Stormwater Detention Reservoirs: An Opportunity for Monitoring and a Potential Site to Prevent the Spread of Urban Microplastics

: Stormwater runo ﬀ carries pollutants from urban areas to rivers and has the potential to be a main contributing source of microplastics (MPs) to the ecosystem. Stormwater detention reservoirs (SDRs) di ﬀ er from ponds and lakes in that SDRs retain most particulate matter and they are emptied after storm events. This paper investigates the occurrence of MPs in the SDR of the Alto-Tiet ê catchment area, Itaim stream in Po á city, S ã o Paulo, Brazil. The MPs found were classiﬁed in di ﬀ erent categories: shapes (fragment, line / ﬁbre, ﬁlm / sheet and pellet); size ( < 0.5 mm, between 0.5 mm and 1 mm and > 1 mm); and polymer composition. Results have shown that most of the MPs found in the samples are fragments (57%), followed by pellets (27%), ﬁbres / lines (9%), and then ﬁlms / sheets (6%). Small particles ( < 0.5 mm) represented 89% of the total MPs, and this category mainly included fragments (62%) and pellets (30%). MPs were found in a vast variety of shapes and colours, which shows a likely variety of sources. Besides the occurrence of MPs in the stormwater samples, the potential of SDRs as a ﬁrst sanitary barrier to retain MPs before they reach the ecosystem has been speculated.


Introduction
Microplastics (MPs) are plastic particles within the range 1 µm-5 mm [1]. They may have been produced at this scale or have originated from the fragmentation of plastic waste (by physical, bio or photo-degradation). Microplastics are released from a large variety of industries, activities and even households and this diversity of sources also translates into heterogeneity of sizes, shapes, colours, compositions and densities of these pollutants in the environment [2].
MPs' impact on the environment is not well defined, in part because of methodological challenges [3], but consensus is that their potential effect is negative [4,5] and, given the high stability of polymers, the negative impact of MP may persist for considerable time periods. The effect of MP's shape on toxicity is yet to be established. Microfibers have been reviewed as the predominant type of microplastic ingested by freshwater biota [3]. Indeed, MPs with sharp edges such as fibres can cause a reduction in food intake, intestinal damage and oxidative stress in organisms such as snails [6]. The effects of MPs on different taxa have been identified, although there are also works describing

General Hydrologic Data
The area's climate is Aw type (Köppen classification) [27]; that is, rainy tropical climate, with rains in summer and dry winter. The average monthly and annual rainfall in the Itaim Stream basin were obtained using data from the E3-091 rain station (23 • 29 00" S and 46 • 22 00" W), between 1944 and 2017.
There are no historical data on the stream flow for this specific case. Thus, to determine their average annual flows, the Equation (1) developed by DAEE (Departamento de Águas e Energia Elétrica-1988) was used.
where Q = mean annual discharge (L/s); P = mean annual rainfall (mm); A = drainage area (km 2 ); and parameters "a" and "b" defined as −26.23 and 0.0278, respectively. The intensity-frequency-duration relationship presented in Equation (2) was based on the extreme data set events registered by a pluviometer at the catchment area [28]. This equation was used to evaluate the frequency of preceding trigger events that washed the urban surface out and end up in the SDR.
The Itaim Stream, in the city of Poá, is one of the tributaries of the Alto-Tietê River catchment with an area of approximately 14.7 km 2 . It is located in SPMA, between 23 • 30 30 and 23 • Figure 1a). The land of the study area is mainly occupied by urban buildings (79.9%), meadow (6.4%), bare soil (3.5%) and trees and shrubs (10.2%) as analysed from aerial photography of the area (Figure 1a).
The SDR located at the Poá city centre (Piscinão da Vila Romana) has~250 m length and 14 m depth, with the capability to store up to 210 million litres of water ( Figure 1b). It started to be used in 2018 to reduce flooding in Poá city centre [30]. It must be stressed that SDRs do not operate as continuous flow, but as an intermittent mode. The studied SDR was only used twice (December 2018 and January 2019) before the sampling for this present study was carried out. During these storm events, it received around 60 million litres and 180 million litres in 2018 and 2019, respectively [31]. Therefore, samples from the SDR sediment represent the sediments carried by the overland flow from the two described events, given that the SDR was not cleaned in between events.
Sediment samples were collected from the empty reservoir from two different points chosen at the same visit, i.e., one sample of approximately 2 kg of sediment from each sample point: SP 1 close to the inlet and SP 2 close to the outlet of the SDR (shown in Figure 1c,d, respectively). Emptying for the SDR depends upon operational criteria, such as the storm duration and stream flow capacity, and it is generally done less than 24 h after the storm. The study samples were collected around both the reservoir inlet and outlet. They were stored in suitable containers and transported immediately to the Environmental Geochemistry Laboratory at the Department of Geography and Environmental Planning-UNESP in Rio Claro. Samples were dried at room temperature and prepared by density separation with ZnCl 2 , adapting a previously published method [26,32,33]. Representative samples (triplicates of 30 g of sediment of each point totalising 6 samples analysed) were added to a glass beaker with 300 mL of ZnCl 2 (≥98%, from Sigma-Aldrich, São Paulo, Brazil) at a dose of 1.6 g ZnCl 2 /cm 3 solution. The mixture was homogenised and left overnight at ambient temperature. The solutions were transferred to an Imhoff cone where, after 24 h of precipitation and sedimentation, the supernatant was collected. This step was repeated twice and the extracted solution was filtered through a 0.45 µm cellulose nitrate membrane filter with 47 mm of diameter (Sartorius, São Bernardo, Brazil) before being left to dry at ambient temperature for 24 h in petri dishes in a closed, low-temperature controlled room (to control any possible airborne contamination) prior to analysis.
Water 2020, 12, x FOR PEER REVIEW 4 of 16 beaker with 300 mL of ZnCl2 (≥98%, from Sigma-Aldrich, São Paulo, Brazil) at a dose of 1.6 g ZnCl2/cm 3 solution. The mixture was homogenised and left overnight at ambient temperature. The solutions were transferred to an Imhoff cone where, after 24 h of precipitation and sedimentation, the supernatant was collected. This step was repeated twice and the extracted solution was filtered through a 0.45 µm cellulose nitrate membrane filter with 47 mm of diameter (Sartorius, São Bernardo, Brazil) before being left to dry at ambient temperature for 24 h in petri dishes in a closed, lowtemperature controlled room (to control any possible airborne contamination) prior to analysis.

Quantification and Characterization of Microplastics
MPs were identified visually using a microscope (Zeiss Discovery V12 SteREO, São Paulo, Brazil) with an integrated camera (AxioCam ERc 5s). They were categorised by shape, size and composition [32] (see Table 1). Particles were counted according to their physical properties (texture, colour and flexibility), and soft organic materials other than plastic sheets (such as small pieces of leaves and insects) were discarded.

Composition
Based on spectroscopy bands Some frequent particle types were selected to try to represent the samples and then were individually separated, cleaned (with distilled water) and analysed by ATR FT-IR (Varian) with a wavelength range from 400-4000 cm −1 or Raman spectroscopy (Olympus BX51, Princeton Instruments Action SP2500, 632 nm HeNe laser, Raman shift 390-600 cm −1 , Princeton, United Kingdom). Spectra were compared to the literature and online databases (Bio-Rad Sadtler, São Paulo, Brazil) [34,35]. Finally, morphology and weathering of the MPs were studied using scanning electrical microscopy with energy dispersive spectroscopy (SEM-EDS, JEOL-JSM-6010 LA).

Hydrological Data Analysis
The monthly rainfall (average of 115.9 mm per month) in the study site peaks in January, whereas August is the driest month, with monthly averages of 241. 5  To calculate the annual rainfall between 1944 and 2017, the annual average flow of the Itaim Stream was used [28]. Thus, through the precipitation-flow ratio it was possible to obtain the average annual flow for those years (Figure 2c). Figure 2d, derived from Equation (2), illustrates the rainfall intensities (mm/h) for different durations (min) and return periods (year). These data were used to characterise stormwater events before sampling. The maximum preceding precipitation event registered resulting in flow going into the detention reservoir was 57.3 mm/h, thus resulting in a return period of 8 years. This indicate a probability P (x ≥ X) of 73% for 10 consecutive years. However, lower intensities (return period of 5 years) have also resulted in flow to the detention reservoir, due to the high impervious surface on the catchment area (80%).

Microplastics Quantification
The number of MPs per size range found in the SDR sediment samples are shown in Table 2. Further information on their distribution by shape and size is displayed in Figure 3. Table 2. Characteristics of the MPs found in the sediment samples (30 g) from the SDR by shape and size (unit/kg). The relative standard deviation (RSD%) from n = X is given in brackets.

Microplastics Quantification
The number of MPs per size range found in the SDR sediment samples are shown in Table 2. Further information on their distribution by shape and size is displayed in Figure 3. Different MP size, colours and materials were identified inside the same shape classification ( Figure 4). The main type of polymers in the samples were polyethylene (PE), polyester (PES), polypropylene (PP), polystyrene (PS), and polyamide (nylon) (PA) according to analysis with FT-IR and Raman. Black particles were identified as tyre fragments by Raman spectroscopy and optical microscopy ( Figure 5). Photomicrographs illustrate a typical rough surface and the etch pits due to the weathering of polyethylene (Figure 6a), the size of pellet (Figure 6b) and the smooth surface of lines (Figure 6c,d).

Discussion
The opportunity of SDRs in retaining MPs as a first barrier against contamination, and the types of MP contamination that could reach ecosystems from urban runoff, are assessed here.

Microplastic Size Distribution
The MPs found in the SDR originated mostly from the urban runoff, as the SDR is designed for stormwater flow peak attenuation. However, some interference with non-regular sewage connections, atmospheric MP contamination [36] and runoff through milder events may also have contributed to the MPs found in the samples. These MPs from the SDR sediment came from the two main rainfall events: one of 164 mm (with average intensity 54.6 mm/h) and another of 172 mm (average intensity of 57.3 mm/h), with return periods of around 8 years (according to IDF, Equation (2)). This means that the likelihood of future events having equal or higher intensities are 49% in the next 5 years and 93% in the next 20 years. Therefore, they cannot be considered extreme events, although they resulted in excess volume of water collected into the stormwater reservoir due to the high impervious catchment area (c.a. 80%).
The most common MPs found in the samples were fragments (57%) followed by pellets (27%), fibres/lines (9%) and then films/sheets (6%). The high number of fragments is result of the degradation and breakage of big plastics in smaller particles [21]. Previous works studying microplastic in urban environment [25] also found fragments to be the most abundant type of MP [37]. In contrast, fibres were the most abundant microplastic in effluent from urban wastewater [19].
The classification highlights the total number of small particles, as 89% of the MPs are between 0.02 mm to 0.5 mm. This finding agrees with the literature [18,38]. The <0.5 mm fraction of the MPs were mainly formed by fragments (62%) and pellets (30%). The fragments were usually debris of bigger plastics such as tyres, bottles and other plastic materials [20]. Some fragments were identified as polystyrene (PS). PS MPs can be easily transported by runoff because of their low density [2] and are commonly used in food packaging.
Pellets (e.g., Figure 4n,o) were only found in sizes smaller than 0.5 mm in this study. This agrees with other works in highly populated and industrial areas [20]. These could be primary MPs, although

Discussion
The opportunity of SDRs in retaining MPs as a first barrier against contamination, and the types of MP contamination that could reach ecosystems from urban runoff, are assessed here.

Microplastic Size Distribution
The MPs found in the SDR originated mostly from the urban runoff, as the SDR is designed for stormwater flow peak attenuation. However, some interference with non-regular sewage connections, atmospheric MP contamination [36] and runoff through milder events may also have contributed to the MPs found in the samples. These MPs from the SDR sediment came from the two main rainfall events: one of 164 mm (with average intensity 54.6 mm/h) and another of 172 mm (average intensity of 57.3 mm/h), with return periods of around 8 years (according to IDF, Equation (2)). This means that the likelihood of future events having equal or higher intensities are 49% in the next 5 years and 93% in the next 20 years. Therefore, they cannot be considered extreme events, although they resulted in excess volume of water collected into the stormwater reservoir due to the high impervious catchment area (c.a. 80%).
The most common MPs found in the samples were fragments (57%) followed by pellets (27%), fibres/lines (9%) and then films/sheets (6%). The high number of fragments is result of the degradation and breakage of big plastics in smaller particles [21]. Previous works studying microplastic in urban environment [25] also found fragments to be the most abundant type of MP [37]. In contrast, fibres were the most abundant microplastic in effluent from urban wastewater [19].
The classification highlights the total number of small particles, as 89% of the MPs are between 0.02 mm to 0.5 mm. This finding agrees with the literature [18,38]. The <0.5 mm fraction of the MPs were mainly formed by fragments (62%) and pellets (30%). The fragments were usually debris of bigger plastics such as tyres, bottles and other plastic materials [20]. Some fragments were identified as polystyrene (PS). PS MPs can be easily transported by runoff because of their low density [2] and are commonly used in food packaging.
Pellets (e.g., Figure 4n,o) were only found in sizes smaller than 0.5 mm in this study. This agrees with other works in highly populated and industrial areas [20]. These could be primary MPs, although some spheres could also be fragments that had been smoothed by physical forces [2,39]. Pellets were not characterised in this study. Previous research has found that they were polyethylene (PE) microbeads used in personal care products, cleaning compounds and industrial sandblasting [17,40].
This smaller MP range (<0.5 mm) can easily be transported through the water cycle, and these particles may be light enough for being transported by the wind and spread. The surface area of this fraction for MP may be larger, and since MPs can interact with other pollutants and act as a carrier [41], collecting them in SDR preventing their further spread is important. These MP particles can also easily enter the food chain as they resemble-in their size and shape-naturally supplied FPOM and consequently, could be ingested (from suspension and deposition) by benthic invertebrates (filterers and gatherers) [9][10][11][12].
The bigger fraction of MPs (from 1 to 5 mm) is mainly fibres/lines (54%). One thing that is important to highlight is that some fibres were found in agglomerates as showed in Figure 4m. These agglomerates were considered one single fragment. The fibres found in the SDR sediments were mostly polyester (PES) and were found in a vast variety of colours, which shows that the high number of fibres are not debris from a unique source, but they possibly came from diverse range of polluters. Previous works have indicated that fibres mainly originate from textile products. Fibres have been reported to be the main source of MPs in the oceans (add 30) and have high prevalence in freshwater environments [3].
Films and sheets were also an important category of the bigger fraction MPs (from 1 to 5 mm) (29%); however, they were mainly found in small sizes in the study sediments (55% < 0.5 mm, 22% 0.5-1 mm, 23% > 1 mm). Films were identified as made from PE for plastics bags or wrappers. Sheets were not identified, but the literature states that they are probably made from acrylic (PMMA) [20].
This research only identified PE, PES, PP, PS and PA as the polymers making up MPs. Given that this study analysed a limited number of samples, other MPs, such as polyethylene terephthalate (PET), polyurethane (PU), polyvinylchloride (PVC), acrylic, acrylamide, polyacrylate, alkyd resin, polyphenylene oxides (PPO), ethylene vinyl acetates (EVA) and others, may be potentially present in the samples, as other works have indicated [26,39,40].
Interestingly, a large number of the MPs found within the fragments fraction were black particles (around 30%). Raman analysis allowed these to be identified as fragments from tyres ( Figure 6). Tyres are made from natural and synthetic rubber (polyisoprene or styrene-butadiene rubber (SBR)) that can get eroded. Tyre debris has been found to be a second major source of MP fragments in oceans [42], and synthetic fibres represent two-thirds of MP at sea [43]. Tyres have been also found to be a main source of MP in sediment samples [25]. The amount of tyre particles varies from each region, and in Brazil, the emissions of tyre wear and tear per capita are of 1.4 kg/year, which means they are a very significant source of MP in the environment [44].
The degradation degree of the samples varies considerably, as is demonstrated by the SEM results ( Figure 6). This might be potentially caused by the stress and friction forces acting on MPs surfaces during transportation, as transportation takes place over a turbulent condition. Therefore, MPs can have their surface damaged or become broken during transportation, depending on their resistance to stress forces and friction. Fibres and pellets were usually found without degradation, while fragments and films had a high degree of degradation.
The low degradation of fibres and pellets can be explained by the direct discharge of these MPs by the wastewater system, while fragments and films are usually carried to the reservoir by the urban runoff. Since the smaller plastics may have a greater degree of toxicity [7,8], policymakers should consider restricting the use of plastics that can degrade to small fraction of fragments. Wastewater treatment must be adjusted to improve the removal of fibres and pellets.
Furthermore, given that SDRs retain MPs, these sites could be used as a preliminary barrier that would prevent ecosystem contamination, as long as regular maintenance is performed between rain events. It is recommended that long-term sampling of the SDR sediment is studied to understand this further.

Comparing the Amount of Microplastic with Other Areas Elsewhere
The amount of microplastics found in the Poá SDR was 57,542 units/kg, which is compared with the MPs found in sediment samples elsewhere in Table 3. The average value obtained here is much larger than previous studies on MPs in lakes, mangrove, creek, rivers and beaches. In most previous studies, the sediment samples were taken with water columns and therefore the amount of debris was smaller. In addition, it is similar to the quantities of MP in sediments at the San Francisco Bay Region (up to 60,000 units/kg) [25], which was explained by wastewater and urban stormwater runoff at the bay. However, a quantity of up to 950,000 units/kg of MPs was found in a stormwater pond in Viborg, Denmark [45]. This high MP concentration was generated by the accumulated sediments of runoffs from urban and highway areas. In Poá SDR, the high amount of MPs can be explained by the way the reservoir works. The reservoir is filled in extreme intermittent rainfall events with urban runoff and then, after rain ceases and the level of river decreases, its water is pumped to the river, leaving behind all the sediments (i.e., settled and floated particles at the reservoir).
Therefore, it is expected that MPs are found in larger quantities in creeks, rivers, bays and SDRs located at urban areas than isolated areas, lakes and ocean. Urban areas are the centre of the contamination sources, and SDRs constitute an opportunity for monitoring and removing MPs from the urban environment.
This work acted as a first case study of MP in SDR, and parallel studies are suggested in various SDRs of the Sao Paulo Metropolitan area. Furthermore, samples from the SDR water layer and sediment should be taken after several events during at least one year for microplastic and microfibers characterization. Other physico-chemical parameters such as plastic degraded products and microbiological characteristics of the water accumulated in SDRs should be analysed.

Conclusions
This case study focused on quantification of MPs in the SDR in Poá, São Paulo Metropolitan Area. Rainfall events such as the ones analysed at this study result in flow to the detention reservoir, which has a probability of 73% to recur within 10 consecutive years. The operational model of the reservoir and its location close to the MP sources generate a high quantity of MPs (57,542 MP units/kg sediment) in different degradation degrees. Specifically, these MPs consisted of fragments <0.5 mm (89%). These MPs can have ecotoxicity and can also act as carrier of other pollutants and microorganisms, which can be ultimately carried to the oceans and enter the food chain.
Most of the particles within the fragments fraction (around 30%) were identified as tyre debris, which represents a significant source of pollution to the environment. The degradation level of the MPs was assessed qualitatively and may be low in fibres and pellets and higher in fragments and films.
Due to the way SDRs operate, it can be speculated that they might work as a preliminary barrier to prevent the further spread of MP in the ecosystem and be used for MP removal, thus having a great ecological benefit. However, more research is needed to confirm what efficiency is expected overall and in what conditions removal may take place, without affecting the primordial purpose of SDR regarding flood control.