Investigating the potential uptake of microplastic-derived carbon into a boreal lake food web using carbon-13 labelled plastic

This experiment took place in two limnocorrals (i.e., in-lake enclosures) in a natural boreal lake at the International Institute for Sustainable Development Experimental Lakes Area (ELA; Ontario, Canada). Our study lake, Lake 378 (49°42′38″N, 93°46′32″W), is an oligotrophic temperate lake with a surface area of 251 579 m² and maximum depth of 16.6 m and has low background levels of microplastics (McIlwraith et al. 2024). This study is one of a series of experiments being conducted under the “pELAstic Project” over several years to understand the fate and effects of microplastics in lake ecosystems (https://thepelasticproject.com/). Because this is the first time, to our knowledge, that ¹³C-plastic was intentionally added to a natural aquatic ecosystem, the study was intended to be exploratory in nature.

In May 2021, two 1 m wide open-bottomed limnocorrals were set up in the littoral zone at a depth of 1 m in ELA Lake 378 (Supporting Information, Part A: Fig. S1). The limnocorrals were constructed with a polyvinylchloride floatation ring attached to a food-grade polyethylene curtain (Curry Industries, Winnipeg, Manitoba, Canada) that was sealed to the sediment using a double ring of sandbags. The food-grade polyethylene curtain was used to limit the potential for plastic C to leach from the curtains into the limnocorral water. These limnocorrals were open to the lake sediments and the atmosphere so that they enclosed the natural sediment, water, and biota from the lake. The food-grade polyethylene “walls” of the limnocorrals were impermeable such that there was no exchange between water in the lake and the limnocorrals once sealed to the lake bottom. The estimated volume of each limnocorral was 1100 L. A tracer was not used in the present study to monitor leakage from the limnocorral or exchange of water between the limnocorral and the lake through sediment. However, based on previous studies using the same design and set up techniques, leakage through these mechanisms was anticipated to be negligible and we do not expect diffusion through sediment to be a relevant route of exchange that would significantly dilute our ¹³C-labelled plastic signature in the current study (Orihel et al. 2006; Graves et al. 2021).

For this experiment, we used 99% ¹³C-labelled PS that was synthesized and ground to a powder of microplastic fragments (Supporting Information, Part A: Fig. S2) by MilliporeSigma (Burlington, Massachusetts, USA). We used carbon-13 labelled PS because it allows us to clearly differentiate between plastic-derived C and natural C that is composed of 98.9% carbon-12. Prior to addition, we counted and characterized the ¹³C-PS fragments in the laboratory. Briefly, the size range and concentration of ¹³C-PS was characterized by first transferring the weighed and ground ¹³C-PS into a clean 1 L nalgene bottle with 100 mL of multi-Q ultrapure water. Two small scoops of Sparkleen™ low foam lab detergent were added to prevent clumping of microplastics. A 0.5 mL aliquot of ¹³C-PS solution was pipetted onto a 0.2 um polycarbonate filter paper. Particles were counted under a microscope at 80x magnification with an eye piece at 10x magnification. The subsampling and counting procedure was done for a total of five replicates, and the counts per replicate were <10% different from the mean. Following counting, photographs of the field of view with microplastics were taken for each replicate and later loaded in ImageJ to measure the length of each particle. Since the fragments were irregularly shaped, measurements were standardized by measuring only the longest side of each particle. Over 200 particles were measured across the five replicates counted and particles ranged in size from 8 to 216 µm (mean: 53.4 µm; median: 50.9 µm) (Supporting Information, Part A: Fig. S3).

To one limnocorral, we added 50 mg (3268 000 particles or 3268 particles/L) of ¹³C-PS. The ¹³C-PS was first mixed into 1 L of lake water in a high-density polyethylene (HDPE) Nalgene container in the laboratory. This solution was then added to the mesocosm by uncapping the 1 L bottle underwater and gently shaking the bottle, held horizontally, to release the plastic into the mesocosm. The bottle was rinsed with mesocosm water and the process repeated twice more. The other limnocorral served as a control where no plastic was added. To ensure that enough ¹³C-labelled PS was being added to the limnocorral to detect degradation of microplastics, prior to beginning the field experiment, we used background C information from the lake and previously calculated rates of microplastic degradation from a previous laboratory study by Zhu et al. (2020) to estimate the projected rate of C leaching from plastic. From this, we estimated the anticipated enrichment of ¹³C in water column dissolved C (see calculations in Supporting Information, Part B). For these calculations, we assumed that microplastics would mix approximately evenly within the water column of the 1 m wide by 2 m deep limnocorral. Polystyrene is a neutrally buoyant microplastic and we assumed, based on this feature, that most PS would remain in the water column. Elagami et al. (2022) estimated the residence time of PS under typical lake conditions and found that particle size had great influence on water residence time. Based on their models, our range of ¹³C-PS particles was expected to remain in the water column up to 10 000 days (Elagami et al. 2022). As such, our sampling efforts focused on collecting water and biota from this compartment within the limnocorrals.

Since the degradation of microplastic is a very slow process (Taipale et al. 2019, 2022, and 2023), we aimed to conduct a long-term (4 month) study to detect the leaching of C from microplastics under natural conditions. The frequency and timing of sampling collection were based on previous work by Taipale et al. (2019) showing that the mineralization of microplastic is a slow process and that in a closed system, the total amount of plastic-derived C, and associated changes in δ¹³C, should increase over time, such that conducting a long experiment is necessary to detect C leaching from plastic. Sampling times throughout the experimental period are referred to as weeks pre- or post-additions, with week 0 being the week of the ¹³C-PS addition. Starting on 17 May (week −1, e.g., 1 week prior to additions), one integrated water sample was taken from the middle of the water column (0.5 m) in each limnocorral using a peristaltic pump for δ¹³C analysis of DOC, DIC, and for water chemistry. These water samples were collected on weeks −1, 1, 4, 8, 12, and 17 of the experimental period. For δ¹³C-DOC and δ¹³C-DIC, an inline filter with stacked 0.7 µm GF/F and 0.2 µm polycarbonate filters were used to minimize biological activity, and to exclude microplastics from water samples. Water (60 mL) was collected into acid-cleaned HDPE bottles with no head space and transported to the laboratory on ice. Water samples were immediately preserved in ZnCl₂ (0.6 mL of 50% w/v) and stored at 4 °C until further analysis. Additional water samples were collected to measure DIC, DOC, alkalinity, and conductivity. For these samples, water was again collected from 0.5 m depth in the limnocorral using a peristaltic pump. For DIC, water was collected into a 150 mL HDPE septum cap bottle with no headspace. For DOC, samples were collected in a 1 L HDPE bottle and transported to the laboratory where they were filtered through 0.7 µm GF/F filter. The DIC and DOC were analyzed in-house following Stainton et al. (1977). Waters for alkalinity and conductivity were collected into 150 mL septum cap HDPE bottles and measured in-house at the IISD-ELA.

In addition to water, one composite sample of plankton and one of periphyton were collected for isotopic analyses at the end of the experiment (i.e., week 17) so that biota had the entire 4 month experimental period to incorporate plastic-derived C. We collected bulk plankton by towing horizontally in each mesocosm with a 53 µm mesh tow net. We also suspended strips (made of the same polyethylene material as the walls; 10 cm wide × 1 m deep) vertically in each limnocorral to collect periphyton, which were subsequently recovered at the end of the experiment. Biota samples were flash-frozen in liquid nitrogen and then stored at −80 °C until further processing for CSIA. To investigate enrichment of ¹³C in biota as evidence of MP-derived C incorporation, we employed amino acid CSIA rather than bulk isotope analysis because organisms could inevitably contain ¹³C labelled microplastics. Bulk isotope analysis would not be able to differentiate microplastic particles and from biota tissue and could inflate our ¹³C value.

To minimize contamination of the ¹³C-labelled PS in the control limnocorral during sampling, limnocorrals were always sampled in the order of control followed by treatment. Separate equipment including different sets of tubing, filter holders, forceps were used for each limnocorral and were kept in separate storage containers. Gloves were worn at all times during sampling and a clean hands/dirty hands approach was used to minimize contamination of equipment with water from the ¹³C-PS limnocorral.

Carbon stable isotope ratios in DIC and DOC were measured at the Environmental Isotope Laboratory at University of Waterloo (UW-EIL) using continuous flow-isotope ratio mass spectrometry (CF-IRMS) according to methods described by St. Jean (2003) and Stainton et al. (1977). Carbon stable isotopes are reported as parts per thousand (per mille or ‰) relative to Vienna PeeDee Belemnite and are expressed using the notation δ¹³C. Quality assurance/quality control measures included duplicate runs of every fourth sample with an average percent difference (±standard deviation) of 1.6% (±1.8%) and an analytical accuracy of ±0.2‰.

Carbon stable isotopes of amino acids (δ¹³CAA) were analyzed at the Dalhousie University Stable Isotope Biogeochemistry Laboratory using methods from Silfer et al. (1991) and modified in Chen et al. (2022). Briefly, ∼5 mg aliquots of the plankton and periphyton samples were hydrolyzed in 2 mL of 6 N HCl at 110 °C for 20 h. Hydrolysates were spiked with a norleucine (Nle) internal standard and purified using cation exchange chromatography (Dowex 50WX8 200-400 H). Amino acids were converted to isopropyl-trifluoroacetyl derivatives and further purified using solvent phase extraction (chloroform and buffered phosphate). Samples were injected in triplicate on a Thermo Trace 1310 gas chromatograph coupled via GC Isolink II and Conflo IV interfaces to a Thermo Delta V isotope ratio mass spectrometer. Sample injections were bracketed by triplicate injections of an external standard consisting of a synthetic mix of 13 amino acids prepared in the same way as the sample hydrolysates. Measured δ¹³CAA values were corrected for instrument drift and linearity effects and calibrated to the external amino acid standard using equations in Silfer et al. (1991) to correct for additional C atoms introduced during derivatization. Analytical reproducibility, as measured from the standard deviation of sample triplicate injections, averaged 0.5‰ across all amino acids. As a check on data accuracy, the calibrated δ¹³C values of Nle, as well as all 13 amino acids in laboratory running standards (chlorella powder and fish muscle) co-prepared with the samples, averaged within 0.5‰ of known values.

The R software environment (version 4.3.2; Posit team, 2024) was used for all statistical analyses and alpha was set a 0.05 for all statistical tests. Normal distribution of data was confirmed using Shapiro–Wilks normality tests. We highlight that the present study was designed to explore the use of ¹³C-labelled microplastic to investigate fate under realistic environmental conditions and that the interpretation of our results is limited given the lack of replication in this experiment. As such, we focused on the enrichment of ¹³C in the treatment limnocorral as an indicator that MP-derived C was present in DIC, DOC, or amino acids. Analysis of covariance was used to determine if there were differences in δ¹³C-DIC or δ¹³C-DOC over time between the control and ¹³C-PS limnocorrals. For amino acids, one-tailed, paired t  tests were used to determine whether treatment values were significantly ¹³C enriched relative to controls. Further, the available literature was used to determine the ranges of values measured for these samples in boreal lakes to estimate ranges of natural variability for these endpoints (Supporting Information, Part A: Tables S1 and S2; Bade et al. 2004; Zigah et al. 2011; Larsen et al. 2013). Enrichment of ¹³C beyond natural variability in DOC, DIC, and amino acids was used as an additional indicator of MP-derived C incorporation. Following the completion of our experiment, we used a more recently published degradation rate (Vesamaki et al. 2022) to re-calculate the projected rate of C leaching from plastic in our limnocorral (see calculations in Supporting Information, Part C) to obtain a second projected rate for comparison to our field study. Finally, we estimated the degradation rate of microplastics in the current study using the relative difference in δ¹³C-DOC in the control versus treatment limnocorral after 4 months (see calculations in Supporting Information, Part D).

Physicochemical parameters varied over the experimental period and the conditions were similar across the control and treatment limnocorrals. Temperature in the water column of the limnocorrals ranged from 13.7 to 28.0 °C in the control and 13.9 to 28.1 °C in the ¹³C-PS treatment over the experimental period (Supporting Information, Part A: Fig. S4). Dissolved inorganic carbon ranged from 127 to 211 µmol/L in the control and 129–182 µmol/L in the treatment limnocorral over time (Table 1; Supporting Information, Part A: Fig. S5). DIC increased over time in both the control and treatment limnocorrals; although DIC appeared to increase more rapidly in the control, the pattern was only marginally nonsignificant (test for homogeneity of slopes, F_3,8 = 6.41, p = 0.06, R² = 0.52, Table 1; Supporting Information, Part A: Fig. S5). DOC ranged from 606 to 762 µmol/L in the control and 593 to 978 µmol/L in the treatment limnocorral (test for homogeneity of slopes, F_3,8 = 33.2, p = 0.005, R² = 0.87, Table 1; Supporting Information, Part A: Fig. S5); DOC increased over time in the control and treatment limnocorrals, but the increase was greater in the treatment limnocorral (Table 1). Alkalinity also increased over time in both limnocorrals (Table 1; Supporting Information, Part A: Fig. S5).

Carbon stable isotope ratios of DIC ranged from –13.4‰ to –8.2‰ in the control and –15.5‰ to –10.2‰ in the treatment limnocorral over the 17-week experiment (Table 1). Similarly, δ¹³C-DOC values changed little over the experimental period, ranging from –28.1‰ to –26.9‰ in the control and –29.3‰ to –25.9‰ in the ¹³C-PS treated limnocorral (Table 1). Within individual collection times, the greatest difference in both δ¹³C-DIC and δ¹³C-DOC between the treatment and control values was 2.1‰ (Fig. 1). There were no statistically significant differences in the slope or intercept of the relationship between δ¹³C and time for DIC (test for homogeneity of slopes, F_3,8 = 0.823, p = 0.517, Fig. 1) or DOC (test for homogeneity of slopes, F_3,8 = 0.491, p = 0.699, Fig. 1).

Based on the amount of ¹³C added to the limnocorral as microplastics, and the measured amount of C in the limnocorrals present as DOC, we estimated that DOC leached from microplastics at a rate of approximately 0.69% year⁻¹ and, based on this leaching rate, the total mass of DOC leached from plastic over the 4-month experiment was 0.11 mg (see calculations in Supporting Information, Part D).

We confirmed that δ¹³CAA values of plankton (Fig. 2) and periphyton (Fig. 3) were most similar to that of bacteria and microalgae by comparison with Larsen et al. (2013; Table S2). We observed no changes in δ¹³C of amino acids that were greater than the range in natural variability expected for these biota (Figs. 2 and 3; see Larsen et al. 2013 for comparison of biota δ¹³CAA values). The largest difference between the ¹³C-PS treated limnocorral and the control was 1.7‰ for Threonine in plankton (Fig. 2), and 1.66‰ for Glutamic acid in periphyton (Fig. 3). Although the differences in δ¹³CAA between control and treatment samples were small (∼0.2 to 2‰; Figs. 2 and 3), the δ¹³CAA values were significantly greater in the ¹³C-PS treated limnocorral relative to the control for both plankton (t  test, t = 2.036, p = 0.031, Fig. 2) and periphyton (t  test, t = 1.810, p = 0.047, Fig. 3). For plankton, 11 out of the 14 measured amino acids were enriched in ¹³C and the mean difference between control and treatment δ¹³CAA was 0.50 ± 0.91‰. Similarly, for periphyton, 10 out of the 14 measured amino acids were enriched in ¹³C and the mean difference between control and treatment δ¹³CAA was 0.49 ± 1.01‰ (Fig. 3).