Snapshot Sampling May Not Be Enough to Obtain Robust Estimates for Riverine Microplastic Loads

Wastewater treatment plants (WWTPs) have been described as key contributors of microplastics (MPs) to aquatic systems, yet temporal fluctuations in MP concentrations and loads downstream are underexplored. This study investigated how different sampling frequencies (hourly, weekly, and monthly) affect MP estimates in a stream linked to a single WWTP. Utilizing fluorescence microscopy and Raman spectroscopy, considerable hourly variations in MP concentrations were discovered, while the polymer composition remained consistent. This temporal variability in MP loads was influenced by MP concentration, discharge rates, or a mix of both. These results show a high uncertainty, as relying on sparse snapshot samples combined with annual discharge data led to significant uncertainties in MP load estimates (over- and/or underestimation of emissions by 3.8 billion MPs annually at this site). Our findings stress the necessity of higher-frequency sampling for better comprehending the hydrodynamic factors influencing MP transport. This improved understanding enables a more accurate quantification of MP dynamics, crucial for downstream impact assessments. Therefore, preliminary reconnaissance campaigns are essential for designing extended, representative site-monitoring programs and ensuring more precise trend predictions on a larger scale.


S1: Hydrograph for discharge
. Discharge reported by Barston WWTP for the period of 14 th June -20 th June 2021.
Mean flow (green), maximum flow (purple) and minimum flow (blue) reported in 15 min intervals.
Black dots represent total discharge values at our sampling location, determined by flowmeter measurements.The offset between Barston WWTP and the sampling site (about 1 km downstream) is roughly 1.5-3 hours depending on flow velocities.It can be seen that most of the water at the downstream sampling point originates from the WWTP. - ] obtained from the triplicate samples, stream discharge [m 3 s -1 ] and MP loads [MP h -1

S3: Sample Preparation, digestion, and staining
The samples were extracted by pouring the contents of the 20 mL vials onto a 63 µm sieve (nylon mesh), after which the vials were rinsed several times with DI-water into the sieve ensuring that all contents were removed.Chemical wet peroxide oxidation was used to digest organic matter (OM) at a ratio of 1:10 [1]; 20 mL 30% H 2 O 2 was used to backwash the contents of the sieve into 250 mL beakers, before adding 2 mL Fe 2+ (aq) (0.05 M) as a catalyst.Samples were covered with loose aluminium foil lids, allowing for any heat/gas to escape, while avoiding airborne contamination, and then left for at least 24 hours at room temperature to allow degradation of the OM.
For Nile Red staining, the oxidized samples were first decanted through clean 63 µm sieves and then backwashed into beakers with DI-water.A stock solution of 1 mg mL -1 Nile Red was added to bring the sample to 5 µg mL -1 Nile Red and left covered on an orbital shaker (Grant-bio, PSU-20i) at 105 rpm to stain for one hour.The samples were then filtered using a glass vacuumfiltration system (Merck) onto glass-fibre GF/D filters (Whatman, diameter 47 mm, pore size 2.7 µm) and rinsed with DI-water.Yellow goggles and UV-light were used to inspect beakers and the filtration unit to ensure all visible particles had been removed.The filters were transferred into pre-labelled clean PP Petri dishes and dried for 24 hours at 50 °C.A set threshold of 100 fluorescence a.u.(arbitrary units) was used to select particles of interest for further consideration.Each putative MP was also observed under bright-field and had to comply with pre-set selection criteria via an identification key [2].Data collected at this stage consisted of longest length (size), colour (with particles distinctly stained as pink due distinctly

S4: Microscopy and spectroscopy
Nile Red categorised as their own category) and morphology.For polymer identification, a minimum of 40% per sampling occasion of counted putative MP was analysed (n total = 729).These particles were transferred into clean glass vials with filtered DI-water with fine tweezers before being filtered onto Whatman Anodiscs (diameter 25 mm, pore size 0.

µm). A Renishaw InVia
Qontor Raman microscope equipped with a 785 nm laser was used for polymer identification with the detailed configuration provided in Supplementary information S2 and further discussed in [3].
For spectral matching, the fingerprint region of 650-1700 cm -1 was assessed for prominent peaks.
The spectral peaks were exported into a custom Python script (Python 3.9) (Alqrinawi, F., & .https://github.com/Fuadqr/Raman-Analyzer-V1.0) and peaks were matched to reference libraries, with >70% match quality index considered as positive identification.The spectral library consisted of the SLOPP, SLOPP-E [4] and an in-house reference library (Table S2).While 88.8% of the particles were confirmed as plastics, no transformation was carried out for the final MP results, due to some organic matter being transferred unintentionally onto the Anodiscs with the suspected MPs, which can lead to added negatives that would skew the apparent accuracy of the fluorescent result [2].
For MP identification, a 5x objective was used, with slit width 65 µm, 1200l/mm, spatial resolution < 1 µm and a spectral resolution < 1cm -1 , laser intensity 10% of the system (approximately 15 mW), 3 accumulations of 5s exposures per MP particle.Cosmic ray removal, baseline subtraction (adaptive polynomial fitting) and smoothing (Savitsky-Golay filter with interval 10 and polynomial order 3) were applied using Spectragryph v1.2.16.1.The polymers tested for with the applied in-house library are listed in Table S2.
Table S2: The in-house polymer library with the polymers and number of samples for each type.
Three particles were measured from each sample with the settings specified above.For testing the extraction efficiency of the method used here, a spiking experiment was conducted.
A researcher picked fibres already identified as MP with Raman spectroscopy from Anodiscs and spiked 15 fibres (diameters 6-20 µm) into 20 mL vials with DI-water.Then a researcher added 10 particles of each of the three different polymers (PE, PLA and PS) between sizes 70 to 500 µm into the vials.These polymers were specifically chosen to represent the full spectrum from very low to high pixel brightness.The samples were extracted and analysed with fluorescence microscopy as described in manuscript section 2.3.1 and the raw results are presented in Table S3.
The researcher counting the filters noted that some original particles seemed to have started to fragment or some of the particles had smaller fragments attached to them (formation of aggregates), however all were counted.Recovery rates from the spiking experiments ranged between 67 and 80 % for fibres, with smaller diameter fibres being lost most frequently (6-8 µm).
The recovery for fragments ranged between 100-113 %, suggesting rather robust fragment capture.
It should be noted, however, that our fibre capture seems to be a slight underestimation, with an average capture rate of 78 %.As we assume that this recovery rate remained relatively stable, no transformation was carried out, as the focus of the paper is the effect of sampling scheme, rather than total number of particles and/or fibres.

S11. Assessing the coefficient of variation for hourly microplastic concentrations
To further assess and understand the variability of the collected hourly MP samples, and to assess how many hourly samples may be needed to characterize the in-stream MP concentration and obtain values close to the daily average (i.e., the mean of the 12 hourly samples per day), we determined all unique combinations of 3, 5, 8, and 10 samples for each of the four 12-hour sampling days.For each of these unique combinations we computed the mean value and then compared these values to the actual daily average (12 samples).To better represent the variation of these combinations of different mean values, we calculated the coefficient of variation (CV) for each of the combinations as follows: (Standard Deviation of CVs divided by the mean) *100%.
Higher CV values suggest higher variability and dispersion of the population around the population mean.This allowed us to visualise the dispersion around the means for the unique combinations and their respective range.

Figure S2 :
Figure S2: The respective colours of MPs extracted from the samples for: A)12-months, B) weekly and C) daily (12 x hourly samples per day over 4 days)

Figure S3 :
Figure S3: The relative fraction and abundance of fragments and fibres extracted from the water column samples for the 12 monthly samples.

Figure S4 :
Figure S4: The relative fraction and abundance of fragments and fibres extracted from the water column samples for the weekly samples.

Figure S5 :
Figure S5: The relative fraction and abundance of fragments and fibres extracted from the water column samples for the hourly samples.

Figure S6 :
Figure S6: The size ranges of the MP fragments extracted from the samples for: A) monthly sampling, B) weekly sampling and C) hourly sampling as a summary.The size fraction bins reflect standard sediment grain sizes between very fine sand and fine gravel.

Figure S9 :
Figure S9: Scatter plot showing microplastic per litre on Y axis (#MP L -1 ) and measured stream discharge on X-axis (m 3 s -1 ).Different sampling frequencies are shown with different shaped markers: triangle: 12-monthly, circular: sub-daily and diamond weekly.

Figure S10 :
Figure S10: The coefficient of variation of MP concentration means for all randomly chosen 3x1h, 5x1h and 10x1h combinations for days when hourly samples were collected.Larger CV values indicate a higher dispersion level around the mean.

Date Sampling time (hour) Stream discharge Average MP con. MP load [MP h -1 ] EC
].

Table S3 :
Results of positive blank spiking experiments.