The study was carried out in Nanjing (31°14′−32°37′N, 118°22′−119°14′E), Jiangsu Province, China. The study site features a subtropical monsoon climate, with an average annual temperature of 15.4°C and an average annual precipitation of 1,106 mm. Numerous rain gardens have been constructed by the local governments in various parts of the city, including residential, traffic, and park areas. In this study, four, six, and four existing rain gardens were randomly chosen from the residential, traffic, and park areas, respectively (Table S1; Fig. 1). The characteristics of MPs (concentration, color, size, and shape) in the rain gardens were evaluated by soil sampling from these rain gardens. Soil samples were also collected from the lawns adjacent to each rain garden. These lawns were located outside the boundaries of the rain gardens at distances ranging from 2 to 10 m. It was assumed that most of the MPs detected in the lawns were sourced from atmospheric deposition, as these locations lacked stormwater inlets and were bordered by curbs that impede runoff [16]. Contribution of atmospheric deposition was assessed by comparing the MP concentrations between the rain gardens and nearby lawns.

A transect was created along the longest straight that connecting the two sides of each rain garden, and three uniformly spaced quadrats (1 ×1 m) were established along each transect [23]. A soil core (5 cm in diameter and 5 cm in depth) was collected from the center of each quadrat using a soil wreath knife after removing visible surface debris, litter, and rocks larger than 5 mm [4]. Similar transects and quadrats were also established in each adjacent lawns. Total soil from 84 quadrats were sampled (14 rain gardens × 3 quadrats + 14 lawns × 3 quadrats). These samples were immediately stored in aluminum containers at 4°C and then transported to the laboratory for the separation and extraction of MPs.

The MPs were extracted from the soil samples by density separation (Fig. S1) [3–4, 24–25]. The soil samples were dried at 65°C for 24 h. Subsequently, 20 g (± 0.1) of the dried soils were randomly weighed, mixed with saturated NaBr solutions (ρ= 1.55 g/mL, 30°C) and centrifuged at 3500 rpm for 10 min. The mixtures were stirred in 400 ml beakers for 10 min, and allowed to settle for 24 h to remove the supernatant. The above steps were repeated three times to completely extract the analogous MPs from the soils. The supernatants were filtered through a 0.45 μm glass fiber filter membrane using a vacuum pump to collect the final supernatants, which were than stored in a glass petri dish. To this, 30% H₂O₂ was added at 50°C for digesting the organic matter for 72 hours. The digested solutions were filtered through a membrane and stored in glass petri dishes for further analysis.

All MPs in the glass petri dishes were examined and photographed under a stereomicroscope (SteREO LEICA Discovery V.20). MPs were initially identified and documented based on their shape, surface, texture, color, and luster [4]. Based on shape, MPs were categorized into fibers, fragments, films, foams, and balls [4]. MPs were grouped into two size ranges: 2–5 mm and < 2 mm, based on their longest dimensions [26]. MP colors were determined by comparison with an online color comparison (https://www.5tu.cn/colors/yansebiao.html) [27].

To identify the MPs across various size ranges, a dual-method approach was implemented [28]: μ-FTIR (Thermo Fisher Scientific Nicolet iS10) with diamond ATR crystal was used for particles > 500 μm, and LDIR (Agilent 8700) was applied to particles < 500 μm. Given the large sample size, 783 (i.e. 19.98% of total collected particles) randomly selected MPs encompassing all morphotypes were analyzed to identify the most common visually identifiable MPs. For μ-FTIR, the spectrum range was set from 4000 to 400 cm⁻¹, with 16 scans and all spectral data were processed using OMNIC software. For LDIR, the spectral range was set to 1800–1000 cm⁻¹ with a spectral resolution of 0.5 cm⁻¹. For both μ-FTIR and LDIR, the results were compared with reference spectra from the libraries of Shanghai WeiPu Testing Technology Group Co., Ltd. and Agilent Technologies. A spectral match of at least 65% was required for identifying the polymer composition.

Strict quality control measures were implemented to prevent plastic contamination during the experimental processes [4, 6]. All sampling tools, laboratory equipment, and containers were thoroughly rinsed with ultrapure water to avoid plastic pollution. During the sample handling, gloves and cotton laboratory coats were worn and the use of plastic tools or materials was avoided. All containers containing soil samples, solutions, and extracted MPs were covered with parafilm and stored in enclosed spaces to prevent plastic pollution. Three blank samples containing ultrapure water taken as control were subjected to the same density separation steps as soil samples. An average of four fibers, one film, one fragment all less than 2 mm, and one fiber 2–5 mm were detected. Consequently, the quantification of MPs in the samples was corrected based on this value [29].

Two-way analysis of variance (ANOVA) was employed to assess the effects of habitat type (rain gardens and lawns) and functional area (residential, traffic, and park areas) on the concentrations of total MPs, 2–5 mm MPs, and < 2 mm MPs. An independent t-test was used to evaluate the differences in MP concentrations between rain gardens and lawns. The Mann-Whitney U test was used to compare the concentrations of MPs with different shapes or colors between rain gardens and lawns in different functional areas. Prior to ANOVA, the Shapiro-Wilk normality test was used to evaluate the normal distribution of the datasets. All statistical analyses were performed using GraphPad Prism 9.0, and the results were considered significant at p < 0.05.

MPs were detected in all the soil samples (Table S2). The overall mean MP concentrations in rain gardens and lawns were 2693 ± 415 items/kg (range 400–5900 items/kg) and 1739 ± 256 items/kg (range 500–3500 items/kg), respectively (Fig. 2a). Habitat type and functional area, when considered independently, did not exert a significant effect on MP concentration (Table 1). However, their interaction had a significant impact on MP concentration (Table 1; Fig. 2b). Specifically, in traffic areas, the MP concentration was significantly higher in rain gardens than in lawns (t = 3.12, P = 0.02). However, in the residential (t = 1.16, P = 0.78) and park areas (t = 1.35, P = 0.57), no significant differences were observed (Fig. 2b).

The size ranges of the MPs detected in the soils of rain gardens and lawns were 35–4855 μm and 50–4963 μm, respectively. The majority of MPs (78.51% in the rain gardens and 65.30% in the lawns) had a size < 2 mm. The concentration of MPs < 2 mm was considerably greater in the rain gardens than in the lawns, whereas no significant difference was found for MPs sized 2–5 mm (Table 1; Fig. 3a). For the 2–5 mm MPs, no significant difference in concentration was detected between the rain gardens and lawns across any of the functional areas (residential areas: t = 1.28, P = 0.25; traffic areas: t = 1.55, P = 0.15; park areas: t = 0.91, P = 0.40; Fig. 3b). In contrast, for MPs < 2 mm, the concentration was significantly higher in the rain gardens than in the lawns in the traffic areas (t = 2.86, P = 0.02), yet no significant differences were found in other functional areas (residential areas: t = 0.98, P = 0.37; park areas: t = 1.52, P = 0.18; Fig. 3c).

Five MP shapes, namely fibers, fragments, films, balls, and foams, were detected in the soil samples (Fig. 4). The MPs detected in the soils from the rain gardens and lawns exhibited similar shape distributions, with fibers being the predominant form, accounting for 58.62% and 67.15% in the rain gardens and lawns, respectively. No significant differences in the concentrations of any of the five MP shapes were observed between the rain gardens and lawns (Table S3; Fig. 5a). In residential areas, the concentrations of most MP shapes were higher in lawns than in rain gardens (Fig. 5b), while traffic and park areas demonstrated an opposite trend (Fig. 5c, d). However, these differences were not statistically significant across any functional area (Table S3).

Seven colors of MPs, (black, red, blue, green, white, transparent, and yellow) were detected in the soil samples. MPs from both rain gardens and lawns displayed similar colors distributions, with black and red being the dominant colors, constituting 38.06% and 23.47% in the rain gardens and 38.19% and 25.26% in the lawns, respectively. No significant differences in the concentrations of any of the seven MP colors were detected between the rain gardens and lawns (Table S3; Fig. 6a). The concentrations of blue and transparent MPs were significantly higher in the rain gardens than in the lawn in the traffic areas but not in the other areas (Table S3; Fig. 6b, c, d).

Of all the selected particles analyzed, 78% were confirmed MPs. In total, 19 polymer types were identified in all the soil samples (Fig. 7). In the rain gardens, the principal polymer types were acrylic (33.5%), PP (15.5%), polyethylene terephthalate (PET, 12.9%), phenolic epoxy resin/phenolic resin PEX/PR (6.9%), and PE (5.1%). In the lawns, the main polymer types were polyamide (PA 39.1%), polyurethane (PU 31.2%), PE (10.6%), PP (4.8%), and polytetrafluoroethylene (PTFE 2.6%). The representative μ-FTIR and LDIR spectra are shown in Fig. S2 and S3 (supplementary materials).

To date, only four studies have investigated MP concentrations in rain gardens, and the reported concentrations vary widely across regions[16, 19–21]. The mean MP concentrations in our study substantially exceed those reported by Mbachu et al. [20] in Australian rain gardens (41 ± 30.5 items/kg) and Jayalakshmamma et al. [19] in New Jersey, USA (469–997 items/kg). However, our measurements were notably lower than those reported by Koutnik et al. [16] in Los Angeles (472000 items/kg) and by Lange et al. [21] in Michigan (14657.8 ± 15769.9 items/kg). Although a direct comparison of MP concentrations is challenging due to the difference in techniques employed in laboratory processing, size fraction analysis, and counting across studies, these findings collectively indicate that rain gardens function as significant MP reservoirs in urban ecosystems. The observed geographical variations in MP concentrations suggest that regional characteristics (e.g., land use and environmental conditions) may affect MP accumulation in rain gardens. This is supported by Jayalakshmamma et al., who reported higher MP concentrations in rain garden situated within a high-density residential area (997 ± 64.3 items/kg) than the commercial (604 ± 91.4 items/kg) and low-density residential (469 ± 89.8 items/kg) areas [19]. Consistent with these findings, our study demonstrated that rain gardens in traffic areas exhibited comparatively higher MP concentrations than those in other areas. This pattern may primarily be attributed to more intensive anthropogenic activities in traffic areas, leading to continuous inputs of MP sources such as tire wear particles.

Overall, the mean MP concentration was comparable between the rain gardens and nearby lawns. This result is consistent with that of Koutnik et al. [16], who reported similar MP concentrations in soil samples from inside and outside the stormwater control measures. These results imply that atmospheric deposition is a significant source of MP accumulation in rain gardens. Interestingly, rain gardens in traffic areas demonstrated significantly higher MP concentrations than that in adjacent lawns, whereas no such differences were observed in the residential and park areas. This indicates that stormwater runoff plays a crucial role in MP accumulation in rain gardens in traffic areas. A possible reason is that frequent traffic activities in traffic areas generate a large quantity of MPs, such as tire–road wear particles and road pavement wear particles, which are deposited on impermeable road surfaces and subsequently transported by stormwater runoff [30–32]. This suggests that the MP retention efficiency of rain gardens is associated with the specific functional area it is located.

We also discovered that the concentration of MPs smaller than 2 mm was considerably higher in rain gardens than in nearby lawns, particularly in traffic areas. This pattern can be explained by the following three factors. First, the absence of taller vegetation and thicker litter layers in lawns enables wind to carry smaller and less dense MPs to the surrounding habitats [33–34]. Second, owing to the greater availability of habitats and food resources, the soils in rain gardens host more invertebrates than lawns [23, 35], that contribute to the fragmentation of MPs [36–37]. Moreover, frequent human activities in rain gardens may promote the mechanical breakdown of MPs, aiding in their fragmentation and diameters reduction [38]. Ecosystems are more vulnerable to smaller MPs because they are more prone to carrying and absorbing contaminants [39–40]. However, due to the limitations of stereomicroscopy in the quantitative analysis of small MPs [41–42], we did not further classify MPs < 2 mm in size. Future studies should employ more advanced methods such as micro-FTIR to investigate the characteristics of MPs, including their size, shape, and color.

Although several studies have examined MP accumulation in rain gardens [16, 19–21], limited information is available regarding MP characteristics such as shape and color in rain gardens [16]. In this study, five MP shapes were identified in all soil samples, with fibers as the predominant shape. Similarly, fibers were also reported in wetlands and stormwater retention ponds with high traffic, commercial activity, and population density [3, 19, 43]. However, Mbachu et al. [20] reported no plastic fibers in the filter media of 20 bioretention systems in three regions of southeast Queensland, Australia. The study attributed this apparent absence to a low population and anthropogenic activities at the sampled sites. We also found that rain gardens had substantially higher fiber concentrations than lawns, particularly in traffic areas. Fibers are widely sourced from textile [44–45]. Their primary source is greywater [46]. In traffic areas, greywater is more likely to remain on the surface and subsequently mix with stormwater runoff and flow into rain gardens because of the high proportion of impervious surfaces. In lawns, however, fibers are more easily removed by the atmosphere because of their large surface-area-to-volume ratio [47], lack of taller vegetation, and thicker litter layers.

In this study, seven MP colors were detected in the soil samples. Black MPs were the most prevalent color, which is consistent with recent studies indicating that the primary MPs removed from stormwater runoff through the purification process of rain gardens are black [17–18]. However, in contrast to our findings, Mbachu et al. [20] did not detect black MPs in rain gardens, which may be partly due to the long distances between rain gardens and traffic areas at their study sites. Black MPs in urban areas are believed to originate from traffic activities, such as tire wear [48–49]. Our results confirm that the concentration of black MPs was higher in traffic areas than in other areas, although the limitations of FTIR analysis prevented definitive identification of these black particles as tire wear particles. On the other hand, the black MPs in rain gardens may also come from recycled asphalt materials, which are widely used in the construction of rain gardens in recent years [50]. Besides black MPs, the concentration of red (23.47%), blue (14.32%), and green MPs (11.54%) were higher than transparent (6.50%) and white MPs (3.45%) in rain garden. Yuan et al. [51] also reported a higher proportion of colored plastics in urban areas, reflecting the extensive use of colored plastics in daily life [52].

Furthermore, we observed that the concentrations of blue and transparent MPs were considerably higher in rain gardens than in the lawns in traffic areas. The blue MPs in traffic areas were likely derived from paint peeled from roadside indication signs [53]. Transparent MPs were likely sourced from daily plastic products, such as transparent plastic bottles and packaging bags [54]. Human activities such as irrigation, fertilization, and gardening are more frequent in rain gardens. Thus, these MPs in traffic areas may be retained in rain gardens by stormwater, which may partly account for the higher concentrations of blue and transparent MPs in rain gardens than in lawns. These findings illustrate the complex interactions among human activities, environmental sources, and stormwater runoff in shaping MP accumulation in urban rain gardens.

Several studies identified polymer types in the soils of rain gardens that indicated variation in MP composition and dominant polymer types [19–21]. For instance, Mbachu et al. [20] reported 9 polymer types in the filter media of all bioretention systems in southeast Queensland, Australia, with LDPE, PP, and Nylon being the dominant components. Lange et al. [21] identified 17 polymer types in the filter media of bioretention systems in Ohio and Michigan, USA, and the most frequently observed polymer types were PP, EVA, PS, and EPDM. Jayalakshmamma et al. [19] found 11 polymer types in the sediments of rain gardens at three sites in Newark, New Jersey, USA; where PP was dominant at all sites. In the present study, 16 polymer types were identified in the soils of all rain gardens. ACR, PP, and PET were the dominant components with relative proportions 33.62%, 15.51%, and 12.93%, respectively. Although the reported MP compositions and dominant polymer types differed, all studied indicated PP as an important polymer in rain gardens. PP is a common polymer in urban ecosystems [51, 55]. ACR is commonly found in housing cleaning products [56]. Through cleaning wastewater it may mix with stormwater runoff before entering rain gardens. The reason for the ACR-dominated MPs in this study is unclear but could be the absence of a pre-sedimentation forebay, which was used to reduce sediment loads. This is supported by the results of Lange et al. [21], who found ACR in forebays but not in rain gardens.

Although rain gardens and lawns exhibited similar overall MP composition, their dominant polymer types showed distinct patterns. This divergence may stem from differences in MP sources between the two habitat types: rain gardens receive MPs through both stormwater runoff and atmospheric deposition, whereas lawns primarily accumulate MPs via atmospheric deposition. Furthermore, the physical properties of the MPs themselves may contribute to this variation. PU (widely used in road paints and vehicle coatings [57]) and PA (commonly used in textiles and packaging materials [58]) were the predominant polymer types in lawns, despite also being present in rain gardens. Their higher proportions in lawns can be partly explained by their aerodynamic properties, which facilitate atmospheric transport and deposition.