London, Canada lies within the Thames River watershed of the Laurentian Great Lakes Basin (Figure 1). In 2024, the estimated population of the London metropolitan area was 626,000, with an area encompassing 2,661 km² (Government of Canada, 2025). The major economic sectors in the region include agri-food, manufacturing, digital media and technology, and health (LEDC, 2025). London’s municipal government regularly updates design standards for stormwater management infrastructure.

We surveyed 28 stormwater ponds for MaP debris between August and November of 2022 (Table 1). The ponds were grouped according to geographic location into northwest (NW), northeast (NE), southwest (SW), and southeast (SE) quadrants, as delineated by major roads within the city (Figure 1). The number of MaP survey quadrats at each pond mirrored the number and location of sites at which bottom sediment was collected for a separate investigation of microplastics as described in Corcoran et al. (2025). The number of quadrats per pond (total n = 138) varied based on pond size, number of inlets and outlets, and accessibility. Directly onshore from each microplastic sampling site, a 10 × 2 m quadrat with the long axis parallel to the shoreline was surveyed for MaPs, and the number and types of MaPs were recorded. At least one quadrat at every pond contained visible, rigid plastic fragments <3 cm long for which the source product could not be identified. These fragments, combined with Styrofoam particles <5 cm long, were not included in the overall MaP counts, as at some ponds, there were too many to count.

A database (Supplementary Table S1) was created with each type of identifiable debris item (e.g., beverage bottle, cigarette butt). Every item was assigned to one of thirty-one categories and was then coded according to one of six applications (Table 2). The items “packaging (non-food)”, “container (non-food)”, and “lid (non-beverage)” were assigned to more than one category depending on their identified use.

The influence of source-driven inputs (human activities at each land use type) and pond-level characteristics (total area, length-to-width ratio, within-pond sampling location) on macroplastic types and abundances was investigated. Land use types were designated based on the following characteristics within 100 m of each pond periphery: (1) residential: mainly houses, (2) construction-proximal: housing and building construction, 3) commercial/industrial: industrial and/or commercial buildings, and (4) open: 10 or fewer buildings. Pond measurements, including the area and the length-to-width ratio of the pond, were determined from specifications provided by Ecosystem Recovery Inc. (2019) and measurements from Google Earth (version 7.3).

Macroplastic abundance at each sampling site was determined as items per 20 m². Nonparametric statistical tests, the Wilcoxon rank-sum test and the Kruskal–Wallis rank-sum test, were used to assess differences in abundances across pond-scale factors of within-pond location (inlet, outlet, open) and basin type (sediment forebay, main basin). The relationships between average MaP abundance and pond characteristics were examined using scatter plots and Pearson correlation analysis. The Wilcoxon rank-sum test and the Kruskal–Wallis rank-sum test were also used to assess variable MaP abundances with regional-scale factors of city quadrant (NW, NE, SW, SE) and land use (residential, construction-proximal, industrial/commercial, open). The associations between the top ten MaP categories and regional-scale factors were determined using the chi-square test with the assumption that the expected frequencies are >5.

Statistical modelling incorporated the MaP count using pond- and regional-scale predictors along with known pond characteristics (e.g., area and length-to-width ratio of the pond). The South River pond was excluded as it functions as a detention pond designed for gradual water reduction rather than long-term retention with sporadic storm-related release. The River Road Industrial pond was also excluded from the statistical analysis due to its function as a stilling pond, which is constructed at the mouth of a river or spillway to reduce the velocity of water flow. All data analyses and modelling were done using R version 4.3.1 (R Core Team, 2023). A more detailed description of the statistical analyses is provided in Supplementary File 2.

The total MaP count in 138 quadrats of all twenty-eight ponds was 1,574, with an average 11.4 ± 11.0 MaPs/20 m² (quadrat) and a range of 0–87 MaPs/20 m² (quadrat). The two highest counts of MaPs within single quadrats were at Kilally North (n = 87) and Murray Marr 3 (n = 86) (Figure 2). Abundances varied significantly across ponds, with average counts ranging from 1.7 ± 1.2 (Duncairn) to 40.0 ± 16.3 (Fox Hollow 1S) (Table 1). Macroplastic concentrations also varied between quadrats of the same pond, with the greatest variabilities between quadrats of Murray Marr 3, Hyde Park 3E, and Kilally North (Figure 2).

The ten most abundant items overall were food-related wrappers, packages, and bags (n = 230), Styrofoam chunks >5 cm in size (n = 182), beverage bottles (n = 181), cigarette butts (n = 181), food- and beverage-related bottle caps (n = 108), non-food bags (n = 100), pet waste bags (n = 56), beverage cups (n = 54), food- and beverage-related lids (51), and straws (n = 41) (Figures 3A–I; Supplementary Table S1). The ten most common categories of MaPs in descending order were beverage packaging (n = 301), food packaging (n = 236), smoking/vaping (n = 213), foam packaging (n = 182), food/beverage lids/caps (n = 160), non-food plastic bags (n = 108), pet waste bags (n = 56), toys and other entertainment (n = 43), sports-related (n = 30), and packaging (non-food) (n = 30).

By grouping similar categories into six application types, the data show that food and beverage packaging (43%) was the most common application, followed by household and garden (25%), leisure and sports (22%), health (5%), building and construction (4%), and transportation (1%) (Figure 4). All ponds contained food and beverage packaging, as well as items from the household and garden application, whereas items comprising the leisure and sports application were found at all but two ponds (Duncairn and Talbot Village). Talbot Village and Duncairn ponds were characterized only by items from the food and beverage packaging and household and garden applications, whereas Warbler Woods and Hyde Park 1 contained items related only to food and beverage packaging, leisure and sports, and household and garden applications (Figure 4).

Testing MaP concentrations with surrounding land use revealed a significant relationship between the two variables (p = 0.021). Pairwise comparisons showed that construction-proximal ponds had greater concentrations of MaPs than ponds in residential areas (p = 0.033), but no significant differences were observed among the other land use categories. At the regional scale, there was strong evidence for associations between the ten most common MaP categories and both land use (p < 0.01) and city quadrant (p < 0.01). Construction-proximal ponds contained greater abundances of smoking/vaping items (e.g., cigarette butts) than ponds in other land use areas (Figure 5). In contrast, the residential ponds contained more non-food plastic bags and pet waste bags. Industrial/commercial ponds contained the most items from the toys and other entertainment category and the second-highest concentration of items from the smoking/vaping category (Figure 5). Finally, packaging and foam packaging categories were most represented at open ponds. Spatially, the smoking/vaping category was most represented at ponds in the SE quadrant of the city, whereas the lowest concentration of smoking/vaping items was found around NE ponds (Figure 6). Foam packaging was most common around the SW ponds, whereas ponds in the NW quadrant contained the greatest concentration of pet waste bags and packaging (non-food) (Figure 6).

Within ponds, there were no significant differences in MaP abundances between the sediment forebay and the main cell basins (p = 0.189). There were, however, greater MaP concentrations near inlets and outlets compared to open areas (p = <0.001). After excluding outlier ponds Fox Hollow 1S and Kilally North, as identified by standardized residuals (|Z| > 2), there was a strong correlation (r = 0.72, p = <0.001) between abundance and pond area (Figure 7A). There was also a moderate correlation (r = 0.5, p = 0.013) between abundance and the length to width ratio of the main basin following exclusion of identified outliers (Fox Hollow 1S and Hyde Park 3E) as well as exclusion of the South River detention pond and the River Road Industrial stilling pond, both of which had no defined forebays or main cells (Figure 7B).

Considerable variability in the concentrations of MaPs between the 28 stormwater ponds as well as between survey quadrats at each pond, demonstrates the complexity of multiple sources, dispersal patterns and accumulation sites. Prevailing wind direction has been shown to be an important factor controlling macroplastic transport (Semcesen et al., 2025), but west winds prevail throughout the entire city of London simultaneously. Even ponds aligned east-west do not show a preferred accumulation of MaPs near their eastern margins. Land use and city quadrant, which are both related to human activities, were better predictors of MaP concentration. The influence of pond-level factors (pond area, length-to-width ratio, proximity of quadrats to inlets and outlets compared to open areas) are statistically supported. Within-pond variability, however, shows that seven of the 28 ponds had outliers with significantly greater MaP concentrations than the other quadrats at the same pond (Figure 8). Unfortunately, these outliers could be attributed to more than one factor; for example, Styrofoam chunks were considerably more concentrated in the White Oaks 2, Murray Marr, and Sunningdale 4 outliers compared to their respective groups of quadrats, but all of the outliers were also located near pond inlets. The concentration of cigarette butts in one of the South River quadrats may account for that specific outlier, whereas the Kilally outlier (highest concentration of MaPs overall) does not appear to be related to any of the source-driven or pond-level factors (Supplementary Table S1). Although we tested for the influence of several source-driven and pond-level effects, we could not incorporate population density nor number of stormwater inflows for each pond area due to a lack of accessible data. The results thus indicate that stormwater ponds are complex settings with many sources and transport pathways for MaP debris.

The macroplastic items determined to be the most abundant in the twenty-eight stormwater ponds reflect larger-scale audits of plastic debris pollution. In this study, food-related items (wrappers, packaging, bags), beverage-related items (bottles, caps), and cigarette butts were within the top five littered items. This aligns with the comprehensive European freshwater analysis by Winton et al. (2020), who, through a weighted meta-analysis of nine studies, identified plastic bottles, food wrappers, and cigarette butts as the most common items. Similarly, an investigation of visible plastic debris on 66 beaches of the Laurentian Great Lakes indicated that, of the identifiable MaPs, cigarette butts, beverage-related items (bottle caps, foam drink cups, plastic straws), and food wrappers were among the top six items identified (Arturo and Corcoran, 2022). In another study of 89,691 plastic items found within 40 survey sites along the Ganges River, Youngblood et al. (2022) discovered that plastic food wrappers, smoking-related items (cigarette butts, tobacco sachets), plastic bags, and beverage-related items (cups, lids) were among the top five identifiable products. Similarly, Rossi et al. (2023) determined that of a total 28,431 litter items identified in the Lower Passaic River watershed and Harbor Estuary of the U.S.A, cigarette butts and food- and drink-related items comprised 45% and 32%, respectively. These results, combined with many other reports in the literature, indicate that: (i) public awareness concerning single-use plastic products in the environment (e.g., wrappers, cigarette butts, straws) is still lacking, (ii) warnings without policies are not prohibitive enough to make a significant reduction in plastic litter, and (iii) plastic reduction and recycling programs will work only if consumers across the globe have the resources to access and manage those programs.

The results from all 28 ponds indicate that of all items, only cigarette butt concentrations varied significantly with land use, being more common at construction- and industrial/commercial-proximal ponds than at their open and residential counterparts (p = 0.001) (Figure 8A). Although only Summerside pond was associated with construction activities, the average abundance of cigarette butts across the eight quadrats at this pond was 30.5/100 m². The industrial/commercial ponds contained an average of 10.0 cigarette butts per 100 m² across 26 quadrats, whereas residential and open ponds contained an average of 2.7 cigarette butts per 100 m² (80 quadrats) and 4.9 cigarette butts per 100 m² (24 quadrats), respectively (Figure 8A). The relatively low cigarette butt concentrations at residential ponds may be explained by residents opting to smoke in their homes or yards rather than walking to a pond to do so. Open ponds are more remote from roads and houses than ponds associated with other land use types, and this may explain the low cigarette butt abundances at these locations. Arat (2024), in a review of cigarette butts as a pollutant, showed that they are most common in public places. Similarly, an investigation of the distribution of cigarette butts across Madrid, Spain, demonstrated that central districts with the greatest number of hospitality venues and public transportation stops (i.e., sites with frequented visitors) contained the greatest butt concentrations (Valiente et al., 2020).

Although no statistically significant differences were found between land use and the concentrations of pet waste bags and fishing gear, these items were most common at residential ponds (Figures 8B,C). Because the greater number of houses in residential areas equates to higher population, and fishing and dog walking are activities conducted by people, the results are not surprising. The link between the second-most abundant item (Styrofoam chunks) and land use, however, is less clear. Although not statistically significant, Figure 8D suggests that open land use areas contain the greatest concentrations of Styrofoam chunks, but the result is skewed by a relatively high abundance at open pond, Murray Marr 3 (n = >25). This pond is located within 200 m of London’s largest public recycling drop-off facility at which Styrofoam is accepted as a non-recyclable form of construction and renovation debris. This result indicates that the influence of surrounding land use on the abundance of specific MaP types can be overshadowed by local activities and that assigning ponds to specific land use types requires careful consideration.

With respect to MaP categories, the association between the highly frequented construction-proximal and industrial/commercial ponds and cigarette butt concentrations results in the smoking/vaping category also being most represented at ponds in these land use types (Figure 5). Based on the item and category results, we therefore expected that the leisure and sports application, to which smoking/vaping was included, would also be most represented at construction-proximal and industrial/commercial ponds. Although the data show that Hyde Park 3E (industrial/commercial) did indeed contain the greatest concentration of items and categories belonging to the leisure and sports application (Supplementary Table S1), the pond with the second-highest concentration of leisure and sports items was Fox Hollow 1S, a residential pond. This discrepancy indicates that combining all leisure and sports items into one application may be too simplistic for a study involving more than a thousand MaPs comprising >200 distinct items.

The average abundance of MaPs around ponds of each city quadrant were not significantly different (NW: 13.9/20 m²; NE: 10.2/20 m²; SW: 8.1/20 m²; SE: 12.2/20 m²). Notwithstanding this result, ponds in the SE quadrant contained the greatest concentration of items in the smoking/vaping category, which may be related back to land use. Of the five studied SE ponds, one was designated a construction-proximal pond, and two as industrial/commercial ponds. Ponds classified with these land use types contained the greatest cigarette butt concentrations, which are items included in the smoking/vaping category. The results also showed that pet waste bags are more common in northern ponds (NE + NW) compared to their southern (SE + SW) counterparts (Figure 6), which may also be related to land use, as 78% of the northern ponds are in residential areas compared to only 35% of the southern ponds being classified as residential. Fishing gear was most abundant at NE ponds where 100% of the fishing-related items found were at residential ponds. Cigarette butts, pet waste bags, and fishing gear all fall within the leisure and sports application, which is most represented at residential ponds, thereby emphasizing the stronger influence of land use versus city quadrant on the abundances of these specific MaP types. Finally, the foam packaging category, represented by Styrofoam chunks, was mostly found at SW ponds (Figure 6). Of the six SW ponds, three are residential, two are open, and one is industrial/commercial. Given that many of the Styrofoam chunks were related to household packaging, it stands to reason that residential pond areas contained these items. The relative abundance of foam packaging at open ponds in the SW could be explained by the flattened shape and very low density of expanded polystyrene (1.04–1.07 g/cm³), which enables Styrofoam chunks to be easily mobilized by wind from residential areas and from open vehicles (i.e., trucks) as individuals drive past Murray Marr 3 pond to transport Styrofoam waste to the landfill site.

Although very few investigations focus on MaP movement in ponds or lakes, studies of MaP transfer in estuaries show that the geomorphology of the basin, meteorological conditions, and density of the plastic can control transport dynamics (Browne et al., 2010; Ledieu et al., 2022). The results of this study highlight that the inlets and outlets of stormwater ponds are preferential sites for MaP deposition compared to open pond areas. Any MaPs captured within runoff and diverted into a stormwater system will enter ponds through natural (tributary) or built inlets. This is true for both microplastics and MaPs. For example, Öborn et al. (2024) and Corcoran et al. (2025) found that sediment located at stormwater pond inlets was the main sink for microplastic particles. If at the inlets, the pond water is tranquil for an extended period, plastics with multiple layers, pockets, or of high density will sink to the pond bottom. If, however, wind blows across the pond surface, low density plastics will become stranded along the pond banks proximal to the entry points (inlets). In a study of tagged plastic and wooden blocks released into the mouths of African rivers, MacLean et al. (2021) showed that 80% were found on beaches proximal to their release points. In the present study, deposition of MaPs proximal to the inlets was visibly obvious at Fox Hollow 1S, Murray Marr 3, and Hyde Park 4S ponds (Supplementary Table S1). In contrast, where a wind-induced current flows down the length of a pond, floating MaPs will be transferred toward the outlet structures and become stranded on the banks. This phenomenon was visibly clear at Sunningdale 6B, Cranbrook South, and Stoney Creek 1N ponds (Supplementary Table S1).

The strong correlation between MaP abundance and pond area is consistent with the expectation that larger water bodies tend to accumulate more MaPs. Interestingly, the smallest average pond area is represented by residential ponds (7261/m²) and although not statistically significant, this land use type also had the lowest average abundance of MaPs (10.0 per 20 m²) compared with open (12.9 per 20 m²), industrial/commercial (12.3 per 20 m²), and construction-proximal (22.0 per 20 m² – one pond only) ponds. This could also support land use as the predominant driver of MaP abundances and types in stormwater ponds.