Microplastics and Anthropogenic Particles in Recreationally Caught Freshwater Fish from an Urbanized Region of the North American Great Lakes

Abstract

Background:

Microplastics are a pervasive contaminant cycling through food webs—leading to concerns regarding exposure and risk to humans.

Objectives:

We aimed to quantify and characterize anthropogenic particle contamination (including microplastics) in fish caught for human consumption from the Humber Bay region of Lake Ontario. We related quantities of anthropogenic particles to other factors (e.g., fish size) that may help in understanding accumulation of microplastics in fish.

Methods:

A total of 45 samples of six fish species collected from Humber Bay in Lake Ontario near Toronto, Ontario, Canada, were examined for anthropogenic particles in their gastrointestinal (GI) tracts and fillets. Using microscopy and spectroscopy, suspected anthropogenic particles were identified and characterized.

Results:

We observed anthropogenic particles in the GI tracts and fillets of all species. Individual fish had a $\text{mean}\pm \text{standard\hspace{0.17em}deviation}$ of $138\pm 231$ anthropogenic particles, with a single fish containing up to 1,508 particles. GI tracts had $93\pm 226$ particles/fish ($9.8\pm 32.6$ particles/gram), and fillets had $56\pm 61$ particles/fish ($0.5\pm 0.8$ particles/gram). Based on a consumption rate of 2 servings/week, the average yearly human exposure through the consumption of these fish fillets would be $\mathrm{12,800}\pm \mathrm{18,300}$ particles.

Discussion:

Our findings suggest that consumption of recreationally caught freshwater fish can be a pathway for human exposure to microplastics. The elevated number of particles observed in fish from Humber Bay highlights the need for large-scale geographic monitoring, especially near sources of microplastics. Currently, it is unclear what the effects of ingesting microplastics are for humans, but given that recreationally caught freshwater fish are one pathway for human exposure, these data can be incorporated into future human health risk assessment frameworks for microplastics. https://doi.org/10.1289/EHP13540

Introduction

It has been well documented that microplastics, pieces of plastic $1\mathrm{\mu }\mathrm{m}–5\mathrm{mm}$ in size, are a pervasive contaminant in the environment globally.¹ More recently, scientists have demonstrated that microplastics are entering and accumulating in humans. Recent studies have documented the presence of microplastics in breast milk ($0–2.72\text{\hspace{0.17em}particles}/\mathrm{g}$ tissue),² placenta (12 particles in 4 women’s placentas),³ lungs ($0.56\text{\hspace{0.17em}particles}/\mathrm{g}$ tissue),⁴ and human blood ($\text{mean}=1.6\mathrm{\mu }\mathrm{g}/\mathrm{mL}$).⁵ Contamination in humans can occur via various exposure pathways, including inhalation (particles in air and dust^6,7) and ingestion (various types of food and beverages^8–10). Some researchers have attempted to estimate microplastic exposure from these sources.¹¹ Recent studies estimate that 39,000–320,000 microplastic particles might be ingested annually by adults.^11,12 However, the total human burden of microplastics from all pathways remains unclear.

For recreational and subsistence fishers, freshwater fish represent one pathway for exposure to microplastics. Currently, the microplastic literature is dominated by a focus on marine systems; however, studies suggest that freshwater bodies are relevant owing to their proximity to sources of plastic pollution, such as wastewater treatment plants, landfills, and litter.^13,14 Freshwater bodies are important vectors of microplastic movement thought to be involved in transporting the majority of plastic that ends up in marine environments, meaning that freshwater fish may be exposed to and consume comparable levels of microplastics compared with marine fish.^13,15,16

Globally, $>200$ species of freshwater fish have been reported to ingest microplastics, but it is unclear how, once ingested, particles move into other fish tissues post-ingestion and if it is limited to certain types or sizes of particles.^17,18 Mechanisms of translocation have been proposed for micro- and nanoplastic particles, including by vesicular processes, such as endocytosis¹⁹ by enterocytes.²⁰ Although the edible tissues of seafood have been a large focus in the literature,²¹ limited data exists on the contamination of tissues in freshwater fish, such as the muscle tissue (fillets) that are typically consumed by humans.^22–24 Moreover, studies suggest that microplastic contamination in the gut content is not a good predictor of contamination in other tissues and cannot be applied to better understand human exposure.²² A better understanding of the concentrations in fish, both the gut contents and muscle, will help us better understand exposure to humans.

In addition to understanding the prevalence of exposures to microplastics in the environment, numerous studies have demonstrated harmful effects of microplastics on organisms. For example, laboratory experiments using fish have found that microplastics can cause stress to the immune system,²⁵ reductions in growth,²⁶ changes in behavior,²⁶ and deformities during development.²⁷ Within the last 10 years, researchers have begun to investigate how microplastics may impact humans.²⁸ In vitro studies suggest that multiple end points are affected by microplastic particles, including cell viability ($5\mathrm{\hspace{0.17em}to\hspace{0.17em}}200\mathrm{\mu }\mathrm{m}$ diameter particles at a concentration of $10\mathrm{\mu }\mathrm{g}/\mathrm{mL}$) and immune response ($0.4\mathrm{\mu }\mathrm{m}$ diameter particles at concentrations of $20\mathrm{\mu }\mathrm{g}/\mathrm{mL}$).²⁹ In vivo studies examining the impacts of microplastics in mammals have also shown effects at high exposure concentrations, including changes in energy and lipid metabolism and oxidative stress (polystyrene particles $5\mathrm{\mu }\mathrm{m}$ in diameter at a concentration of $1.46\times {10}^6\text{\hspace{0.17em}particles}/0.5\mathrm{mL}$ and $20\mathrm{\mu }\mathrm{m}$ in diameter at a concentration of $2.27\times {10}^4\text{\hspace{0.17em}particles}/0.5\mathrm{mL}$)³⁰ and metabolic disorders ($5\text{-}\mathrm{\mu }\mathrm{m}$ polystyrene microplastics at concentrations of $1.456\times {10}^6\text{\hspace{0.17em}particles}/\mathrm{L}$ and $1.456\times {10}^7\text{\hspace{0.17em}particles}/\mathrm{L}$).³¹ These studies suggest that microplastics may harm humans.

Although we are starting to understand how certain kinds and sizes of microplastics can affect humans, a recent study suggests we do not yet know enough to assess total risk from environmental exposure, noting concerns about the quantity and quality of the studies to date.²⁸ For example, the studies testing hypotheses about the effects of microplastics on mammalian models and cells are often not a direct reflection of potential risks from environmental exposure because of their limited testing conditions (i.e., using a single polymer, size, or morphology). Exposure in humans includes a diversity of particles present in environmental samples (e.g., numerous polymer types, morphologies, additives, sizes).^32,33 The diversity of microplastics in the environment poses a challenge for establishing a risk assessment framework.³³ There is still much uncertainty regarding the mechanisms of toxicity across organisms, as well as how they vary by taxa, types of microplastics, end points, and the threshold at which effects occur.²⁸ As such, no risk framework for human exposure exists, but efforts are being made to develop this kind of framework.³⁴ As researchers continue to study the effects of microplastics in humans and work toward building a risk assessment framework, studies are needed to better understand human exposure. Given that eating fish can be a potential exposure route, understanding the toxicokinetics of microplastics (e.g., translocation, bioaccumulation, biomagnification) and the variability in contamination among the parts of the fish generally consumed by humans is important.

In the present study, we examined the gastrointestinal (GI) tracts and fillets of six species of commonly recreationally caught fish collected from the Humber Bay region of Lake Ontario adjacent to Toronto, Ontario, Canada for the presence of microplastic and other anthropogenic particles. From here forward, we use “anthropogenic particles” to refer to microparticles of plastic and anthropogenic origin (e.g., dyed cellulose used in textiles). Specifically, we aimed to address the following knowledge gaps: contamination across different tissues, differences in patterns of contamination among species, the occurrence of bioaccumulation, and biomagnification. We also attempted to better understand microplastic exposure to humans from consuming sportfish locally from the Humber Bay region of Lake Ontario, which is known for its world class fishery for salmon, trout, pike, bass, and walleye species. Our study focuses on this region specifically because studies have demonstrated relatively high levels of microplastics in the Toronto region of the Laurentian Great Lakes,³⁵ likely as a result of discharges from tributaries that flow through areas of high urbanization and industrial activity, including several manufacturers of plastic items^36–38 There is also a wastewater treatment plant upstream that empties into Humber Bay.^39,40 Fish consumption advisories have been issued for this and other parts of the Great Lakes for about a half a century now; however, the advisories are due to other contaminants (e.g., mercury, polychlorinated biphenyls).^41–43 At present, microplastics are yet to be incorporated into any advisory framework.

Methods

Fish Sample Collection and Processing

Fish were collected as a part of the Fish Contaminant Monitoring Program of the Ontario Ministry of the Environment, Conservation and Parks (MECP) from the Humber Bay region of Lake Ontario, Ontario, Canada, during two periods: November 2018 and November 2021. Collections were approved by Ontario Ministry of Natural Resources and Forestry (MNRF) Southern Region (permit nos. 1089191 and 1097677) and followed the animal use protocol of the MNRF’s Aquatic Research and Development Section Animal Care Committee. A total of 45 fish across six species were caught: brown bullhead (Ameiurus nebulosus; $n=10$), white sucker (Catostomus commersonii; $n=6$), largemouth bass (Micropterus salmoides; $n=12$), smallmouth bass (Micropterus dolomieu; $n=6$), rock bass (Ambloplites rupestris; $n=1$), and northern pike (Esox lucius; $n=10$). The species were selected to represent popularity for consumption as well as sentinel species for microplastic uptake from water and sediments. All fish were sacrificed by a blow to the head and were frozen until shortly before the dissection procedure occurred at the MECP laboratory. Prior to dissection, each fish was measured for total length and weight (see Table S1 for further details). A single fillet and the GI tract were individually removed from each fish, weighed, and then stored in Whirl-Pak bags at $-20°\mathrm{C}$ until further processing.

Each sample (either a whole fillet or the GI tract including gut contents) was individually digested in a 20% potassium hydroxide (KOH) solution (Fisher Scientific catalogue no. P250-500) at $45º\mathrm{C}$ for at least 24 h to break down the fish tissue.^44,45 For this procedure, the methods outlined by Dehaut et al.⁴⁵ for KOH digestion were followed, with modifications to increase the concentration of KOH from 10% to 20% to improve the breakdown of natural materials, and reduction of temperature from $60°\mathrm{C}\text{\hspace{0.17em}to\hspace{0.17em}}45°\mathrm{C}$ based on findings of high temperatures reducing recoveries of microplastics, as described by Munno et al.⁴⁴ Some GI tract samples were further processed to remove undigested organic debris with a wet peroxide oxidation procedure using sulfuric acid (${\mathrm{Fe}\left(\text{II}\right)\mathrm{SO}}_4$ from VWR International catalogue no. 470045-604) and ${\mathrm{Fe}\left(\text{II}\right)\mathrm{SO}}_4$·heptahydrate (from Fisher Scientific catalogue no. 1146-500) as a catalyst and 30% hydrogen peroxide (${\mathrm{H}}_2{\mathrm{O}}_2$; Fisher Scientific catalogue no. H325-500) in a 1:5 ratio while using an ice bath to maintain a temperature $<50º\mathrm{C}$ to minimize the loss of microplastic and anthropogenic particles (following the method outlined by Munno et al.).⁴⁴ Following recommendations described by Lusher et al.,⁴⁶ some GI tract samples (Excel Table S1) also underwent a density separation procedure using $1.4{\mathrm{g}/\mathrm{cm}}^3$ calcium chloride (${\text{CaCl}}_2$), if required, to separate dense natural material from microplastics and anthropogenic particles (method adapted from Munno et al. and Claessens et al.) (VWR catalogue no. 97062-590).^44,46–48 Samples were then sieved through a $45\text{-}\mathrm{\mu }\mathrm{m}$ stainless steel mesh and returned to their original jars with 10% Alcojet detergent solution in a 1:1 ratio to remove remaining fatty material (Decon Conrad 70, Fisher Scientific catalogue no. 16-000-111). Fillets were split into two equal subsamples by repeatedly stirring and pouring the liquefied sample mixed with detergent into two identical precleaned beakers. One subsample was used for microscopy (discussed here) and the other to be processed using pyrolysis–gas chromatography/mass-spectrometry (to be discussed in a separate forthcoming paper). To ensure that subsamples were representative, we quantified and characterized particles from subsamples created from seven GI tract samples using the same methods to test the covariance of the split samples. Before quantification and characterization via microscopy, the samples were rinsed through stacked 45-, 125-, and $355\text{-}\mathrm{\mu }\mathrm{m}$ sieves to size fraction. A diagram overviewing the methods used can be found in Figure S1.

Anthropogenic Particle Sorting and Quantification

Samples were examined using a dissecting microscope ($10–80\times \text{magnification}$; Leica S8 APO Stereozoom; Leica Microsystems), and the first 10 suspected anthropogenic particles of each morphology and color observed in each size fraction (e.g., blue fragments in the $>355\text{-}\mathrm{\mu }\mathrm{m}$ size fraction) were picked out using forceps and mounted onto double-sided tape in a Petri dish. All other suspected anthropogenic particles were characterized to color and morphology and counted. The morphologies used to categorize particles were fragment, fiber, film, rubber, fiber bundle, sphere, and foam. Particles were categorized into these morphology categories according to the images and descriptions provided by Rochman et al. and Werbowski et al. (Table S2).^49,50 All picked particles were photographed and measured for length and width using OMAX Toupview software (version 3.7; ToupTek).

Identification and Characterization Using Spectroscopy

A subsample of all particles (before blank-correction) were chemically analyzed to determine the proportion of suspected anthropogenic particles that were anthropogenic and microplastic. For chemical identification, we predominantly used Raman spectroscopy (Horiba Raman XploRA PLUS confocal microscope) in LabSpec6 software (version 6.7) and equipped with a charge-coupled device detector ($-60°\mathrm{C}$, $\mathrm{1,024}\times 256$ pixels Raman spectra) were obtained using a $100\times$ long working distance objective (0.8 numerical aperture), resulting in laser powers of 15.0 and $17.8\text{\hspace{0.17em}mW}$ at 100% filter for the 532- and $785\text{-nm}$ lasers, respectively. Spectral resolution ranged from $1.3{\mathrm{cm}}^{-1}$ ($785\text{\hspace{0.17em}nm}$ excitation laser, $600\text{\hspace{0.17em}grooves}/\mathrm{mm}$) to $3.3{\mathrm{cm}}^{-1}$ ($532\text{\hspace{0.17em}nm}$ excitation laser, $\mathrm{1,200}\text{\hspace{0.17em}grooves}/\mathrm{mm}$). Spectra were matched to library spectra using the Wiley KnowItAll and ID Expert spectral matching software from the KnowItAll Raman Spectral Library (ID Expert version 23.1.45.0), as well as the Spectral Library of Plastic Particles (SloPP and SloPP-E).⁵¹ Minimal manual corrections to spectra were made, including baseline correction and vertical clipping. Spectra matches were categorized into one of nine material type categories as outlined by Munno et al.⁴⁷: plastic, anthropogenic (synthetic), anthropogenic (cellulosic), anthropogenic (inorganic), anthropogenic (unknown base), cellulosic, natural (inorganic), natural (organic), and unknown (see Appendix S7 in Munno et al. for full descriptions of these classifications).⁴⁷

The suspected rubber particles were analyzed using $\mathrm{\mu }\text{-Fourier}$ Transform Infrared ($\mathrm{\mu }\text{-FTIR}$) instead of Raman spectroscopy to achieve better chemical identification of particles of this suspected morphology. For suspected rubber particles, spectra were collected with a Nicolet iN10 infrared microscope [Thermo Fisher Scientific in attenuated total reflectance (ATR) mode ($15\times \text{objective}$, 0.7 numerical aperture), using a germanium ATR crystal].

The subsampling method was as follows: at least 30% of fish within each species were chosen at random ($n=17$), within each sample at least 10% of particles of each color, morphology, and size fraction were chosen at random and tested (to a maximum of 10 particles of each grouping).⁵² Using this method, 8% of observed sample particles ($n=566$ microparticles) and 64% of laboratory blank particles ($n=223$ microparticles) were chemically analyzed.

Blanks and Quality Assurance/Quality Control

Throughout all the processes, quality assurance/quality control (QA/QC) procedures to reduce contamination were followed. Specifically, all materials were triple-rinsed with reverse osmosis (RO) water, work was done in a clean cabinet (enclosed space with reduced airflow to reduce potential contamination) where possible (i.e., when KOH was being added to samples), a white cotton lab coat was worn, and all materials were kept covered when not being used. Our laboratory is equipped with a high-efficiency particulate air (HEPA) filter intended to reduce airborne contamination.

At least one laboratory blank was run per every 10 samples, and each was taken through the same full procedure (either for GI tract or fillet) as actual samples from dissection onward to measure and account for any microplastic contamination from the laboratory processes ($n=17$). During the dissection step, blanks started as either a Whirl-Pak bag left open for the duration of the dissection of a fish ($n=13$), or as a plate of RO water left on the lab bench during the dissection of a fish ($n=4$). To be conservative, all blanks run through the GI tract procedure ($n=8$) were taken through the wet peroxide oxidation procedure and half ($n=4$) were taken through density separation procedures (despite not all samples undergoing either or both procedures).

Blank sample counts for each tissue (particles found in the blank samples are listed in Tables S3 and S4) were used to calculate a limit of detection (LOD) for each color and morphology combination within a size fraction (e.g., blue fragments in the $>355\text{-}\mathrm{\mu }\mathrm{m}$ size fraction). The LOD was calculated as the mean plus three times the standard deviation (SD) and was used to exclude particles from samples that were below the LOD.⁵³ For groupings of particle morphologies, colors, and sizes present in quantities above the LOD, the mean calculated for each of the same grouping of particles if found in the blanks was subtracted from sample particle counts of the same classifications. For example, for blue fragments in the $>355\text{-}\mathrm{\mu }\mathrm{m}$ size fraction, if the LOD was calculated as 9 particles and a sample had 10 particles of that grouping, it would be considered above the LOD, and therefore it would be corrected by subtracting the mean number of blank particles. In contrast, if a sample had $<9$ particles, it would be considered below the LOD, and thus corrected to 0. This correction method was applied to all samples and for all groupings of particles found in the blanks.

For a further measure of our accuracy of identifying and picking out anthropogenic particles instead of natural particles, six particles thought to be natural were picked from each of six fillet and five GI tract samples. The 66 picked suspected natural particles were chemically identified using Raman spectroscopy to assess the rate of false-negative identification of particles (particles thought to be natural that were instead anthropogenic).

To measure the recovery of our extraction procedure, six spike and recovery samples were created, processed, and analyzed following the same procedure used for each tissue ($n=3$ for each tissue). Predetermined spike and recovery particles were added to GI tracts of lab-raised rainbow trout provided by the MECP and fillets of extra brown bullhead samples. Ten particles of each of five different kinds of microplastics were used in each spiked sample: orange nylon microfibers ($181–\mathrm{3,130}\mathrm{\mu }\mathrm{m}$), clear polyethylene terephthalate fragments ($318–\mathrm{1,910}\mathrm{\mu }\mathrm{m}$), green polypropylene fragments ($392–\mathrm{1,457}\mathrm{\mu }\mathrm{m}$), Cospheric green polyethylene microspheres ($63–75\mathrm{\mu }\mathrm{m}$), and Cospheric clear polyethylene microspheres ($180–212\mathrm{\mu }\mathrm{m}$) (Table S5).

Data Analysis

All statistical analyses were performed using R statistical software (Rstudio version 5.12.10; R Development Core Team) with a significance level of $\mathrm{\alpha }=0.05$. For all reports of fillet and combined tissue counts, the original fillet counts were multiplied by 4 to account for the subsampling (using 1 of the 2 fillets, and then splitting the sample) so they were representative of the amount of fillet tissue in the full fish. Kruskal–Wallis tests were used to determine differences in particle counts between species for each tissue and for the combined tissue count per fish. For tissues found to be statistically significant, a post hoc Dunn’s test with Bonferroni correction was used to determine which species differed from one another. To assess differences in particle counts across habitat types, the nonparametric Mann–Whitney test was used. To explore differences between tissues in terms of assemblages of particle size fractions and morphologies, we used nonmetric multidimensional scaling (nMDS), using Euclidian distance to visualize particle arrangements, and statistically analyzed for significant differences across tissues using permutational analyses of variance (PERMANOVAs) (nMDS and PERMANOVAs were run using the vegan package).⁵⁴ To test hypotheses about the occurrence of bioaccumulation, we used fish length as a proxy for age. We ran linear regressions for each species individually using particle count concentrations in fillets as the dependent variable and fish lengths as the independent variable. To assess biomagnification, we ran linear regressions using log-transformed standardized fillet counts as the dependent variable and trophic position as the independent variable. We assigned trophic positions using data from FishBase (version 02/2024), which calculates a value for trophic level based on the mean trophic levels of the food items that the species consumes.⁵⁵ Here, trophic levels were as follows: northern pike, 4.1; smallmouth bass, 3.6; largemouth bass, 3.8; brown bullhead, 3.7; white sucker, 2.8; and rock bass, 3.4.

To assess potential human exposure, we multiplied the observed contamination in each fillet sample (as particles per gram) by a serving size of $227\mathrm{g}$ of fish tissue consumed twice a week. Fish consumption advisories for the waterbody studied here are issued by Government of Ontario’s Fish Contaminant Monitoring Program through Guide to Eating Ontario Fish, which uses $227\text{\hspace{0.17em}grams}$ as an average meal size for an average adult weighing $70\mathrm{kg}$.⁴³ Surveys conducted by the Government of Ontario has shown that using 2 fish meals/wk covers the maximum consumption of $>90\%$ of recreational fishers.⁵⁶ As such, our use of twice a week of fish consumption provides reasonable exposure estimates useful for protecting health of recreational fishers. We scaled the number of particles ingested per week to a yearly amount to produce a range of annual consumption values. To assess exposure taking into account the variability among fish, we prepared a cumulative distribution function by bootstrapping sampling 1,000 times from the observed number of particles per gram fillet tissue and scaling this to an annual value, assuming a serving of $227\mathrm{g}$ consumed twice weekly. We did not weight species differently based on their popularity owing to limited data availability.

Results

QA/QC

To ensure that subsamples were representative, we quantified and characterized particles from seven pairs of subsamples from seven GI tract samples. For each of these seven pairs, we measured the covariance of the split samples (Table S6). We observed a mean coefficient of variability of 21% across seven split samples, which suggests that the subsampling procedure was representative (Table S7). Our samples were blank-corrected using a conservative LOD method (the mean plus three times the standard deviation for each particle type) that helped in assuring that particles reported were not laboratory contamination. Following blank-correction, 33% of the total number of suspected microplastic particles ($N=\mathrm{2,282}$) were considered outside of the LOD and were removed from the dataset (predominantly blue and gray fibers, likely resulting from airborne contamination during processing). For further details about the particles observed in the blank samples and the LOD calculations, refer to Tables S3 and S4. Based on our spike and recovery tests, which involved running samples where known particles were added, the recoveries were $>70\%$ (Table S5). Furthermore, recoveries were similar across particle types. Of the 566 subsampled particles, 80% ($N=455$) were determined to be anthropogenic in origin, 36% ($N=203$) of these anthorpogenic in origin particles were confirmed to be plastic, 11% ($N=61$) were natural, and 9% ($N=52$) were unknown (Figure S2A). The high proportion of anthropogenic particles from our spectroscopic analyses suggest that our anthropogenic particle characterization was accurate. We further tested our accuracy by picking particles that we thought were natural. Using Raman spectroscopy, $<10\%$ of particles thought to be natural were anthropogenic [62% ($N=41$) natural, 6% ($N=4$) anthropogenic, and 32% ($N=21$) unknown] (Figure S2B). Our subsampling procedure for spectroscopy was representative to allow for confidence about broad material types (i.e., plastic, anthropogenic, and natural).³⁴ Particles were categorized by material types outlined by Munno et al.⁴⁷ For more information on the different material types and polymers identified using spectroscopy, refer to Figure S2.

Body Burden

Herein we use the total number of anthropogenic particles per fish to describe the sum of anthropogenic particles in the GI tract and fillet after blank subtraction, with fillet counts extrapolated to account for both fillets in a fish. Across all individual fish from all six species, the total number of anthropogenic particles per fish ranged from 2 (a largemouth bass) to 1,503 (a white sucker) particles. On average, $\text{number}\pm \text{SD}$ of anthropogenic particles per fish (both tissues combined) was $138\pm 231$ particles. Anthropogenic particles were observed in all tissue samples, except for three fillets and one GI tract. In the GI tract, the number of anthropogenic particles ranged from 0 to 1,400 ($\text{mean}=93\pm 234$), and in the fillets, the number of particles ranged from 0 to 284 ($\text{mean}=56\pm 61$) (Figure 1). Adjusted to particles per gram of tissue, there was a mean of $9.8\pm 32.6\text{\hspace{0.17em}particles}/\mathrm{g}$ in the GI tract and a mean of $0.5\pm 0.8\text{\hspace{0.17em}particles}/\mathrm{g}$ in the fillets. Microplastic body burden was significantly different among species for total amount ($p=0.02$) and within GI tracts ($p=0.01$), but not for fillets ($p=0.1$). Brown bullheads had significantly more particles in total, and within their GI tracts, than largemouth bass. For detailed information on the particles obtained in each sample fish tissue, see Excel Tables S1–S3.

Figure 1.

Number of anthropogenic particles in two tissues, gastrointestinal (GI) tract and fillet, for six species of freshwater fish sampled from Humber Bay in Lake Ontario. Sample sizes for each species were $n=10$ for brown bullhead (BB), $n=12$ for largemouth bass (LB), $n=10$ for northern pike (NP), $n=1$ for rock bass (RB), $n=6$ for smallmouth bass (SB), and $n=6$ for white sucker (WS). Each box represents the distance from the first quartile to third quartile of the data, with a horizontal line going through the box to show the median. Vertical lines extend from either end of each box to the minimum and maximum values, with any outliers shown as dots. Corresponding summary data, medians, and quartiles are presented in Excel Table S1.

Patterns in Particle Characteristics

Six different particle morphologies were observed in fish samples: fibers, fiber bundles, films, foams, fragments, and rubber. For the GI tract, the mean particle length was $1.3\pm 3.5\mathrm{mm}$, and in the fillet, mean particle length was $0.7\pm 1.0\mathrm{mm}$ (Figure S3). In the GI tracts, rubber particles were the most common particle type (70%; $N=\mathrm{2,898}$), and in the fillets, nonrubbery fragments were most common (49%; $N=283$). (More information can be found in the supplementary files Raw Data – Excel Table S3.) However, the observation of rubber particles being the most common in GI tracts was driven by two species: brown bullhead and white sucker. Fibers were the most common morphology in the GI tracts of the four other species.

Two nMDS tests were run; one to visualize differences between tissues in terms of particle size fractions (i.e., 45–125, 125–355, and $>355\mathrm{\mu }\mathrm{m}$; Figure S4A), and another for particle morphologies (Figure S4B). A PERMANOVA detected a significant difference among morphologies between the GI tract and fillets ($p=0.002$; Figure S4B), likely driven by the rubber particles in the GI tracts. The assemblage of particle sizes was not significantly different between tissues ($p=0.4$; Figure S4A).

Accumulation of Anthropogenic Particles

We observed no relationship between counts in the two tissues; counts of anthropogenic particles in the GI tracts were not a predictor of counts in the fillets (Figure S5). To look at the potential for accumulation of anthropogenic particles over time, fish total length was used as a proxy for age (bioaccumulation is defined as the accumulation of a contaminant over time). We observed a significantly negative relationship between fish length and particle concentration in the fillet (wet weight) for two species (white sucker $p=0.02$, and brown bullhead $p=0.01$; Figure 2). A visually similar, but nonsignificant, trend was observed for the other three species examined. This suggests that bioaccumulation is not occurring, and instead other mechanisms (such as growth dilution or biliary excretion) may be driving this pattern.^12,22 Further, comparing concentrations of anthropogenic particles across species of different trophic levels showed no significant trends ($p=0.25$). The lack of increasing concentration of anthropogenic particles at higher trophic levels indicates no biomagnification, although we looked at only a narrow trophic range (Figure S6).

Figure 2.

Total fish length (cm) compared with counts of anthropogenic particles in the fillets standardized by the wet weight of the fillet tissue (g). Rock bass was excluded from this analysis because there were not enough individuals sampled to produce a regression ($n=1$). Sample sizes for each species were $n=10$ for brown bullhead (BB), $n=12$ for largemouth bass (LB), $n=10$ for northern pike (NP), $n=6$ for smallmouth bass (SB), and $n=6$ for white sucker (WS). Corresponding summary data are presented in Excel Table S1. Note: WW, wet weight.

Estimated Human Consumption of Anthropogenic Particles by Eating Sportfish

In the fillets, we found a mean of $0.5\pm 0.8$ particles/g of tissue. The Ontario government, specifically the MECP, uses $227\mathrm{g}$ as an average meal size for an average adult weighing $70\mathrm{kg}$.⁴³ Based on the number of anthropogenic particles observed in the edible portion (fillets) of these sportfish samples (the GI tract is typically not consumed), the average number of particles consumed in a single serving was estimated to be $123\pm 176$ (Figure S7). The number of particles that would be consumed in a serving range from 0 (individual largemouth bass and brown bullhead) to 854 (a smallmouth bass). Scaled up to a yearly consumption amount using a rate of 2 servings/wk (i.e., $454\mathrm{g}/\mathrm{wk}$), the average yearly consumption of anthropogenic particles from these freshwater fish was estimated to be $\mathrm{12,800}\pm \mathrm{18,300}$ (Figure 3; Figure S8). The high variability observed in the particle count would translate into a yearly consumption range of 0–89,000 particles. We conducted a probabilistic exposure assessment of annual ingestion based on all samples here to assess the variability in exposure and the frequency at which different exposure amounts may occur (Figure 3). Given the high variability between individual fish and the lack of statistical differences between species for fillet counts ($p=0.1$), all species were considered together for an exposure estimate, instead of individually.

Figure 3.

(A) Histogram showing annual exposure simulation for ingestion of fish using an average consumption of $227\mathrm{g}$ per serving, twice a week, scaled to yearly consumption (sampled randomly from the observed contamination per gram observed in the fillets 1,000 times to produce this distribution). (B) Cumulative distribution functions for the annual exposure simulation. The mean number of microplastics (MPs) ingested annually is shown by a red dashed line ($\text{mean}=\mathrm{13,400}\pm \mathrm{18,400}$ microplastic particles). The median number of MPs ingested annually is shown by a green dotted line ($\text{median}=\mathrm{6,370}$). Corresponding summary data are presented in Excel Table S1.

Discussion

Measure of Contamination in Lake Ontario Fish

Although literature on the ingestion of microplastics by fish in marine environments exceeds that of freshwater, freshwater fish have been observed to have a higher body burden of microplastics than marine fish ($\text{mean}=8.0$ particles per freshwater fish, compared with a mean of 2.7 in marine fish).¹⁶ The average reported here, $138\pm 231$ anthropogenic particles/fish greatly exceeds these averages. These results may be attributed to the high contamination in the sediment and water in Humber Bay (where the fish were caught) owing to the proximity of the site to Toronto, a large urban center.^37,57 For example, we observed relatively large quantities of rubber particles (thought to be associated with road runoff) in the GI tracts of demersal fish (brown bullhead and white sucker).^58,59 We also report higher contamination of microplastics and other anthropogenic particles in fish than a previous study focusing on the same kinds of fish caught at the same location, Humber Bay.⁴⁷ The disparity may be due to the inclusion of fillets in this study which contributes to a higher body burden of particles compared with most studies that focus only on GI tracts.^16,60,61 As such, these findings likely represent a worst-case scenario in terms of microplastic contamination of freshwater fish.

The presence of anthropogenic particles in the fillets is concerning because it indicates that freshwater fish can be a pathway for human consumption of microplastics and other anthropogenic particles. Despite observing high anthropogenic particle body burdens relative to other studies of fish, the amount humans could be exposed to through these freshwater fish (and others) is within reported ranges for other comparable exposure sources, including seafood. In the fillets, we found a mean of $0.5\pm 0.8$ particles/g of tissue, whereas a systematic review of studies looking at microplastics in seafood intended for human consumption reported $0–2.9\text{\hspace{0.17em}microplastics}/\mathrm{g}$ in fish.²¹ Moreover, fish have been reported as a less significant pathway of exposure relative to other sources, including commonly consumed food, drinking water, and inhalation from the air.¹¹ For instance, we estimated an average consumption of 12,800 particles annually from freshwater fish, but through bottled water alone, individuals were estimated to ingest 90,000 microplastic particles each year.¹¹

Direct and Indirect Human Health Risks

Concerns surrounding the direct harm of consuming microplastics arises owing to the potential for microplastics to cause negative physical effects (e.g., food dilution in Daphnia magna)⁶² and chemical effects (e.g., plastic additives reduce ecophysiological function in Arenicola marina).^63,64 Chemical effects include those caused by additives in plastics^64,65 and contaminants sorbed from the environment (e.g., heavy metals, persistent organic pollutants).^66–68 However, it is unclear if toxicological effects from either of these mechanisms are occurring in humans, and at what thresholds effects occur. Although some in vitro and in vivo studies examining impacts of microplastics on mammals show effects, such as intestinal barrier disfunction and alteration of biomarkers,^30,31 there are also conflicting reports of no acute toxicity from microplastics.^69,70 Discrepancies are likely due to exposure experiments using different concentrations, but also these variations in results could be relevant to the polymer type, morphology, or size used in the study.²⁸ These studies are less environmentally relevant, in that they tend to use one polymer type or one size. They also tend to use smaller particles than those we know we are exposed to. For example, most studies looking at environmental samples are constrained by methods that detect larger particles (e.g., LODs of spectroscopy are generally $>10\mathrm{\mu }\mathrm{m}$),⁷¹ many laboratory studies testing effects look at smaller particles.⁷² Given the many uncertainties, no risk assessment framework for human health currently exists.²⁸

Here, we observed a high diversity of particle morphologies, polymers, and sizes. The diversity of anthropogenic particles in environmental samples contributes to challenges in estimating risk. Of particular concern here may be the high quantities of rubber particles observed primarily in white sucker (mean number of rubber particles per $\text{fish}=297\pm 491$) and brown bullhead (mean number of rubber particles per $\text{fish}=91\pm 132$) owing to the harmful effects that tire rubber microparticles have been demonstrated to have on fish. For instance, fish exposed to tire particles and their leachate have been observed to show altered behavior and growth,²⁶ as well as decreased heart rates and reduced hatching success.²⁷ Further, chemicals from tire rubber have been demonstrated to be directly connected to the mortality of coho salmon (Oncorhynchus kisutch).⁷³ However, there is little information on effects of tire rubber on humans. Only one study has investigated the impacts of tire rubber on humans, an in vitro study using intestinal epithelial cells, and it found no effects on inflammatory end points. In our samples, few rubber particles were found in the fillets, suggesting a low threat for human exposure to rubber particles by consuming recreationally caught freshwater fish in Humber Bay.

Also relevant to the impacts of microplastics on human is any change in the nutritional value or food security (or availability) of consumed organisms, such as fish. Laboratory studies have demonstrated microplastics can cause a variety of detrimental effects on fish, including negative effects on growth,⁷⁴ changes in reproduction, and reduced survival.⁷³ These, and other impacts on fish, represent a concern in terms of reducing the quality and quantity of fish as a food source.⁷⁵

Patterns of Particle Fate Relevant to Human Exposure

Understanding trends of anthropogenic particle contamination in fish is useful for understanding how to minimize exposure for anglers. Here, similar to findings reported in other studies, we did not observe evidence of bioaccumulation^22,23 or biomagnification.^23,76 Using exploratory analysis techniques, we observed a negative relationship between fish length (a proxy for age) and fillet concentrations (Figure 2). This supports the possibility that translocated anthropogenic particles can be excreted over time (e.g., biliary excretion) or there may be growth dilution.^12,22 Growth dilution is the phenomenon of larger fish having lower concentrations of a chemical than smaller fish owing to their fast growth and resulting high biomass. This effect has been reported to occur in fish for other contaminants, including methylmercury.⁷⁷ As such, this indicates that larger fish may be less contaminated than smaller fish on a per mass basis (although larger fish have more anthropogenic particles on a per fish basis), a consideration relevant to exposure and risk in humans.

Based on the species used in this study, which occupy a narrow range of trophic positions, we observed no evidence of biomagnification; that is, there was no positive relationship between anthropogenic particle concentration and trophic level (Figure S6). In terms of human exposure to microplastics from fish, the lack of observed differences between fish of differing trophic levels suggests fish position on the food chain is not a consideration for accumulation of microplastics and other anthropogenic particles. Instead, it is possible that other factors (e.g., proximity to sources of microplastics into the environment, ecological traits)^76,78 might be more important to consider in predicting contamination of fish and the resulting exposure in humans. Further work should focus more closely on investigating the potential for biomagnification to occur using a larger number of species occupying more varied trophic positions.

We observed significant differences in contamination of microplastic and anthropogenic particles between species; brown bullhead had significantly more particles overall and in the GI tract than largemouth bass. This may be due to differences in, for example, feeding strategy, habitat niche, ecology, or morphology. For instance, brown bullhead are demersal species, occupying the bottom of the lake, meaning that they may be exposed to higher quantities of particles than species occupying the water column.⁵⁵ This was affirmed by the similarly high quantities of particles observed in white sucker, another demersal species.⁵⁵ Better understanding these drivers of contamination between species can help to inform which species are most vulnerable to microplastic pollution.

In the present study, we did not measure the retention of microplastics and anthropogenic particles within the GI tract or external tissues. The lack of a relationship between GI tracts and fillets (Figure S5) suggests that the number of particles found in the GI tract is not a predictor of the amount that translocates to the muscle tissue—suggesting that microplastics in the GI tract are transitory. Gut retention time, based on laboratory experiments that measured microplastics in feces and/or in animals sampled over time after dietary exposure, is reported to range from $\sim 1$ to 6 d in fish.^79–82 The presence of anthropogenic particles in other tissues, such as livers and fillets, suggests the possibility that anthropogenic particles can accumulate and persist in fish over longer time spans.²² Further work is needed to understand the toxicokinetics of microplastics.

Limitations

Given the varied diets of the sampled fish, there were differences in how well the GI tract tissue samples were digested using KOH. This led to a need for other methods to process some samples prior to microscopy. To try to account for this heterogeneity of extraction methods, all GI tract spike and recovery samples were taken through all the subsequent procedures (which not all the GI tract samples went through). In addition, a subset of laboratory blank samples were processed with these additional processing steps to account for any contamination acquired through these methods. Although the recoveries for all of the spike and recovery samples were $>70\%$, there still may have been differences in the recovery among samples. Overall, although accounted for in our recovery and blank-correction, these different methods may have affected the accuracy of our results.

The demanding nature of sample analysis meant that we had a limited sample size of fish ($N=45$) and number of trophic positions, which limited our ability to examine certain trends. In avoiding overanalysis of the data, we also had limited power to investigate bioaccumulation in this study based on the data and used regressions just as an exploratory tool to begin to understand if there was evidence for the occurrence of this effect. Moreover, sampling fish from one region within Lake Ontario limits our ability to generalize about contamination in fish globally. Although these species should be largely local to the Humber Bay area of the fish collection, their potential foraging in other parts of Lake Ontario could have contributed to the variability in the microplastic levels.⁸³ Our study builds on other studies focused on investigating spatial differences and variation across species, including Munno et al. and Mcilwraith et al.^22,47 Instead of further pursuing those questions, we conducted deeper investigations into differences across fish tissues under a worst-case scenario in an area of the Laurentian Great Lakes known to have high contamination owing to its proximity to Toronto (Ontario, Canada), a large urban center and source of microplastic pollution.⁴⁷ As such, the exposure we observed here may be relevant to freshwater bodies adjacent to heavily urbanized regions and for future more in-depth studies.

Conclusions and Considerations for Future Risk Assessments

Our work shows prevalent translocation of microplastics and other anthropogenic particles in wild-caught fish, demonstrating fish fillets as a pathway for human exposure. Our work also suggests that bioaccumulation and biomagnification of microplastics do not occur in fish from the Humber Bay region, which is relevant to understanding the exposure landscape for fish. Currently, there are critical gaps in knowledge surrounding the toxicokinetics and mechanisms of toxicity for environmental mixtures of microplastics, as well as the amount of microplastics that are likely to cause an adverse effect in humans.²⁸ As such, future research should attempt to address these gaps so as to be able to contextualize exposure data.

Conclusions and opinions are those of the individual authors and do not necessarily reflect the policies or views of EHP Publishing or the National Institute of Environmental Health Sciences.