Microplastics in Muskoka-Haliburton Headwater Lakes, Ontario, Canada

10 Microplastics (mp) are a growing environmental concern due to their ubiquity in terrestrial and aquatic 11 environments. Nonetheless, there is limited knowledge on their abundance in lakes in rural areas. In this 12 study, we surveyed 14 headwater lakes in Muskoka-Haliburton, Ontario, to assess the spatial and 13 temporal variability of microplastics. The average microplastic concentration across the study lakes was 14 1.78 mp/L during May–June 2019, with limited spatial variability (coefficient of variation = 22%). Further, 15 microplastic abundance was weakly correlated with lake area (rs: 0.469), the number of shoreline 16 residences (rs: 0.399), and watershed area (rs: 0.350), suggesting that diffusive inputs, such as 17 atmospheric deposition, were the dominant source of microplastics to the study lakes. In contrast, 18 microplastics showed a distinct temporal (seasonal) variability, as the average concentration in August 19 2019 (0.91 mp/L) was significantly lower (p<0.05) compared with May and June 2019. While microplastic 20 abundance was generally higher in the metalimnion (0.70 mp/L) and epilimnion (0.67 mp/L), there was 21 no significant difference by stratified layer. The annual percent removal of microplastics in lake sediment 22 was estimated to be 14%, suggesting that for most of the study lakes, sediment burial was not a dominant 23 sink for microplastics. Effective management of microplastic pollution requires an understanding of the 24 interlinkages between microplastics in the atmosphere, lake water, and sediment. In rural areas, 25 microplastic abundance appears to be dominated by atmospheric inputs, suggesting limited need for 26 spatial monitoring. Temporal monitoring however is required to understand seasonal changes and long27 term trends in microplastic abundance and delivery. 28


38
The mass production of plastic materials coupled with the mismanagement of waste has led to an Microplastics are of widespread concern due to the potential risks they pose to organisms (Wright et al. temporal variability of microplastics. The objective of this study was to quantify the abundance of 70 microplastics in headwater lakes in Muskoka-Haliburton. During summer 2019, a regional assessment of 71 microplastics was conducted in 14 headwater lakes with water samples being collected on multiple 72 occasions. At a subset of the study sites, lake sediment samples and water samples from each stratified 73 layer were also collected. We specifically wanted to assess the spatial and temporal variability in 74 microplastic abundance in small, rural lakes to understand their role in the microplastic cycle. 75 The Muskoka-Haliburton region is comprised of two adjacent municipalities in south-central Ontario, 78 Canada, the District Municipality of Muskoka, more commonly referred to as 'Muskoka', and Haliburton 79

Materials and Method
County. The region, which is located approximately 2.5 hours (250 km) north of Toronto, is commonly 80 referred to as cottage country as it sees more than 2.1 million tourists and visitors each year during the 81 summer period. The Muskoka-Haliburton region spans 8,041 km 2 , has a year-round population of 78,661 82 and contains more than 1,200 lakes. The region has a humid continental climate with a 30-year annual 83 precipitation average of 1,008 mm, of which 30% falls as snow (Yao et al. 2009). 84

85
The study focused on 14 small rural lakes in Muskoka-Haliburton ( Figure 1; 2007). Samples from these lakes were collected from the main and east basin for Red Chalk Lake and from 98 Hammell's Bay for Three Mile Lake. All lakes are headwater lakes with the exception of Red Chalk Lake 99 which receives discharge from Blue Chalk Lake. 100

101
Lake water samples (1.0-1.5 L) were collected by the Ministry of the Environment, Conservation and Parks 102 between May and October 2019, from the deepest point of each of the study lakes, 1-metre below the 103 surface with replicate samples (n=3) at 5 lakes (Table 1). At a subset of lakes (n=10), 1 L samples were also 104 collected from the epilimnion, metalimnion and hypolimnion during August 2020 (Table 1). These samples 105 were collected in a volume-weighted fashion from discreet depths (1 m, 3 m, 5 m, etc., throughout the 106 epi-, meta-, or hypolimnion), with thermal layers determined from a temperature profile. All samples were 107 collected using a peristaltic pump with Tygon tubing and stored in a dark refrigerator until processed. 108

109
Sediment cores were collected from a subset of lakes (n=7) between June and September 2019 (Table 1)  110 with duplicate sediment cores collected at 3 lakes. The sediment cores were collected from the deepest 111 point of each lake using a Glew gravity corer attached to a clear Lucite™ core tube with an internal 112 diameter of 7.62 cm. The sediment cores were extruded on shore using a Glew vertical extruder, an 113 extruding tray and a 4-inch stainless steel paint scraper. Before extrusion, any water present on the 114 surface of the sediment core was removed using a pipette and/or narrow plastic tubing and the first 3 cm 115 of the sediment was extruded and placed into a sterile Whirl-Pak® bag. The middle section of the sediment 116 core was discarded, and the 30-33 cm section of the sediment core was extruded and placed into a sterile 117 Whirl-Pak® bag. In three of the lakes, the sediment cores were greater than 20 cm but less than 33 cm, so 118 the last 3 cm of the sediment core was extruded. The 30-33 cm section, or the last 3 cm of the core, 119 corresponds temporally to the pre-industrial era, prior to the invention of plastic; a number of lakes in the 120 study area have been dated using Pb 210 and the pre-industrial era was determined to be at a depth of 121 between 15 and 20 cm in the sediment column (Mills et al. 2009). The stainless-steel paint scraper and  122  extruding tray were rinsed in lake water, cleaned with Kimwipes™ and rinsed three times with filtered  123  deionized (DI) water before sectioning the upper 3 cm and the 30-33 cm section (or deepest 3  five criteria used were: (i) unnaturally coloured (blue, red, purple, etc.) relative to the sample, (ii) no visible 151 cellular structure or offshoots and appears homogenous in texture and material; (iii) not brittle and 152 remains intact when poked, compressed or tugged with fine tweezers; (iv) shiny or glossy appearance; 153 and (v) no similarities to natural fibres and has limited fraying. If two or more of the criteria were met, the 154 particle was classified as an anthropogenic particle (i.e., synthetic but not necessarily plastic), the colour 155 and proportion of criteria each particle met was recorded and each anthropogenic particle was 156 photographed and measured using an image processing software (ImageJ; URL: imagej.nih.gov/ij). 157 Particles were identified down to a size class of 50 µm but smaller particles were identified where possible. 158 The size (length) of particles were manually estimated by converting the number of pixels measured to a 159 known length in millimetres. All non-fibres were referred to as fragments as there were few to no films or 160 foams identified; see Supporting Information (Section A) for further details on the size, shape, and colour 161 of anthropogenic particles. A minimum of 20% of the anthropogenic particles were randomly selected and 162 tested using the hot needle test to determine the proportion that were plastic (i.e., polymers with a 163 against the edge of the selected particle and if the particle melted, it was classified as plastic. Within the 165 lake water and sediment samples, the proportion of plastic (i.e., the proportion of anthropogenic particles 166 that melted) were used to report the number of microplastics. This study was built onto an existing long-term monitoring program that routinely used plastic containers 175 for collection. To determine potential contamination, lake blanks (n=12) were collected in the field using 176 filtered B-pure™ or DI water with the routine sample containers; plastic particles were not identified in 177 the lake blanks. Furthermore, procedural open-air blanks, i.e., open petri dishes with filter papers, were 178 routinely collected and used to determine potential contamination during the extraction and analysis 179 process in the laboratory. Open-air blanks were exposed during filtering and particle identification and 180 the period of exposure was recorded. The average potential contamination was less than one microplastic 181 per sample based on the average contamination from the open-air blanks and the time samples were 182 exposed to the air (Table 2). Accordingly, samples were not blank corrected as potential contamination 183 was low. Procedural B-pure™ water, DI water, zinc chloride and hydrogen peroxide blanks were also 184 collected by vacuum filtering a known quantity through 1.6 µm Fisherbrand™ G6 glass-fibre filter papers 185 and analyzing them following the same method as the field samples ( Table 2). The average potential 186 contamination in these blanks ranged from 0.16 mp/L to 7.38 mp/L, therefore, all solutions (i.e., DI water, 187 B-pure™ water, zinc chloride, hydrogen peroxide) were vacuum filtered prior to use in the Fe (II) solution, 188 sediment digestion, density separation, rinsing and cleaning to prevent or limit potential contamination. 189 Approximately 21 microplastics were identified in the blanks, nine were found in solutions prior to 190 filtration (i.e., all solutions were filtered prior to use), 11 were found in open air blanks with an average 191 exposure of 6.5 hours and one was found in a microscope blank with an average exposure of 42 hours. 192 The field samples had an average exposure of 0.60 hours during the filtering process and 0.17 hours during 193 microscope analysis. Although some plastic was used during the sample procedure, i.e., the sample bottles 194 were comprised of polyethylene terephthalate, the petri dishes and weigh boats were comprised of 195 polystyrene, and the Whirl-Pak® bags were comprised of polyethylene, potential contamination was 196 limited. There were no microplastics identified in the lake field blanks (n=12) and the polystyrene, 197 polyethylene and polyethylene terephthalate polymers identified in the field samples were coloured 198 indicating that they did not originate from laboratory sources. In addition, all equipment was triple rinsed 199 with filtered B-pure™ or filtered DI water prior to use and immediately covered in aluminium foil when 200 not in use. The Buchner funnel was triple rinsed with filtered B-pure™ or filtered DI water, cleaned with 201 Kimwipes™ to remove any remaining material, and triple rinsed again in between each sample. Finally, all 202 samples were collected, processed and analyzed using nitrile gloves and 100% cotton laboratory coats 203 were worn when processing and analyzing the samples. 204

205
The concentration of microplastics in lake water is reported as the number of microplastics per litre (mp/L) 206 and for sediment as the number of microplastics per gram of dried sediment (mp/g). For the lake water 207 samples collected 1-metre below the surface between May and June 2019, the average concentration of 208 microplastics in the lake was multiplied by the total lake volume to determine the total number of 209 microplastics in the lake. A Spearman's correlation test was used to determine the association between 210 the average microplastic concentration and lake attributes (number of shoreline residences, lake area, 211 watershed area, mean depth and max depth; Table 1). The variability between replicate (n=3) lake water 212 samples was determined as the coefficient of variation. For the temporal samples collected between May 213 and October 2019, the average concentration of microplastics was determined for each month by lake. 214 The concentration of microplastics in the epilimnion, metalimnion and hypolimnion for each lake was 215 calculated and multiplied by the percent of the total lake volume that each stratum comprised to 216 determine the volume-weighted concentration of microplastics. Statistical comparisons between 217 temporal periods or lake strata were generally based on a paired concentration was used to determine a rough microplastic accumulation rate (mp/m 2 /day) in the top 224 section of the sediment core for each lake. The percent of microplastics that are removed annually via 225 sedimentation was determined using the microplastic accumulation rate, lake area and total number of 226 microplastics in the lake. A Spearman's correlation test was then used to determine if there was a 227 correlation between the average microplastic concentration and lake attributes (number of shoreline  228 residences, lake area, watershed area, mean depth and max depth). 229

231
The concentration of microplastics in the study lakes (n=12) ranged from 1.02-2.39 mp/L between May 232 and June 2019 with an average concentration of 1.78 mp/L (Table 3; Table SI-5); microplastics were  233 normally distributed across the lakes ( Figure 2) with a coefficient of variation of 22%. The average 234 coefficient of variation for replicate samples (n=3) was generally consistent across the study lakes at 53%. 235 When lake volume was taken into account, the total number of microplastics in lakes across the region 236 ranged from 0.45 billion to 34.59 billion microplastics, with an average of 8.18 billion microplastics per 237 lake, and a coefficient of variation of 115%. 238 The abundance of microplastic (mp/L) in the study lakes was weakly correlated with lake area (rs: 0.469), 239 the number of shoreline residences (rs: 0.399), and watershed area (rs: 0.350). Given the low variability 240 in the concentration of microplastics between lakes (coefficient of variation of 22%), this suggests that 241 diffuse inputs such as atmospheric deposition, which is influenced by lake and catchment size, was the 242 dominant source of microplastics to the study lakes. While the number of shoreline residences may be an 243 indicator of direct anthropogenic inputs, the number of residences is also correlated to lake size, which is 244 the likely association in the study lakes given the weaker correlation to microplastic abundance. 245 Furthermore, lakes which had very few or no residences (see Table 1) and consequently limited road 246 access, had similar microplastic concentrations compared with lakes that had numerous homes and 247 cottages and year-round road access (Table SI- suggesting a slight preferential settling of microplastics within the study lakes based on plastic type (fibre 282 versus fragment) or density. 283 The average volume-weighted microplastic concentration (0.66 mp/L) was very similar to the 284 concentration in samples collected at a depth of 1-m (0.68 mp/L), suggesting that a 1-m depth sample is 285 representative of the concentration of microplastics in the lake, although the 1-m samples showed greater 286 variability between lakes (49% versus 26%; Figure 4b). It is worth noting that the abundance of 287 microplastics in the study lakes was significantly lower (p<0.05) in August 2020 compared with August 288 2019, again high lighting the importance of capturing temporal observations (Figure 4c; Table SI-8). 289 The lack of a distinct vertical distribution in the concentration of microplastics in the study lakes is similar 290 to Tamminga and Fischer (2020), who examined a dimictic lake in northern Germany and found that the 291 concentration of fibres did not display a noticeable vertical distribution. They did, however, find that the 292 concentration of irregular particles displayed a distinct vertical pattern, decreasing with depth, but this 293 vertical difference was not significant (Tamminga and Fischer 2020). Similarly, Lenaker et al. (2019) found 294 that the concentration of microplastics at five out of the six river and lake sites, were not significantly 295 different between the surface and subsurface. 296

297
In the top 0-3 cm section, the concentration of microplastics ranged from 2.17 to 5.45 mp/g with an 298 average of 3.66 mp/g and a coefficient of variation of 35% ( Figure 5; Table SI-9). 299 There was a weak correlation between the concentration of microplastics in the top 0-3 cm section and 300 the number of cottages on the lake (rs: 0.487). In the bottom section of the sediment core, the abundance 301 of microplastics was significantly lower (p<0.05) than the top section, with an average concentration of 302 0.72 mp/g and a coefficient of variation of 47%. The presence of microplastics in the bottom 30-33 cm 303 section of the sediment core, which corresponds temporally to the pre-industrial era, suggests that 304 microplastics can migrate or mobilize down the sediment column. 305 The average relative percent difference between duplicate cores (n = 3) for the top 0-3 cm section was 306 20% and for the bottom 30-33 cm section it was 42%, suggesting slightly higher variability in microplastics 307 abundance in sediment between the lakes than between duplicate cores. Furthermore, the bottom 308 section of the sediment core had greater variability in microplastic abundance compared to the top. Given 309 an average rate of sedimentation for the study lakes (Mills et al. 2009), the average accumulation of 310 microplastics in the top lake sediment was 1.78 mp/m 2 /day (1.06-2.65 mp/m 2 /day) with a coefficient of 311 variation of 34%. As such, the percent of microplastics buried annually into the lake sediment ranged from 312 1.1-63.6% with an average of 14% and a median of 5.3%. This suggests that for most lakes (Table SI-9),  313 lake sediment is a small sink for microplastics with the majority of microplastics leaving the lake via the 314 outflow. 315 The concentration of microplastics in the top section of the sediment core for the study lakes (2,170-316 5,450 mp/kg) is much higher than the concentration of microplastics in sediment observed in urban lakes 317 such as Lake Simcoe (1.24-160 mp/kg), Lake Erie (0-391 mp/kg), Lake Ontario (40-4,270 mp/kg) and lakes 318 located in Lake Mead National Recreation Area (87.

328
In the lake water and bottom section of the sediment core, fibres were the dominant anthropogenic 329 particle shape identified (>70%) whereas in the top section of the sediment core, fibres were only slightly 330 more common than fragments (57%; see Supporting Information Section A). There was no noticeable 331 difference in the proportion of fibres and fragments across the lakes depths with fibres being relatively 332 evenly distributed (51-67%). 333 Of the anthropogenic particles that were identified in the lake samples and tested using a hot needle, 14% 334 melted and were identified as plastic. In the lake sediment, 36% of the anthropogenic particles identified 335 in the top 0-3 cm section melted while only 13% in the bottom 30-33 cm section melted. Thermoplastics 336 were the predominant plastic type identified in the study lakes with polypropylene being most abundant 337 (26%) followed by acrylonitrile butadiene styrene (21%) and polyethylene (16%). In the lake samples, 338 there was an equal proportion (22%) of polypropylene, polyethylene and polyisoprene identified with 339 polyamide, polyacrylamide and polyethylene terephthalate also being identified. In the lake sediment, the 340 majority of particles were acrylonitrile butadiene (40%) followed by polypropylene (30%). Polyamide, 341 polyethylene, and styrene acrylonitrile were also identified but were less common. 342

343
There was limited spatial variability in the concentration of microplastics across the study lakes with 344 microplastic abundance being weakly correlated with lake area (rs: 0.469), the number of shoreline 345 residences (rs: 0.399), and watershed area (rs: 0.350). This suggests that atmospheric deposition is the 346 dominant source of microplastics to the study lakes, and therefore, limited spatial monitoring can 347 adequately quantify microplastic abundance in headwater catchments in rural regions. In contrast, 348 temporal monitoring is required as a distinct seasonal variability in microplastics abundance was 349 observed, with August 2019 having a significantly lower concentration compared with May and June 2019. 350 This suggests that seasonal changes in diffuse sources or hydrological processes influence the abundance 351 of microplastics within headwater lakes in rural areas. 352 Sediment has been identified as an important long-term sink for microplastics in freshwater 353 environments; however, in this study, the percent of microplastics buried in the lake sediment annually 354 was relatively low. This suggests that lake sediment is not a dominant sink for microplastics, but rather 355 that small rural lakes are reservoirs for export to downstream rivers or lakes. Few studies, however, have 356 evaluated the residence time of microplastics in lakes, which is an important factor in the microplastic 357 cycle as it identifies the length of time microplastics can stay (reside) in a lake. Future monitoring and 358 research in rural lakes should focus on a catchment balance approach for microplastics, including 359 atmospheric inputs, in order to better understand the fluxes and cycle of microplastics in lakes. This will 360 help to improve our understanding on microplastics and will allow for effective policy decisions to address 361 microplastic pollution. 362

363
Allen S, Allen D, Phoenix VR, Le Roux G, Durántez Jiménez P, Simonneau A, Binet S, Galop D (2019) 364 Atmospheric transport and deposition of microplastics in a remote mountain catchment. Nat 365 Geosci  Box-plot: The black line represents the median, the box represents the 25th and 75th percentile and the 522 whiskers represent 1.5 times the interquartile range 523 Fibres were the dominant shape identified in lake water and in the bottom 30-33 cm section of the 559 sediment core with more than 70% of the identified anthropogenic particles being fibres ( Figure SI-1). In 560 the top 0 -3 cm section however, fibres were only slightly more dominant than fragments (57%). Across 561 the different lake depths, there was no noticeable difference in the proportion of fibres and fragments 562 across the water column with fibres being relatively evenly distributed (51-67%). The even distribution of 563 fibres across the water column suggests that the specific density of polymers may be less important for 564 fibres (UNEP 2020). 565 Anthropogenic particle size ranged from 20 µm to 4,990 µm across the study sites, with lake water having 566 the smallest fibre and fragment sizes while lake sediment had the largest (Table SI-1). This suggests that 567 large sized particles have a tendency to sink to the lake bottom while smaller particles remain suspended 568 in the lake. Within the lake sediment, the size of fibres in the 0-3 cm and 30-33 cm sections were generally 569 similar; however, the size of fragments in the 30-33 cm section were about 2 times greater than the size 570 of fragments in the 0-3 cm section. The dominance of smaller particles (fibres and fragments) in the study 571 suggests that particles are likely secondary anthropogenic particles that arose through the weathering 572 and breakdown of large plastic pieces currently in the environment (Kooi and Koelmans 2019). 573 Across the depth samples, the mean and median anthropogenic particle size generally increased down 574 the water column (Table SI-2). The proportion of anthropogenic particles that had a size less than 400 µm 575 also decreased down the water column while the proportion of particles that had a size greater than 1,000 576 µm increased (Table SI-3a). At a depth of 1-metre and in the epilimnion, approximately 64% of 577 anthropogenic particles had a size less than 400 µm while only 14% of particles had a size greater than 578 1,000 µm while in the metalimnion and hypolimnion, less than 62% of particles had a size less than 400 579 µm and more than 17% had a size greater than 1,000 µm. There was also a significant difference in the 580 proportion go anthropogenic. There did not appear to be a difference in the proportion of fragments that 581 fell within each size across the lake depths (Table SI-3b) but for fibres, there was a larger proportion of 582 fibres which had a size less than 400 µm in the epilimnion (57%) and at 1-metre (46%) whereas there was 583 a larger proportion of fibres that had a size greater than 1,000 µm in the metalimnion (33%) and 584 hypolimnion (43% ; Table SI-3c). 585 Blue was the most common coloured anthropogenic particle identified across all sample media with more 586 than 72% of fibres and 60% of fragments being blue (Table SI-4). White was also commonly identified for 587 fragments (15%) but was less common for fibres (0.12%). White fragments were common in the sediment 588 samples with 21% of the fragments identified in the 0-3 cm section and 52% in the 30-33 cm section 589 being white. This could suggest that white fragments are a specific type of polymer that tends to sink or 590 that the fragment lose their colour during the digestion process. Bright eye-catching colours, such as blue, 591 may be more easily recognized during visual identification compared to dull colours and thus 592 overestimated  -3cm section and  595 bottom 30-33 cm section) and depth samples (1-metre, epilimnion, metalimnion, and hypolimnion) 596 collected from the study headwater lakes in Muskoka-Haliburton, Canada 597 Table SI-1 The mean and median size (µm) as well as the range (µm) and coefficient of variation (CV) (%) 598 for anthropogenic particles (microfibres and fragments combined) microfibres only, and fragments only, 599 collected from atmospheric deposition, lake water, and sediment collected from headwater lakes in 600 Muskoka-Haliburton, Canada 601 Table SI-2 The mean (µm), median (µm) and range (µm) in anthropogenic particles observed in the depth 602 samples collected from the epilimnion, metalimnion, hypolimnion and 1-metre below the surface from 603 each of the study lakes in Muskoka-Haliburton, Canada as well the coefficient of variation (%) 604 Table SI-3a The proportion (%) of anthropogenic particles identified in lake water samples collected from 605 the epilimnion, metalimnion, hypolimnion and 1-metre below the surface that fall within each size class 606 Table SI-3b The proportion (%) of fragments identified in lake water samples collected from the 607 epilimnion, metalimnion, hypolimnion and 1-metre below the surface that fell within in each size class 608 Table SI-3c The proportion (%) of fibres identified in lake water samples collected from the epilimnion, 609 metalimnion, hypolimnion, and 1-metre below the surface that fell withing in each size class 610 Table SI     Muskoka-Haliburton, Canada (taking into account the percent total volume that each stratum comprises) 661 Table SI