in aquatic systems that receive large quantities of microplastics.

Additionally, each cyanobacteria response to the presence of UV-aged fibers differed, with no significant changes found between M. aeruginosa controls and cultures with UV-aged fibers, but T. variabilis saw a drop in chlorophyll-a content in the presence of large UV-aged fibers. Notably, this drop was not seen in the UV trial for conditions containing non-aged, large fibers, polypropylene despite this drop being present in the same experimental conditions within the polymer trial. Nevertheless, when comparing between experimental conditions containing either non-aged or UV-aged fiber impacts, T. variabilis and M. aeruginosa again exhibited different responses, with larger UV-aged fibers driving increases in chlorophyll content compared to non-aged conditions in M. aeruginosa, but driving decreases in T. variabilis. Concentration of polymers did not play a significant role in driving chlorophyll-a content across any conditions for M. aeruginosa communities, indicating that the range of polymer concentrations used in this study was insufficient to capture a meaningful impact, though it is possible that variations could be observed at a lower or higher concentration range.

Variations between responses of M. aeruginosa and T. variabilis to polymer presence and the influence of polymer characteristics on growth are likely attributed to differing interactions between the cyanobacteria and the microplastics. As T. variabilis is filamentous and floccular, it settles in culture and mostly remains at the base of the culture flasks. In contrast, M. aeruginosa disperses throughout the liquid media and culture flask, allowing for more potential for interaction, and interference, by the polymers. Of the three polymers used in this study, both polyethylene and polypropylene are less dense than the liquid media and tended to float along the cultures surface, while cellulosic fibers are more dense and tended to sink alongside T. variabilis, though daily agitation of the cultures facilitated interactions and resuspended the polymers throughout the culture. During the course of the study, T. variabilis was noted to have bound to each of the polymer types trapped on the surface of the culture and draped down into the media (Fig. S1). Cellulosic fibers, which predominantly rested at the bottom of the culture, were also occasionally resuspended by gas bubbles, and T. variabilis again was seen attached to these newly suspended fibers. While M. aeruginosa was not visibly noted to attach to polymers, our 16S amplicon sequencing indicated that adhesion was still occurring.

Trends in microcystin-LR production by M. aeruginosa did not follow the same trend as growth rate, with most communities displaying increased average toxin content after 4 weeks compared to controls, which likely coincides with secondary metabolite production. This further suggests that growth and toxin production are decoupled processes in these microbial communities. However, studies have identified the ability for microcystins to adhere to the surface of plastics (Pestana et al. 2  021, Wan et al. 2023), and the potential impacts of co-exposure of microcystins and microplastics (Liu et al. 2023, Xiao et al. 2024), few have looked at how their presence may impact the production of microcystins. However, studies have found plastic presence and characteristics to influence growth and toxin production in a similar manner within microalgae (Liu et al. 2021, Li et al. 2023). In this study, our findings suggest that polymer presence increases the production of microcystin-LR, though this change is likely not driven by a specific polymer characteristic. In order to identify what causes this increased production, future studies should aim to identify changes in transcription of the microcystin synthetase gene cluster in the presence of plastics, as well as how the interactions between M. aeruginosa and other members of the bloom-associated community shift in the presence of these plastics. Possible causes could include microbial community response to the additives and other small molecules and oligomers that can leach from plastics in aquatic environments, as it is known that microbes preferentially consume these molecules when exposed to plastics (Kim et al. 2023, Rivera-Kohr et al. 2024).

In conclusion, as agricultural and urban runoff containing contaminants of emerging concern continue to accumulate in aquatic systems, our ability to monitor and predict changes to environmental phenomena is crucial. Environmentally relevant and applicable conditions should be replicated in laboratory experiments in order to best understand anthropogenic influences on the environment and mitigate negative impacts, particularly with groups of chemically distinct contaminants like microplastics. As highlighted by this study using design of experiments, specific microplastic properties exhibit different influences onto bloom-causing cyanobacteria, and future experiments on the interactions within the plastisphere of aquatic systems should take into account the variable responses to different microplastics and which forms are relevant to the particular system of interest.

Acknowledgments

The authors gratefully acknowledge the laboratory of Dr. Greg Boyer in the Department of Chemistry at SUNY Environmental Science and Forestry in Syracuse, NY for providing the initial cultures of both cHAB cyanobacterial communities and for their advice on culturing. We also thank James Lazarcik and the staff of the University of Wisconsin–Madison Water Science and Engineering Lab Core for providing guidance on instrumental analysis of toxin production throughout this study. This research was partially supported by the University of Wisconsin–Madison College of Engineering Shared Research Facilities and the NSF through the Materials Science Research and Engineering Center (DMR-1720415) using instrumentation provided at the UW-Madison Materials Science Center.

Author contributions

Fuad J. Shatara (Conceptualization [equal], Formal analysis [lead], Investigation [lead], Methodology [lead], Validation [equal], Visualization [lead], Writing – original draft [lead], Writing – review & editing [equal]), Azul Kothari (Formal analysis [equal], Investigation [equal], Methodology [equal], Validation [equal]), Liyuan Hou (Conceptualization [equal], Funding acquisition [equal], Writing – review & editing [equal]), Kiyoko Yokota (Conceptualization [equal], Funding acquisition [equal], Methodology [equal], Project administration [equal], Resources [equal], Validation [equal], Writing – review & editing [equal]), and Erica L.-W. Majumder (Conceptualization [equal], Funding acquisition [lead], Methodology [equal], Project administration [lead], Resources [equal], Supervision [lead], Validation [equal], Writing – review & editing [lead])

Conflict of interest

The authors declare no conflict of interest.

Funding

This work was primarily sponsored by the Great Lakes Research Consortium (GLRC) small grants award for “Effects of Great Lakes—Isolated Microplastics and their Associated Microbial Communities and Small Molecules on the Growth of Harmful Algal Bloom-Causing Species.” The fund came from the New York State Environmental Protection Fund’s Ocean-Great Lakes Ecosystem Conservation Act programs and was administered by the GLRC.

Data availability

Sequencing reads are publicly available from the NCBI sequencing read archive under BioProject accession: PRJNA1197465. https://www.ncbi.nlm.nih.gov/bioproject/1197465. Metagenomic reads and analysis are publicly available from the DoE KBase Narrative 201678. https://kbase.us/n/201678/2/. All other data are provided in the manuscript or supplemental materials.