Microplastics are ubiquitous in our environment and are found in rivers, streams, oceans, and even tap water. Riverine microplastics are relatively understudied compared to those in marine ecosystems. In Oregon, we sampled eight sites along four freshwater rivers spanning rural to urban areas to quantify microplastics. Plankton tow samples from sites along the Columbia, Willamette, Deschutes, and Rogue Rivers were analyzed using traditional light microscopy for initial microplastic counts. Application of Nile Red dye to validate microplastics improved microplastic identification, particularly for particles (Wilcox Test; p-value=0.001).  Nile Red-corrected microfiber abundance was correlated with human population within five kilometers of the sample site (R²=0.554), though no such relationship was observed between microparticles and population (R²=0.183). This study finds plastics present in all samples from all sites, despite the range from undeveloped, remote stretches of river in rural areas to metropolitan sites within Portland, demonstrating the pervasive presence of plastic pollution in freshwater environments.

Eight study sites spanned four rivers in Oregon, ranging from the Columbia River mainstem at the Washington border in the north to the Rogue River in Southern Oregon (Figure 1). Rivers vary in length from the Columbia River spanning 2,010 km to the Deschutes River spanning 280 km (Appendix Table 1). Plastic pollution from point and nonpoint sources is an emerging concern in all four rivers, especially given the cultural and ecological importance of fish species that rely on these habitats for spawning, rearing, and migration. Samples were collected at eight locations: the Columbia River near St. Helens, the Willamette River near Fall Creek (upstream), Albany (midstream), and Portland (Oregon Museum of Science and Industry [OMSI] dock; downstream), the Rogue River near Woodruff Bridge (upstream) and in Grants Pass (downstream), and at Big River (upstream) and Tumulo (downstream) on the Deschutes River (Figure 1). Sites were chosen based on a) their proximity to WWTP and urban centers: OMSI in downtown Portland, Albany on the Willamette River, and Grants Pass on the Rogue River or b) their remote locations: Fall Creek on the Willamette, Woodruff Bridge on the Rogue, and Big River on the Deschutes.

Between September 7^th and 14^th, 2018, three samples plus a field control were collected from each of the eight sites (n=24) using a General Oceanics plankton tow net with a 0.5m mouth and 200 µm mesh size equipped with a flow meter. Based on previous research studying riverine surface water microplastics using a sample depth of 0.15m (Yonkos et al. 2014), our net was submerged in the river approximately 0.3 to 1m below the river surface (depending on river depth and access conditions) for 15 min to sample subsurface flow (Lenaker et al. 2019). Excess water exited through the mesh netting while debris and plankton were trapped in the cod-end (Vinzant 2016). Samples were poured and the cod-end rinsed with deionized (DI) water into pre-rinsed glass jars for transport back to the Applied Coastal Ecology lab at Portland State University. At each site, a control jar was open during sampling to collect any airborne plastic particles. The total water volume sampled from each river varied greatly, however water volume of each sample collected and rinsed from the net cod-end was approximately 680 mL. Flow was recorded as rotor revolutions (converted to “counts”) at each sampling event (see Appendix Table 2), and this value was used to calculate sample volume (m³) using the equation: area [(3.14 x (net diameter)²)/4] x distance [(counts x rotor constant)/999999], based on the General Oceanics flow meter user guidelines.

The plankton tow samples contained a significant amount of biological material making microscope inspection of microplastics difficult. To avoid misidentification, a potassium hydroxide (KOH) digestion was performed to remove naturally-occurring biological material from the samples (Rochman et al. 2015). Each sample was filtered with a 200 µm strainer and the remnants were rinsed into a beaker with 400 mL of filtered DI water and a 10% potassium hydroxide (KOH) solution (Rochman et al. 2015). Covered beakers sat on a 60^oC hotplate with a stir bar for 24 hrs before being filtered into a petri dish. Samples that remained murky after the first digestion were split into two petri dishes and were digested a second time to increase clarity. Despite these extra steps, many samples remained muddy so density separation was utilized to effectively isolate the plastics from the biological material (Masura et al. 2015).

Samples were rehydrated, scraped with a shucking tool to loosen the sample from the bottom of the dish, then added to a hypersaline solution with a ratio of 168.4 g of salt (NaCl) to 2 L of water. Jars were sealed and shaken vigorously for 60 sec, then returned to the lab bench for the contents to separate and stratify. Since the hypersaline solution causes heavier sediment particles to sink to the bottom of the jar, while the lighter plastic particles floated to the top (Thompson et al. 2004), heavy plastic particles may have been lost during this step (Crichton et al. 2017). Once the solution had stratified, the liquid was removed using a vacuum filtration set up: a 2 L glass Erlenmeyer flask connected to the sink faucet by a rubber tube (see Appendix Figure 1) with a glass filter (Whatman 1820-047 Glass Microfiber Binder Free Filter, 1.6 Micron, 4.3 s/100mL Flow Rate, Grade GF/A, 4.7cm Diameter, Amazon) atop. The quart sample jar was then opened and the top layer was poured out to ensure that the plastics were filtered but no sediment was included. Once the water in the beaker was sucked into the Erlenmeyer flask, and the plastic particles were left on the filter paper, it was lifted and transferred to a new petri dish using Excelta 5-SA stainless steel precision tweezers. Petri dish lids were secured with two rubber bands, and the filter papers were stored in a cardboard box for microscope analysis. All glassware in the vacuum setup was rinsed twice with DI water between samples. Nitrile gloves and cotton lab attire were worn during processing to minimize contamination.

Initial microscope analysis (methodology adopted from the Marine & Environmental Research Institute “Guide to Microplastic Identification” nd) differentiated the suspected microplastics by color.  Each filter was viewed on a Leica MZ6 light microscope using 40x magnification. Per method protocol, each filter was scanned in its entirety.  The physical characteristics of each suspected microplastic were assessed using precision tweezers to test malleability. Parameters including thickness, homogeneity of color, and presence/absence of cellular structures were assessed visually to differentiate plastic from natural materials (“Guide to Microplastic Identification” nd, Masura et al 2015). Each suspected microplastic was photographed and shape and color were recorded. While assessing each filter, a petri dish with DI water sat at the back of the microscope to collect any potential contamination from the microscope lab room. Each control dish was analyzed under the scope after its corresponding filter paper sample, and contamination was recorded. This procedure was repeated for each filter paper and control pre- and post-Nile Red dye application (April-May, 2019). Field and lab controls were calculated and reported as average microplastic contamination per site (See Appendix Tables 3 and 4).

One mg Nile Red (Santa Cruz Biotechnology, SC-203747C) was mixed with 1 mL acetone to create a stock solution, that was diluted with 100 mL of hexane to create a working solution of 10 µg Nile Red/mL (Wiggin & Holland 2019). After thorough mixing with a stir bar for 3 to 5 hours, the working solution was transferred into an amber dropper bottle, and the solution was applied to each filter paper until coated (about nine drops) and allowed to dry on a 12-hr, 30°C cycle in a drying oven (Wiggin & Holland 2019). Microscope analysis was repeated for each filter paper and microscope control post Nile Red dye application. To create proper light conditions for fluorescence, the lab room was completely dark and orange safety goggles were taped under the microscope lens to create an orange viewing environment. A 455 nm LED flashlight (Arrowhead Forensics PART NO: A-6994FK) was used to illuminate the samples (Figure 2), causing fluorescence (Wiggin & Holland 2019).

All glassware and lids were rinsed twice with DI water to avoid microplastic contamination. Glassware was inverted or covered if not in use, and controls were used both in the field and lab to quantify contamination. Proper lab attire included nitrile gloves, 100% cotton t-shirt and lab coat to avoid contamination. The following controls were included to account for microplastic contamination during field collection and lab processing: a mason jar was left open during field sampling, and again in lab during the hypersaline procedure; a petri dish was left open during each microscope analysis, and an open dish was left in the oven during the drying cycle (Baechler et al. 2019). Contamination in the above controls was summed and reported per site (Appendix Tables 3, 4).

All Nile Red statistical analyses were conducted in R Studio version 1.1.453.  To test for significant differences, nonparametric t-tests were run between the number of fibers and particles before and after dye application. Shapiro Tests revealed data were abnormal, thus the non-parametric Wilcox Test was used to compare microplastic counts before and after dye application (significance level of <0.05). Tests revealed more microplastics post Nile Red dye than initially counted, so Nile Red “after” counts were used for the site population comparison. Collection site GPS coordinates were used to determine population estimates within a 5 km radius of each location using Population Estimation Service, a web-based GIS tool developed by NASA’s Center for International Earth Science Information Network (CIESIN) (CIESEN, 2019). Population estimates were derived from the Gridded Population of the World (GPW) v4.11 developed by the Socioeconomic Data and Applications Center (CIESIN, 2019). Population and site information were projected onto a map using ArcGIS Desktop version 15.5.1. (ESRI, 2017). Aggregated daily data from the collection date based on United States Geological Survey (USGS) or Oregon Water Resources Department (OWRD) flow meters near each sampling location were used to identify flow at or near the sample locations. Microplastics concentration data after Nile Red and the flow data converted into m³/sec were used to calculate microfibers/sec or particles/sec. We multiplied the per second counts by 3600 (60 sec/min X 60 min/hr) to calculate the number of microfibers/microplastics flowing through the sampling location hourly. Microfiber and microparticle per hour data were regressed onto the 5 km radius human population estimate data using simple linear regression in R Studio. Variables were plotted and fitted with regression lines to explore the strength of linear relationships.

Two samples were excluded from data analysis. "Eugene 1" did not have a flow value recorded at sampling and the flow value for "W. Rogue 2" was recorded incorrectly.

Several intermediary steps are needed to calculate water volume from the rotor revolutions flow value. First, the rotor revolution value is multiplied by 10 to get "counts." The next value needed is distance in meters, which can be found by the equation [(counts x rotor constant)/999999]. The provided rotor constant value is 26,873. Finally, the equation for volume in cubic meters is {[(3.14 x net diameter squared)/4] x distance}. The net diameter would be 0.5 meters. This is how the "cubic meters of water" column was generated from the "rotor revolutions" column. Values for average fiber or particle per cubic meter were plotted by site to create Figure 3 and Table 1. The calculated flows were also used to create the average contamination per cubic meter values per site used in Appendix Table 3.

The flow meter values in cubic feet per second were used to calculate the number of microfibers/microparticles flowing through the sample location hourly. Microfiber and microparticle per hour data were regressed onto the 5 km radius human population estimate data using simple linear regression in R Studio. Variables were plotted and fitted with regression lines to explore the strength of linear relationships in Figure 4.